diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,21443 +3,254900 @@
   "best_model_checkpoint": null,
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 3060,
+  "global_step": 36411,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.000326797385620915,
-      "grad_norm": 210.05616760253906,
-      "learning_rate": 2.173913043478261e-07,
-      "loss": 6.8656,
+      "epoch": 2.746422784323419e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298261665141813e-08,
+      "loss": 1.5936,
       "step": 1
     },
     {
-      "epoch": 0.00065359477124183,
-      "grad_norm": 203.4049835205078,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 6.8633,
+      "epoch": 5.492845568646838e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 3.659652333028363e-08,
+      "loss": 1.5551,
       "step": 2
     },
     {
-      "epoch": 0.000980392156862745,
-      "grad_norm": 210.0348663330078,
-      "learning_rate": 6.521739130434783e-07,
-      "loss": 6.9027,
+      "epoch": 8.239268352970256e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 5.489478499542544e-08,
+      "loss": 1.6017,
       "step": 3
     },
     {
-      "epoch": 0.00130718954248366,
-      "grad_norm": 204.6519317626953,
-      "learning_rate": 8.695652173913044e-07,
-      "loss": 6.766,
+      "epoch": 0.00010985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319304666056725e-08,
+      "loss": 1.595,
       "step": 4
     },
     {
-      "epoch": 0.0016339869281045752,
-      "grad_norm": 193.91259765625,
-      "learning_rate": 1.0869565217391306e-06,
-      "loss": 6.6258,
+      "epoch": 0.00013732113921617094,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149130832570908e-08,
+      "loss": 1.5455,
       "step": 5
     },
     {
-      "epoch": 0.00196078431372549,
-      "grad_norm": 167.2730712890625,
-      "learning_rate": 1.3043478260869566e-06,
-      "loss": 6.2645,
+      "epoch": 0.00016478536705940512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0978956999085087e-07,
+      "loss": 1.6211,
       "step": 6
     },
     {
-      "epoch": 0.0022875816993464053,
-      "grad_norm": 163.7445068359375,
-      "learning_rate": 1.521739130434783e-06,
-      "loss": 6.0457,
+      "epoch": 0.00019224959490263933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2808783165599268e-07,
+      "loss": 1.5092,
       "step": 7
     },
     {
-      "epoch": 0.00261437908496732,
-      "grad_norm": 119.4488754272461,
-      "learning_rate": 1.7391304347826088e-06,
-      "loss": 5.0,
+      "epoch": 0.0002197138227458735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.463860933211345e-07,
+      "loss": 1.5684,
       "step": 8
     },
     {
-      "epoch": 0.0029411764705882353,
-      "grad_norm": 107.94153594970703,
-      "learning_rate": 1.956521739130435e-06,
-      "loss": 4.8355,
+      "epoch": 0.0002471780505891077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646843549862763e-07,
+      "loss": 1.6221,
       "step": 9
     },
     {
-      "epoch": 0.0032679738562091504,
-      "grad_norm": 1034.4962158203125,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 4.5201,
+      "epoch": 0.00027464227843234187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298261665141815e-07,
+      "loss": 1.3969,
       "step": 10
     },
     {
-      "epoch": 0.003594771241830065,
-      "grad_norm": 73.4400863647461,
-      "learning_rate": 2.391304347826087e-06,
-      "loss": 3.9979,
+      "epoch": 0.00030210650627557605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0128087831655995e-07,
+      "loss": 1.563,
       "step": 11
     },
     {
-      "epoch": 0.00392156862745098,
-      "grad_norm": 80.6120834350586,
-      "learning_rate": 2.6086956521739132e-06,
-      "loss": 3.9287,
+      "epoch": 0.00032957073411881023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1957913998170175e-07,
+      "loss": 1.6204,
       "step": 12
     },
     {
-      "epoch": 0.004248366013071895,
-      "grad_norm": 67.04983520507812,
-      "learning_rate": 2.8260869565217393e-06,
-      "loss": 3.7238,
+      "epoch": 0.0003570349619620444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3787740164684357e-07,
+      "loss": 1.6775,
       "step": 13
     },
     {
-      "epoch": 0.004575163398692811,
-      "grad_norm": 46.324153900146484,
-      "learning_rate": 3.043478260869566e-06,
-      "loss": 3.3391,
+      "epoch": 0.00038449918980527865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5617566331198537e-07,
+      "loss": 1.7063,
       "step": 14
     },
     {
-      "epoch": 0.004901960784313725,
-      "grad_norm": 47.31280517578125,
-      "learning_rate": 3.2608695652173914e-06,
-      "loss": 3.184,
+      "epoch": 0.00041196341764851283,
+      "grad_norm": 0.0,
+      "learning_rate": 2.744739249771272e-07,
+      "loss": 1.6238,
       "step": 15
     },
     {
-      "epoch": 0.00522875816993464,
-      "grad_norm": 39.3209228515625,
-      "learning_rate": 3.4782608695652175e-06,
-      "loss": 3.0863,
+      "epoch": 0.000439427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.92772186642269e-07,
+      "loss": 1.6914,
       "step": 16
     },
     {
-      "epoch": 0.005555555555555556,
-      "grad_norm": 32.045310974121094,
-      "learning_rate": 3.6956521739130436e-06,
-      "loss": 3.0131,
+      "epoch": 0.0004668918733349812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1107044830741084e-07,
+      "loss": 1.5363,
       "step": 17
     },
     {
-      "epoch": 0.0058823529411764705,
-      "grad_norm": 24.695682525634766,
-      "learning_rate": 3.91304347826087e-06,
-      "loss": 2.8705,
+      "epoch": 0.0004943561011782154,
+      "grad_norm": 0.0,
+      "learning_rate": 3.293687099725526e-07,
+      "loss": 1.5259,
       "step": 18
     },
     {
-      "epoch": 0.006209150326797385,
-      "grad_norm": 23.613971710205078,
-      "learning_rate": 4.130434782608696e-06,
-      "loss": 2.7795,
+      "epoch": 0.0005218203290214496,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4766697163769443e-07,
+      "loss": 1.6575,
       "step": 19
     },
     {
-      "epoch": 0.006535947712418301,
-      "grad_norm": 17.700233459472656,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 2.6077,
+      "epoch": 0.0005492845568646837,
+      "grad_norm": 0.0,
+      "learning_rate": 3.659652333028363e-07,
+      "loss": 1.4421,
       "step": 20
     },
     {
-      "epoch": 0.006862745098039216,
-      "grad_norm": 14.554231643676758,
-      "learning_rate": 4.565217391304348e-06,
-      "loss": 2.5355,
+      "epoch": 0.000576748784707918,
+      "grad_norm": 0.0,
+      "learning_rate": 3.842634949679781e-07,
+      "loss": 1.5054,
       "step": 21
     },
     {
-      "epoch": 0.00718954248366013,
-      "grad_norm": 12.106937408447266,
-      "learning_rate": 4.782608695652174e-06,
-      "loss": 2.4457,
+      "epoch": 0.0006042130125511521,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025617566331199e-07,
+      "loss": 1.3625,
       "step": 22
     },
     {
-      "epoch": 0.007516339869281046,
-      "grad_norm": 11.585206985473633,
-      "learning_rate": 5e-06,
-      "loss": 2.3672,
+      "epoch": 0.0006316772403943863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2086001829826167e-07,
+      "loss": 1.4928,
       "step": 23
     },
     {
-      "epoch": 0.00784313725490196,
-      "grad_norm": 15.738335609436035,
-      "learning_rate": 5.2173913043478265e-06,
-      "loss": 2.3678,
+      "epoch": 0.0006591414682376205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391582799634035e-07,
+      "loss": 1.4324,
       "step": 24
     },
     {
-      "epoch": 0.008169934640522876,
-      "grad_norm": 13.756026268005371,
-      "learning_rate": 5.4347826086956525e-06,
-      "loss": 2.2731,
+      "epoch": 0.0006866056960808547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5745654162854537e-07,
+      "loss": 1.3921,
       "step": 25
     },
     {
-      "epoch": 0.00849673202614379,
-      "grad_norm": 11.496369361877441,
-      "learning_rate": 5.652173913043479e-06,
-      "loss": 2.2354,
+      "epoch": 0.0007140699239240888,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7575480329368714e-07,
+      "loss": 1.502,
       "step": 26
     },
     {
-      "epoch": 0.008823529411764706,
-      "grad_norm": 9.002495765686035,
-      "learning_rate": 5.8695652173913055e-06,
-      "loss": 2.1686,
+      "epoch": 0.0007415341517673231,
+      "grad_norm": 0.0,
+      "learning_rate": 4.94053064958829e-07,
+      "loss": 1.4207,
       "step": 27
     },
     {
-      "epoch": 0.009150326797385621,
-      "grad_norm": 9.574803352355957,
-      "learning_rate": 6.086956521739132e-06,
-      "loss": 2.1118,
+      "epoch": 0.0007689983796105573,
+      "grad_norm": 0.0,
+      "learning_rate": 5.123513266239707e-07,
+      "loss": 1.3965,
       "step": 28
     },
     {
-      "epoch": 0.009477124183006535,
-      "grad_norm": 8.496610641479492,
-      "learning_rate": 6.304347826086958e-06,
-      "loss": 2.0922,
+      "epoch": 0.0007964626074537914,
+      "grad_norm": 0.0,
+      "learning_rate": 5.306495882891126e-07,
+      "loss": 1.4606,
       "step": 29
     },
     {
-      "epoch": 0.00980392156862745,
-      "grad_norm": 9.064095497131348,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 2.0979,
+      "epoch": 0.0008239268352970257,
+      "grad_norm": 0.0,
+      "learning_rate": 5.489478499542544e-07,
+      "loss": 1.4561,
       "step": 30
     },
     {
-      "epoch": 0.010130718954248366,
-      "grad_norm": 6.706709384918213,
-      "learning_rate": 6.739130434782609e-06,
-      "loss": 2.0572,
+      "epoch": 0.0008513910631402598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.672461116193962e-07,
+      "loss": 1.5658,
       "step": 31
     },
     {
-      "epoch": 0.01045751633986928,
-      "grad_norm": 7.099653244018555,
-      "learning_rate": 6.956521739130435e-06,
-      "loss": 2.0112,
+      "epoch": 0.000878855290983494,
+      "grad_norm": 0.0,
+      "learning_rate": 5.85544373284538e-07,
+      "loss": 1.3806,
       "step": 32
     },
     {
-      "epoch": 0.010784313725490196,
-      "grad_norm": 6.086899280548096,
-      "learning_rate": 7.173913043478261e-06,
-      "loss": 1.9646,
+      "epoch": 0.0009063195188267282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.038426349496798e-07,
+      "loss": 1.4355,
       "step": 33
     },
     {
-      "epoch": 0.011111111111111112,
-      "grad_norm": 5.762439727783203,
-      "learning_rate": 7.391304347826087e-06,
-      "loss": 1.9934,
+      "epoch": 0.0009337837466699624,
+      "grad_norm": 0.0,
+      "learning_rate": 6.221408966148217e-07,
+      "loss": 1.3265,
       "step": 34
     },
     {
-      "epoch": 0.011437908496732025,
-      "grad_norm": 5.084736347198486,
-      "learning_rate": 7.608695652173914e-06,
-      "loss": 1.8997,
+      "epoch": 0.0009612479745131965,
+      "grad_norm": 0.0,
+      "learning_rate": 6.404391582799636e-07,
+      "loss": 1.4443,
       "step": 35
     },
     {
-      "epoch": 0.011764705882352941,
-      "grad_norm": 4.223413467407227,
-      "learning_rate": 7.82608695652174e-06,
-      "loss": 1.9332,
+      "epoch": 0.0009887122023564308,
+      "grad_norm": 0.0,
+      "learning_rate": 6.587374199451052e-07,
+      "loss": 1.3655,
       "step": 36
     },
     {
-      "epoch": 0.012091503267973857,
-      "grad_norm": 3.7073795795440674,
-      "learning_rate": 8.043478260869566e-06,
-      "loss": 1.897,
+      "epoch": 0.0010161764301996649,
+      "grad_norm": 0.0,
+      "learning_rate": 6.770356816102471e-07,
+      "loss": 1.4266,
       "step": 37
     },
     {
-      "epoch": 0.01241830065359477,
-      "grad_norm": 3.709670066833496,
-      "learning_rate": 8.260869565217392e-06,
-      "loss": 1.9031,
+      "epoch": 0.0010436406580428992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.953339432753889e-07,
+      "loss": 1.3354,
       "step": 38
     },
     {
-      "epoch": 0.012745098039215686,
-      "grad_norm": 3.622067928314209,
-      "learning_rate": 8.478260869565218e-06,
-      "loss": 1.8794,
+      "epoch": 0.0010711048858861334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.136322049405307e-07,
+      "loss": 1.3412,
       "step": 39
     },
     {
-      "epoch": 0.013071895424836602,
-      "grad_norm": 3.0796477794647217,
-      "learning_rate": 8.695652173913044e-06,
-      "loss": 1.86,
+      "epoch": 0.0010985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319304666056726e-07,
+      "loss": 1.3121,
       "step": 40
     },
     {
-      "epoch": 0.013398692810457516,
-      "grad_norm": 3.2281718254089355,
-      "learning_rate": 8.91304347826087e-06,
-      "loss": 1.8603,
+      "epoch": 0.0011260333415726016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.502287282708143e-07,
+      "loss": 1.2714,
       "step": 41
     },
     {
-      "epoch": 0.013725490196078431,
-      "grad_norm": 2.808539867401123,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 1.8536,
+      "epoch": 0.001153497569415836,
+      "grad_norm": 0.0,
+      "learning_rate": 7.685269899359562e-07,
+      "loss": 1.3658,
       "step": 42
     },
     {
-      "epoch": 0.014052287581699347,
-      "grad_norm": 2.9279775619506836,
-      "learning_rate": 9.347826086956523e-06,
-      "loss": 1.8416,
+      "epoch": 0.00118096179725907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.868252516010979e-07,
+      "loss": 1.321,
       "step": 43
     },
     {
-      "epoch": 0.01437908496732026,
-      "grad_norm": 2.902385950088501,
-      "learning_rate": 9.565217391304349e-06,
-      "loss": 1.8263,
+      "epoch": 0.0012084260251023042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.051235132662398e-07,
+      "loss": 1.279,
       "step": 44
     },
     {
-      "epoch": 0.014705882352941176,
-      "grad_norm": 2.69392728805542,
-      "learning_rate": 9.782608695652175e-06,
-      "loss": 1.7682,
+      "epoch": 0.0012358902529455383,
+      "grad_norm": 0.0,
+      "learning_rate": 8.234217749313817e-07,
+      "loss": 1.4259,
       "step": 45
     },
     {
-      "epoch": 0.015032679738562092,
-      "grad_norm": 3.4571120738983154,
-      "learning_rate": 1e-05,
-      "loss": 1.8336,
+      "epoch": 0.0012633544807887727,
+      "grad_norm": 0.0,
+      "learning_rate": 8.417200365965233e-07,
+      "loss": 1.2783,
       "step": 46
     },
     {
-      "epoch": 0.015359477124183006,
-      "grad_norm": 2.6960256099700928,
-      "learning_rate": 1.0217391304347829e-05,
-      "loss": 1.7726,
+      "epoch": 0.0012908187086320068,
+      "grad_norm": 0.0,
+      "learning_rate": 8.600182982616652e-07,
+      "loss": 1.2484,
       "step": 47
     },
     {
-      "epoch": 0.01568627450980392,
-      "grad_norm": 2.7725367546081543,
-      "learning_rate": 1.0434782608695653e-05,
-      "loss": 1.8145,
+      "epoch": 0.001318282936475241,
+      "grad_norm": 0.0,
+      "learning_rate": 8.78316559926807e-07,
+      "loss": 1.4684,
       "step": 48
     },
     {
-      "epoch": 0.016013071895424835,
-      "grad_norm": 2.9043965339660645,
-      "learning_rate": 1.0652173913043479e-05,
-      "loss": 1.7543,
+      "epoch": 0.0013457471643184753,
+      "grad_norm": 0.0,
+      "learning_rate": 8.966148215919489e-07,
+      "loss": 1.343,
       "step": 49
     },
     {
-      "epoch": 0.016339869281045753,
-      "grad_norm": 2.426875352859497,
-      "learning_rate": 1.0869565217391305e-05,
-      "loss": 1.7672,
+      "epoch": 0.0013732113921617094,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149130832570907e-07,
+      "loss": 1.3803,
       "step": 50
     },
     {
-      "epoch": 0.016666666666666666,
-      "grad_norm": 2.266655206680298,
-      "learning_rate": 1.1086956521739131e-05,
-      "loss": 1.7506,
+      "epoch": 0.0014006756200049435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.332113449222324e-07,
+      "loss": 1.2894,
       "step": 51
     },
     {
-      "epoch": 0.01699346405228758,
-      "grad_norm": 2.4189774990081787,
-      "learning_rate": 1.1304347826086957e-05,
-      "loss": 1.729,
+      "epoch": 0.0014281398478481777,
+      "grad_norm": 0.0,
+      "learning_rate": 9.515096065873743e-07,
+      "loss": 1.3667,
       "step": 52
     },
     {
-      "epoch": 0.017320261437908498,
-      "grad_norm": 3.3211092948913574,
-      "learning_rate": 1.1521739130434783e-05,
-      "loss": 1.7938,
+      "epoch": 0.001455604075691412,
+      "grad_norm": 0.0,
+      "learning_rate": 9.698078682525162e-07,
+      "loss": 1.3089,
       "step": 53
     },
     {
-      "epoch": 0.01764705882352941,
-      "grad_norm": 2.3615641593933105,
-      "learning_rate": 1.1739130434782611e-05,
-      "loss": 1.7711,
+      "epoch": 0.0014830683035346461,
+      "grad_norm": 0.0,
+      "learning_rate": 9.88106129917658e-07,
+      "loss": 1.3933,
       "step": 54
     },
     {
-      "epoch": 0.017973856209150325,
-      "grad_norm": 2.277372360229492,
-      "learning_rate": 1.1956521739130435e-05,
-      "loss": 1.7624,
+      "epoch": 0.0015105325313778803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064043915827997e-06,
+      "loss": 1.3556,
       "step": 55
     },
     {
-      "epoch": 0.018300653594771243,
-      "grad_norm": 2.696061134338379,
-      "learning_rate": 1.2173913043478263e-05,
-      "loss": 1.8405,
+      "epoch": 0.0015379967592211146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247026532479415e-06,
+      "loss": 1.3142,
       "step": 56
     },
     {
-      "epoch": 0.018627450980392157,
-      "grad_norm": 2.259239435195923,
-      "learning_rate": 1.2391304347826088e-05,
-      "loss": 1.7327,
+      "epoch": 0.0015654609870643487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0430009149130832e-06,
+      "loss": 1.2067,
       "step": 57
     },
     {
-      "epoch": 0.01895424836601307,
-      "grad_norm": 2.4332070350646973,
-      "learning_rate": 1.2608695652173915e-05,
-      "loss": 1.7102,
+      "epoch": 0.0015929252149075829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0612991765782252e-06,
+      "loss": 1.2626,
       "step": 58
     },
     {
-      "epoch": 0.019281045751633988,
-      "grad_norm": 2.413156270980835,
-      "learning_rate": 1.282608695652174e-05,
-      "loss": 1.7696,
+      "epoch": 0.001620389442750817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079597438243367e-06,
+      "loss": 1.2023,
       "step": 59
     },
     {
-      "epoch": 0.0196078431372549,
-      "grad_norm": 1.8719111680984497,
-      "learning_rate": 1.3043478260869566e-05,
-      "loss": 1.7289,
+      "epoch": 0.0016478536705940513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0978956999085088e-06,
+      "loss": 1.303,
       "step": 60
     },
     {
-      "epoch": 0.019934640522875816,
-      "grad_norm": 2.288964033126831,
-      "learning_rate": 1.3260869565217392e-05,
-      "loss": 1.6652,
+      "epoch": 0.0016753178984372855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1161939615736505e-06,
+      "loss": 1.2014,
       "step": 61
     },
     {
-      "epoch": 0.020261437908496733,
-      "grad_norm": 1.6563140153884888,
-      "learning_rate": 1.3478260869565218e-05,
-      "loss": 1.7031,
+      "epoch": 0.0017027821262805196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1344922232387923e-06,
+      "loss": 1.2105,
       "step": 62
     },
     {
-      "epoch": 0.020588235294117647,
-      "grad_norm": 2.2754271030426025,
-      "learning_rate": 1.3695652173913046e-05,
-      "loss": 1.7008,
+      "epoch": 0.0017302463541237537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1527904849039343e-06,
+      "loss": 1.2161,
       "step": 63
     },
     {
-      "epoch": 0.02091503267973856,
-      "grad_norm": 1.679484486579895,
-      "learning_rate": 1.391304347826087e-05,
-      "loss": 1.68,
+      "epoch": 0.001757710581966988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.171088746569076e-06,
+      "loss": 1.28,
       "step": 64
     },
     {
-      "epoch": 0.021241830065359478,
-      "grad_norm": 1.6373448371887207,
-      "learning_rate": 1.4130434782608698e-05,
-      "loss": 1.6756,
+      "epoch": 0.0017851748098102222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1893870082342178e-06,
+      "loss": 1.2353,
       "step": 65
     },
     {
-      "epoch": 0.021568627450980392,
-      "grad_norm": 1.6868451833724976,
-      "learning_rate": 1.4347826086956522e-05,
-      "loss": 1.7329,
+      "epoch": 0.0018126390376534563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2076852698993596e-06,
+      "loss": 1.2277,
       "step": 66
     },
     {
-      "epoch": 0.021895424836601306,
-      "grad_norm": 1.9283547401428223,
-      "learning_rate": 1.456521739130435e-05,
-      "loss": 1.722,
+      "epoch": 0.0018401032654966907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2259835315645016e-06,
+      "loss": 1.3584,
       "step": 67
     },
     {
-      "epoch": 0.022222222222222223,
-      "grad_norm": 2.1388015747070312,
-      "learning_rate": 1.4782608695652174e-05,
-      "loss": 1.721,
+      "epoch": 0.0018675674933399248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2442817932296434e-06,
+      "loss": 1.3751,
       "step": 68
     },
     {
-      "epoch": 0.022549019607843137,
-      "grad_norm": 2.186607837677002,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 1.6734,
+      "epoch": 0.001895031721183159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2625800548947851e-06,
+      "loss": 1.3365,
       "step": 69
     },
     {
-      "epoch": 0.02287581699346405,
-      "grad_norm": 1.642698049545288,
-      "learning_rate": 1.5217391304347828e-05,
-      "loss": 1.7176,
+      "epoch": 0.001922495949026393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.280878316559927e-06,
+      "loss": 1.2231,
       "step": 70
     },
     {
-      "epoch": 0.023202614379084968,
-      "grad_norm": 2.054966449737549,
-      "learning_rate": 1.5434782608695654e-05,
-      "loss": 1.7281,
+      "epoch": 0.0019499601768696274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2991765782250687e-06,
+      "loss": 1.2771,
       "step": 71
     },
     {
-      "epoch": 0.023529411764705882,
-      "grad_norm": 1.8186839818954468,
-      "learning_rate": 1.565217391304348e-05,
-      "loss": 1.675,
+      "epoch": 0.0019774244047128615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3174748398902104e-06,
+      "loss": 1.2168,
       "step": 72
     },
     {
-      "epoch": 0.023856209150326796,
-      "grad_norm": 2.2886812686920166,
-      "learning_rate": 1.5869565217391306e-05,
-      "loss": 1.6677,
+      "epoch": 0.002004888632556096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3357731015553524e-06,
+      "loss": 1.2792,
       "step": 73
     },
     {
-      "epoch": 0.024183006535947713,
-      "grad_norm": 2.062208652496338,
-      "learning_rate": 1.6086956521739132e-05,
-      "loss": 1.6818,
+      "epoch": 0.0020323528603993298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3540713632204942e-06,
+      "loss": 1.2772,
       "step": 74
     },
     {
-      "epoch": 0.024509803921568627,
-      "grad_norm": 2.038698673248291,
-      "learning_rate": 1.630434782608696e-05,
-      "loss": 1.6611,
+      "epoch": 0.002059817088242564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3723696248856362e-06,
+      "loss": 1.1987,
       "step": 75
     },
     {
-      "epoch": 0.02483660130718954,
-      "grad_norm": 2.7376365661621094,
-      "learning_rate": 1.6521739130434785e-05,
-      "loss": 1.67,
+      "epoch": 0.0020872813160857985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3906678865507777e-06,
+      "loss": 1.2405,
       "step": 76
     },
     {
-      "epoch": 0.02516339869281046,
-      "grad_norm": 2.177619218826294,
-      "learning_rate": 1.673913043478261e-05,
-      "loss": 1.6363,
+      "epoch": 0.0021147455439290324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4089661482159195e-06,
+      "loss": 1.2092,
       "step": 77
     },
     {
-      "epoch": 0.025490196078431372,
-      "grad_norm": 2.448625087738037,
-      "learning_rate": 1.6956521739130437e-05,
-      "loss": 1.6665,
+      "epoch": 0.0021422097717722667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4272644098810615e-06,
+      "loss": 1.2432,
       "step": 78
     },
     {
-      "epoch": 0.025816993464052286,
-      "grad_norm": 3.226773977279663,
-      "learning_rate": 1.7173913043478263e-05,
-      "loss": 1.6303,
+      "epoch": 0.0021696739996155006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455626715462032e-06,
+      "loss": 1.2573,
       "step": 79
     },
     {
-      "epoch": 0.026143790849673203,
-      "grad_norm": 3.243680000305176,
-      "learning_rate": 1.739130434782609e-05,
-      "loss": 1.6363,
+      "epoch": 0.002197138227458735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4638609332113452e-06,
+      "loss": 1.3247,
       "step": 80
     },
     {
-      "epoch": 0.026470588235294117,
-      "grad_norm": 2.7989518642425537,
-      "learning_rate": 1.7608695652173915e-05,
-      "loss": 1.6392,
+      "epoch": 0.0022246024553019693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4821591948764868e-06,
+      "loss": 1.2956,
       "step": 81
     },
     {
-      "epoch": 0.02679738562091503,
-      "grad_norm": 5.386751174926758,
-      "learning_rate": 1.782608695652174e-05,
-      "loss": 1.6396,
+      "epoch": 0.0022520666831452032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5004574565416286e-06,
+      "loss": 1.2766,
       "step": 82
     },
     {
-      "epoch": 0.02712418300653595,
-      "grad_norm": 5.94257926940918,
-      "learning_rate": 1.8043478260869567e-05,
-      "loss": 1.6021,
+      "epoch": 0.0022795309109884376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5187557182067705e-06,
+      "loss": 1.2224,
       "step": 83
     },
     {
-      "epoch": 0.027450980392156862,
-      "grad_norm": 4.704083442687988,
-      "learning_rate": 1.8260869565217393e-05,
-      "loss": 1.6903,
+      "epoch": 0.002306995138831672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5370539798719123e-06,
+      "loss": 1.0414,
       "step": 84
     },
     {
-      "epoch": 0.027777777777777776,
-      "grad_norm": 5.924014091491699,
-      "learning_rate": 1.847826086956522e-05,
-      "loss": 1.6313,
+      "epoch": 0.002334459366674906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5553522415370543e-06,
+      "loss": 1.2047,
       "step": 85
     },
     {
-      "epoch": 0.028104575163398694,
-      "grad_norm": 4.046575546264648,
-      "learning_rate": 1.8695652173913045e-05,
-      "loss": 1.6064,
+      "epoch": 0.00236192359451814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5736505032021959e-06,
+      "loss": 1.2291,
       "step": 86
     },
     {
-      "epoch": 0.028431372549019607,
-      "grad_norm": 6.7253193855285645,
-      "learning_rate": 1.891304347826087e-05,
-      "loss": 1.6141,
+      "epoch": 0.0023893878223613745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5919487648673376e-06,
+      "loss": 1.2036,
       "step": 87
     },
     {
-      "epoch": 0.02875816993464052,
-      "grad_norm": 4.927245140075684,
-      "learning_rate": 1.9130434782608697e-05,
-      "loss": 1.6199,
+      "epoch": 0.0024168520502046084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6102470265324796e-06,
+      "loss": 1.1955,
       "step": 88
     },
     {
-      "epoch": 0.02908496732026144,
-      "grad_norm": 4.71911096572876,
-      "learning_rate": 1.9347826086956523e-05,
-      "loss": 1.6538,
+      "epoch": 0.0024443162780478428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6285452881976214e-06,
+      "loss": 1.2741,
       "step": 89
     },
     {
-      "epoch": 0.029411764705882353,
-      "grad_norm": 3.2200610637664795,
-      "learning_rate": 1.956521739130435e-05,
-      "loss": 1.6665,
+      "epoch": 0.0024717805058910767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6468435498627634e-06,
+      "loss": 1.2816,
       "step": 90
     },
     {
-      "epoch": 0.029738562091503266,
-      "grad_norm": 2.535007953643799,
-      "learning_rate": 1.9782608695652176e-05,
-      "loss": 1.566,
+      "epoch": 0.002499244733734311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665141811527905e-06,
+      "loss": 1.3589,
       "step": 91
     },
     {
-      "epoch": 0.030065359477124184,
-      "grad_norm": 5.191009998321533,
-      "learning_rate": 2e-05,
-      "loss": 1.6346,
+      "epoch": 0.0025267089615775454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834400731930467e-06,
+      "loss": 1.2254,
       "step": 92
     },
     {
-      "epoch": 0.030392156862745098,
-      "grad_norm": 4.016279697418213,
-      "learning_rate": 1.9999994398015332e-05,
-      "loss": 1.6625,
+      "epoch": 0.0025541731894207793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7017383348581887e-06,
+      "loss": 1.2515,
       "step": 93
     },
     {
-      "epoch": 0.03071895424836601,
-      "grad_norm": 3.715398073196411,
-      "learning_rate": 1.9999977592067603e-05,
-      "loss": 1.6367,
+      "epoch": 0.0025816374172640136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7200365965233304e-06,
+      "loss": 1.2297,
       "step": 94
     },
     {
-      "epoch": 0.03104575163398693,
-      "grad_norm": 4.49213981628418,
-      "learning_rate": 1.9999949582175637e-05,
-      "loss": 1.6313,
+      "epoch": 0.002609101645107248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7383348581884724e-06,
+      "loss": 1.1483,
       "step": 95
     },
     {
-      "epoch": 0.03137254901960784,
-      "grad_norm": 4.44793701171875,
-      "learning_rate": 1.9999910368370826e-05,
-      "loss": 1.6295,
+      "epoch": 0.002636565872950482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756633119853614e-06,
+      "loss": 1.2375,
       "step": 96
     },
     {
-      "epoch": 0.03169934640522876,
-      "grad_norm": 3.7609996795654297,
-      "learning_rate": 1.9999859950697097e-05,
-      "loss": 1.613,
+      "epoch": 0.002664030100793716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7749313815187558e-06,
+      "loss": 1.2148,
       "step": 97
     },
     {
-      "epoch": 0.03202614379084967,
-      "grad_norm": 4.505050182342529,
-      "learning_rate": 1.9999798329210938e-05,
-      "loss": 1.6134,
+      "epoch": 0.0026914943286369506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7932296431838977e-06,
+      "loss": 1.1724,
       "step": 98
     },
     {
-      "epoch": 0.03235294117647059,
-      "grad_norm": 4.256154537200928,
-      "learning_rate": 1.9999725503981394e-05,
-      "loss": 1.669,
+      "epoch": 0.0027189585564801845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8115279048490395e-06,
+      "loss": 1.1733,
       "step": 99
     },
     {
-      "epoch": 0.032679738562091505,
-      "grad_norm": 4.464376926422119,
-      "learning_rate": 1.999964147509006e-05,
-      "loss": 1.6296,
+      "epoch": 0.002746422784323419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298261665141815e-06,
+      "loss": 1.3197,
       "step": 100
     },
     {
-      "epoch": 0.03300653594771242,
-      "grad_norm": 3.8376145362854004,
-      "learning_rate": 1.9999546242631073e-05,
-      "loss": 1.6121,
+      "epoch": 0.0027738870121666527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848124428179323e-06,
+      "loss": 1.1848,
       "step": 101
     },
     {
-      "epoch": 0.03333333333333333,
-      "grad_norm": 4.058324337005615,
-      "learning_rate": 1.9999439806711137e-05,
-      "loss": 1.5615,
+      "epoch": 0.002801351240009887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8664226898444648e-06,
+      "loss": 1.1934,
       "step": 102
     },
     {
-      "epoch": 0.03366013071895425,
-      "grad_norm": 4.652215003967285,
-      "learning_rate": 1.9999322167449503e-05,
-      "loss": 1.5895,
+      "epoch": 0.0028288154678531214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8847209515096068e-06,
+      "loss": 1.2139,
       "step": 103
     },
     {
-      "epoch": 0.03398692810457516,
-      "grad_norm": 3.8210885524749756,
-      "learning_rate": 1.9999193324977974e-05,
-      "loss": 1.6212,
+      "epoch": 0.0028562796956963553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030192131747486e-06,
+      "loss": 1.2678,
       "step": 104
     },
     {
-      "epoch": 0.03431372549019608,
-      "grad_norm": 2.9106106758117676,
-      "learning_rate": 1.99990532794409e-05,
-      "loss": 1.5943,
+      "epoch": 0.0028837439235395897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213174748398906e-06,
+      "loss": 1.2308,
       "step": 105
     },
     {
-      "epoch": 0.034640522875816995,
-      "grad_norm": 2.740485906600952,
-      "learning_rate": 1.999890203099519e-05,
-      "loss": 1.5698,
+      "epoch": 0.002911208151382824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396157365050323e-06,
+      "loss": 1.2008,
       "step": 106
     },
     {
-      "epoch": 0.03496732026143791,
-      "grad_norm": 2.9900577068328857,
-      "learning_rate": 1.9998739579810304e-05,
-      "loss": 1.5671,
+      "epoch": 0.002938672379226058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957913998170174e-06,
+      "loss": 1.101,
       "step": 107
     },
     {
-      "epoch": 0.03529411764705882,
-      "grad_norm": 4.3122239112854,
-      "learning_rate": 1.9998565926068253e-05,
-      "loss": 1.5651,
+      "epoch": 0.0029661366070692923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976212259835316e-06,
+      "loss": 1.2411,
       "step": 108
     },
     {
-      "epoch": 0.03562091503267974,
-      "grad_norm": 2.7479448318481445,
-      "learning_rate": 1.9998381069963593e-05,
-      "loss": 1.5779,
+      "epoch": 0.0029936008349125266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945105215004576e-06,
+      "loss": 1.2816,
       "step": 109
     },
     {
-      "epoch": 0.03594771241830065,
-      "grad_norm": 3.574004888534546,
-      "learning_rate": 1.999818501170344e-05,
-      "loss": 1.5692,
+      "epoch": 0.0030210650627557605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0128087831655994e-06,
+      "loss": 1.1897,
       "step": 110
     },
     {
-      "epoch": 0.03627450980392157,
-      "grad_norm": 3.529456615447998,
-      "learning_rate": 1.9997977751507456e-05,
-      "loss": 1.5703,
+      "epoch": 0.003048529290598995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.031107044830741e-06,
+      "loss": 1.2439,
       "step": 111
     },
     {
-      "epoch": 0.036601307189542485,
-      "grad_norm": 3.098120927810669,
-      "learning_rate": 1.9997759289607854e-05,
-      "loss": 1.5391,
+      "epoch": 0.003075993518442229,
+      "grad_norm": 0.0,
+      "learning_rate": 2.049405306495883e-06,
+      "loss": 1.1527,
       "step": 112
     },
     {
-      "epoch": 0.0369281045751634,
-      "grad_norm": 4.0015177726745605,
-      "learning_rate": 1.99975296262494e-05,
-      "loss": 1.573,
+      "epoch": 0.003103457746285463,
+      "grad_norm": 0.0,
+      "learning_rate": 2.067703568161025e-06,
+      "loss": 1.2227,
       "step": 113
     },
     {
-      "epoch": 0.03725490196078431,
-      "grad_norm": 3.4221723079681396,
-      "learning_rate": 1.9997288761689404e-05,
-      "loss": 1.5192,
+      "epoch": 0.0031309219741286975,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0860018298261665e-06,
+      "loss": 1.2294,
       "step": 114
     },
     {
-      "epoch": 0.03758169934640523,
-      "grad_norm": 4.231026649475098,
-      "learning_rate": 1.9997036696197735e-05,
-      "loss": 1.5536,
+      "epoch": 0.0031583862019719314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1043000914913083e-06,
+      "loss": 1.1948,
       "step": 115
     },
     {
-      "epoch": 0.03790849673202614,
-      "grad_norm": 2.8599634170532227,
-      "learning_rate": 1.9996773430056806e-05,
-      "loss": 1.5277,
+      "epoch": 0.0031858504298151657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1225983531564504e-06,
+      "loss": 1.0785,
       "step": 116
     },
     {
-      "epoch": 0.03823529411764706,
-      "grad_norm": 3.7466394901275635,
-      "learning_rate": 1.9996498963561572e-05,
-      "loss": 1.5785,
+      "epoch": 0.0032133146576584,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1408966148215922e-06,
+      "loss": 1.1322,
       "step": 117
     },
     {
-      "epoch": 0.038562091503267976,
-      "grad_norm": 3.190620183944702,
-      "learning_rate": 1.9996213297019556e-05,
-      "loss": 1.6036,
+      "epoch": 0.003240778885501634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.159194876486734e-06,
+      "loss": 1.2626,
       "step": 118
     },
     {
-      "epoch": 0.03888888888888889,
-      "grad_norm": 2.778001070022583,
-      "learning_rate": 1.999591643075081e-05,
-      "loss": 1.5435,
+      "epoch": 0.0032682431133448683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1774931381518758e-06,
+      "loss": 1.2365,
       "step": 119
     },
     {
-      "epoch": 0.0392156862745098,
-      "grad_norm": 3.7060232162475586,
-      "learning_rate": 1.9995608365087945e-05,
-      "loss": 1.5915,
+      "epoch": 0.0032957073411881027,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1957913998170175e-06,
+      "loss": 1.1459,
       "step": 120
     },
     {
-      "epoch": 0.03954248366013072,
-      "grad_norm": 3.376969337463379,
-      "learning_rate": 1.9995289100376113e-05,
-      "loss": 1.5214,
+      "epoch": 0.0033231715690313366,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2140896614821593e-06,
+      "loss": 1.1844,
       "step": 121
     },
     {
-      "epoch": 0.03986928104575163,
-      "grad_norm": 3.651099681854248,
-      "learning_rate": 1.9994958636973024e-05,
-      "loss": 1.59,
+      "epoch": 0.003350635796874571,
+      "grad_norm": 0.0,
+      "learning_rate": 2.232387923147301e-06,
+      "loss": 1.3112,
       "step": 122
     },
     {
-      "epoch": 0.04019607843137255,
-      "grad_norm": 4.168360233306885,
-      "learning_rate": 1.9994616975248925e-05,
-      "loss": 1.5412,
+      "epoch": 0.0033781000247178053,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2506861848124433e-06,
+      "loss": 1.1517,
       "step": 123
     },
     {
-      "epoch": 0.040522875816993466,
-      "grad_norm": 4.079196929931641,
-      "learning_rate": 1.999426411558661e-05,
-      "loss": 1.5411,
+      "epoch": 0.003405564252561039,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2689844464775846e-06,
+      "loss": 1.2363,
       "step": 124
     },
     {
-      "epoch": 0.04084967320261438,
-      "grad_norm": 3.6665360927581787,
-      "learning_rate": 1.9993900058381426e-05,
-      "loss": 1.5065,
+      "epoch": 0.0034330284804042735,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2872827081427264e-06,
+      "loss": 1.242,
       "step": 125
     },
     {
-      "epoch": 0.041176470588235294,
-      "grad_norm": 3.9368391036987305,
-      "learning_rate": 1.9993524804041255e-05,
-      "loss": 1.5557,
+      "epoch": 0.0034604927082475074,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3055809698078686e-06,
+      "loss": 1.2004,
       "step": 126
     },
     {
-      "epoch": 0.04150326797385621,
-      "grad_norm": 4.691778659820557,
-      "learning_rate": 1.9993138352986538e-05,
-      "loss": 1.5633,
+      "epoch": 0.0034879569360907418,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3238792314730103e-06,
+      "loss": 1.2158,
       "step": 127
     },
     {
-      "epoch": 0.04183006535947712,
-      "grad_norm": 3.3509023189544678,
-      "learning_rate": 1.9992740705650252e-05,
-      "loss": 1.4952,
+      "epoch": 0.003515421163933976,
+      "grad_norm": 0.0,
+      "learning_rate": 2.342177493138152e-06,
+      "loss": 1.1562,
       "step": 128
     },
     {
-      "epoch": 0.04215686274509804,
-      "grad_norm": 3.7970926761627197,
-      "learning_rate": 1.9992331862477922e-05,
-      "loss": 1.5708,
+      "epoch": 0.00354288539177721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.360475754803294e-06,
+      "loss": 1.2055,
       "step": 129
     },
     {
-      "epoch": 0.042483660130718956,
-      "grad_norm": 5.200839042663574,
-      "learning_rate": 1.9991911823927607e-05,
-      "loss": 1.54,
+      "epoch": 0.0035703496196204444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3787740164684357e-06,
+      "loss": 1.1396,
       "step": 130
     },
     {
-      "epoch": 0.04281045751633987,
-      "grad_norm": 4.036410331726074,
-      "learning_rate": 1.9991480590469922e-05,
-      "loss": 1.5031,
+      "epoch": 0.0035978138474636787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3970722781335774e-06,
+      "loss": 1.179,
       "step": 131
     },
     {
-      "epoch": 0.043137254901960784,
-      "grad_norm": 3.6340575218200684,
-      "learning_rate": 1.9991038162588018e-05,
-      "loss": 1.5174,
+      "epoch": 0.0036252780753069126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.415370539798719e-06,
+      "loss": 1.2169,
       "step": 132
     },
     {
-      "epoch": 0.0434640522875817,
-      "grad_norm": 3.8800292015075684,
-      "learning_rate": 1.9990584540777595e-05,
-      "loss": 1.5632,
+      "epoch": 0.003652742303150147,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4336688014638614e-06,
+      "loss": 1.1897,
       "step": 133
     },
     {
-      "epoch": 0.04379084967320261,
-      "grad_norm": 4.397945880889893,
-      "learning_rate": 1.999011972554688e-05,
-      "loss": 1.5247,
+      "epoch": 0.0036802065309933813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.451967063129003e-06,
+      "loss": 1.1252,
       "step": 134
     },
     {
-      "epoch": 0.04411764705882353,
-      "grad_norm": 4.1392340660095215,
-      "learning_rate": 1.998964371741666e-05,
-      "loss": 1.6018,
+      "epoch": 0.0037076707588366152,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4702653247941445e-06,
+      "loss": 1.0972,
       "step": 135
     },
     {
-      "epoch": 0.044444444444444446,
-      "grad_norm": 4.526089191436768,
-      "learning_rate": 1.9989156516920248e-05,
-      "loss": 1.5112,
+      "epoch": 0.0037351349866798496,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4885635864592867e-06,
+      "loss": 1.1776,
       "step": 136
     },
     {
-      "epoch": 0.04477124183006536,
-      "grad_norm": 3.1959755420684814,
-      "learning_rate": 1.9988658124603502e-05,
-      "loss": 1.5496,
+      "epoch": 0.0037625992145230835,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5068618481244285e-06,
+      "loss": 1.2098,
       "step": 137
     },
     {
-      "epoch": 0.045098039215686274,
-      "grad_norm": 4.511390686035156,
-      "learning_rate": 1.998814854102482e-05,
-      "loss": 1.5413,
+      "epoch": 0.003790063442366318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5251601097895702e-06,
+      "loss": 1.1565,
       "step": 138
     },
     {
-      "epoch": 0.04542483660130719,
-      "grad_norm": 3.0931382179260254,
-      "learning_rate": 1.998762776675514e-05,
-      "loss": 1.5322,
+      "epoch": 0.003817527670209552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5434583714547116e-06,
+      "loss": 1.1798,
       "step": 139
     },
     {
-      "epoch": 0.0457516339869281,
-      "grad_norm": 3.336494207382202,
-      "learning_rate": 1.9987095802377933e-05,
-      "loss": 1.5455,
+      "epoch": 0.003844991898052786,
+      "grad_norm": 0.0,
+      "learning_rate": 2.561756633119854e-06,
+      "loss": 1.2996,
       "step": 140
     },
     {
-      "epoch": 0.04607843137254902,
-      "grad_norm": 3.366914749145508,
-      "learning_rate": 1.9986552648489212e-05,
-      "loss": 1.5151,
+      "epoch": 0.0038724561258960204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5800548947849956e-06,
+      "loss": 1.1483,
       "step": 141
     },
     {
-      "epoch": 0.046405228758169936,
-      "grad_norm": 3.560352087020874,
-      "learning_rate": 1.998599830569752e-05,
-      "loss": 1.517,
+      "epoch": 0.0038999203537392548,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5983531564501373e-06,
+      "loss": 1.2255,
       "step": 142
     },
     {
-      "epoch": 0.04673202614379085,
-      "grad_norm": 3.575732469558716,
-      "learning_rate": 1.9985432774623947e-05,
-      "loss": 1.49,
+      "epoch": 0.003927384581582489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.616651418115279e-06,
+      "loss": 1.2207,
       "step": 143
     },
     {
-      "epoch": 0.047058823529411764,
-      "grad_norm": 4.391687393188477,
-      "learning_rate": 1.998485605590211e-05,
-      "loss": 1.5016,
+      "epoch": 0.003954848809425723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.634949679780421e-06,
+      "loss": 1.2468,
       "step": 144
     },
     {
-      "epoch": 0.04738562091503268,
-      "grad_norm": 3.3899121284484863,
-      "learning_rate": 1.998426815017817e-05,
-      "loss": 1.5442,
+      "epoch": 0.003982313037268957,
+      "grad_norm": 0.0,
+      "learning_rate": 2.653247941445563e-06,
+      "loss": 1.1663,
       "step": 145
     },
     {
-      "epoch": 0.04771241830065359,
-      "grad_norm": 3.1382956504821777,
-      "learning_rate": 1.99836690581108e-05,
-      "loss": 1.5102,
+      "epoch": 0.004009777265112192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.671546203110705e-06,
+      "loss": 1.1552,
       "step": 146
     },
     {
-      "epoch": 0.04803921568627451,
-      "grad_norm": 3.5683422088623047,
-      "learning_rate": 1.998305878037123e-05,
-      "loss": 1.5111,
+      "epoch": 0.004037241492955425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6898444647758466e-06,
+      "loss": 1.2036,
       "step": 147
     },
     {
-      "epoch": 0.048366013071895426,
-      "grad_norm": 3.5064756870269775,
-      "learning_rate": 1.9982437317643218e-05,
-      "loss": 1.4669,
+      "epoch": 0.0040647057207986595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7081427264409884e-06,
+      "loss": 1.1792,
       "step": 148
     },
     {
-      "epoch": 0.04869281045751634,
-      "grad_norm": 3.473219394683838,
-      "learning_rate": 1.998180467062304e-05,
-      "loss": 1.5044,
+      "epoch": 0.004092169948641894,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7264409881061297e-06,
+      "loss": 1.21,
       "step": 149
     },
     {
-      "epoch": 0.049019607843137254,
-      "grad_norm": 3.32301926612854,
-      "learning_rate": 1.9981160840019513e-05,
-      "loss": 1.5013,
+      "epoch": 0.004119634176485128,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7447392497712723e-06,
+      "loss": 1.1571,
       "step": 150
     },
     {
-      "epoch": 0.04934640522875817,
-      "grad_norm": 3.7283339500427246,
-      "learning_rate": 1.9980505826553984e-05,
-      "loss": 1.5269,
+      "epoch": 0.0041470984043283626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7630375114364137e-06,
+      "loss": 1.1296,
       "step": 151
     },
     {
-      "epoch": 0.04967320261437908,
-      "grad_norm": 3.357790470123291,
-      "learning_rate": 1.9979839630960333e-05,
-      "loss": 1.4882,
+      "epoch": 0.004174562632171597,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7813357731015555e-06,
+      "loss": 1.1715,
       "step": 152
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 3.874324083328247,
-      "learning_rate": 1.9979162253984956e-05,
-      "loss": 1.4958,
+      "epoch": 0.00420202686001483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7996340347666972e-06,
+      "loss": 1.1208,
       "step": 153
     },
     {
-      "epoch": 0.05032679738562092,
-      "grad_norm": 3.685762643814087,
-      "learning_rate": 1.9978473696386787e-05,
-      "loss": 1.5239,
+      "epoch": 0.004229491087858065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.817932296431839e-06,
+      "loss": 1.0857,
       "step": 154
     },
     {
-      "epoch": 0.05065359477124183,
-      "grad_norm": 3.1720829010009766,
-      "learning_rate": 1.997777395893729e-05,
-      "loss": 1.4632,
+      "epoch": 0.004256955315701299,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836230558096981e-06,
+      "loss": 1.1745,
       "step": 155
     },
     {
-      "epoch": 0.050980392156862744,
-      "grad_norm": 3.5734663009643555,
-      "learning_rate": 1.9977063042420438e-05,
-      "loss": 1.477,
+      "epoch": 0.004284419543544533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.854528819762123e-06,
+      "loss": 1.3831,
       "step": 156
     },
     {
-      "epoch": 0.05130718954248366,
-      "grad_norm": 3.401782989501953,
-      "learning_rate": 1.9976340947632743e-05,
-      "loss": 1.5088,
+      "epoch": 0.004311883771387768,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8728270814272647e-06,
+      "loss": 1.1889,
       "step": 157
     },
     {
-      "epoch": 0.05163398692810457,
-      "grad_norm": 3.2480790615081787,
-      "learning_rate": 1.9975607675383243e-05,
-      "loss": 1.4856,
+      "epoch": 0.004339347999231001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8911253430924065e-06,
+      "loss": 1.1417,
       "step": 158
     },
     {
-      "epoch": 0.05196078431372549,
-      "grad_norm": 3.5424041748046875,
-      "learning_rate": 1.997486322649349e-05,
-      "loss": 1.5191,
+      "epoch": 0.004366812227074236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.909423604757548e-06,
+      "loss": 1.2154,
       "step": 159
     },
     {
-      "epoch": 0.05228758169934641,
-      "grad_norm": 3.332812547683716,
-      "learning_rate": 1.997410760179756e-05,
-      "loss": 1.5211,
+      "epoch": 0.00439427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9277218664226905e-06,
+      "loss": 1.2657,
       "step": 160
     },
     {
-      "epoch": 0.05261437908496732,
-      "grad_norm": 3.576366662979126,
-      "learning_rate": 1.997334080214206e-05,
-      "loss": 1.4736,
+      "epoch": 0.004421740682760704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.946020128087832e-06,
+      "loss": 1.1462,
       "step": 161
     },
     {
-      "epoch": 0.052941176470588235,
-      "grad_norm": 3.2843410968780518,
-      "learning_rate": 1.99725628283861e-05,
-      "loss": 1.4989,
+      "epoch": 0.004449204910603939,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9643183897529736e-06,
+      "loss": 1.2156,
       "step": 162
     },
     {
-      "epoch": 0.05326797385620915,
-      "grad_norm": 3.600156784057617,
-      "learning_rate": 1.997177368140133e-05,
-      "loss": 1.4919,
+      "epoch": 0.004476669138447173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9826166514181154e-06,
+      "loss": 1.1558,
       "step": 163
     },
     {
-      "epoch": 0.05359477124183006,
-      "grad_norm": 3.436350107192993,
-      "learning_rate": 1.99709733620719e-05,
-      "loss": 1.4727,
+      "epoch": 0.0045041333662904064,
+      "grad_norm": 0.0,
+      "learning_rate": 3.000914913083257e-06,
+      "loss": 1.1626,
       "step": 164
     },
     {
-      "epoch": 0.05392156862745098,
-      "grad_norm": 3.4844353199005127,
-      "learning_rate": 1.997016187129449e-05,
-      "loss": 1.4711,
+      "epoch": 0.004531597594133641,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0192131747483993e-06,
+      "loss": 1.1611,
       "step": 165
     },
     {
-      "epoch": 0.0542483660130719,
-      "grad_norm": 2.8937063217163086,
-      "learning_rate": 1.9969339209978287e-05,
-      "loss": 1.4857,
+      "epoch": 0.004559061821976875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.037511436413541e-06,
+      "loss": 1.1149,
       "step": 166
     },
     {
-      "epoch": 0.05457516339869281,
-      "grad_norm": 3.423212766647339,
-      "learning_rate": 1.9968505379045e-05,
-      "loss": 1.5011,
+      "epoch": 0.0045865260498201095,
+      "grad_norm": 0.0,
+      "learning_rate": 3.055809698078683e-06,
+      "loss": 1.1115,
       "step": 167
     },
     {
-      "epoch": 0.054901960784313725,
-      "grad_norm": 4.952045440673828,
-      "learning_rate": 1.9967660379428855e-05,
-      "loss": 1.5131,
+      "epoch": 0.004613990277663344,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0741079597438246e-06,
+      "loss": 1.1837,
       "step": 168
     },
     {
-      "epoch": 0.05522875816993464,
-      "grad_norm": 4.23706579208374,
-      "learning_rate": 1.9966804212076583e-05,
-      "loss": 1.4846,
+      "epoch": 0.004641454505506577,
+      "grad_norm": 0.0,
+      "learning_rate": 3.092406221408966e-06,
+      "loss": 1.2615,
       "step": 169
     },
     {
-      "epoch": 0.05555555555555555,
-      "grad_norm": 3.1214308738708496,
-      "learning_rate": 1.996593687794743e-05,
-      "loss": 1.4886,
+      "epoch": 0.004668918733349812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1107044830741086e-06,
+      "loss": 1.0778,
       "step": 170
     },
     {
-      "epoch": 0.05588235294117647,
-      "grad_norm": 2.8994576930999756,
-      "learning_rate": 1.9965058378013157e-05,
-      "loss": 1.4988,
+      "epoch": 0.004696382961193046,
+      "grad_norm": 0.0,
+      "learning_rate": 3.12900274473925e-06,
+      "loss": 1.0592,
       "step": 171
     },
     {
-      "epoch": 0.05620915032679739,
-      "grad_norm": 3.648244619369507,
-      "learning_rate": 1.996416871325803e-05,
-      "loss": 1.5495,
+      "epoch": 0.00472384718903628,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1473010064043917e-06,
+      "loss": 1.1002,
       "step": 172
     },
     {
-      "epoch": 0.0565359477124183,
-      "grad_norm": 4.249721527099609,
-      "learning_rate": 1.996326788467883e-05,
-      "loss": 1.4651,
+      "epoch": 0.004751311416879515,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1655992680695335e-06,
+      "loss": 1.2027,
       "step": 173
     },
     {
-      "epoch": 0.056862745098039215,
-      "grad_norm": 3.194725513458252,
-      "learning_rate": 1.996235589328484e-05,
-      "loss": 1.4503,
+      "epoch": 0.004778775644722749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1838975297346752e-06,
+      "loss": 1.177,
       "step": 174
     },
     {
-      "epoch": 0.05718954248366013,
-      "grad_norm": 2.9750545024871826,
-      "learning_rate": 1.9961432740097856e-05,
-      "loss": 1.4869,
+      "epoch": 0.0048062398725659825,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2021957913998174e-06,
+      "loss": 1.139,
       "step": 175
     },
     {
-      "epoch": 0.05751633986928104,
-      "grad_norm": 2.762216329574585,
-      "learning_rate": 1.996049842615217e-05,
-      "loss": 1.5063,
+      "epoch": 0.004833704100409217,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2204940530649592e-06,
+      "loss": 1.1938,
       "step": 176
     },
     {
-      "epoch": 0.05784313725490196,
-      "grad_norm": 3.516991138458252,
-      "learning_rate": 1.9959552952494593e-05,
-      "loss": 1.5252,
+      "epoch": 0.004861168328252451,
+      "grad_norm": 0.0,
+      "learning_rate": 3.238792314730101e-06,
+      "loss": 1.1224,
       "step": 177
     },
     {
-      "epoch": 0.05816993464052288,
-      "grad_norm": 2.6470696926116943,
-      "learning_rate": 1.995859632018442e-05,
-      "loss": 1.504,
+      "epoch": 0.0048886325560956855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2570905763952428e-06,
+      "loss": 1.1071,
       "step": 178
     },
     {
-      "epoch": 0.05849673202614379,
-      "grad_norm": 3.3341174125671387,
-      "learning_rate": 1.995762853029347e-05,
-      "loss": 1.4933,
+      "epoch": 0.00491609678393892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2753888380603845e-06,
+      "loss": 1.1241,
       "step": 179
     },
     {
-      "epoch": 0.058823529411764705,
-      "grad_norm": 2.9031410217285156,
-      "learning_rate": 1.995664958390604e-05,
-      "loss": 1.487,
+      "epoch": 0.004943561011782153,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2936870997255267e-06,
+      "loss": 1.2053,
       "step": 180
     },
     {
-      "epoch": 0.05915032679738562,
-      "grad_norm": 3.1738388538360596,
-      "learning_rate": 1.995565948211895e-05,
-      "loss": 1.4744,
+      "epoch": 0.004971025239625388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3119853613906685e-06,
+      "loss": 1.1533,
       "step": 181
     },
     {
-      "epoch": 0.05947712418300653,
-      "grad_norm": 2.730618476867676,
-      "learning_rate": 1.99546582260415e-05,
-      "loss": 1.4765,
+      "epoch": 0.004998489467468622,
+      "grad_norm": 0.0,
+      "learning_rate": 3.33028362305581e-06,
+      "loss": 1.1522,
       "step": 182
     },
     {
-      "epoch": 0.059803921568627454,
-      "grad_norm": 2.9981024265289307,
-      "learning_rate": 1.99536458167955e-05,
-      "loss": 1.4611,
+      "epoch": 0.005025953695311856,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3485818847209516e-06,
+      "loss": 1.1573,
       "step": 183
     },
     {
-      "epoch": 0.06013071895424837,
-      "grad_norm": 3.4508137702941895,
-      "learning_rate": 1.995262225551524e-05,
-      "loss": 1.5346,
+      "epoch": 0.005053417923155091,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3668801463860934e-06,
+      "loss": 1.0895,
       "step": 184
     },
     {
-      "epoch": 0.06045751633986928,
-      "grad_norm": 2.8818087577819824,
-      "learning_rate": 1.9951587543347527e-05,
-      "loss": 1.4791,
+      "epoch": 0.005080882150998325,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3851784080512356e-06,
+      "loss": 1.0415,
       "step": 185
     },
     {
-      "epoch": 0.060784313725490195,
-      "grad_norm": 2.9819560050964355,
-      "learning_rate": 1.9950541681451644e-05,
-      "loss": 1.4741,
+      "epoch": 0.0051083463788415585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4034766697163773e-06,
+      "loss": 1.1145,
       "step": 186
     },
     {
-      "epoch": 0.06111111111111111,
-      "grad_norm": 2.9156906604766846,
-      "learning_rate": 1.9949484670999367e-05,
-      "loss": 1.402,
+      "epoch": 0.005135810606684793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.421774931381519e-06,
+      "loss": 1.1654,
       "step": 187
     },
     {
-      "epoch": 0.06143790849673202,
-      "grad_norm": 4.016847133636475,
-      "learning_rate": 1.9948416513174976e-05,
-      "loss": 1.4896,
+      "epoch": 0.005163274834528027,
+      "grad_norm": 0.0,
+      "learning_rate": 3.440073193046661e-06,
+      "loss": 1.2862,
       "step": 188
     },
     {
-      "epoch": 0.061764705882352944,
-      "grad_norm": 3.464527130126953,
-      "learning_rate": 1.9947337209175223e-05,
-      "loss": 1.4632,
+      "epoch": 0.005190739062371262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4583714547118027e-06,
+      "loss": 1.1634,
       "step": 189
     },
     {
-      "epoch": 0.06209150326797386,
-      "grad_norm": 3.1484811305999756,
-      "learning_rate": 1.994624676020936e-05,
-      "loss": 1.4579,
+      "epoch": 0.005218203290214496,
+      "grad_norm": 0.0,
+      "learning_rate": 3.476669716376945e-06,
+      "loss": 1.1613,
       "step": 190
     },
     {
-      "epoch": 0.06241830065359477,
-      "grad_norm": 2.9161295890808105,
-      "learning_rate": 1.994514516749913e-05,
-      "loss": 1.4748,
+      "epoch": 0.005245667518057729,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4949679780420866e-06,
+      "loss": 1.1861,
       "step": 191
     },
     {
-      "epoch": 0.06274509803921569,
-      "grad_norm": 3.3965725898742676,
-      "learning_rate": 1.9944032432278743e-05,
-      "loss": 1.4745,
+      "epoch": 0.005273131745900964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.513266239707228e-06,
+      "loss": 1.2162,
       "step": 192
     },
     {
-      "epoch": 0.0630718954248366,
-      "grad_norm": 3.274447441101074,
-      "learning_rate": 1.9942908555794912e-05,
-      "loss": 1.3999,
+      "epoch": 0.005300595973744198,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5315645013723697e-06,
+      "loss": 1.1166,
       "step": 193
     },
     {
-      "epoch": 0.06339869281045751,
-      "grad_norm": 3.3372459411621094,
-      "learning_rate": 1.9941773539306818e-05,
-      "loss": 1.5094,
+      "epoch": 0.005328060201587432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5498627630375115e-06,
+      "loss": 1.0579,
       "step": 194
     },
     {
-      "epoch": 0.06372549019607843,
-      "grad_norm": 3.7964842319488525,
-      "learning_rate": 1.9940627384086135e-05,
-      "loss": 1.4634,
+      "epoch": 0.005355524429430667,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5681610247026537e-06,
+      "loss": 1.1235,
       "step": 195
     },
     {
-      "epoch": 0.06405228758169934,
-      "grad_norm": 2.9186158180236816,
-      "learning_rate": 1.9939470091417012e-05,
-      "loss": 1.4745,
+      "epoch": 0.005382988657273901,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5864592863677955e-06,
+      "loss": 1.1003,
       "step": 196
     },
     {
-      "epoch": 0.06437908496732025,
-      "grad_norm": 3.1373605728149414,
-      "learning_rate": 1.9938301662596075e-05,
-      "loss": 1.45,
+      "epoch": 0.005410452885117135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6047575480329372e-06,
+      "loss": 1.1634,
       "step": 197
     },
     {
-      "epoch": 0.06470588235294118,
-      "grad_norm": 3.7945210933685303,
-      "learning_rate": 1.9937122098932428e-05,
-      "loss": 1.4868,
+      "epoch": 0.005437917112960369,
+      "grad_norm": 0.0,
+      "learning_rate": 3.623055809698079e-06,
+      "loss": 1.2423,
       "step": 198
     },
     {
-      "epoch": 0.0650326797385621,
-      "grad_norm": 3.216769218444824,
-      "learning_rate": 1.993593140174765e-05,
-      "loss": 1.4804,
+      "epoch": 0.005465381340803603,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6413540713632208e-06,
+      "loss": 1.1833,
       "step": 199
     },
     {
-      "epoch": 0.06535947712418301,
-      "grad_norm": 3.021155834197998,
-      "learning_rate": 1.9934729572375792e-05,
-      "loss": 1.4612,
+      "epoch": 0.005492845568646838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.659652333028363e-06,
+      "loss": 1.1599,
       "step": 200
     },
     {
-      "epoch": 0.06568627450980392,
-      "grad_norm": 2.927077531814575,
-      "learning_rate": 1.9933516612163385e-05,
-      "loss": 1.4467,
+      "epoch": 0.005520309796490072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6779505946935047e-06,
+      "loss": 1.1855,
       "step": 201
     },
     {
-      "epoch": 0.06601307189542484,
-      "grad_norm": 3.3251519203186035,
-      "learning_rate": 1.9932292522469424e-05,
-      "loss": 1.4685,
+      "epoch": 0.0055477740243333054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.696248856358646e-06,
+      "loss": 1.1403,
       "step": 202
     },
     {
-      "epoch": 0.06633986928104575,
-      "grad_norm": 3.198406457901001,
-      "learning_rate": 1.9931057304665376e-05,
-      "loss": 1.451,
+      "epoch": 0.00557523825217654,
+      "grad_norm": 0.0,
+      "learning_rate": 3.714547118023788e-06,
+      "loss": 1.1366,
       "step": 203
     },
     {
-      "epoch": 0.06666666666666667,
-      "grad_norm": 3.436568021774292,
-      "learning_rate": 1.992981096013517e-05,
-      "loss": 1.467,
+      "epoch": 0.005602702480019774,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7328453796889296e-06,
+      "loss": 1.1817,
       "step": 204
     },
     {
-      "epoch": 0.06699346405228758,
-      "grad_norm": 2.9751980304718018,
-      "learning_rate": 1.9928553490275212e-05,
-      "loss": 1.4161,
+      "epoch": 0.0056301667078630085,
+      "grad_norm": 0.0,
+      "learning_rate": 3.751143641354072e-06,
+      "loss": 1.2004,
       "step": 205
     },
     {
-      "epoch": 0.0673202614379085,
-      "grad_norm": 3.3083081245422363,
-      "learning_rate": 1.992728489649437e-05,
-      "loss": 1.4595,
+      "epoch": 0.005657630935706243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7694419030192136e-06,
+      "loss": 1.12,
       "step": 206
     },
     {
-      "epoch": 0.06764705882352941,
-      "grad_norm": 3.3727641105651855,
-      "learning_rate": 1.9926005180213966e-05,
-      "loss": 1.43,
+      "epoch": 0.005685095163549477,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7877401646843554e-06,
+      "loss": 1.0684,
       "step": 207
     },
     {
-      "epoch": 0.06797385620915032,
-      "grad_norm": 3.001987934112549,
-      "learning_rate": 1.992471434286779e-05,
-      "loss": 1.484,
+      "epoch": 0.005712559391392711,
+      "grad_norm": 0.0,
+      "learning_rate": 3.806038426349497e-06,
+      "loss": 1.2816,
       "step": 208
     },
     {
-      "epoch": 0.06830065359477124,
-      "grad_norm": 3.224729537963867,
-      "learning_rate": 1.99234123859021e-05,
-      "loss": 1.4842,
+      "epoch": 0.005740023619235945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8243366880146385e-06,
+      "loss": 1.1233,
       "step": 209
     },
     {
-      "epoch": 0.06862745098039216,
-      "grad_norm": 2.9690120220184326,
-      "learning_rate": 1.99220993107756e-05,
-      "loss": 1.4356,
+      "epoch": 0.005767487847079179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.842634949679781e-06,
+      "loss": 1.186,
       "step": 210
     },
     {
-      "epoch": 0.06895424836601308,
-      "grad_norm": 3.071007251739502,
-      "learning_rate": 1.9920775118959454e-05,
-      "loss": 1.4513,
+      "epoch": 0.005794952074922414,
+      "grad_norm": 0.0,
+      "learning_rate": 3.860933211344923e-06,
+      "loss": 1.1672,
       "step": 211
     },
     {
-      "epoch": 0.06928104575163399,
-      "grad_norm": 3.344125270843506,
-      "learning_rate": 1.9919439811937283e-05,
-      "loss": 1.4453,
+      "epoch": 0.005822416302765648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.879231473010065e-06,
+      "loss": 1.2286,
       "step": 212
     },
     {
-      "epoch": 0.0696078431372549,
-      "grad_norm": 3.1145882606506348,
-      "learning_rate": 1.991809339120516e-05,
-      "loss": 1.4321,
+      "epoch": 0.005849880530608882,
+      "grad_norm": 0.0,
+      "learning_rate": 3.897529734675206e-06,
+      "loss": 1.1808,
       "step": 213
     },
     {
-      "epoch": 0.06993464052287582,
-      "grad_norm": 2.8141672611236572,
-      "learning_rate": 1.9916735858271615e-05,
-      "loss": 1.4152,
+      "epoch": 0.005877344758452116,
+      "grad_norm": 0.0,
+      "learning_rate": 3.915827996340348e-06,
+      "loss": 1.1252,
       "step": 214
     },
     {
-      "epoch": 0.07026143790849673,
-      "grad_norm": 2.5612850189208984,
-      "learning_rate": 1.991536721465762e-05,
-      "loss": 1.4167,
+      "epoch": 0.00590480898629535,
+      "grad_norm": 0.0,
+      "learning_rate": 3.934126258005489e-06,
+      "loss": 1.1938,
       "step": 215
     },
     {
-      "epoch": 0.07058823529411765,
-      "grad_norm": 3.221402645111084,
-      "learning_rate": 1.9913987461896597e-05,
-      "loss": 1.4418,
+      "epoch": 0.0059322732141385845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.952424519670632e-06,
+      "loss": 1.1943,
       "step": 216
     },
     {
-      "epoch": 0.07091503267973856,
-      "grad_norm": 2.714503526687622,
-      "learning_rate": 1.9912596601534422e-05,
-      "loss": 1.4162,
+      "epoch": 0.005959737441981819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9707227813357735e-06,
+      "loss": 1.1385,
       "step": 217
     },
     {
-      "epoch": 0.07124183006535947,
-      "grad_norm": 3.2948598861694336,
-      "learning_rate": 1.991119463512941e-05,
-      "loss": 1.4326,
+      "epoch": 0.005987201669825053,
+      "grad_norm": 0.0,
+      "learning_rate": 3.989021043000915e-06,
+      "loss": 1.1805,
       "step": 218
     },
     {
-      "epoch": 0.07156862745098039,
-      "grad_norm": 2.6665005683898926,
-      "learning_rate": 1.9909781564252314e-05,
-      "loss": 1.4507,
+      "epoch": 0.006014665897668287,
+      "grad_norm": 0.0,
+      "learning_rate": 4.007319304666057e-06,
+      "loss": 1.1771,
       "step": 219
     },
     {
-      "epoch": 0.0718954248366013,
-      "grad_norm": 2.9566078186035156,
-      "learning_rate": 1.9908357390486342e-05,
-      "loss": 1.4738,
+      "epoch": 0.006042130125511521,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025617566331199e-06,
+      "loss": 1.0819,
       "step": 220
     },
     {
-      "epoch": 0.07222222222222222,
-      "grad_norm": 2.4681146144866943,
-      "learning_rate": 1.990692211542713e-05,
-      "loss": 1.4624,
+      "epoch": 0.006069594353354755,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0439158279963406e-06,
+      "loss": 1.2445,
       "step": 221
     },
     {
-      "epoch": 0.07254901960784314,
-      "grad_norm": 2.896775960922241,
-      "learning_rate": 1.9905475740682756e-05,
-      "loss": 1.409,
+      "epoch": 0.00609705858119799,
+      "grad_norm": 0.0,
+      "learning_rate": 4.062214089661482e-06,
+      "loss": 1.2114,
       "step": 222
     },
     {
-      "epoch": 0.07287581699346406,
-      "grad_norm": 3.0612738132476807,
-      "learning_rate": 1.9904018267873733e-05,
-      "loss": 1.4476,
+      "epoch": 0.006124522809041224,
+      "grad_norm": 0.0,
+      "learning_rate": 4.080512351326624e-06,
+      "loss": 1.0655,
       "step": 223
     },
     {
-      "epoch": 0.07320261437908497,
-      "grad_norm": 2.7236390113830566,
-      "learning_rate": 1.990254969863301e-05,
-      "loss": 1.3894,
+      "epoch": 0.006151987036884458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.098810612991766e-06,
+      "loss": 1.0993,
       "step": 224
     },
     {
-      "epoch": 0.07352941176470588,
-      "grad_norm": 2.8606607913970947,
-      "learning_rate": 1.9901070034605968e-05,
-      "loss": 1.4643,
+      "epoch": 0.006179451264727692,
+      "grad_norm": 0.0,
+      "learning_rate": 4.117108874656908e-06,
+      "loss": 1.1742,
       "step": 225
     },
     {
-      "epoch": 0.0738562091503268,
-      "grad_norm": 3.1209611892700195,
-      "learning_rate": 1.9899579277450417e-05,
-      "loss": 1.4574,
+      "epoch": 0.006206915492570926,
+      "grad_norm": 0.0,
+      "learning_rate": 4.13540713632205e-06,
+      "loss": 1.0986,
       "step": 226
     },
     {
-      "epoch": 0.07418300653594771,
-      "grad_norm": 2.8395328521728516,
-      "learning_rate": 1.98980774288366e-05,
-      "loss": 1.3938,
+      "epoch": 0.006234379720414161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.153705397987192e-06,
+      "loss": 1.1749,
       "step": 227
     },
     {
-      "epoch": 0.07450980392156863,
-      "grad_norm": 2.8646841049194336,
-      "learning_rate": 1.989656449044718e-05,
-      "loss": 1.505,
+      "epoch": 0.006261843948257395,
+      "grad_norm": 0.0,
+      "learning_rate": 4.172003659652333e-06,
+      "loss": 1.1092,
       "step": 228
     },
     {
-      "epoch": 0.07483660130718954,
-      "grad_norm": 2.8520190715789795,
-      "learning_rate": 1.989504046397725e-05,
-      "loss": 1.4695,
+      "epoch": 0.006289308176100629,
+      "grad_norm": 0.0,
+      "learning_rate": 4.190301921317475e-06,
+      "loss": 1.1416,
       "step": 229
     },
     {
-      "epoch": 0.07516339869281045,
-      "grad_norm": 2.583080768585205,
-      "learning_rate": 1.9893505351134322e-05,
-      "loss": 1.3985,
+      "epoch": 0.006316772403943863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2086001829826165e-06,
+      "loss": 1.1833,
       "step": 230
     },
     {
-      "epoch": 0.07549019607843137,
-      "grad_norm": 2.7279202938079834,
-      "learning_rate": 1.9891959153638332e-05,
-      "loss": 1.4693,
+      "epoch": 0.006344236631787097,
+      "grad_norm": 0.0,
+      "learning_rate": 4.226898444647759e-06,
+      "loss": 1.1022,
       "step": 231
     },
     {
-      "epoch": 0.07581699346405228,
-      "grad_norm": 3.1004648208618164,
-      "learning_rate": 1.9890401873221642e-05,
-      "loss": 1.4327,
+      "epoch": 0.0063717008596303314,
+      "grad_norm": 0.0,
+      "learning_rate": 4.245196706312901e-06,
+      "loss": 1.1034,
       "step": 232
     },
     {
-      "epoch": 0.0761437908496732,
-      "grad_norm": 2.718640089035034,
-      "learning_rate": 1.9888833511629017e-05,
-      "loss": 1.4246,
+      "epoch": 0.006399165087473566,
+      "grad_norm": 0.0,
+      "learning_rate": 4.263494967978043e-06,
+      "loss": 1.1933,
       "step": 233
     },
     {
-      "epoch": 0.07647058823529412,
-      "grad_norm": 3.3611106872558594,
-      "learning_rate": 1.9887254070617644e-05,
-      "loss": 1.478,
+      "epoch": 0.0064266293153168,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2817932296431844e-06,
+      "loss": 1.1765,
       "step": 234
     },
     {
-      "epoch": 0.07679738562091504,
-      "grad_norm": 3.02630352973938,
-      "learning_rate": 1.9885663551957128e-05,
-      "loss": 1.4266,
+      "epoch": 0.0064540935431600345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.300091491308326e-06,
+      "loss": 1.1736,
       "step": 235
     },
     {
-      "epoch": 0.07712418300653595,
-      "grad_norm": 3.1517412662506104,
-      "learning_rate": 1.988406195742948e-05,
-      "loss": 1.4094,
+      "epoch": 0.006481557771003268,
+      "grad_norm": 0.0,
+      "learning_rate": 4.318389752973468e-06,
+      "loss": 1.1617,
       "step": 236
     },
     {
-      "epoch": 0.07745098039215687,
-      "grad_norm": 2.9395110607147217,
-      "learning_rate": 1.988244928882912e-05,
-      "loss": 1.4231,
+      "epoch": 0.006509021998846502,
+      "grad_norm": 0.0,
+      "learning_rate": 4.33668801463861e-06,
+      "loss": 1.1404,
       "step": 237
     },
     {
-      "epoch": 0.07777777777777778,
-      "grad_norm": 3.008352518081665,
-      "learning_rate": 1.9880825547962882e-05,
-      "loss": 1.4623,
+      "epoch": 0.006536486226689737,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3549862763037515e-06,
+      "loss": 1.1012,
       "step": 238
     },
     {
-      "epoch": 0.07810457516339869,
-      "grad_norm": 3.165929079055786,
-      "learning_rate": 1.9879190736649992e-05,
-      "loss": 1.3888,
+      "epoch": 0.006563950454532971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.373284537968893e-06,
+      "loss": 1.1377,
       "step": 239
     },
     {
-      "epoch": 0.0784313725490196,
-      "grad_norm": 3.5827269554138184,
-      "learning_rate": 1.9877544856722094e-05,
-      "loss": 1.4153,
+      "epoch": 0.006591414682376205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391582799634035e-06,
+      "loss": 1.1825,
       "step": 240
     },
     {
-      "epoch": 0.07875816993464052,
-      "grad_norm": 3.088611602783203,
-      "learning_rate": 1.987588791002322e-05,
-      "loss": 1.3896,
+      "epoch": 0.006618878910219439,
+      "grad_norm": 0.0,
+      "learning_rate": 4.409881061299177e-06,
+      "loss": 1.1163,
       "step": 241
     },
     {
-      "epoch": 0.07908496732026143,
-      "grad_norm": 3.3514695167541504,
-      "learning_rate": 1.987421989840982e-05,
-      "loss": 1.4514,
+      "epoch": 0.006646343138062673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.428179322964319e-06,
+      "loss": 1.1724,
       "step": 242
     },
     {
-      "epoch": 0.07941176470588235,
-      "grad_norm": 2.849977731704712,
-      "learning_rate": 1.9872540823750718e-05,
-      "loss": 1.4093,
+      "epoch": 0.0066738073659059075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.44647758462946e-06,
+      "loss": 1.1242,
       "step": 243
     },
     {
-      "epoch": 0.07973856209150326,
-      "grad_norm": 3.2350876331329346,
-      "learning_rate": 1.987085068792715e-05,
-      "loss": 1.4376,
+      "epoch": 0.006701271593749142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.464775846294602e-06,
+      "loss": 1.1963,
       "step": 244
     },
     {
-      "epoch": 0.08006535947712418,
-      "grad_norm": 2.9238381385803223,
-      "learning_rate": 1.9869149492832737e-05,
-      "loss": 1.4137,
+      "epoch": 0.006728735821592376,
+      "grad_norm": 0.0,
+      "learning_rate": 4.483074107959744e-06,
+      "loss": 1.2042,
       "step": 245
     },
     {
-      "epoch": 0.0803921568627451,
-      "grad_norm": 2.8743152618408203,
-      "learning_rate": 1.986743724037349e-05,
-      "loss": 1.4252,
+      "epoch": 0.0067562000494356105,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5013723696248865e-06,
+      "loss": 1.1319,
       "step": 246
     },
     {
-      "epoch": 0.08071895424836602,
-      "grad_norm": 3.2635719776153564,
-      "learning_rate": 1.9865713932467815e-05,
-      "loss": 1.4254,
+      "epoch": 0.006783664277278844,
+      "grad_norm": 0.0,
+      "learning_rate": 4.519670631290028e-06,
+      "loss": 1.1329,
       "step": 247
     },
     {
-      "epoch": 0.08104575163398693,
-      "grad_norm": 3.0876240730285645,
-      "learning_rate": 1.98639795710465e-05,
-      "loss": 1.4414,
+      "epoch": 0.006811128505122078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.537968892955169e-06,
+      "loss": 1.145,
       "step": 248
     },
     {
-      "epoch": 0.08137254901960785,
-      "grad_norm": 3.2986855506896973,
-      "learning_rate": 1.986223415805272e-05,
-      "loss": 1.3958,
+      "epoch": 0.006838592732965313,
+      "grad_norm": 0.0,
+      "learning_rate": 4.556267154620311e-06,
+      "loss": 1.072,
       "step": 249
     },
     {
-      "epoch": 0.08169934640522876,
-      "grad_norm": 2.645822048187256,
-      "learning_rate": 1.986047769544203e-05,
-      "loss": 1.4206,
+      "epoch": 0.006866056960808547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.574565416285453e-06,
+      "loss": 1.1544,
       "step": 250
     },
     {
-      "epoch": 0.08202614379084967,
-      "grad_norm": 2.919750928878784,
-      "learning_rate": 1.985871018518236e-05,
-      "loss": 1.406,
+      "epoch": 0.006893521188651781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.592863677950595e-06,
+      "loss": 1.1995,
       "step": 251
     },
     {
-      "epoch": 0.08235294117647059,
-      "grad_norm": 2.6684274673461914,
-      "learning_rate": 1.9856931629254032e-05,
-      "loss": 1.4022,
+      "epoch": 0.006920985416495015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.611161939615737e-06,
+      "loss": 1.1395,
       "step": 252
     },
     {
-      "epoch": 0.0826797385620915,
-      "grad_norm": 3.007178783416748,
-      "learning_rate": 1.9855142029649725e-05,
-      "loss": 1.4471,
+      "epoch": 0.006948449644338249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.629460201280879e-06,
+      "loss": 1.0445,
       "step": 253
     },
     {
-      "epoch": 0.08300653594771241,
-      "grad_norm": 2.6720900535583496,
-      "learning_rate": 1.9853341388374504e-05,
-      "loss": 1.4057,
+      "epoch": 0.0069759138721814835,
+      "grad_norm": 0.0,
+      "learning_rate": 4.647758462946021e-06,
+      "loss": 1.1965,
       "step": 254
     },
     {
-      "epoch": 0.08333333333333333,
-      "grad_norm": 2.7071261405944824,
-      "learning_rate": 1.9851529707445808e-05,
-      "loss": 1.5067,
+      "epoch": 0.007003378100024718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6660567246111625e-06,
+      "loss": 1.103,
       "step": 255
     },
     {
-      "epoch": 0.08366013071895424,
-      "grad_norm": 2.8128342628479004,
-      "learning_rate": 1.9849706988893433e-05,
-      "loss": 1.4132,
+      "epoch": 0.007030842327867952,
+      "grad_norm": 0.0,
+      "learning_rate": 4.684354986276304e-06,
+      "loss": 1.1757,
       "step": 256
     },
     {
-      "epoch": 0.08398692810457517,
-      "grad_norm": 2.910353899002075,
-      "learning_rate": 1.9847873234759548e-05,
-      "loss": 1.428,
+      "epoch": 0.007058306555711187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.702653247941446e-06,
+      "loss": 1.2317,
       "step": 257
     },
     {
-      "epoch": 0.08431372549019608,
-      "grad_norm": 2.662872076034546,
-      "learning_rate": 1.9846028447098687e-05,
-      "loss": 1.4125,
+      "epoch": 0.00708577078355442,
+      "grad_norm": 0.0,
+      "learning_rate": 4.720951509606588e-06,
+      "loss": 1.1733,
       "step": 258
     },
     {
-      "epoch": 0.084640522875817,
-      "grad_norm": 2.836163282394409,
-      "learning_rate": 1.9844172627977744e-05,
-      "loss": 1.4392,
+      "epoch": 0.007113235011397654,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7392497712717295e-06,
+      "loss": 1.1459,
       "step": 259
     },
     {
-      "epoch": 0.08496732026143791,
-      "grad_norm": 3.216170310974121,
-      "learning_rate": 1.984230577947597e-05,
-      "loss": 1.3991,
+      "epoch": 0.007140699239240889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.757548032936871e-06,
+      "loss": 1.1281,
       "step": 260
     },
     {
-      "epoch": 0.08529411764705883,
-      "grad_norm": 2.433295726776123,
-      "learning_rate": 1.984042790368498e-05,
-      "loss": 1.3687,
+      "epoch": 0.007168163467084123,
+      "grad_norm": 0.0,
+      "learning_rate": 4.775846294602013e-06,
+      "loss": 1.0634,
       "step": 261
     },
     {
-      "epoch": 0.08562091503267974,
-      "grad_norm": 3.244344711303711,
-      "learning_rate": 1.983853900270874e-05,
-      "loss": 1.4227,
+      "epoch": 0.007195627694927357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.794144556267155e-06,
+      "loss": 1.0823,
       "step": 262
     },
     {
-      "epoch": 0.08594771241830065,
-      "grad_norm": 2.5063343048095703,
-      "learning_rate": 1.983663907866357e-05,
-      "loss": 1.3938,
+      "epoch": 0.007223091922770591,
+      "grad_norm": 0.0,
+      "learning_rate": 4.812442817932297e-06,
+      "loss": 1.0497,
       "step": 263
     },
     {
-      "epoch": 0.08627450980392157,
-      "grad_norm": 2.6966018676757812,
-      "learning_rate": 1.983472813367813e-05,
-      "loss": 1.4331,
+      "epoch": 0.007250556150613825,
+      "grad_norm": 0.0,
+      "learning_rate": 4.830741079597438e-06,
+      "loss": 1.1079,
       "step": 264
     },
     {
-      "epoch": 0.08660130718954248,
-      "grad_norm": 2.6378235816955566,
-      "learning_rate": 1.983280616989345e-05,
-      "loss": 1.4301,
+      "epoch": 0.00727802037845706,
+      "grad_norm": 0.0,
+      "learning_rate": 4.84903934126258e-06,
+      "loss": 1.1727,
       "step": 265
     },
     {
-      "epoch": 0.0869281045751634,
-      "grad_norm": 2.534935712814331,
-      "learning_rate": 1.9830873189462886e-05,
-      "loss": 1.3629,
+      "epoch": 0.007305484606300294,
+      "grad_norm": 0.0,
+      "learning_rate": 4.867337602927723e-06,
+      "loss": 1.0744,
       "step": 266
     },
     {
-      "epoch": 0.08725490196078431,
-      "grad_norm": 2.6496570110321045,
-      "learning_rate": 1.9828929194552143e-05,
-      "loss": 1.4021,
+      "epoch": 0.007332948834143528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8856358645928646e-06,
+      "loss": 1.0855,
       "step": 267
     },
     {
-      "epoch": 0.08758169934640522,
-      "grad_norm": 2.3720784187316895,
-      "learning_rate": 1.9826974187339267e-05,
-      "loss": 1.4151,
+      "epoch": 0.007360413061986763,
+      "grad_norm": 0.0,
+      "learning_rate": 4.903934126258006e-06,
+      "loss": 1.1646,
       "step": 268
     },
     {
-      "epoch": 0.08790849673202615,
-      "grad_norm": 3.1974165439605713,
-      "learning_rate": 1.9825008170014643e-05,
-      "loss": 1.4148,
+      "epoch": 0.007387877289829996,
+      "grad_norm": 0.0,
+      "learning_rate": 4.922232387923147e-06,
+      "loss": 1.1924,
       "step": 269
     },
     {
-      "epoch": 0.08823529411764706,
-      "grad_norm": 2.8328723907470703,
-      "learning_rate": 1.982303114478099e-05,
-      "loss": 1.4108,
+      "epoch": 0.0074153415176732304,
+      "grad_norm": 0.0,
+      "learning_rate": 4.940530649588289e-06,
+      "loss": 1.0686,
       "step": 270
     },
     {
-      "epoch": 0.08856209150326798,
-      "grad_norm": 3.3349967002868652,
-      "learning_rate": 1.982104311385336e-05,
-      "loss": 1.4262,
+      "epoch": 0.007442805745516465,
+      "grad_norm": 0.0,
+      "learning_rate": 4.958828911253432e-06,
+      "loss": 1.0571,
       "step": 271
     },
     {
-      "epoch": 0.08888888888888889,
-      "grad_norm": 2.82424259185791,
-      "learning_rate": 1.9819044079459142e-05,
-      "loss": 1.4125,
+      "epoch": 0.007470269973359699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977127172918573e-06,
+      "loss": 1.1096,
       "step": 272
     },
     {
-      "epoch": 0.0892156862745098,
-      "grad_norm": 3.481797456741333,
-      "learning_rate": 1.9817034043838044e-05,
-      "loss": 1.439,
+      "epoch": 0.0074977342012029335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.995425434583715e-06,
+      "loss": 1.1412,
       "step": 273
     },
     {
-      "epoch": 0.08954248366013072,
-      "grad_norm": 3.4965014457702637,
-      "learning_rate": 1.9815013009242103e-05,
-      "loss": 1.4343,
+      "epoch": 0.007525198429046167,
+      "grad_norm": 0.0,
+      "learning_rate": 5.013723696248857e-06,
+      "loss": 1.1759,
       "step": 274
     },
     {
-      "epoch": 0.08986928104575163,
-      "grad_norm": 3.3743388652801514,
-      "learning_rate": 1.981298097793568e-05,
-      "loss": 1.4592,
+      "epoch": 0.007552662656889401,
+      "grad_norm": 0.0,
+      "learning_rate": 5.032021957913999e-06,
+      "loss": 1.1808,
       "step": 275
     },
     {
-      "epoch": 0.09019607843137255,
-      "grad_norm": 3.3719773292541504,
-      "learning_rate": 1.981093795219546e-05,
-      "loss": 1.4129,
+      "epoch": 0.007580126884732636,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0503202195791405e-06,
+      "loss": 1.0981,
       "step": 276
     },
     {
-      "epoch": 0.09052287581699346,
-      "grad_norm": 2.9949405193328857,
-      "learning_rate": 1.9808883934310436e-05,
-      "loss": 1.3759,
+      "epoch": 0.00760759111257587,
+      "grad_norm": 0.0,
+      "learning_rate": 5.068618481244281e-06,
+      "loss": 1.1174,
       "step": 277
     },
     {
-      "epoch": 0.09084967320261438,
-      "grad_norm": 3.1619081497192383,
-      "learning_rate": 1.980681892658193e-05,
-      "loss": 1.451,
+      "epoch": 0.007635055340419104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.086916742909423e-06,
+      "loss": 1.1232,
       "step": 278
     },
     {
-      "epoch": 0.09117647058823529,
-      "grad_norm": 2.5450901985168457,
-      "learning_rate": 1.980474293132357e-05,
-      "loss": 1.4471,
+      "epoch": 0.007662519568262339,
+      "grad_norm": 0.0,
+      "learning_rate": 5.105215004574567e-06,
+      "loss": 1.1825,
       "step": 279
     },
     {
-      "epoch": 0.0915032679738562,
-      "grad_norm": 3.339644432067871,
-      "learning_rate": 1.980265595086129e-05,
-      "loss": 1.4764,
+      "epoch": 0.007689983796105572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.123513266239708e-06,
+      "loss": 1.1703,
       "step": 280
     },
     {
-      "epoch": 0.09183006535947713,
-      "grad_norm": 2.903198003768921,
-      "learning_rate": 1.980055798753334e-05,
-      "loss": 1.4321,
+      "epoch": 0.0077174480239488065,
+      "grad_norm": 0.0,
+      "learning_rate": 5.141811527904849e-06,
+      "loss": 1.1295,
       "step": 281
     },
     {
-      "epoch": 0.09215686274509804,
-      "grad_norm": 3.0291531085968018,
-      "learning_rate": 1.9798449043690272e-05,
-      "loss": 1.394,
+      "epoch": 0.007744912251792041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.160109789569991e-06,
+      "loss": 1.2668,
       "step": 282
     },
     {
-      "epoch": 0.09248366013071896,
-      "grad_norm": 2.741492986679077,
-      "learning_rate": 1.979632912169494e-05,
-      "loss": 1.4147,
+      "epoch": 0.007772376479635275,
+      "grad_norm": 0.0,
+      "learning_rate": 5.178408051235133e-06,
+      "loss": 1.1536,
       "step": 283
     },
     {
-      "epoch": 0.09281045751633987,
-      "grad_norm": 2.5195438861846924,
-      "learning_rate": 1.9794198223922496e-05,
-      "loss": 1.4158,
+      "epoch": 0.0077998407074785095,
+      "grad_norm": 0.0,
+      "learning_rate": 5.196706312900275e-06,
+      "loss": 1.169,
       "step": 284
     },
     {
-      "epoch": 0.09313725490196079,
-      "grad_norm": 2.6308178901672363,
-      "learning_rate": 1.9792056352760392e-05,
-      "loss": 1.4385,
+      "epoch": 0.007827304935321744,
+      "grad_norm": 0.0,
+      "learning_rate": 5.215004574565416e-06,
+      "loss": 1.2267,
       "step": 285
     },
     {
-      "epoch": 0.0934640522875817,
-      "grad_norm": 2.765836238861084,
-      "learning_rate": 1.9789903510608374e-05,
-      "loss": 1.4155,
+      "epoch": 0.007854769163164977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.233302836230558e-06,
+      "loss": 1.1351,
       "step": 286
     },
     {
-      "epoch": 0.09379084967320261,
-      "grad_norm": 2.6371757984161377,
-      "learning_rate": 1.9787739699878483e-05,
-      "loss": 1.4303,
+      "epoch": 0.007882233391008213,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2516010978957e-06,
+      "loss": 1.1339,
       "step": 287
     },
     {
-      "epoch": 0.09411764705882353,
-      "grad_norm": 3.0152199268341064,
-      "learning_rate": 1.9785564922995042e-05,
-      "loss": 1.4486,
+      "epoch": 0.007909697618851446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.269899359560842e-06,
+      "loss": 1.0414,
       "step": 288
     },
     {
-      "epoch": 0.09444444444444444,
-      "grad_norm": 2.787240743637085,
-      "learning_rate": 1.9783379182394668e-05,
-      "loss": 1.3534,
+      "epoch": 0.00793716184669468,
+      "grad_norm": 0.0,
+      "learning_rate": 5.288197621225984e-06,
+      "loss": 1.0904,
       "step": 289
     },
     {
-      "epoch": 0.09477124183006536,
-      "grad_norm": 2.762385845184326,
-      "learning_rate": 1.978118248052625e-05,
-      "loss": 1.3885,
+      "epoch": 0.007964626074537915,
+      "grad_norm": 0.0,
+      "learning_rate": 5.306495882891126e-06,
+      "loss": 1.1382,
       "step": 290
     },
     {
-      "epoch": 0.09509803921568627,
-      "grad_norm": 2.8326668739318848,
-      "learning_rate": 1.9778974819850973e-05,
-      "loss": 1.3907,
+      "epoch": 0.007992090302381148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.324794144556268e-06,
+      "loss": 1.2112,
       "step": 291
     },
     {
-      "epoch": 0.09542483660130718,
-      "grad_norm": 2.678194046020508,
-      "learning_rate": 1.9776756202842297e-05,
-      "loss": 1.3857,
+      "epoch": 0.008019554530224383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.34309240622141e-06,
+      "loss": 1.0925,
       "step": 292
     },
     {
-      "epoch": 0.09575163398692811,
-      "grad_norm": 3.155956506729126,
-      "learning_rate": 1.9774526631985942e-05,
-      "loss": 1.4027,
+      "epoch": 0.008047018758067617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3613906678865514e-06,
+      "loss": 1.1296,
       "step": 293
     },
     {
-      "epoch": 0.09607843137254903,
-      "grad_norm": 2.983729362487793,
-      "learning_rate": 1.977228610977992e-05,
-      "loss": 1.4222,
+      "epoch": 0.00807448298591085,
+      "grad_norm": 0.0,
+      "learning_rate": 5.379688929551693e-06,
+      "loss": 1.0937,
       "step": 294
     },
     {
-      "epoch": 0.09640522875816994,
-      "grad_norm": 2.5394527912139893,
-      "learning_rate": 1.9770034638734506e-05,
-      "loss": 1.4149,
+      "epoch": 0.008101947213754086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.397987191216835e-06,
+      "loss": 1.0829,
       "step": 295
     },
     {
-      "epoch": 0.09673202614379085,
-      "grad_norm": 2.633219003677368,
-      "learning_rate": 1.976777222137224e-05,
-      "loss": 1.3349,
+      "epoch": 0.008129411441597319,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416285452881977e-06,
+      "loss": 1.1942,
       "step": 296
     },
     {
-      "epoch": 0.09705882352941177,
-      "grad_norm": 2.6908907890319824,
-      "learning_rate": 1.976549886022793e-05,
-      "loss": 1.4082,
+      "epoch": 0.008156875669440554,
+      "grad_norm": 0.0,
+      "learning_rate": 5.434583714547118e-06,
+      "loss": 1.1516,
       "step": 297
     },
     {
-      "epoch": 0.09738562091503268,
-      "grad_norm": 3.091236114501953,
-      "learning_rate": 1.9763214557848634e-05,
-      "loss": 1.394,
+      "epoch": 0.008184339897283788,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4528819762122594e-06,
+      "loss": 1.1746,
       "step": 298
     },
     {
-      "epoch": 0.0977124183006536,
-      "grad_norm": 2.495039463043213,
-      "learning_rate": 1.9760919316793684e-05,
-      "loss": 1.3972,
+      "epoch": 0.008211804125127021,
+      "grad_norm": 0.0,
+      "learning_rate": 5.471180237877403e-06,
+      "loss": 1.0529,
       "step": 299
     },
     {
-      "epoch": 0.09803921568627451,
-      "grad_norm": 2.849370241165161,
-      "learning_rate": 1.9758613139634662e-05,
-      "loss": 1.4063,
+      "epoch": 0.008239268352970256,
+      "grad_norm": 0.0,
+      "learning_rate": 5.489478499542545e-06,
+      "loss": 1.1154,
       "step": 300
     },
     {
-      "epoch": 0.09836601307189542,
-      "grad_norm": 2.545807123184204,
-      "learning_rate": 1.9756296028955396e-05,
-      "loss": 1.3913,
+      "epoch": 0.00826673258081349,
+      "grad_norm": 0.0,
+      "learning_rate": 5.507776761207686e-06,
+      "loss": 1.0875,
       "step": 301
     },
     {
-      "epoch": 0.09869281045751634,
-      "grad_norm": 2.9218556880950928,
-      "learning_rate": 1.975396798735198e-05,
-      "loss": 1.4061,
+      "epoch": 0.008294196808656725,
+      "grad_norm": 0.0,
+      "learning_rate": 5.526075022872827e-06,
+      "loss": 1.1081,
       "step": 302
     },
     {
-      "epoch": 0.09901960784313725,
-      "grad_norm": 2.97839093208313,
-      "learning_rate": 1.9751629017432733e-05,
-      "loss": 1.4186,
+      "epoch": 0.008321661036499959,
+      "grad_norm": 0.0,
+      "learning_rate": 5.544373284537969e-06,
+      "loss": 1.1856,
       "step": 303
     },
     {
-      "epoch": 0.09934640522875816,
-      "grad_norm": 2.9594669342041016,
-      "learning_rate": 1.9749279121818235e-05,
-      "loss": 1.4223,
+      "epoch": 0.008349125264343194,
+      "grad_norm": 0.0,
+      "learning_rate": 5.562671546203111e-06,
+      "loss": 1.1754,
       "step": 304
     },
     {
-      "epoch": 0.09967320261437909,
-      "grad_norm": 2.3152008056640625,
-      "learning_rate": 1.9746918303141306e-05,
-      "loss": 1.3663,
+      "epoch": 0.008376589492186427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.580969807868253e-06,
+      "loss": 1.1003,
       "step": 305
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 2.813594341278076,
-      "learning_rate": 1.9744546564046995e-05,
-      "loss": 1.4215,
+      "epoch": 0.00840405372002966,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5992680695333944e-06,
+      "loss": 1.0606,
       "step": 306
     },
     {
-      "epoch": 0.10032679738562092,
-      "grad_norm": 3.1636452674865723,
-      "learning_rate": 1.974216390719259e-05,
-      "loss": 1.4452,
+      "epoch": 0.008431517947872896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.617566331198536e-06,
+      "loss": 1.0421,
       "step": 307
     },
     {
-      "epoch": 0.10065359477124183,
-      "grad_norm": 2.9237940311431885,
-      "learning_rate": 1.9739770335247616e-05,
-      "loss": 1.4106,
+      "epoch": 0.00845898217571613,
+      "grad_norm": 0.0,
+      "learning_rate": 5.635864592863678e-06,
+      "loss": 1.1219,
       "step": 308
     },
     {
-      "epoch": 0.10098039215686275,
-      "grad_norm": 2.9429140090942383,
-      "learning_rate": 1.973736585089382e-05,
-      "loss": 1.3684,
+      "epoch": 0.008486446403559365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.654162854528821e-06,
+      "loss": 1.1296,
       "step": 309
     },
     {
-      "epoch": 0.10130718954248366,
-      "grad_norm": 2.8573365211486816,
-      "learning_rate": 1.9734950456825187e-05,
-      "loss": 1.3906,
+      "epoch": 0.008513910631402598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.672461116193962e-06,
+      "loss": 1.0713,
       "step": 310
     },
     {
-      "epoch": 0.10163398692810457,
-      "grad_norm": 3.0197410583496094,
-      "learning_rate": 1.9732524155747906e-05,
-      "loss": 1.3955,
+      "epoch": 0.008541374859245832,
+      "grad_norm": 0.0,
+      "learning_rate": 5.690759377859104e-06,
+      "loss": 1.0714,
       "step": 311
     },
     {
-      "epoch": 0.10196078431372549,
-      "grad_norm": 2.6138863563537598,
-      "learning_rate": 1.9730086950380404e-05,
-      "loss": 1.3877,
+      "epoch": 0.008568839087089067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.709057639524246e-06,
+      "loss": 1.0625,
       "step": 312
     },
     {
-      "epoch": 0.1022875816993464,
-      "grad_norm": 2.9914348125457764,
-      "learning_rate": 1.9727638843453313e-05,
-      "loss": 1.4158,
+      "epoch": 0.0085963033149323,
+      "grad_norm": 0.0,
+      "learning_rate": 5.727355901189388e-06,
+      "loss": 1.1181,
       "step": 313
     },
     {
-      "epoch": 0.10261437908496732,
-      "grad_norm": 2.574997663497925,
-      "learning_rate": 1.9725179837709494e-05,
-      "loss": 1.4056,
+      "epoch": 0.008623767542775536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7456541628545295e-06,
+      "loss": 1.2025,
       "step": 314
     },
     {
-      "epoch": 0.10294117647058823,
-      "grad_norm": 2.9144539833068848,
-      "learning_rate": 1.9722709935904005e-05,
-      "loss": 1.4006,
+      "epoch": 0.008651231770618769,
+      "grad_norm": 0.0,
+      "learning_rate": 5.763952424519671e-06,
+      "loss": 1.1424,
       "step": 315
     },
     {
-      "epoch": 0.10326797385620914,
-      "grad_norm": 3.114650011062622,
-      "learning_rate": 1.972022914080411e-05,
-      "loss": 1.4526,
+      "epoch": 0.008678695998462002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.782250686184813e-06,
+      "loss": 1.1202,
       "step": 316
     },
     {
-      "epoch": 0.10359477124183007,
-      "grad_norm": 2.4231269359588623,
-      "learning_rate": 1.9717737455189293e-05,
-      "loss": 1.4329,
+      "epoch": 0.008706160226305238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.800548947849955e-06,
+      "loss": 1.1774,
       "step": 317
     },
     {
-      "epoch": 0.10392156862745099,
-      "grad_norm": 2.910346031188965,
-      "learning_rate": 1.9715234881851226e-05,
-      "loss": 1.4341,
+      "epoch": 0.008733624454148471,
+      "grad_norm": 0.0,
+      "learning_rate": 5.818847209515096e-06,
+      "loss": 1.1004,
       "step": 318
     },
     {
-      "epoch": 0.1042483660130719,
-      "grad_norm": 2.672971487045288,
-      "learning_rate": 1.9712721423593787e-05,
-      "loss": 1.433,
+      "epoch": 0.008761088681991706,
+      "grad_norm": 0.0,
+      "learning_rate": 5.837145471180239e-06,
+      "loss": 1.0948,
       "step": 319
     },
     {
-      "epoch": 0.10457516339869281,
-      "grad_norm": 2.7275373935699463,
-      "learning_rate": 1.9710197083233044e-05,
-      "loss": 1.4337,
+      "epoch": 0.00878855290983494,
+      "grad_norm": 0.0,
+      "learning_rate": 5.855443732845381e-06,
+      "loss": 1.1798,
       "step": 320
     },
     {
-      "epoch": 0.10490196078431373,
-      "grad_norm": 2.474569797515869,
-      "learning_rate": 1.9707661863597264e-05,
-      "loss": 1.3501,
+      "epoch": 0.008816017137678173,
+      "grad_norm": 0.0,
+      "learning_rate": 5.873741994510523e-06,
+      "loss": 1.1943,
       "step": 321
     },
     {
-      "epoch": 0.10522875816993464,
-      "grad_norm": 3.1499836444854736,
-      "learning_rate": 1.9705115767526894e-05,
-      "loss": 1.4598,
+      "epoch": 0.008843481365521409,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892040256175664e-06,
+      "loss": 1.1416,
       "step": 322
     },
     {
-      "epoch": 0.10555555555555556,
-      "grad_norm": 2.45200514793396,
-      "learning_rate": 1.970255879787458e-05,
-      "loss": 1.3753,
+      "epoch": 0.008870945593364642,
+      "grad_norm": 0.0,
+      "learning_rate": 5.910338517840805e-06,
+      "loss": 1.1185,
       "step": 323
     },
     {
-      "epoch": 0.10588235294117647,
-      "grad_norm": 2.7671656608581543,
-      "learning_rate": 1.9699990957505136e-05,
-      "loss": 1.4286,
+      "epoch": 0.008898409821207877,
+      "grad_norm": 0.0,
+      "learning_rate": 5.928636779505947e-06,
+      "loss": 1.1128,
       "step": 324
     },
     {
-      "epoch": 0.10620915032679738,
-      "grad_norm": 3.0369880199432373,
-      "learning_rate": 1.9697412249295567e-05,
-      "loss": 1.4489,
+      "epoch": 0.00892587404905111,
+      "grad_norm": 0.0,
+      "learning_rate": 5.946935041171089e-06,
+      "loss": 1.088,
       "step": 325
     },
     {
-      "epoch": 0.1065359477124183,
-      "grad_norm": 2.623159408569336,
-      "learning_rate": 1.9694822676135045e-05,
-      "loss": 1.3614,
+      "epoch": 0.008953338276894346,
+      "grad_norm": 0.0,
+      "learning_rate": 5.965233302836231e-06,
+      "loss": 1.2142,
       "step": 326
     },
     {
-      "epoch": 0.10686274509803921,
-      "grad_norm": 2.9030966758728027,
-      "learning_rate": 1.9692222240924926e-05,
-      "loss": 1.4096,
+      "epoch": 0.00898080250473758,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9835315645013725e-06,
+      "loss": 1.1305,
       "step": 327
     },
     {
-      "epoch": 0.10718954248366012,
-      "grad_norm": 2.7304348945617676,
-      "learning_rate": 1.9689610946578725e-05,
-      "loss": 1.4266,
+      "epoch": 0.009008266732580813,
+      "grad_norm": 0.0,
+      "learning_rate": 6.001829826166514e-06,
+      "loss": 1.0944,
       "step": 328
     },
     {
-      "epoch": 0.10751633986928105,
-      "grad_norm": 2.490604877471924,
-      "learning_rate": 1.968698879602213e-05,
-      "loss": 1.3865,
+      "epoch": 0.009035730960424048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.020128087831657e-06,
+      "loss": 1.1139,
       "step": 329
     },
     {
-      "epoch": 0.10784313725490197,
-      "grad_norm": 2.8761608600616455,
-      "learning_rate": 1.968435579219299e-05,
-      "loss": 1.3661,
+      "epoch": 0.009063195188267282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.038426349496799e-06,
+      "loss": 1.0875,
       "step": 330
     },
     {
-      "epoch": 0.10816993464052288,
-      "grad_norm": 2.6761343479156494,
-      "learning_rate": 1.968171193804131e-05,
-      "loss": 1.3467,
+      "epoch": 0.009090659416110517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.05672461116194e-06,
+      "loss": 1.1429,
       "step": 331
     },
     {
-      "epoch": 0.1084967320261438,
-      "grad_norm": 2.5821239948272705,
-      "learning_rate": 1.9679057236529266e-05,
-      "loss": 1.3138,
+      "epoch": 0.00911812364395375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.075022872827082e-06,
+      "loss": 1.0479,
       "step": 332
     },
     {
-      "epoch": 0.10882352941176471,
-      "grad_norm": 2.7348897457122803,
-      "learning_rate": 1.967639169063117e-05,
-      "loss": 1.3749,
+      "epoch": 0.009145587871796984,
+      "grad_norm": 0.0,
+      "learning_rate": 6.093321134492224e-06,
+      "loss": 1.02,
       "step": 333
     },
     {
-      "epoch": 0.10915032679738562,
-      "grad_norm": 2.7133071422576904,
-      "learning_rate": 1.9673715303333495e-05,
-      "loss": 1.4057,
+      "epoch": 0.009173052099640219,
+      "grad_norm": 0.0,
+      "learning_rate": 6.111619396157366e-06,
+      "loss": 1.1081,
       "step": 334
     },
     {
-      "epoch": 0.10947712418300654,
-      "grad_norm": 2.5769052505493164,
-      "learning_rate": 1.967102807763485e-05,
-      "loss": 1.412,
+      "epoch": 0.009200516327483452,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1299176578225075e-06,
+      "loss": 1.0781,
       "step": 335
     },
     {
-      "epoch": 0.10980392156862745,
-      "grad_norm": 2.809353828430176,
-      "learning_rate": 1.9668330016546004e-05,
-      "loss": 1.403,
+      "epoch": 0.009227980555326688,
+      "grad_norm": 0.0,
+      "learning_rate": 6.148215919487649e-06,
+      "loss": 1.0494,
       "step": 336
     },
     {
-      "epoch": 0.11013071895424836,
-      "grad_norm": 2.455536127090454,
-      "learning_rate": 1.966562112308985e-05,
-      "loss": 1.4188,
+      "epoch": 0.009255444783169921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.166514181152791e-06,
+      "loss": 1.159,
       "step": 337
     },
     {
-      "epoch": 0.11045751633986928,
-      "grad_norm": 2.340277671813965,
-      "learning_rate": 1.9662901400301432e-05,
-      "loss": 1.3494,
+      "epoch": 0.009282909011013155,
+      "grad_norm": 0.0,
+      "learning_rate": 6.184812442817932e-06,
+      "loss": 1.1321,
       "step": 338
     },
     {
-      "epoch": 0.11078431372549019,
-      "grad_norm": 2.729403257369995,
-      "learning_rate": 1.9660170851227908e-05,
-      "loss": 1.3865,
+      "epoch": 0.00931037323885639,
+      "grad_norm": 0.0,
+      "learning_rate": 6.203110704483074e-06,
+      "loss": 1.1869,
       "step": 339
     },
     {
-      "epoch": 0.1111111111111111,
-      "grad_norm": 2.639981985092163,
-      "learning_rate": 1.965742947892858e-05,
-      "loss": 1.3931,
+      "epoch": 0.009337837466699623,
+      "grad_norm": 0.0,
+      "learning_rate": 6.221408966148217e-06,
+      "loss": 1.1788,
       "step": 340
     },
     {
-      "epoch": 0.11143790849673203,
-      "grad_norm": 2.729423761367798,
-      "learning_rate": 1.965467728647488e-05,
-      "loss": 1.4022,
+      "epoch": 0.009365301694542858,
+      "grad_norm": 0.0,
+      "learning_rate": 6.239707227813359e-06,
+      "loss": 1.2104,
       "step": 341
     },
     {
-      "epoch": 0.11176470588235295,
-      "grad_norm": 2.665289878845215,
-      "learning_rate": 1.965191427695035e-05,
-      "loss": 1.3709,
+      "epoch": 0.009392765922386092,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2580054894785e-06,
+      "loss": 1.1486,
       "step": 342
     },
     {
-      "epoch": 0.11209150326797386,
-      "grad_norm": 2.597684860229492,
-      "learning_rate": 1.964914045345066e-05,
-      "loss": 1.3472,
+      "epoch": 0.009420230150229327,
+      "grad_norm": 0.0,
+      "learning_rate": 6.276303751143642e-06,
+      "loss": 1.1144,
       "step": 343
     },
     {
-      "epoch": 0.11241830065359477,
-      "grad_norm": 2.8944318294525146,
-      "learning_rate": 1.964635581908359e-05,
-      "loss": 1.3826,
+      "epoch": 0.00944769437807256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.294602012808783e-06,
+      "loss": 1.1456,
       "step": 344
     },
     {
-      "epoch": 0.11274509803921569,
-      "grad_norm": 2.562507152557373,
-      "learning_rate": 1.9643560376969035e-05,
-      "loss": 1.3743,
+      "epoch": 0.009475158605915794,
+      "grad_norm": 0.0,
+      "learning_rate": 6.312900274473925e-06,
+      "loss": 1.0535,
       "step": 345
     },
     {
-      "epoch": 0.1130718954248366,
-      "grad_norm": 2.751328468322754,
-      "learning_rate": 1.9640754130239004e-05,
-      "loss": 1.3744,
+      "epoch": 0.00950262283375903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.331198536139067e-06,
+      "loss": 1.0894,
       "step": 346
     },
     {
-      "epoch": 0.11339869281045752,
-      "grad_norm": 2.762190580368042,
-      "learning_rate": 1.963793708203761e-05,
-      "loss": 1.3466,
+      "epoch": 0.009530087061602263,
+      "grad_norm": 0.0,
+      "learning_rate": 6.349496797804209e-06,
+      "loss": 1.0774,
       "step": 347
     },
     {
-      "epoch": 0.11372549019607843,
-      "grad_norm": 2.848928928375244,
-      "learning_rate": 1.9635109235521057e-05,
-      "loss": 1.3556,
+      "epoch": 0.009557551289445498,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3677950594693505e-06,
+      "loss": 1.1392,
       "step": 348
     },
     {
-      "epoch": 0.11405228758169934,
-      "grad_norm": 2.5845608711242676,
-      "learning_rate": 1.9632270593857662e-05,
-      "loss": 1.3474,
+      "epoch": 0.009585015517288731,
+      "grad_norm": 0.0,
+      "learning_rate": 6.386093321134492e-06,
+      "loss": 1.0812,
       "step": 349
     },
     {
-      "epoch": 0.11437908496732026,
-      "grad_norm": 2.916745901107788,
-      "learning_rate": 1.9629421160227825e-05,
-      "loss": 1.3691,
+      "epoch": 0.009612479745131965,
+      "grad_norm": 0.0,
+      "learning_rate": 6.404391582799635e-06,
+      "loss": 1.0724,
       "step": 350
     },
     {
-      "epoch": 0.11470588235294117,
-      "grad_norm": 3.103421211242676,
-      "learning_rate": 1.9626560937824047e-05,
-      "loss": 1.397,
+      "epoch": 0.0096399439729752,
+      "grad_norm": 0.0,
+      "learning_rate": 6.422689844464777e-06,
+      "loss": 1.1321,
       "step": 351
     },
     {
-      "epoch": 0.11503267973856209,
-      "grad_norm": 2.9455201625823975,
-      "learning_rate": 1.962368992985091e-05,
-      "loss": 1.3401,
+      "epoch": 0.009667408200818434,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4409881061299184e-06,
+      "loss": 1.192,
       "step": 352
     },
     {
-      "epoch": 0.11535947712418301,
-      "grad_norm": 2.9365615844726562,
-      "learning_rate": 1.9620808139525082e-05,
-      "loss": 1.3739,
+      "epoch": 0.009694872428661669,
+      "grad_norm": 0.0,
+      "learning_rate": 6.45928636779506e-06,
+      "loss": 1.156,
       "step": 353
     },
     {
-      "epoch": 0.11568627450980393,
-      "grad_norm": 3.0864813327789307,
-      "learning_rate": 1.9617915570075317e-05,
-      "loss": 1.3354,
+      "epoch": 0.009722336656504902,
+      "grad_norm": 0.0,
+      "learning_rate": 6.477584629460202e-06,
+      "loss": 1.1307,
       "step": 354
     },
     {
-      "epoch": 0.11601307189542484,
-      "grad_norm": 2.456505537033081,
-      "learning_rate": 1.9615012224742436e-05,
-      "loss": 1.3148,
+      "epoch": 0.009749800884348136,
+      "grad_norm": 0.0,
+      "learning_rate": 6.495882891125344e-06,
+      "loss": 1.1056,
       "step": 355
     },
     {
-      "epoch": 0.11633986928104575,
-      "grad_norm": 3.1606554985046387,
-      "learning_rate": 1.961209810677934e-05,
-      "loss": 1.3102,
+      "epoch": 0.009777265112191371,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5141811527904855e-06,
+      "loss": 1.1079,
       "step": 356
     },
     {
-      "epoch": 0.11666666666666667,
-      "grad_norm": 2.7732672691345215,
-      "learning_rate": 1.9609173219450998e-05,
-      "loss": 1.3068,
+      "epoch": 0.009804729340034605,
+      "grad_norm": 0.0,
+      "learning_rate": 6.532479414455627e-06,
+      "loss": 1.1653,
       "step": 357
     },
     {
-      "epoch": 0.11699346405228758,
-      "grad_norm": 3.1660280227661133,
-      "learning_rate": 1.9606237566034443e-05,
-      "loss": 1.2702,
+      "epoch": 0.00983219356787784,
+      "grad_norm": 0.0,
+      "learning_rate": 6.550777676120769e-06,
+      "loss": 1.1686,
       "step": 358
     },
     {
-      "epoch": 0.1173202614379085,
-      "grad_norm": 2.936166763305664,
-      "learning_rate": 1.9603291149818774e-05,
-      "loss": 1.2275,
+      "epoch": 0.009859657795721073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.56907593778591e-06,
+      "loss": 1.3115,
       "step": 359
     },
     {
-      "epoch": 0.11764705882352941,
-      "grad_norm": 4.320514678955078,
-      "learning_rate": 1.9600333974105145e-05,
-      "loss": 1.2735,
+      "epoch": 0.009887122023564307,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5873741994510534e-06,
+      "loss": 1.0395,
       "step": 360
     },
     {
-      "epoch": 0.11797385620915032,
-      "grad_norm": 6.2742156982421875,
-      "learning_rate": 1.959736604220677e-05,
-      "loss": 1.2611,
+      "epoch": 0.009914586251407542,
+      "grad_norm": 0.0,
+      "learning_rate": 6.605672461116195e-06,
+      "loss": 1.261,
       "step": 361
     },
     {
-      "epoch": 0.11830065359477124,
-      "grad_norm": 3.4271345138549805,
-      "learning_rate": 1.9594387357448908e-05,
-      "loss": 1.2129,
+      "epoch": 0.009942050479250775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.623970722781337e-06,
+      "loss": 1.0651,
       "step": 362
     },
     {
-      "epoch": 0.11862745098039215,
-      "grad_norm": 3.924644947052002,
-      "learning_rate": 1.9591397923168868e-05,
-      "loss": 1.2549,
+      "epoch": 0.00996951470709401,
+      "grad_norm": 0.0,
+      "learning_rate": 6.642268984446478e-06,
+      "loss": 1.0784,
       "step": 363
     },
     {
-      "epoch": 0.11895424836601307,
-      "grad_norm": 3.5456173419952393,
-      "learning_rate": 1.9588397742716004e-05,
-      "loss": 1.23,
+      "epoch": 0.009996978934937244,
+      "grad_norm": 0.0,
+      "learning_rate": 6.66056724611162e-06,
+      "loss": 1.1669,
       "step": 364
     },
     {
-      "epoch": 0.119281045751634,
-      "grad_norm": 3.434445381164551,
-      "learning_rate": 1.9585386819451707e-05,
-      "loss": 1.2205,
+      "epoch": 0.01002444316278048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6788655077767614e-06,
+      "loss": 1.1372,
       "step": 365
     },
     {
-      "epoch": 0.11960784313725491,
-      "grad_norm": 3.265918016433716,
-      "learning_rate": 1.958236515674941e-05,
-      "loss": 1.2688,
+      "epoch": 0.010051907390623713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.697163769441903e-06,
+      "loss": 1.1578,
       "step": 366
     },
     {
-      "epoch": 0.11993464052287582,
-      "grad_norm": 3.146692991256714,
-      "learning_rate": 1.9579332757994573e-05,
-      "loss": 1.1935,
+      "epoch": 0.010079371618466946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.715462031107045e-06,
+      "loss": 1.1105,
       "step": 367
     },
     {
-      "epoch": 0.12026143790849673,
-      "grad_norm": 2.4543609619140625,
-      "learning_rate": 1.9576289626584685e-05,
-      "loss": 1.2055,
+      "epoch": 0.010106835846310181,
+      "grad_norm": 0.0,
+      "learning_rate": 6.733760292772187e-06,
+      "loss": 1.0366,
       "step": 368
     },
     {
-      "epoch": 0.12058823529411765,
-      "grad_norm": 3.48191237449646,
-      "learning_rate": 1.957323576592926e-05,
-      "loss": 1.1998,
+      "epoch": 0.010134300074153415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7520585544373285e-06,
+      "loss": 1.2154,
       "step": 369
     },
     {
-      "epoch": 0.12091503267973856,
-      "grad_norm": 2.518322706222534,
-      "learning_rate": 1.957017117944984e-05,
-      "loss": 1.1338,
+      "epoch": 0.01016176430199665,
+      "grad_norm": 0.0,
+      "learning_rate": 6.770356816102471e-06,
+      "loss": 1.1219,
       "step": 370
     },
     {
-      "epoch": 0.12124183006535948,
-      "grad_norm": 2.6334097385406494,
-      "learning_rate": 1.9567095870579973e-05,
-      "loss": 1.171,
+      "epoch": 0.010189228529839884,
+      "grad_norm": 0.0,
+      "learning_rate": 6.788655077767613e-06,
+      "loss": 1.1031,
       "step": 371
     },
     {
-      "epoch": 0.12156862745098039,
-      "grad_norm": 3.1855249404907227,
-      "learning_rate": 1.9564009842765225e-05,
-      "loss": 1.1817,
+      "epoch": 0.010216692757683117,
+      "grad_norm": 0.0,
+      "learning_rate": 6.806953339432755e-06,
+      "loss": 1.0647,
       "step": 372
     },
     {
-      "epoch": 0.1218954248366013,
-      "grad_norm": 2.3599770069122314,
-      "learning_rate": 1.9560913099463176e-05,
-      "loss": 1.1486,
+      "epoch": 0.010244156985526352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8252516010978965e-06,
+      "loss": 1.1259,
       "step": 373
     },
     {
-      "epoch": 0.12222222222222222,
-      "grad_norm": 2.510519504547119,
-      "learning_rate": 1.9557805644143408e-05,
-      "loss": 1.143,
+      "epoch": 0.010271621213369586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.843549862763038e-06,
+      "loss": 1.0842,
       "step": 374
     },
     {
-      "epoch": 0.12254901960784313,
-      "grad_norm": 2.753664970397949,
-      "learning_rate": 1.9554687480287497e-05,
-      "loss": 1.1575,
+      "epoch": 0.010299085441212821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.86184812442818e-06,
+      "loss": 1.1506,
       "step": 375
     },
     {
-      "epoch": 0.12287581699346405,
-      "grad_norm": 2.234624147415161,
-      "learning_rate": 1.955155861138903e-05,
-      "loss": 1.1389,
+      "epoch": 0.010326549669056054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.880146386093322e-06,
+      "loss": 1.0303,
       "step": 376
     },
     {
-      "epoch": 0.12320261437908497,
-      "grad_norm": 2.498760223388672,
-      "learning_rate": 1.9548419040953585e-05,
-      "loss": 1.1351,
+      "epoch": 0.010354013896899288,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8984446477584635e-06,
+      "loss": 1.1472,
       "step": 377
     },
     {
-      "epoch": 0.12352941176470589,
-      "grad_norm": 2.4905800819396973,
-      "learning_rate": 1.9545268772498723e-05,
-      "loss": 1.1365,
+      "epoch": 0.010381478124742523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.916742909423605e-06,
+      "loss": 1.0455,
       "step": 378
     },
     {
-      "epoch": 0.1238562091503268,
-      "grad_norm": 2.5865540504455566,
-      "learning_rate": 1.9542107809553995e-05,
-      "loss": 1.1684,
+      "epoch": 0.010408942352585757,
+      "grad_norm": 0.0,
+      "learning_rate": 6.935041171088746e-06,
+      "loss": 1.0761,
       "step": 379
     },
     {
-      "epoch": 0.12418300653594772,
-      "grad_norm": 2.3196511268615723,
-      "learning_rate": 1.9538936155660934e-05,
-      "loss": 1.1534,
+      "epoch": 0.010436406580428992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95333943275389e-06,
+      "loss": 1.16,
       "step": 380
     },
     {
-      "epoch": 0.12450980392156863,
-      "grad_norm": 2.281812906265259,
-      "learning_rate": 1.9535753814373053e-05,
-      "loss": 1.1159,
+      "epoch": 0.010463870808272225,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9716376944190315e-06,
+      "loss": 1.1306,
       "step": 381
     },
     {
-      "epoch": 0.12483660130718954,
-      "grad_norm": 2.8912973403930664,
-      "learning_rate": 1.9532560789255838e-05,
-      "loss": 1.1415,
+      "epoch": 0.010491335036115459,
+      "grad_norm": 0.0,
+      "learning_rate": 6.989935956084173e-06,
+      "loss": 1.1476,
       "step": 382
     },
     {
-      "epoch": 0.12516339869281046,
-      "grad_norm": 2.0186479091644287,
-      "learning_rate": 1.9529357083886744e-05,
-      "loss": 1.0956,
+      "epoch": 0.010518799263958694,
+      "grad_norm": 0.0,
+      "learning_rate": 7.008234217749314e-06,
+      "loss": 1.1554,
       "step": 383
     },
     {
-      "epoch": 0.12549019607843137,
-      "grad_norm": 2.4610164165496826,
-      "learning_rate": 1.9526142701855187e-05,
-      "loss": 1.1146,
+      "epoch": 0.010546263491801927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.026532479414456e-06,
+      "loss": 1.2032,
       "step": 384
     },
     {
-      "epoch": 0.12581699346405228,
-      "grad_norm": 2.4622392654418945,
-      "learning_rate": 1.9522917646762556e-05,
-      "loss": 1.1229,
+      "epoch": 0.010573727719645163,
+      "grad_norm": 0.0,
+      "learning_rate": 7.044830741079598e-06,
+      "loss": 1.2022,
       "step": 385
     },
     {
-      "epoch": 0.1261437908496732,
-      "grad_norm": 2.119061231613159,
-      "learning_rate": 1.9519681922222195e-05,
-      "loss": 1.1276,
+      "epoch": 0.010601191947488396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0631290027447395e-06,
+      "loss": 1.134,
       "step": 386
     },
     {
-      "epoch": 0.1264705882352941,
-      "grad_norm": 2.183105707168579,
-      "learning_rate": 1.9516435531859398e-05,
-      "loss": 1.1225,
+      "epoch": 0.010628656175331631,
+      "grad_norm": 0.0,
+      "learning_rate": 7.081427264409881e-06,
+      "loss": 1.1182,
       "step": 387
     },
     {
-      "epoch": 0.12679738562091503,
-      "grad_norm": 2.2047040462493896,
-      "learning_rate": 1.951317847931141e-05,
-      "loss": 1.1182,
+      "epoch": 0.010656120403174865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.099725526075023e-06,
+      "loss": 1.0266,
       "step": 388
     },
     {
-      "epoch": 0.12712418300653594,
-      "grad_norm": 2.4372596740722656,
-      "learning_rate": 1.9509910768227423e-05,
-      "loss": 1.1141,
+      "epoch": 0.010683584631018098,
+      "grad_norm": 0.0,
+      "learning_rate": 7.118023787740165e-06,
+      "loss": 1.1895,
       "step": 389
     },
     {
-      "epoch": 0.12745098039215685,
-      "grad_norm": 2.218967914581299,
-      "learning_rate": 1.950663240226857e-05,
-      "loss": 1.1311,
+      "epoch": 0.010711048858861334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.136322049405307e-06,
+      "loss": 1.2248,
       "step": 390
     },
     {
-      "epoch": 0.12777777777777777,
-      "grad_norm": 2.2395217418670654,
-      "learning_rate": 1.950334338510792e-05,
-      "loss": 1.111,
+      "epoch": 0.010738513086704567,
+      "grad_norm": 0.0,
+      "learning_rate": 7.154620311070449e-06,
+      "loss": 1.1049,
       "step": 391
     },
     {
-      "epoch": 0.12810457516339868,
-      "grad_norm": 2.2699081897735596,
-      "learning_rate": 1.9500043720430484e-05,
-      "loss": 1.1199,
+      "epoch": 0.010765977314547802,
+      "grad_norm": 0.0,
+      "learning_rate": 7.172918572735591e-06,
+      "loss": 1.0527,
       "step": 392
     },
     {
-      "epoch": 0.1284313725490196,
-      "grad_norm": 2.132289409637451,
-      "learning_rate": 1.949673341193319e-05,
-      "loss": 1.0742,
+      "epoch": 0.010793441542391036,
+      "grad_norm": 0.0,
+      "learning_rate": 7.191216834400733e-06,
+      "loss": 1.2294,
       "step": 393
     },
     {
-      "epoch": 0.1287581699346405,
-      "grad_norm": 2.321000814437866,
-      "learning_rate": 1.9493412463324903e-05,
-      "loss": 1.1081,
+      "epoch": 0.01082090577023427,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2095150960658745e-06,
+      "loss": 1.09,
       "step": 394
     },
     {
-      "epoch": 0.12908496732026145,
-      "grad_norm": 2.8936045169830322,
-      "learning_rate": 1.9490080878326397e-05,
-      "loss": 1.1448,
+      "epoch": 0.010848369998077504,
+      "grad_norm": 0.0,
+      "learning_rate": 7.227813357731016e-06,
+      "loss": 1.2212,
       "step": 395
     },
     {
-      "epoch": 0.12941176470588237,
-      "grad_norm": 2.059569835662842,
-      "learning_rate": 1.9486738660670373e-05,
-      "loss": 1.0703,
+      "epoch": 0.010875834225920738,
+      "grad_norm": 0.0,
+      "learning_rate": 7.246111619396158e-06,
+      "loss": 1.0251,
       "step": 396
     },
     {
-      "epoch": 0.12973856209150328,
-      "grad_norm": 2.3990120887756348,
-      "learning_rate": 1.9483385814101444e-05,
-      "loss": 1.0975,
+      "epoch": 0.010903298453763973,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2644098810613e-06,
+      "loss": 1.0366,
       "step": 397
     },
     {
-      "epoch": 0.1300653594771242,
-      "grad_norm": 2.119702100753784,
-      "learning_rate": 1.9480022342376125e-05,
-      "loss": 1.0893,
+      "epoch": 0.010930762681607207,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2827081427264416e-06,
+      "loss": 1.1177,
       "step": 398
     },
     {
-      "epoch": 0.1303921568627451,
-      "grad_norm": 2.1689648628234863,
-      "learning_rate": 1.947664824926284e-05,
-      "loss": 1.1434,
+      "epoch": 0.01095822690945044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.301006404391583e-06,
+      "loss": 1.1044,
       "step": 399
     },
     {
-      "epoch": 0.13071895424836602,
-      "grad_norm": 2.288529396057129,
-      "learning_rate": 1.9473263538541916e-05,
-      "loss": 1.141,
+      "epoch": 0.010985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319304666056726e-06,
+      "loss": 1.1979,
       "step": 400
     },
     {
-      "epoch": 0.13104575163398693,
-      "grad_norm": 2.141195297241211,
-      "learning_rate": 1.9469868214005568e-05,
-      "loss": 1.0827,
+      "epoch": 0.011013155365136909,
+      "grad_norm": 0.0,
+      "learning_rate": 7.337602927721868e-06,
+      "loss": 0.9887,
       "step": 401
     },
     {
-      "epoch": 0.13137254901960785,
-      "grad_norm": 2.1519768238067627,
-      "learning_rate": 1.946646227945791e-05,
-      "loss": 1.122,
+      "epoch": 0.011040619592980144,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3559011893870095e-06,
+      "loss": 1.1738,
       "step": 402
     },
     {
-      "epoch": 0.13169934640522876,
-      "grad_norm": 2.2068448066711426,
-      "learning_rate": 1.946304573871494e-05,
-      "loss": 1.1122,
+      "epoch": 0.011068083820823377,
+      "grad_norm": 0.0,
+      "learning_rate": 7.374199451052151e-06,
+      "loss": 1.2031,
       "step": 403
     },
     {
-      "epoch": 0.13202614379084968,
-      "grad_norm": 2.019348621368408,
-      "learning_rate": 1.945961859560454e-05,
-      "loss": 1.1572,
+      "epoch": 0.011095548048666611,
+      "grad_norm": 0.0,
+      "learning_rate": 7.392497712717292e-06,
+      "loss": 1.1228,
       "step": 404
     },
     {
-      "epoch": 0.1323529411764706,
-      "grad_norm": 2.281096935272217,
-      "learning_rate": 1.9456180853966475e-05,
-      "loss": 1.0671,
+      "epoch": 0.011123012276509846,
+      "grad_norm": 0.0,
+      "learning_rate": 7.410795974382434e-06,
+      "loss": 1.0339,
       "step": 405
     },
     {
-      "epoch": 0.1326797385620915,
-      "grad_norm": 2.279888868331909,
-      "learning_rate": 1.9452732517652366e-05,
-      "loss": 1.0982,
+      "epoch": 0.01115047650435308,
+      "grad_norm": 0.0,
+      "learning_rate": 7.429094236047576e-06,
+      "loss": 1.0858,
       "step": 406
     },
     {
-      "epoch": 0.13300653594771242,
-      "grad_norm": 2.0907647609710693,
-      "learning_rate": 1.9449273590525736e-05,
-      "loss": 1.0946,
+      "epoch": 0.011177940732196315,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4473924977127175e-06,
+      "loss": 1.0754,
       "step": 407
     },
     {
-      "epoch": 0.13333333333333333,
-      "grad_norm": 2.2389535903930664,
-      "learning_rate": 1.9445804076461946e-05,
-      "loss": 1.1199,
+      "epoch": 0.011205404960039548,
+      "grad_norm": 0.0,
+      "learning_rate": 7.465690759377859e-06,
+      "loss": 1.1747,
       "step": 408
     },
     {
-      "epoch": 0.13366013071895425,
-      "grad_norm": 2.27432918548584,
-      "learning_rate": 1.9442323979348226e-05,
-      "loss": 1.1136,
+      "epoch": 0.011232869187882783,
+      "grad_norm": 0.0,
+      "learning_rate": 7.483989021043001e-06,
+      "loss": 1.1652,
       "step": 409
     },
     {
-      "epoch": 0.13398692810457516,
-      "grad_norm": 2.219472646713257,
-      "learning_rate": 1.9438833303083677e-05,
-      "loss": 1.1074,
+      "epoch": 0.011260333415726017,
+      "grad_norm": 0.0,
+      "learning_rate": 7.502287282708144e-06,
+      "loss": 1.1842,
       "step": 410
     },
     {
-      "epoch": 0.13431372549019607,
-      "grad_norm": 2.263427495956421,
-      "learning_rate": 1.9435332051579235e-05,
-      "loss": 1.0952,
+      "epoch": 0.01128779764356925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.520585544373285e-06,
+      "loss": 1.0657,
       "step": 411
     },
     {
-      "epoch": 0.134640522875817,
-      "grad_norm": 2.3284103870391846,
-      "learning_rate": 1.943182022875769e-05,
-      "loss": 1.0789,
+      "epoch": 0.011315261871412486,
+      "grad_norm": 0.0,
+      "learning_rate": 7.538883806038427e-06,
+      "loss": 1.1514,
       "step": 412
     },
     {
-      "epoch": 0.1349673202614379,
-      "grad_norm": 2.051105499267578,
-      "learning_rate": 1.9428297838553683e-05,
-      "loss": 1.0676,
+      "epoch": 0.011342726099255719,
+      "grad_norm": 0.0,
+      "learning_rate": 7.557182067703569e-06,
+      "loss": 1.0731,
       "step": 413
     },
     {
-      "epoch": 0.13529411764705881,
-      "grad_norm": 2.2306723594665527,
-      "learning_rate": 1.942476488491369e-05,
-      "loss": 1.0226,
+      "epoch": 0.011370190327098954,
+      "grad_norm": 0.0,
+      "learning_rate": 7.575480329368711e-06,
+      "loss": 1.1253,
       "step": 414
     },
     {
-      "epoch": 0.13562091503267973,
-      "grad_norm": 2.298214912414551,
-      "learning_rate": 1.9421221371796016e-05,
-      "loss": 1.0824,
+      "epoch": 0.011397654554942188,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5937785910338525e-06,
+      "loss": 1.1044,
       "step": 415
     },
     {
-      "epoch": 0.13594771241830064,
-      "grad_norm": 2.422027587890625,
-      "learning_rate": 1.9417667303170803e-05,
-      "loss": 1.0686,
+      "epoch": 0.011425118782785421,
+      "grad_norm": 0.0,
+      "learning_rate": 7.612076852698994e-06,
+      "loss": 1.1577,
       "step": 416
     },
     {
-      "epoch": 0.13627450980392156,
-      "grad_norm": 2.1885769367218018,
-      "learning_rate": 1.9414102683020016e-05,
-      "loss": 1.1023,
+      "epoch": 0.011452583010628656,
+      "grad_norm": 0.0,
+      "learning_rate": 7.630375114364136e-06,
+      "loss": 1.0702,
       "step": 417
     },
     {
-      "epoch": 0.13660130718954247,
-      "grad_norm": 2.295722723007202,
-      "learning_rate": 1.941052751533745e-05,
-      "loss": 1.0719,
+      "epoch": 0.01148004723847189,
+      "grad_norm": 0.0,
+      "learning_rate": 7.648673376029277e-06,
+      "loss": 1.093,
       "step": 418
     },
     {
-      "epoch": 0.1369281045751634,
-      "grad_norm": 2.1816556453704834,
-      "learning_rate": 1.9406941804128714e-05,
-      "loss": 1.0885,
+      "epoch": 0.011507511466315125,
+      "grad_norm": 0.0,
+      "learning_rate": 7.66697163769442e-06,
+      "loss": 1.0846,
       "step": 419
     },
     {
-      "epoch": 0.13725490196078433,
-      "grad_norm": 2.272871971130371,
-      "learning_rate": 1.940334555341122e-05,
-      "loss": 1.0871,
+      "epoch": 0.011534975694158359,
+      "grad_norm": 0.0,
+      "learning_rate": 7.685269899359562e-06,
+      "loss": 1.0654,
       "step": 420
     },
     {
-      "epoch": 0.13758169934640524,
-      "grad_norm": 2.128847599029541,
-      "learning_rate": 1.93997387672142e-05,
-      "loss": 1.1302,
+      "epoch": 0.011562439922001592,
+      "grad_norm": 0.0,
+      "learning_rate": 7.703568161024703e-06,
+      "loss": 1.067,
       "step": 421
     },
     {
-      "epoch": 0.13790849673202615,
-      "grad_norm": 2.3815624713897705,
-      "learning_rate": 1.9396121449578694e-05,
-      "loss": 1.0938,
+      "epoch": 0.011589904149844827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.721866422689846e-06,
+      "loss": 1.2257,
       "step": 422
     },
     {
-      "epoch": 0.13823529411764707,
-      "grad_norm": 2.2410409450531006,
-      "learning_rate": 1.939249360455752e-05,
-      "loss": 1.084,
+      "epoch": 0.01161736837768806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.740164684354987e-06,
+      "loss": 1.0793,
       "step": 423
     },
     {
-      "epoch": 0.13856209150326798,
-      "grad_norm": 2.1650991439819336,
-      "learning_rate": 1.938885523621531e-05,
-      "loss": 1.0756,
+      "epoch": 0.011644832605531296,
+      "grad_norm": 0.0,
+      "learning_rate": 7.75846294602013e-06,
+      "loss": 1.2566,
       "step": 424
     },
     {
-      "epoch": 0.1388888888888889,
-      "grad_norm": 2.180767059326172,
-      "learning_rate": 1.9385206348628483e-05,
-      "loss": 1.0772,
+      "epoch": 0.01167229683337453,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77676120768527e-06,
+      "loss": 1.0791,
       "step": 425
     },
     {
-      "epoch": 0.1392156862745098,
-      "grad_norm": 1.9977585077285767,
-      "learning_rate": 1.9381546945885236e-05,
-      "loss": 1.0966,
+      "epoch": 0.011699761061217765,
+      "grad_norm": 0.0,
+      "learning_rate": 7.795059469350413e-06,
+      "loss": 1.0902,
       "step": 426
     },
     {
-      "epoch": 0.13954248366013072,
-      "grad_norm": 2.211203098297119,
-      "learning_rate": 1.937787703208556e-05,
-      "loss": 1.0755,
+      "epoch": 0.011727225289060998,
+      "grad_norm": 0.0,
+      "learning_rate": 7.813357731015554e-06,
+      "loss": 1.12,
       "step": 427
     },
     {
-      "epoch": 0.13986928104575164,
-      "grad_norm": 2.5023608207702637,
-      "learning_rate": 1.9374196611341212e-05,
-      "loss": 1.0781,
+      "epoch": 0.011754689516904232,
+      "grad_norm": 0.0,
+      "learning_rate": 7.831655992680696e-06,
+      "loss": 1.1301,
       "step": 428
     },
     {
-      "epoch": 0.14019607843137255,
-      "grad_norm": 1.983900785446167,
-      "learning_rate": 1.937050568777572e-05,
-      "loss": 1.1257,
+      "epoch": 0.011782153744747467,
+      "grad_norm": 0.0,
+      "learning_rate": 7.849954254345837e-06,
+      "loss": 1.0629,
       "step": 429
     },
     {
-      "epoch": 0.14052287581699346,
-      "grad_norm": 2.2858686447143555,
-      "learning_rate": 1.9366804265524388e-05,
-      "loss": 1.0755,
+      "epoch": 0.0118096179725907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.868252516010978e-06,
+      "loss": 1.1594,
       "step": 430
     },
     {
-      "epoch": 0.14084967320261438,
-      "grad_norm": 1.9975996017456055,
-      "learning_rate": 1.9363092348734282e-05,
-      "loss": 1.0694,
+      "epoch": 0.011837082200433936,
+      "grad_norm": 0.0,
+      "learning_rate": 7.886550777676123e-06,
+      "loss": 1.1732,
       "step": 431
     },
     {
-      "epoch": 0.1411764705882353,
-      "grad_norm": 2.230672836303711,
-      "learning_rate": 1.935936994156421e-05,
-      "loss": 1.0854,
+      "epoch": 0.011864546428277169,
+      "grad_norm": 0.0,
+      "learning_rate": 7.904849039341263e-06,
+      "loss": 1.0484,
       "step": 432
     },
     {
-      "epoch": 0.1415032679738562,
-      "grad_norm": 2.1306443214416504,
-      "learning_rate": 1.935563704818476e-05,
-      "loss": 1.0825,
+      "epoch": 0.011892010656120403,
+      "grad_norm": 0.0,
+      "learning_rate": 7.923147301006406e-06,
+      "loss": 1.0276,
       "step": 433
     },
     {
-      "epoch": 0.14183006535947712,
-      "grad_norm": 2.07985782623291,
-      "learning_rate": 1.935189367277824e-05,
-      "loss": 1.0678,
+      "epoch": 0.011919474883963638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.941445562671547e-06,
+      "loss": 1.0932,
       "step": 434
     },
     {
-      "epoch": 0.14215686274509803,
-      "grad_norm": 2.2018582820892334,
-      "learning_rate": 1.934813981953873e-05,
-      "loss": 1.1285,
+      "epoch": 0.011946939111806871,
+      "grad_norm": 0.0,
+      "learning_rate": 7.959743824336688e-06,
+      "loss": 1.2897,
       "step": 435
     },
     {
-      "epoch": 0.14248366013071895,
-      "grad_norm": 1.9055745601654053,
-      "learning_rate": 1.9344375492672024e-05,
-      "loss": 1.0987,
+      "epoch": 0.011974403339650106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.97804208600183e-06,
+      "loss": 1.1136,
       "step": 436
     },
     {
-      "epoch": 0.14281045751633986,
-      "grad_norm": 1.9491794109344482,
-      "learning_rate": 1.9340600696395672e-05,
-      "loss": 1.0613,
+      "epoch": 0.01200186756749334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.996340347666971e-06,
+      "loss": 1.0561,
       "step": 437
     },
     {
-      "epoch": 0.14313725490196078,
-      "grad_norm": 2.091907262802124,
-      "learning_rate": 1.9336815434938935e-05,
-      "loss": 1.0772,
+      "epoch": 0.012029331795336573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.014638609332114e-06,
+      "loss": 1.2742,
       "step": 438
     },
     {
-      "epoch": 0.1434640522875817,
-      "grad_norm": 1.9911099672317505,
-      "learning_rate": 1.933301971254282e-05,
-      "loss": 1.0703,
+      "epoch": 0.012056796023179809,
+      "grad_norm": 0.0,
+      "learning_rate": 8.032936870997255e-06,
+      "loss": 1.1256,
       "step": 439
     },
     {
-      "epoch": 0.1437908496732026,
-      "grad_norm": 2.1567203998565674,
-      "learning_rate": 1.932921353346003e-05,
-      "loss": 1.0983,
+      "epoch": 0.012084260251023042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.051235132662398e-06,
+      "loss": 1.1754,
       "step": 440
     },
     {
-      "epoch": 0.14411764705882352,
-      "grad_norm": 2.1625213623046875,
-      "learning_rate": 1.9325396901955003e-05,
-      "loss": 1.0657,
+      "epoch": 0.012111724478866277,
+      "grad_norm": 0.0,
+      "learning_rate": 8.06953339432754e-06,
+      "loss": 1.1093,
       "step": 441
     },
     {
-      "epoch": 0.14444444444444443,
-      "grad_norm": 2.0096545219421387,
-      "learning_rate": 1.932156982230388e-05,
-      "loss": 1.0697,
+      "epoch": 0.01213918870670951,
+      "grad_norm": 0.0,
+      "learning_rate": 8.087831655992681e-06,
+      "loss": 1.1612,
       "step": 442
     },
     {
-      "epoch": 0.14477124183006537,
-      "grad_norm": 2.2691714763641357,
-      "learning_rate": 1.9317732298794508e-05,
-      "loss": 1.0829,
+      "epoch": 0.012166652934552744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.106129917657824e-06,
+      "loss": 1.1297,
       "step": 443
     },
     {
-      "epoch": 0.1450980392156863,
-      "grad_norm": 2.0852127075195312,
-      "learning_rate": 1.9313884335726443e-05,
-      "loss": 1.0961,
+      "epoch": 0.01219411716239598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124428179322965e-06,
+      "loss": 1.2053,
       "step": 444
     },
     {
-      "epoch": 0.1454248366013072,
-      "grad_norm": 2.139044761657715,
-      "learning_rate": 1.9310025937410924e-05,
-      "loss": 1.1021,
+      "epoch": 0.012221581390239213,
+      "grad_norm": 0.0,
+      "learning_rate": 8.142726440988107e-06,
+      "loss": 1.166,
       "step": 445
     },
     {
-      "epoch": 0.14575163398692811,
-      "grad_norm": 2.1012144088745117,
-      "learning_rate": 1.9306157108170892e-05,
-      "loss": 1.0857,
+      "epoch": 0.012249045618082448,
+      "grad_norm": 0.0,
+      "learning_rate": 8.161024702653248e-06,
+      "loss": 1.135,
       "step": 446
     },
     {
-      "epoch": 0.14607843137254903,
-      "grad_norm": 1.8932533264160156,
-      "learning_rate": 1.930227785234097e-05,
-      "loss": 1.1173,
+      "epoch": 0.012276509845925682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.179322964318391e-06,
+      "loss": 1.0687,
       "step": 447
     },
     {
-      "epoch": 0.14640522875816994,
-      "grad_norm": 1.8762909173965454,
-      "learning_rate": 1.9298388174267463e-05,
-      "loss": 1.0992,
+      "epoch": 0.012303974073768917,
+      "grad_norm": 0.0,
+      "learning_rate": 8.197621225983532e-06,
+      "loss": 0.9941,
       "step": 448
     },
     {
-      "epoch": 0.14673202614379086,
-      "grad_norm": 2.0766425132751465,
-      "learning_rate": 1.9294488078308355e-05,
-      "loss": 1.0641,
+      "epoch": 0.01233143830161215,
+      "grad_norm": 0.0,
+      "learning_rate": 8.215919487648674e-06,
+      "loss": 1.0218,
       "step": 449
     },
     {
-      "epoch": 0.14705882352941177,
-      "grad_norm": 2.115201711654663,
-      "learning_rate": 1.9290577568833305e-05,
-      "loss": 1.0777,
+      "epoch": 0.012358902529455384,
+      "grad_norm": 0.0,
+      "learning_rate": 8.234217749313815e-06,
+      "loss": 1.0146,
       "step": 450
     },
     {
-      "epoch": 0.14738562091503268,
-      "grad_norm": 2.003546714782715,
-      "learning_rate": 1.928665665022363e-05,
-      "loss": 1.0831,
+      "epoch": 0.012386366757298619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.252516010978958e-06,
+      "loss": 1.0087,
       "step": 451
     },
     {
-      "epoch": 0.1477124183006536,
-      "grad_norm": 2.0829083919525146,
-      "learning_rate": 1.9282725326872324e-05,
-      "loss": 1.1198,
+      "epoch": 0.012413830985141852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.2708142726441e-06,
+      "loss": 1.1233,
       "step": 452
     },
     {
-      "epoch": 0.1480392156862745,
-      "grad_norm": 2.1793618202209473,
-      "learning_rate": 1.9278783603184023e-05,
-      "loss": 1.1068,
+      "epoch": 0.012441295212985088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.289112534309241e-06,
+      "loss": 1.0727,
       "step": 453
     },
     {
-      "epoch": 0.14836601307189543,
-      "grad_norm": 2.053628444671631,
-      "learning_rate": 1.927483148357502e-05,
-      "loss": 1.0767,
+      "epoch": 0.012468759440828321,
+      "grad_norm": 0.0,
+      "learning_rate": 8.307410795974384e-06,
+      "loss": 1.0584,
       "step": 454
     },
     {
-      "epoch": 0.14869281045751634,
-      "grad_norm": 1.9892603158950806,
-      "learning_rate": 1.9270868972473263e-05,
-      "loss": 1.0742,
+      "epoch": 0.012496223668671555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.325709057639525e-06,
+      "loss": 1.0406,
       "step": 455
     },
     {
-      "epoch": 0.14901960784313725,
-      "grad_norm": 1.9060319662094116,
-      "learning_rate": 1.9266896074318335e-05,
-      "loss": 1.0933,
+      "epoch": 0.01252368789651479,
+      "grad_norm": 0.0,
+      "learning_rate": 8.344007319304666e-06,
+      "loss": 1.1141,
       "step": 456
     },
     {
-      "epoch": 0.14934640522875817,
-      "grad_norm": 2.035867929458618,
-      "learning_rate": 1.926291279356146e-05,
-      "loss": 1.0724,
+      "epoch": 0.012551152124358023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.362305580969809e-06,
+      "loss": 1.1128,
       "step": 457
     },
     {
-      "epoch": 0.14967320261437908,
-      "grad_norm": 1.8956961631774902,
-      "learning_rate": 1.9258919134665494e-05,
-      "loss": 1.0659,
+      "epoch": 0.012578616352201259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.38060384263495e-06,
+      "loss": 1.0663,
       "step": 458
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 1.9418853521347046,
-      "learning_rate": 1.9254915102104918e-05,
-      "loss": 1.0979,
+      "epoch": 0.012606080580044492,
+      "grad_norm": 0.0,
+      "learning_rate": 8.398902104300092e-06,
+      "loss": 1.0608,
       "step": 459
     },
     {
-      "epoch": 0.1503267973856209,
-      "grad_norm": 1.8926935195922852,
-      "learning_rate": 1.9250900700365837e-05,
-      "loss": 1.095,
+      "epoch": 0.012633544807887725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.417200365965233e-06,
+      "loss": 1.0624,
       "step": 460
     },
     {
-      "epoch": 0.15065359477124182,
-      "grad_norm": 1.9522119760513306,
-      "learning_rate": 1.9246875933945977e-05,
-      "loss": 1.0481,
+      "epoch": 0.01266100903573096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.435498627630376e-06,
+      "loss": 1.1125,
       "step": 461
     },
     {
-      "epoch": 0.15098039215686274,
-      "grad_norm": 1.9014664888381958,
-      "learning_rate": 1.9242840807354676e-05,
-      "loss": 1.0734,
+      "epoch": 0.012688473263574194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.453796889295518e-06,
+      "loss": 1.125,
       "step": 462
     },
     {
-      "epoch": 0.15130718954248365,
-      "grad_norm": 1.9642069339752197,
-      "learning_rate": 1.9238795325112867e-05,
-      "loss": 1.0653,
+      "epoch": 0.01271593749141743,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47209515096066e-06,
+      "loss": 1.1604,
       "step": 463
     },
     {
-      "epoch": 0.15163398692810456,
-      "grad_norm": 2.090040445327759,
-      "learning_rate": 1.923473949175311e-05,
-      "loss": 1.0818,
+      "epoch": 0.012743401719260663,
+      "grad_norm": 0.0,
+      "learning_rate": 8.490393412625802e-06,
+      "loss": 1.2139,
       "step": 464
     },
     {
-      "epoch": 0.15196078431372548,
-      "grad_norm": 1.8883121013641357,
-      "learning_rate": 1.923067331181954e-05,
-      "loss": 1.0741,
+      "epoch": 0.012770865947103896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.508691674290943e-06,
+      "loss": 1.184,
       "step": 465
     },
     {
-      "epoch": 0.1522875816993464,
-      "grad_norm": 1.9398940801620483,
-      "learning_rate": 1.9226596789867895e-05,
-      "loss": 1.0731,
+      "epoch": 0.012798330174947132,
+      "grad_norm": 0.0,
+      "learning_rate": 8.526989935956085e-06,
+      "loss": 1.1196,
       "step": 466
     },
     {
-      "epoch": 0.15261437908496733,
-      "grad_norm": 1.9694336652755737,
-      "learning_rate": 1.9222509930465495e-05,
-      "loss": 1.0656,
+      "epoch": 0.012825794402790365,
+      "grad_norm": 0.0,
+      "learning_rate": 8.545288197621226e-06,
+      "loss": 1.1046,
       "step": 467
     },
     {
-      "epoch": 0.15294117647058825,
-      "grad_norm": 1.9781056642532349,
-      "learning_rate": 1.921841273819125e-05,
-      "loss": 1.0813,
+      "epoch": 0.0128532586306336,
+      "grad_norm": 0.0,
+      "learning_rate": 8.563586459286369e-06,
+      "loss": 1.0825,
       "step": 468
     },
     {
-      "epoch": 0.15326797385620916,
-      "grad_norm": 2.079289674758911,
-      "learning_rate": 1.9214305217635636e-05,
-      "loss": 1.068,
+      "epoch": 0.012880722858476834,
+      "grad_norm": 0.0,
+      "learning_rate": 8.58188472095151e-06,
+      "loss": 1.1666,
       "step": 469
     },
     {
-      "epoch": 0.15359477124183007,
-      "grad_norm": 1.8668270111083984,
-      "learning_rate": 1.921018737340071e-05,
-      "loss": 1.1037,
+      "epoch": 0.012908187086320069,
+      "grad_norm": 0.0,
+      "learning_rate": 8.600182982616652e-06,
+      "loss": 1.0309,
       "step": 470
     },
     {
-      "epoch": 0.153921568627451,
-      "grad_norm": 1.868107557296753,
-      "learning_rate": 1.920605921010009e-05,
+      "epoch": 0.012935651314163302,
+      "grad_norm": 0.0,
+      "learning_rate": 8.618481244281795e-06,
       "loss": 1.107,
       "step": 471
     },
     {
-      "epoch": 0.1542483660130719,
-      "grad_norm": 1.8840105533599854,
-      "learning_rate": 1.920192073235896e-05,
-      "loss": 1.0786,
+      "epoch": 0.012963115542006536,
+      "grad_norm": 0.0,
+      "learning_rate": 8.636779505946936e-06,
+      "loss": 1.1352,
       "step": 472
     },
     {
-      "epoch": 0.15457516339869282,
-      "grad_norm": 2.1253013610839844,
-      "learning_rate": 1.9197771944814053e-05,
-      "loss": 1.0454,
+      "epoch": 0.012990579769849771,
+      "grad_norm": 0.0,
+      "learning_rate": 8.655077767612079e-06,
+      "loss": 1.1299,
       "step": 473
     },
     {
-      "epoch": 0.15490196078431373,
-      "grad_norm": 2.1251184940338135,
-      "learning_rate": 1.9193612852113666e-05,
-      "loss": 1.0658,
+      "epoch": 0.013018043997693005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67337602927722e-06,
+      "loss": 1.1337,
       "step": 474
     },
     {
-      "epoch": 0.15522875816993464,
-      "grad_norm": 2.0038583278656006,
-      "learning_rate": 1.9189443458917623e-05,
-      "loss": 1.0824,
+      "epoch": 0.01304550822553624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.69167429094236e-06,
+      "loss": 1.1982,
       "step": 475
     },
     {
-      "epoch": 0.15555555555555556,
-      "grad_norm": 2.0118515491485596,
-      "learning_rate": 1.918526376989731e-05,
-      "loss": 1.0496,
+      "epoch": 0.013072972453379473,
+      "grad_norm": 0.0,
+      "learning_rate": 8.709972552607503e-06,
+      "loss": 1.112,
       "step": 476
     },
     {
-      "epoch": 0.15588235294117647,
-      "grad_norm": 2.073882579803467,
-      "learning_rate": 1.918107378973563e-05,
-      "loss": 1.0883,
+      "epoch": 0.013100436681222707,
+      "grad_norm": 0.0,
+      "learning_rate": 8.728270814272644e-06,
+      "loss": 1.0905,
       "step": 477
     },
     {
-      "epoch": 0.15620915032679739,
-      "grad_norm": 2.261235237121582,
-      "learning_rate": 1.9176873523127025e-05,
-      "loss": 1.0549,
+      "epoch": 0.013127900909065942,
+      "grad_norm": 0.0,
+      "learning_rate": 8.746569075937787e-06,
+      "loss": 1.1716,
       "step": 478
     },
     {
-      "epoch": 0.1565359477124183,
-      "grad_norm": 2.082690954208374,
-      "learning_rate": 1.917266297477747e-05,
-      "loss": 1.0907,
+      "epoch": 0.013155365136909175,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764867337602928e-06,
+      "loss": 1.1954,
       "step": 479
     },
     {
-      "epoch": 0.1568627450980392,
-      "grad_norm": 1.9154715538024902,
-      "learning_rate": 1.9168442149404438e-05,
-      "loss": 1.0986,
+      "epoch": 0.01318282936475241,
+      "grad_norm": 0.0,
+      "learning_rate": 8.78316559926807e-06,
+      "loss": 1.114,
       "step": 480
     },
     {
-      "epoch": 0.15718954248366013,
-      "grad_norm": 2.145210027694702,
-      "learning_rate": 1.916421105173693e-05,
-      "loss": 1.0604,
+      "epoch": 0.013210293592595644,
+      "grad_norm": 0.0,
+      "learning_rate": 8.801463860933213e-06,
+      "loss": 1.1398,
       "step": 481
     },
     {
-      "epoch": 0.15751633986928104,
-      "grad_norm": 2.035308361053467,
-      "learning_rate": 1.9159969686515462e-05,
-      "loss": 1.0632,
+      "epoch": 0.013237757820438878,
+      "grad_norm": 0.0,
+      "learning_rate": 8.819762122598354e-06,
+      "loss": 1.1361,
       "step": 482
     },
     {
-      "epoch": 0.15784313725490196,
-      "grad_norm": 1.9597710371017456,
-      "learning_rate": 1.9155718058492043e-05,
-      "loss": 1.0594,
+      "epoch": 0.013265222048282113,
+      "grad_norm": 0.0,
+      "learning_rate": 8.838060384263496e-06,
+      "loss": 1.0065,
       "step": 483
     },
     {
-      "epoch": 0.15816993464052287,
-      "grad_norm": 2.1178691387176514,
-      "learning_rate": 1.9151456172430186e-05,
-      "loss": 1.0774,
+      "epoch": 0.013292686276125346,
+      "grad_norm": 0.0,
+      "learning_rate": 8.856358645928637e-06,
+      "loss": 1.2231,
       "step": 484
     },
     {
-      "epoch": 0.15849673202614378,
-      "grad_norm": 1.9065663814544678,
-      "learning_rate": 1.9147184033104892e-05,
-      "loss": 1.075,
+      "epoch": 0.013320150503968581,
+      "grad_norm": 0.0,
+      "learning_rate": 8.87465690759378e-06,
+      "loss": 1.1344,
       "step": 485
     },
     {
-      "epoch": 0.1588235294117647,
-      "grad_norm": 1.9835939407348633,
-      "learning_rate": 1.9142901645302654e-05,
-      "loss": 1.1073,
+      "epoch": 0.013347614731811815,
+      "grad_norm": 0.0,
+      "learning_rate": 8.89295516925892e-06,
+      "loss": 0.9965,
       "step": 486
     },
     {
-      "epoch": 0.1591503267973856,
-      "grad_norm": 2.2415246963500977,
-      "learning_rate": 1.9138609013821448e-05,
-      "loss": 1.0555,
+      "epoch": 0.013375078959655048,
+      "grad_norm": 0.0,
+      "learning_rate": 8.911253430924063e-06,
+      "loss": 1.1442,
       "step": 487
     },
     {
-      "epoch": 0.15947712418300652,
-      "grad_norm": 1.9360538721084595,
-      "learning_rate": 1.9134306143470722e-05,
-      "loss": 1.0676,
+      "epoch": 0.013402543187498284,
+      "grad_norm": 0.0,
+      "learning_rate": 8.929551692589204e-06,
+      "loss": 1.1366,
       "step": 488
     },
     {
-      "epoch": 0.15980392156862744,
-      "grad_norm": 2.2619616985321045,
-      "learning_rate": 1.9129993039071403e-05,
-      "loss": 1.0659,
+      "epoch": 0.013430007415341517,
+      "grad_norm": 0.0,
+      "learning_rate": 8.947849954254347e-06,
+      "loss": 1.2277,
       "step": 489
     },
     {
-      "epoch": 0.16013071895424835,
-      "grad_norm": 1.909492015838623,
-      "learning_rate": 1.912566970545587e-05,
-      "loss": 1.1042,
+      "epoch": 0.013457471643184752,
+      "grad_norm": 0.0,
+      "learning_rate": 8.966148215919488e-06,
+      "loss": 1.0955,
       "step": 490
     },
     {
-      "epoch": 0.1604575163398693,
-      "grad_norm": 1.9304612874984741,
-      "learning_rate": 1.9121336147467987e-05,
-      "loss": 1.0618,
+      "epoch": 0.013484935871027986,
+      "grad_norm": 0.0,
+      "learning_rate": 8.98444647758463e-06,
+      "loss": 1.0126,
       "step": 491
     },
     {
-      "epoch": 0.1607843137254902,
-      "grad_norm": 1.9612163305282593,
-      "learning_rate": 1.911699236996305e-05,
-      "loss": 1.064,
+      "epoch": 0.013512400098871221,
+      "grad_norm": 0.0,
+      "learning_rate": 9.002744739249773e-06,
+      "loss": 1.1127,
       "step": 492
     },
     {
-      "epoch": 0.16111111111111112,
-      "grad_norm": 1.874570608139038,
-      "learning_rate": 1.9112638377807816e-05,
-      "loss": 1.0695,
+      "epoch": 0.013539864326714455,
+      "grad_norm": 0.0,
+      "learning_rate": 9.021043000914914e-06,
+      "loss": 1.1368,
       "step": 493
     },
     {
-      "epoch": 0.16143790849673204,
-      "grad_norm": 2.1484639644622803,
-      "learning_rate": 1.9108274175880483e-05,
-      "loss": 1.0792,
+      "epoch": 0.013567328554557688,
+      "grad_norm": 0.0,
+      "learning_rate": 9.039341262580057e-06,
+      "loss": 1.0303,
       "step": 494
     },
     {
-      "epoch": 0.16176470588235295,
-      "grad_norm": 2.0952961444854736,
-      "learning_rate": 1.9103899769070692e-05,
-      "loss": 1.0561,
+      "epoch": 0.013594792782400923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.057639524245198e-06,
+      "loss": 1.0304,
       "step": 495
     },
     {
-      "epoch": 0.16209150326797386,
-      "grad_norm": 2.0613839626312256,
-      "learning_rate": 1.9099515162279515e-05,
-      "loss": 1.0546,
+      "epoch": 0.013622257010244157,
+      "grad_norm": 0.0,
+      "learning_rate": 9.075937785910338e-06,
+      "loss": 1.1729,
       "step": 496
     },
     {
-      "epoch": 0.16241830065359478,
-      "grad_norm": 2.144468307495117,
-      "learning_rate": 1.909512036041945e-05,
-      "loss": 1.0498,
+      "epoch": 0.013649721238087392,
+      "grad_norm": 0.0,
+      "learning_rate": 9.094236047575481e-06,
+      "loss": 1.1046,
       "step": 497
     },
     {
-      "epoch": 0.1627450980392157,
-      "grad_norm": 2.1842846870422363,
-      "learning_rate": 1.909071536841442e-05,
-      "loss": 1.0461,
+      "epoch": 0.013677185465930625,
+      "grad_norm": 0.0,
+      "learning_rate": 9.112534309240622e-06,
+      "loss": 1.0938,
       "step": 498
     },
     {
-      "epoch": 0.1630718954248366,
-      "grad_norm": 1.9348655939102173,
-      "learning_rate": 1.9086300191199765e-05,
-      "loss": 1.0907,
+      "epoch": 0.013704649693773859,
+      "grad_norm": 0.0,
+      "learning_rate": 9.130832570905765e-06,
+      "loss": 1.1035,
       "step": 499
     },
     {
-      "epoch": 0.16339869281045752,
-      "grad_norm": 2.1271426677703857,
-      "learning_rate": 1.9081874833722234e-05,
-      "loss": 1.0703,
+      "epoch": 0.013732113921617094,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149130832570906e-06,
+      "loss": 1.1687,
       "step": 500
     },
     {
-      "epoch": 0.16372549019607843,
-      "grad_norm": 2.0723514556884766,
-      "learning_rate": 1.907743930093999e-05,
-      "loss": 1.0764,
+      "epoch": 0.013759578149460328,
+      "grad_norm": 0.0,
+      "learning_rate": 9.167429094236048e-06,
+      "loss": 0.9672,
       "step": 501
     },
     {
-      "epoch": 0.16405228758169935,
-      "grad_norm": 1.8847447633743286,
-      "learning_rate": 1.9072993597822586e-05,
-      "loss": 1.0742,
+      "epoch": 0.013787042377303563,
+      "grad_norm": 0.0,
+      "learning_rate": 9.18572735590119e-06,
+      "loss": 1.0415,
       "step": 502
     },
     {
-      "epoch": 0.16437908496732026,
-      "grad_norm": 1.9019173383712769,
-      "learning_rate": 1.906853772935097e-05,
-      "loss": 1.0535,
+      "epoch": 0.013814506605146796,
+      "grad_norm": 0.0,
+      "learning_rate": 9.204025617566332e-06,
+      "loss": 1.0899,
       "step": 503
     },
     {
-      "epoch": 0.16470588235294117,
-      "grad_norm": 2.102416515350342,
-      "learning_rate": 1.906407170051749e-05,
-      "loss": 1.1021,
+      "epoch": 0.01384197083299003,
+      "grad_norm": 0.0,
+      "learning_rate": 9.222323879231474e-06,
+      "loss": 1.1692,
       "step": 504
     },
     {
-      "epoch": 0.1650326797385621,
-      "grad_norm": 1.9088590145111084,
-      "learning_rate": 1.905959551632587e-05,
-      "loss": 1.0421,
+      "epoch": 0.013869435060833265,
+      "grad_norm": 0.0,
+      "learning_rate": 9.240622140896615e-06,
+      "loss": 1.0685,
       "step": 505
     },
     {
-      "epoch": 0.165359477124183,
-      "grad_norm": 1.9134483337402344,
-      "learning_rate": 1.905510918179121e-05,
-      "loss": 1.0462,
+      "epoch": 0.013896899288676498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.258920402561758e-06,
+      "loss": 1.1523,
       "step": 506
     },
     {
-      "epoch": 0.16568627450980392,
-      "grad_norm": 1.997748613357544,
-      "learning_rate": 1.9050612701939988e-05,
-      "loss": 1.096,
+      "epoch": 0.013924363516519734,
+      "grad_norm": 0.0,
+      "learning_rate": 9.277218664226899e-06,
+      "loss": 1.2092,
       "step": 507
     },
     {
-      "epoch": 0.16601307189542483,
-      "grad_norm": 1.8503550291061401,
-      "learning_rate": 1.9046106081810047e-05,
-      "loss": 1.0732,
+      "epoch": 0.013951827744362967,
+      "grad_norm": 0.0,
+      "learning_rate": 9.295516925892041e-06,
+      "loss": 1.1728,
       "step": 508
     },
     {
-      "epoch": 0.16633986928104574,
-      "grad_norm": 1.774874210357666,
-      "learning_rate": 1.9041589326450585e-05,
-      "loss": 1.0832,
+      "epoch": 0.013979291972206202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.313815187557182e-06,
+      "loss": 1.1227,
       "step": 509
     },
     {
-      "epoch": 0.16666666666666666,
-      "grad_norm": 1.909637451171875,
-      "learning_rate": 1.903706244092217e-05,
-      "loss": 1.047,
+      "epoch": 0.014006756200049436,
+      "grad_norm": 0.0,
+      "learning_rate": 9.332113449222325e-06,
+      "loss": 1.132,
       "step": 510
     },
     {
-      "epoch": 0.16699346405228757,
-      "grad_norm": 2.1369361877441406,
-      "learning_rate": 1.90325254302967e-05,
-      "loss": 1.0812,
+      "epoch": 0.01403422042789267,
+      "grad_norm": 0.0,
+      "learning_rate": 9.350411710887466e-06,
+      "loss": 1.1444,
       "step": 511
     },
     {
-      "epoch": 0.16732026143790849,
-      "grad_norm": 2.1417055130004883,
-      "learning_rate": 1.9027978299657436e-05,
-      "loss": 1.047,
+      "epoch": 0.014061684655735904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.368709972552608e-06,
+      "loss": 1.0914,
       "step": 512
     },
     {
-      "epoch": 0.1676470588235294,
-      "grad_norm": 1.9455406665802002,
-      "learning_rate": 1.9023421054098967e-05,
-      "loss": 1.0512,
+      "epoch": 0.014089148883579138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.387008234217751e-06,
+      "loss": 1.1141,
       "step": 513
     },
     {
-      "epoch": 0.16797385620915034,
-      "grad_norm": 1.8990036249160767,
-      "learning_rate": 1.9018853698727216e-05,
-      "loss": 1.0696,
+      "epoch": 0.014116613111422373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.405306495882892e-06,
+      "loss": 1.179,
       "step": 514
     },
     {
-      "epoch": 0.16830065359477125,
-      "grad_norm": 2.2404239177703857,
-      "learning_rate": 1.9014276238659434e-05,
-      "loss": 1.0725,
+      "epoch": 0.014144077339265607,
+      "grad_norm": 0.0,
+      "learning_rate": 9.423604757548035e-06,
+      "loss": 1.1824,
       "step": 515
     },
     {
-      "epoch": 0.16862745098039217,
-      "grad_norm": 2.035701274871826,
-      "learning_rate": 1.900968867902419e-05,
-      "loss": 1.0863,
+      "epoch": 0.01417154156710884,
+      "grad_norm": 0.0,
+      "learning_rate": 9.441903019213176e-06,
+      "loss": 1.072,
       "step": 516
     },
     {
-      "epoch": 0.16895424836601308,
-      "grad_norm": 1.9236302375793457,
-      "learning_rate": 1.900509102496138e-05,
-      "loss": 1.0343,
+      "epoch": 0.014199005794952075,
+      "grad_norm": 0.0,
+      "learning_rate": 9.460201280878316e-06,
+      "loss": 1.0786,
       "step": 517
     },
     {
-      "epoch": 0.169281045751634,
-      "grad_norm": 1.9199105501174927,
-      "learning_rate": 1.9000483281622198e-05,
-      "loss": 1.0461,
+      "epoch": 0.014226470022795309,
+      "grad_norm": 0.0,
+      "learning_rate": 9.478499542543459e-06,
+      "loss": 1.1263,
       "step": 518
     },
     {
-      "epoch": 0.1696078431372549,
-      "grad_norm": 1.8855175971984863,
-      "learning_rate": 1.8995865454169142e-05,
-      "loss": 1.0793,
+      "epoch": 0.014253934250638544,
+      "grad_norm": 0.0,
+      "learning_rate": 9.4967978042086e-06,
+      "loss": 1.1624,
       "step": 519
     },
     {
-      "epoch": 0.16993464052287582,
-      "grad_norm": 1.9111987352371216,
-      "learning_rate": 1.8991237547776014e-05,
-      "loss": 1.0987,
+      "epoch": 0.014281398478481777,
+      "grad_norm": 0.0,
+      "learning_rate": 9.515096065873743e-06,
+      "loss": 1.0445,
       "step": 520
     },
     {
-      "epoch": 0.17026143790849674,
-      "grad_norm": 2.052130937576294,
-      "learning_rate": 1.8986599567627904e-05,
-      "loss": 1.0462,
+      "epoch": 0.014308862706325011,
+      "grad_norm": 0.0,
+      "learning_rate": 9.533394327538884e-06,
+      "loss": 1.0809,
       "step": 521
     },
     {
-      "epoch": 0.17058823529411765,
-      "grad_norm": 2.1426286697387695,
-      "learning_rate": 1.8981951518921194e-05,
-      "loss": 1.0261,
+      "epoch": 0.014336326934168246,
+      "grad_norm": 0.0,
+      "learning_rate": 9.551692589204026e-06,
+      "loss": 1.0471,
       "step": 522
     },
     {
-      "epoch": 0.17091503267973857,
-      "grad_norm": 2.1951632499694824,
-      "learning_rate": 1.8977293406863545e-05,
-      "loss": 1.0567,
+      "epoch": 0.01436379116201148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.569990850869169e-06,
+      "loss": 0.957,
       "step": 523
     },
     {
-      "epoch": 0.17124183006535948,
-      "grad_norm": 2.289877414703369,
-      "learning_rate": 1.8972625236673887e-05,
-      "loss": 1.0514,
+      "epoch": 0.014391255389854715,
+      "grad_norm": 0.0,
+      "learning_rate": 9.58828911253431e-06,
+      "loss": 1.1519,
       "step": 524
     },
     {
-      "epoch": 0.1715686274509804,
-      "grad_norm": 2.232743501663208,
-      "learning_rate": 1.8967947013582428e-05,
-      "loss": 1.0577,
+      "epoch": 0.014418719617697948,
+      "grad_norm": 0.0,
+      "learning_rate": 9.606587374199452e-06,
+      "loss": 1.1639,
       "step": 525
     },
     {
-      "epoch": 0.1718954248366013,
-      "grad_norm": 2.154771327972412,
-      "learning_rate": 1.896325874283063e-05,
-      "loss": 1.1032,
+      "epoch": 0.014446183845541182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.624885635864593e-06,
+      "loss": 1.1105,
       "step": 526
     },
     {
-      "epoch": 0.17222222222222222,
-      "grad_norm": 2.2819595336914062,
-      "learning_rate": 1.8958560429671217e-05,
-      "loss": 1.0291,
+      "epoch": 0.014473648073384417,
+      "grad_norm": 0.0,
+      "learning_rate": 9.643183897529736e-06,
+      "loss": 1.1614,
       "step": 527
     },
     {
-      "epoch": 0.17254901960784313,
-      "grad_norm": 2.013746976852417,
-      "learning_rate": 1.895385207936817e-05,
-      "loss": 1.0764,
+      "epoch": 0.01450111230122765,
+      "grad_norm": 0.0,
+      "learning_rate": 9.661482159194877e-06,
+      "loss": 1.0829,
       "step": 528
     },
     {
-      "epoch": 0.17287581699346405,
-      "grad_norm": 1.769188642501831,
-      "learning_rate": 1.894913369719671e-05,
-      "loss": 1.0936,
+      "epoch": 0.014528576529070886,
+      "grad_norm": 0.0,
+      "learning_rate": 9.67978042086002e-06,
+      "loss": 1.0879,
       "step": 529
     },
     {
-      "epoch": 0.17320261437908496,
-      "grad_norm": 2.014901876449585,
-      "learning_rate": 1.894440528844329e-05,
-      "loss": 1.0475,
+      "epoch": 0.01455604075691412,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69807868252516e-06,
+      "loss": 1.1412,
       "step": 530
     },
     {
-      "epoch": 0.17352941176470588,
-      "grad_norm": 1.973423957824707,
-      "learning_rate": 1.8939666858405613e-05,
-      "loss": 1.0938,
+      "epoch": 0.014583504984757354,
+      "grad_norm": 0.0,
+      "learning_rate": 9.716376944190303e-06,
+      "loss": 1.0612,
       "step": 531
     },
     {
-      "epoch": 0.1738562091503268,
-      "grad_norm": 2.170497417449951,
-      "learning_rate": 1.8934918412392596e-05,
-      "loss": 1.0509,
+      "epoch": 0.014610969212600588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.734675205855446e-06,
+      "loss": 1.0485,
       "step": 532
     },
     {
-      "epoch": 0.1741830065359477,
-      "grad_norm": 2.072681188583374,
-      "learning_rate": 1.8930159955724386e-05,
-      "loss": 1.0485,
+      "epoch": 0.014638433440443821,
+      "grad_norm": 0.0,
+      "learning_rate": 9.752973467520586e-06,
+      "loss": 1.2066,
       "step": 533
     },
     {
-      "epoch": 0.17450980392156862,
-      "grad_norm": 1.8526345491409302,
-      "learning_rate": 1.8925391493732346e-05,
-      "loss": 1.0845,
+      "epoch": 0.014665897668287057,
+      "grad_norm": 0.0,
+      "learning_rate": 9.771271729185729e-06,
+      "loss": 1.2123,
       "step": 534
     },
     {
-      "epoch": 0.17483660130718953,
-      "grad_norm": 1.9210079908370972,
-      "learning_rate": 1.892061303175904e-05,
-      "loss": 1.0506,
+      "epoch": 0.01469336189613029,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78956999085087e-06,
+      "loss": 1.1908,
       "step": 535
     },
     {
-      "epoch": 0.17516339869281045,
-      "grad_norm": 1.8590030670166016,
-      "learning_rate": 1.891582457515825e-05,
-      "loss": 1.0694,
+      "epoch": 0.014720826123973525,
+      "grad_norm": 0.0,
+      "learning_rate": 9.807868252516013e-06,
+      "loss": 1.151,
       "step": 536
     },
     {
-      "epoch": 0.17549019607843136,
-      "grad_norm": 1.9291212558746338,
-      "learning_rate": 1.891102612929494e-05,
-      "loss": 1.0964,
+      "epoch": 0.014748290351816759,
+      "grad_norm": 0.0,
+      "learning_rate": 9.826166514181154e-06,
+      "loss": 1.0673,
       "step": 537
     },
     {
-      "epoch": 0.1758169934640523,
-      "grad_norm": 1.8585822582244873,
-      "learning_rate": 1.890621769954528e-05,
-      "loss": 1.082,
+      "epoch": 0.014775754579659992,
+      "grad_norm": 0.0,
+      "learning_rate": 9.844464775846294e-06,
+      "loss": 1.1224,
       "step": 538
     },
     {
-      "epoch": 0.17614379084967322,
-      "grad_norm": 1.8422821760177612,
-      "learning_rate": 1.890139929129661e-05,
-      "loss": 1.0729,
+      "epoch": 0.014803218807503227,
+      "grad_norm": 0.0,
+      "learning_rate": 9.862763037511437e-06,
+      "loss": 1.0863,
       "step": 539
     },
     {
-      "epoch": 0.17647058823529413,
-      "grad_norm": 2.0134506225585938,
-      "learning_rate": 1.8896570909947477e-05,
-      "loss": 1.0582,
+      "epoch": 0.014830683035346461,
+      "grad_norm": 0.0,
+      "learning_rate": 9.881061299176578e-06,
+      "loss": 1.1022,
       "step": 540
     },
     {
-      "epoch": 0.17679738562091504,
-      "grad_norm": 1.8755661249160767,
-      "learning_rate": 1.8891732560907567e-05,
-      "loss": 1.0695,
+      "epoch": 0.014858147263189696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.89935956084172e-06,
+      "loss": 1.0009,
       "step": 541
     },
     {
-      "epoch": 0.17712418300653596,
-      "grad_norm": 1.9807038307189941,
-      "learning_rate": 1.8886884249597763e-05,
-      "loss": 1.0542,
+      "epoch": 0.01488561149103293,
+      "grad_norm": 0.0,
+      "learning_rate": 9.917657822506863e-06,
+      "loss": 1.1115,
       "step": 542
     },
     {
-      "epoch": 0.17745098039215687,
-      "grad_norm": 1.954964518547058,
-      "learning_rate": 1.8882025981450095e-05,
-      "loss": 1.0391,
+      "epoch": 0.014913075718876163,
+      "grad_norm": 0.0,
+      "learning_rate": 9.935956084172004e-06,
+      "loss": 1.1578,
       "step": 543
     },
     {
-      "epoch": 0.17777777777777778,
-      "grad_norm": 1.976402997970581,
-      "learning_rate": 1.887715776190775e-05,
-      "loss": 1.1235,
+      "epoch": 0.014940539946719398,
+      "grad_norm": 0.0,
+      "learning_rate": 9.954254345837147e-06,
+      "loss": 1.1396,
       "step": 544
     },
     {
-      "epoch": 0.1781045751633987,
-      "grad_norm": 2.189208745956421,
-      "learning_rate": 1.8872279596425068e-05,
-      "loss": 1.0994,
+      "epoch": 0.014968004174562632,
+      "grad_norm": 0.0,
+      "learning_rate": 9.972552607502288e-06,
+      "loss": 1.1135,
       "step": 545
     },
     {
-      "epoch": 0.1784313725490196,
-      "grad_norm": 1.8535302877426147,
-      "learning_rate": 1.8867391490467534e-05,
-      "loss": 1.0242,
+      "epoch": 0.014995468402405867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99085086916743e-06,
+      "loss": 1.0934,
       "step": 546
     },
     {
-      "epoch": 0.17875816993464053,
-      "grad_norm": 1.978800892829895,
-      "learning_rate": 1.8862493449511757e-05,
-      "loss": 1.0434,
+      "epoch": 0.0150229326302491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0009149130832571e-05,
+      "loss": 1.0767,
       "step": 547
     },
     {
-      "epoch": 0.17908496732026144,
-      "grad_norm": 2.0189826488494873,
-      "learning_rate": 1.8857585479045493e-05,
-      "loss": 1.0212,
+      "epoch": 0.015050396858092334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0027447392497714e-05,
+      "loss": 1.1591,
       "step": 548
     },
     {
-      "epoch": 0.17941176470588235,
-      "grad_norm": 1.8798962831497192,
-      "learning_rate": 1.885266758456762e-05,
-      "loss": 1.0485,
+      "epoch": 0.015077861085935569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0045745654162855e-05,
+      "loss": 1.1989,
       "step": 549
     },
     {
-      "epoch": 0.17973856209150327,
-      "grad_norm": 1.9316891431808472,
-      "learning_rate": 1.884773977158813e-05,
-      "loss": 1.0518,
+      "epoch": 0.015105325313778803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064043915827997e-05,
+      "loss": 1.1375,
       "step": 550
     },
     {
-      "epoch": 0.18006535947712418,
-      "grad_norm": 2.1212732791900635,
-      "learning_rate": 1.8842802045628127e-05,
-      "loss": 1.0425,
+      "epoch": 0.015132789541622038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0082342177493138e-05,
+      "loss": 1.0478,
       "step": 551
     },
     {
-      "epoch": 0.1803921568627451,
-      "grad_norm": 2.141771078109741,
-      "learning_rate": 1.8837854412219828e-05,
-      "loss": 1.0608,
+      "epoch": 0.015160253769465271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0100640439158281e-05,
+      "loss": 1.2038,
       "step": 552
     },
     {
-      "epoch": 0.180718954248366,
-      "grad_norm": 1.8706334829330444,
-      "learning_rate": 1.883289687690654e-05,
-      "loss": 1.0638,
+      "epoch": 0.015187717997308506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0118938700823422e-05,
+      "loss": 1.0006,
       "step": 553
     },
     {
-      "epoch": 0.18104575163398692,
-      "grad_norm": 1.8672336339950562,
-      "learning_rate": 1.8827929445242678e-05,
-      "loss": 1.0711,
+      "epoch": 0.01521518222515174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0137236962488563e-05,
+      "loss": 1.1087,
       "step": 554
     },
     {
-      "epoch": 0.18137254901960784,
-      "grad_norm": 2.017338275909424,
-      "learning_rate": 1.882295212279373e-05,
-      "loss": 1.0596,
+      "epoch": 0.015242646452994973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0155535224153705e-05,
+      "loss": 1.1065,
       "step": 555
     },
     {
-      "epoch": 0.18169934640522875,
-      "grad_norm": 1.762457013130188,
-      "learning_rate": 1.8817964915136277e-05,
-      "loss": 1.0879,
+      "epoch": 0.015270110680838209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0173833485818846e-05,
+      "loss": 1.1435,
       "step": 556
     },
     {
-      "epoch": 0.18202614379084966,
-      "grad_norm": 2.0531821250915527,
-      "learning_rate": 1.8812967827857972e-05,
-      "loss": 1.0859,
+      "epoch": 0.015297574908681442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019213174748399e-05,
+      "loss": 1.2084,
       "step": 557
     },
     {
-      "epoch": 0.18235294117647058,
-      "grad_norm": 2.0638985633850098,
-      "learning_rate": 1.8807960866557535e-05,
-      "loss": 1.0502,
+      "epoch": 0.015325039136524677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0210430009149133e-05,
+      "loss": 1.205,
       "step": 558
     },
     {
-      "epoch": 0.1826797385620915,
-      "grad_norm": 1.79250967502594,
-      "learning_rate": 1.8802944036844747e-05,
-      "loss": 1.0452,
+      "epoch": 0.01535250336436791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0228728270814274e-05,
+      "loss": 1.1865,
       "step": 559
     },
     {
-      "epoch": 0.1830065359477124,
-      "grad_norm": 1.9440603256225586,
-      "learning_rate": 1.8797917344340453e-05,
-      "loss": 1.1083,
+      "epoch": 0.015379967592211144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247026532479417e-05,
+      "loss": 1.1472,
       "step": 560
     },
     {
-      "epoch": 0.18333333333333332,
-      "grad_norm": 1.928031325340271,
-      "learning_rate": 1.8792880794676546e-05,
-      "loss": 1.0584,
+      "epoch": 0.01540743182005438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0265324794144558e-05,
+      "loss": 1.0334,
       "step": 561
     },
     {
-      "epoch": 0.18366013071895426,
-      "grad_norm": 2.0655455589294434,
-      "learning_rate": 1.8787834393495952e-05,
-      "loss": 1.0926,
+      "epoch": 0.015434896047897613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0283623055809699e-05,
+      "loss": 1.2318,
       "step": 562
     },
     {
-      "epoch": 0.18398692810457518,
-      "grad_norm": 1.97580087184906,
-      "learning_rate": 1.878277814645265e-05,
+      "epoch": 0.015462360275740848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0301921317474841e-05,
       "loss": 1.0533,
       "step": 563
     },
     {
-      "epoch": 0.1843137254901961,
-      "grad_norm": 2.0451083183288574,
-      "learning_rate": 1.8777712059211643e-05,
-      "loss": 1.0942,
+      "epoch": 0.015489824503584082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0320219579139982e-05,
+      "loss": 1.0425,
       "step": 564
     },
     {
-      "epoch": 0.184640522875817,
-      "grad_norm": 1.8486709594726562,
-      "learning_rate": 1.877263613744896e-05,
-      "loss": 1.0623,
+      "epoch": 0.015517288731427315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0338517840805125e-05,
+      "loss": 1.1456,
       "step": 565
     },
     {
-      "epoch": 0.18496732026143792,
-      "grad_norm": 1.769727110862732,
-      "learning_rate": 1.8767550386851646e-05,
-      "loss": 1.0657,
+      "epoch": 0.01554475295927055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0356816102470266e-05,
+      "loss": 1.1198,
       "step": 566
     },
     {
-      "epoch": 0.18529411764705883,
-      "grad_norm": 1.9201351404190063,
-      "learning_rate": 1.8762454813117758e-05,
-      "loss": 1.038,
+      "epoch": 0.015572217187113784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0375114364135408e-05,
+      "loss": 1.064,
       "step": 567
     },
     {
-      "epoch": 0.18562091503267975,
-      "grad_norm": 1.8433293104171753,
-      "learning_rate": 1.875734942195637e-05,
-      "loss": 1.074,
+      "epoch": 0.015599681414957019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.039341262580055e-05,
+      "loss": 1.0765,
       "step": 568
     },
     {
-      "epoch": 0.18594771241830066,
-      "grad_norm": 1.7927290201187134,
-      "learning_rate": 1.8752234219087538e-05,
-      "loss": 1.0604,
+      "epoch": 0.015627145642800253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0411710887465692e-05,
+      "loss": 1.1238,
       "step": 569
     },
     {
-      "epoch": 0.18627450980392157,
-      "grad_norm": 1.863419771194458,
-      "learning_rate": 1.8747109210242326e-05,
-      "loss": 1.0489,
+      "epoch": 0.015654609870643488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0430009149130833e-05,
+      "loss": 1.1685,
       "step": 570
     },
     {
-      "epoch": 0.1866013071895425,
-      "grad_norm": 1.8806010484695435,
-      "learning_rate": 1.874197440116277e-05,
-      "loss": 1.0712,
+      "epoch": 0.01568207409848672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0448307410795975e-05,
+      "loss": 1.1176,
       "step": 571
     },
     {
-      "epoch": 0.1869281045751634,
-      "grad_norm": 1.8666801452636719,
-      "learning_rate": 1.8736829797601903e-05,
-      "loss": 1.0731,
+      "epoch": 0.015709538326329955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0466605672461116e-05,
+      "loss": 1.1141,
       "step": 572
     },
     {
-      "epoch": 0.18725490196078431,
-      "grad_norm": 1.9314799308776855,
-      "learning_rate": 1.873167540532372e-05,
-      "loss": 1.0595,
+      "epoch": 0.01573700255417319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0484903934126259e-05,
+      "loss": 1.0494,
       "step": 573
     },
     {
-      "epoch": 0.18758169934640523,
-      "grad_norm": 1.7790790796279907,
-      "learning_rate": 1.8726511230103183e-05,
-      "loss": 1.0647,
+      "epoch": 0.015764466782016425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.05032021957914e-05,
+      "loss": 1.1573,
       "step": 574
     },
     {
-      "epoch": 0.18790849673202614,
-      "grad_norm": 1.9712653160095215,
-      "learning_rate": 1.872133727772622e-05,
-      "loss": 1.0729,
+      "epoch": 0.015791931009859657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0521500457456541e-05,
+      "loss": 1.0947,
       "step": 575
     },
     {
-      "epoch": 0.18823529411764706,
-      "grad_norm": 2.059629201889038,
-      "learning_rate": 1.8716153553989716e-05,
-      "loss": 1.0583,
+      "epoch": 0.015819395237702892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0539798719121683e-05,
+      "loss": 1.1157,
       "step": 576
     },
     {
-      "epoch": 0.18856209150326797,
-      "grad_norm": 1.9768999814987183,
-      "learning_rate": 1.8710960064701494e-05,
-      "loss": 1.0726,
+      "epoch": 0.015846859465546127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0558096980786828e-05,
+      "loss": 1.1666,
       "step": 577
     },
     {
-      "epoch": 0.18888888888888888,
-      "grad_norm": 1.9182180166244507,
-      "learning_rate": 1.8705756815680324e-05,
-      "loss": 1.07,
+      "epoch": 0.01587432369338936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0576395242451969e-05,
+      "loss": 1.1404,
       "step": 578
     },
     {
-      "epoch": 0.1892156862745098,
-      "grad_norm": 1.6762070655822754,
-      "learning_rate": 1.8700543812755912e-05,
-      "loss": 1.0569,
+      "epoch": 0.015901787921232594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0594693504117111e-05,
+      "loss": 1.075,
       "step": 579
     },
     {
-      "epoch": 0.1895424836601307,
-      "grad_norm": 1.8501369953155518,
-      "learning_rate": 1.8695321061768886e-05,
-      "loss": 1.0495,
+      "epoch": 0.01592925214907583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0612991765782252e-05,
+      "loss": 1.0893,
       "step": 580
     },
     {
-      "epoch": 0.18986928104575163,
-      "grad_norm": 1.9762897491455078,
-      "learning_rate": 1.869008856857081e-05,
-      "loss": 1.0274,
+      "epoch": 0.01595671637691906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0631290027447395e-05,
+      "loss": 1.225,
       "step": 581
     },
     {
-      "epoch": 0.19019607843137254,
-      "grad_norm": 1.827672004699707,
-      "learning_rate": 1.8684846339024145e-05,
-      "loss": 1.0302,
+      "epoch": 0.015984180604762296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0649588289112536e-05,
+      "loss": 1.0723,
       "step": 582
     },
     {
-      "epoch": 0.19052287581699345,
-      "grad_norm": 1.8619896173477173,
-      "learning_rate": 1.8679594379002267e-05,
-      "loss": 1.0794,
+      "epoch": 0.01601164483260553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0667886550777677e-05,
+      "loss": 1.0982,
       "step": 583
     },
     {
-      "epoch": 0.19084967320261437,
-      "grad_norm": 1.989200472831726,
-      "learning_rate": 1.8674332694389464e-05,
-      "loss": 1.0819,
+      "epoch": 0.016039109060448767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.068618481244282e-05,
+      "loss": 1.0916,
       "step": 584
     },
     {
-      "epoch": 0.19117647058823528,
-      "grad_norm": 1.82257878780365,
-      "learning_rate": 1.8669061291080904e-05,
-      "loss": 1.0682,
+      "epoch": 0.016066573288292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070448307410796e-05,
+      "loss": 0.9717,
       "step": 585
     },
     {
-      "epoch": 0.19150326797385622,
-      "grad_norm": 1.9591268301010132,
-      "learning_rate": 1.8663780174982656e-05,
-      "loss": 1.084,
+      "epoch": 0.016094037516135234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0722781335773103e-05,
+      "loss": 1.0495,
       "step": 586
     },
     {
-      "epoch": 0.19183006535947714,
-      "grad_norm": 1.780016303062439,
-      "learning_rate": 1.8658489352011668e-05,
-      "loss": 1.0324,
+      "epoch": 0.01612150174397847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0741079597438244e-05,
+      "loss": 0.9789,
       "step": 587
     },
     {
-      "epoch": 0.19215686274509805,
-      "grad_norm": 1.9490388631820679,
-      "learning_rate": 1.8653188828095754e-05,
-      "loss": 1.053,
+      "epoch": 0.0161489659718217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0759377859103386e-05,
+      "loss": 1.1611,
       "step": 588
     },
     {
-      "epoch": 0.19248366013071896,
-      "grad_norm": 1.8499828577041626,
-      "learning_rate": 1.8647878609173612e-05,
-      "loss": 1.0734,
+      "epoch": 0.016176430199664936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0777676120768527e-05,
+      "loss": 1.1117,
       "step": 589
     },
     {
-      "epoch": 0.19281045751633988,
-      "grad_norm": 1.9767959117889404,
-      "learning_rate": 1.8642558701194787e-05,
-      "loss": 1.0374,
+      "epoch": 0.01620389442750817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079597438243367e-05,
+      "loss": 1.0757,
       "step": 590
     },
     {
-      "epoch": 0.1931372549019608,
-      "grad_norm": 1.8135459423065186,
-      "learning_rate": 1.86372291101197e-05,
-      "loss": 1.0633,
+      "epoch": 0.016231358655351406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0814272644098811e-05,
+      "loss": 1.1422,
       "step": 591
     },
     {
-      "epoch": 0.1934640522875817,
-      "grad_norm": 2.0285465717315674,
-      "learning_rate": 1.8631889841919596e-05,
-      "loss": 1.0221,
+      "epoch": 0.016258822883194638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0832570905763953e-05,
+      "loss": 1.0802,
       "step": 592
     },
     {
-      "epoch": 0.19379084967320262,
-      "grad_norm": 1.8673619031906128,
-      "learning_rate": 1.8626540902576582e-05,
-      "loss": 1.0447,
+      "epoch": 0.016286287111037873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0850869167429094e-05,
+      "loss": 1.0415,
       "step": 593
     },
     {
-      "epoch": 0.19411764705882353,
-      "grad_norm": 2.0938851833343506,
-      "learning_rate": 1.862118229808359e-05,
-      "loss": 1.039,
+      "epoch": 0.01631375133888111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0869167429094235e-05,
+      "loss": 1.0819,
       "step": 594
     },
     {
-      "epoch": 0.19444444444444445,
-      "grad_norm": 2.026275396347046,
-      "learning_rate": 1.8615814034444385e-05,
-      "loss": 1.0765,
+      "epoch": 0.01634121556672434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0887465690759378e-05,
+      "loss": 1.0217,
       "step": 595
     },
     {
-      "epoch": 0.19477124183006536,
-      "grad_norm": 1.935767412185669,
-      "learning_rate": 1.8610436117673557e-05,
-      "loss": 1.0523,
+      "epoch": 0.016368679794567575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0905763952424519e-05,
+      "loss": 1.0768,
       "step": 596
     },
     {
-      "epoch": 0.19509803921568628,
-      "grad_norm": 1.9800009727478027,
-      "learning_rate": 1.86050485537965e-05,
-      "loss": 1.0572,
+      "epoch": 0.01639614402241081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0924062214089661e-05,
+      "loss": 1.0909,
       "step": 597
     },
     {
-      "epoch": 0.1954248366013072,
-      "grad_norm": 1.8137407302856445,
-      "learning_rate": 1.859965134884943e-05,
-      "loss": 1.0715,
+      "epoch": 0.016423608250254042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0942360475754806e-05,
+      "loss": 1.1365,
       "step": 598
     },
     {
-      "epoch": 0.1957516339869281,
-      "grad_norm": 1.7272435426712036,
-      "learning_rate": 1.859424450887936e-05,
-      "loss": 1.0677,
+      "epoch": 0.016451072478097278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0960658737419947e-05,
+      "loss": 1.032,
       "step": 599
     },
     {
-      "epoch": 0.19607843137254902,
-      "grad_norm": 1.8572869300842285,
-      "learning_rate": 1.8588828039944086e-05,
-      "loss": 1.0602,
+      "epoch": 0.016478536705940513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.097895699908509e-05,
+      "loss": 1.2051,
       "step": 600
     },
     {
-      "epoch": 0.19640522875816993,
-      "grad_norm": 1.7301594018936157,
-      "learning_rate": 1.8583401948112216e-05,
-      "loss": 1.0343,
+      "epoch": 0.016506000933783748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.099725526075023e-05,
+      "loss": 1.0412,
       "step": 601
     },
     {
-      "epoch": 0.19673202614379084,
-      "grad_norm": 1.9606674909591675,
-      "learning_rate": 1.857796623946312e-05,
-      "loss": 1.0676,
+      "epoch": 0.01653346516162698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1015553522415371e-05,
+      "loss": 1.1021,
       "step": 602
     },
     {
-      "epoch": 0.19705882352941176,
-      "grad_norm": 1.8914992809295654,
-      "learning_rate": 1.857252092008695e-05,
-      "loss": 1.0156,
+      "epoch": 0.016560929389470215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1033851784080514e-05,
+      "loss": 1.122,
       "step": 603
     },
     {
-      "epoch": 0.19738562091503267,
-      "grad_norm": 1.8392646312713623,
-      "learning_rate": 1.8567065996084628e-05,
-      "loss": 1.0705,
+      "epoch": 0.01658839361731345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1052150045745655e-05,
+      "loss": 1.157,
       "step": 604
     },
     {
-      "epoch": 0.1977124183006536,
-      "grad_norm": 1.8458361625671387,
-      "learning_rate": 1.8561601473567828e-05,
-      "loss": 1.0692,
+      "epoch": 0.016615857845156682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1070448307410797e-05,
+      "loss": 1.1319,
       "step": 605
     },
     {
-      "epoch": 0.1980392156862745,
-      "grad_norm": 1.8705992698669434,
-      "learning_rate": 1.8556127358658995e-05,
-      "loss": 1.0391,
+      "epoch": 0.016643322072999917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1088746569075938e-05,
+      "loss": 1.1317,
       "step": 606
     },
     {
-      "epoch": 0.19836601307189541,
-      "grad_norm": 1.864869475364685,
-      "learning_rate": 1.85506436574913e-05,
-      "loss": 1.0427,
+      "epoch": 0.016670786300843152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1107044830741081e-05,
+      "loss": 1.1201,
       "step": 607
     },
     {
-      "epoch": 0.19869281045751633,
-      "grad_norm": 1.817996859550476,
-      "learning_rate": 1.854515037620867e-05,
-      "loss": 1.0805,
+      "epoch": 0.016698250528686388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1125343092406222e-05,
+      "loss": 1.2271,
       "step": 608
     },
     {
-      "epoch": 0.19901960784313724,
-      "grad_norm": 1.907495379447937,
-      "learning_rate": 1.8539647520965757e-05,
-      "loss": 1.0659,
+      "epoch": 0.01672571475652962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1143641354071364e-05,
+      "loss": 1.1299,
       "step": 609
     },
     {
-      "epoch": 0.19934640522875818,
-      "grad_norm": 1.8292343616485596,
-      "learning_rate": 1.853413509792795e-05,
-      "loss": 1.018,
+      "epoch": 0.016753178984372855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1161939615736505e-05,
+      "loss": 1.069,
       "step": 610
     },
     {
-      "epoch": 0.1996732026143791,
-      "grad_norm": 1.8932121992111206,
-      "learning_rate": 1.8528613113271344e-05,
-      "loss": 1.0328,
+      "epoch": 0.01678064321221609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1180237877401648e-05,
+      "loss": 1.1117,
       "step": 611
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 1.7829105854034424,
-      "learning_rate": 1.8523081573182754e-05,
-      "loss": 1.0315,
+      "epoch": 0.01680810744005932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1198536139066789e-05,
+      "loss": 1.177,
       "step": 612
     },
     {
-      "epoch": 0.20032679738562092,
-      "grad_norm": 1.8915088176727295,
-      "learning_rate": 1.8517540483859707e-05,
-      "loss": 1.0395,
+      "epoch": 0.016835571667902557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1216834400731932e-05,
+      "loss": 1.1188,
       "step": 613
     },
     {
-      "epoch": 0.20065359477124184,
-      "grad_norm": 1.8347907066345215,
-      "learning_rate": 1.8511989851510416e-05,
-      "loss": 1.0376,
+      "epoch": 0.016863035895745792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1235132662397072e-05,
+      "loss": 1.2456,
       "step": 614
     },
     {
-      "epoch": 0.20098039215686275,
-      "grad_norm": 1.7669236660003662,
-      "learning_rate": 1.8506429682353793e-05,
-      "loss": 1.0487,
+      "epoch": 0.016890500123589024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1253430924062213e-05,
+      "loss": 1.1498,
       "step": 615
     },
     {
-      "epoch": 0.20130718954248367,
-      "grad_norm": 1.9599943161010742,
-      "learning_rate": 1.8500859982619438e-05,
-      "loss": 1.0462,
+      "epoch": 0.01691796435143226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1271729185727356e-05,
+      "loss": 1.1175,
       "step": 616
     },
     {
-      "epoch": 0.20163398692810458,
-      "grad_norm": 1.9591596126556396,
-      "learning_rate": 1.849528075854762e-05,
-      "loss": 1.0484,
+      "epoch": 0.016945428579275494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1290027447392497e-05,
+      "loss": 1.1024,
       "step": 617
     },
     {
-      "epoch": 0.2019607843137255,
-      "grad_norm": 1.7656697034835815,
-      "learning_rate": 1.848969201638929e-05,
-      "loss": 1.0543,
+      "epoch": 0.01697289280711873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1308325709057641e-05,
+      "loss": 1.0838,
       "step": 618
     },
     {
-      "epoch": 0.2022875816993464,
-      "grad_norm": 1.9272146224975586,
-      "learning_rate": 1.848409376240606e-05,
-      "loss": 1.0781,
+      "epoch": 0.01700035703496196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1326623970722784e-05,
+      "loss": 1.2339,
       "step": 619
     },
     {
-      "epoch": 0.20261437908496732,
-      "grad_norm": 1.7910609245300293,
-      "learning_rate": 1.847848600287019e-05,
-      "loss": 1.0838,
+      "epoch": 0.017027821262805196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1344922232387925e-05,
+      "loss": 1.1278,
       "step": 620
     },
     {
-      "epoch": 0.20294117647058824,
-      "grad_norm": 1.7562174797058105,
-      "learning_rate": 1.84728687440646e-05,
-      "loss": 1.0478,
+      "epoch": 0.01705528549064843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1363220494053067e-05,
+      "loss": 1.0377,
       "step": 621
     },
     {
-      "epoch": 0.20326797385620915,
-      "grad_norm": 1.9585661888122559,
-      "learning_rate": 1.8467241992282842e-05,
-      "loss": 1.0871,
+      "epoch": 0.017082749718491663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1381518755718208e-05,
+      "loss": 1.0669,
       "step": 622
     },
     {
-      "epoch": 0.20359477124183006,
-      "grad_norm": 1.7822569608688354,
-      "learning_rate": 1.8461605753829125e-05,
-      "loss": 1.0534,
+      "epoch": 0.0171102139463349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.139981701738335e-05,
+      "loss": 1.1265,
       "step": 623
     },
     {
-      "epoch": 0.20392156862745098,
-      "grad_norm": 1.8685938119888306,
-      "learning_rate": 1.845596003501826e-05,
-      "loss": 1.0563,
+      "epoch": 0.017137678174178134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1418115279048492e-05,
+      "loss": 1.016,
       "step": 624
     },
     {
-      "epoch": 0.2042483660130719,
-      "grad_norm": 1.9384000301361084,
-      "learning_rate": 1.8450304842175707e-05,
-      "loss": 1.0307,
+      "epoch": 0.01716514240202137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1436413540713633e-05,
+      "loss": 1.2626,
       "step": 625
     },
     {
-      "epoch": 0.2045751633986928,
-      "grad_norm": 1.8621412515640259,
-      "learning_rate": 1.8444640181637514e-05,
-      "loss": 1.0488,
+      "epoch": 0.0171926066298646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1454711802378775e-05,
+      "loss": 1.0905,
       "step": 626
     },
     {
-      "epoch": 0.20490196078431372,
-      "grad_norm": 1.8028227090835571,
-      "learning_rate": 1.8438966059750352e-05,
-      "loss": 1.055,
+      "epoch": 0.017220070857707836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1473010064043916e-05,
+      "loss": 1.1247,
       "step": 627
     },
     {
-      "epoch": 0.20522875816993463,
-      "grad_norm": 1.7884498834609985,
-      "learning_rate": 1.8433282482871497e-05,
-      "loss": 1.0311,
+      "epoch": 0.01724753508555107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1491308325709059e-05,
+      "loss": 1.1783,
       "step": 628
     },
     {
-      "epoch": 0.20555555555555555,
-      "grad_norm": 1.9722820520401,
-      "learning_rate": 1.8427589457368806e-05,
-      "loss": 1.0634,
+      "epoch": 0.017274999313394303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15096065873742e-05,
+      "loss": 1.1636,
       "step": 629
     },
     {
-      "epoch": 0.20588235294117646,
-      "grad_norm": 1.919356346130371,
-      "learning_rate": 1.8421886989620724e-05,
-      "loss": 1.065,
+      "epoch": 0.017302463541237538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1527904849039342e-05,
+      "loss": 1.1716,
       "step": 630
     },
     {
-      "epoch": 0.20620915032679737,
-      "grad_norm": 1.9422677755355835,
-      "learning_rate": 1.8416175086016284e-05,
-      "loss": 1.0744,
+      "epoch": 0.017329927769080773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1546203110704483e-05,
+      "loss": 1.0789,
       "step": 631
     },
     {
-      "epoch": 0.2065359477124183,
-      "grad_norm": 2.0933480262756348,
-      "learning_rate": 1.841045375295508e-05,
-      "loss": 1.0365,
+      "epoch": 0.017357391996924005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1564501372369626e-05,
+      "loss": 1.1255,
       "step": 632
     },
     {
-      "epoch": 0.2068627450980392,
-      "grad_norm": 2.0303518772125244,
-      "learning_rate": 1.8404722996847282e-05,
-      "loss": 1.0533,
+      "epoch": 0.01738485622476724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1582799634034767e-05,
+      "loss": 1.1931,
       "step": 633
     },
     {
-      "epoch": 0.20718954248366014,
-      "grad_norm": 1.8719193935394287,
-      "learning_rate": 1.839898282411361e-05,
-      "loss": 1.0543,
+      "epoch": 0.017412320452610475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.160109789569991e-05,
+      "loss": 1.1804,
       "step": 634
     },
     {
-      "epoch": 0.20751633986928106,
-      "grad_norm": 1.9279483556747437,
-      "learning_rate": 1.8393233241185333e-05,
-      "loss": 1.0483,
+      "epoch": 0.01743978468045371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.161939615736505e-05,
+      "loss": 1.232,
       "step": 635
     },
     {
-      "epoch": 0.20784313725490197,
-      "grad_norm": 1.8710894584655762,
-      "learning_rate": 1.8387474254504265e-05,
-      "loss": 1.058,
+      "epoch": 0.017467248908296942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1637694419030191e-05,
+      "loss": 1.1425,
       "step": 636
     },
     {
-      "epoch": 0.20816993464052289,
-      "grad_norm": 1.9775340557098389,
-      "learning_rate": 1.838170587052276e-05,
-      "loss": 1.0585,
+      "epoch": 0.017494713136140178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1655992680695334e-05,
+      "loss": 1.0942,
       "step": 637
     },
     {
-      "epoch": 0.2084967320261438,
-      "grad_norm": 1.794744849205017,
-      "learning_rate": 1.8375928095703703e-05,
-      "loss": 1.0573,
+      "epoch": 0.017522177363983413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1674290942360478e-05,
+      "loss": 1.0238,
       "step": 638
     },
     {
-      "epoch": 0.2088235294117647,
-      "grad_norm": 2.074080228805542,
-      "learning_rate": 1.8370140936520486e-05,
-      "loss": 1.0759,
+      "epoch": 0.017549641591826644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.169258920402562e-05,
+      "loss": 1.1532,
       "step": 639
     },
     {
-      "epoch": 0.20915032679738563,
-      "grad_norm": 1.6575369834899902,
-      "learning_rate": 1.8364344399457028e-05,
-      "loss": 1.071,
+      "epoch": 0.01757710581966988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1710887465690762e-05,
+      "loss": 1.1199,
       "step": 640
     },
     {
-      "epoch": 0.20947712418300654,
-      "grad_norm": 1.8823904991149902,
-      "learning_rate": 1.8358538491007748e-05,
-      "loss": 1.0763,
+      "epoch": 0.017604570047513115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1729185727355903e-05,
+      "loss": 1.0663,
       "step": 641
     },
     {
-      "epoch": 0.20980392156862746,
-      "grad_norm": 1.8569639921188354,
-      "learning_rate": 1.835272321767758e-05,
-      "loss": 1.0291,
+      "epoch": 0.017632034275356347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1747483989021045e-05,
+      "loss": 1.1314,
       "step": 642
     },
     {
-      "epoch": 0.21013071895424837,
-      "grad_norm": 1.8166108131408691,
-      "learning_rate": 1.8346898585981923e-05,
-      "loss": 1.0453,
+      "epoch": 0.017659498503199582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1765782250686186e-05,
+      "loss": 1.1037,
       "step": 643
     },
     {
-      "epoch": 0.21045751633986928,
-      "grad_norm": 1.7722333669662476,
-      "learning_rate": 1.8341064602446686e-05,
-      "loss": 1.0314,
+      "epoch": 0.017686962731042817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1784080512351327e-05,
+      "loss": 0.9713,
       "step": 644
     },
     {
-      "epoch": 0.2107843137254902,
-      "grad_norm": 1.7922865152359009,
-      "learning_rate": 1.833522127360824e-05,
-      "loss": 1.0677,
+      "epoch": 0.017714426958886052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.180237877401647e-05,
+      "loss": 1.1409,
       "step": 645
     },
     {
-      "epoch": 0.2111111111111111,
-      "grad_norm": 1.8785959482192993,
-      "learning_rate": 1.832936860601344e-05,
-      "loss": 1.0548,
+      "epoch": 0.017741891186729284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.182067703568161e-05,
+      "loss": 1.1707,
       "step": 646
     },
     {
-      "epoch": 0.21143790849673202,
-      "grad_norm": 1.7716997861862183,
-      "learning_rate": 1.832350660621959e-05,
-      "loss": 1.0487,
+      "epoch": 0.01776935541457252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1838975297346753e-05,
+      "loss": 1.0543,
       "step": 647
     },
     {
-      "epoch": 0.21176470588235294,
-      "grad_norm": 1.829897403717041,
-      "learning_rate": 1.8317635280794466e-05,
-      "loss": 1.0173,
+      "epoch": 0.017796819642415754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1857273559011894e-05,
+      "loss": 1.1811,
       "step": 648
     },
     {
-      "epoch": 0.21209150326797385,
-      "grad_norm": 1.9153690338134766,
-      "learning_rate": 1.831175463631627e-05,
-      "loss": 1.0649,
+      "epoch": 0.017824283870258986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1875571820677037e-05,
+      "loss": 1.1361,
       "step": 649
     },
     {
-      "epoch": 0.21241830065359477,
-      "grad_norm": 1.846561312675476,
-      "learning_rate": 1.8305864679373667e-05,
-      "loss": 1.0368,
+      "epoch": 0.01785174809810222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1893870082342178e-05,
+      "loss": 1.131,
       "step": 650
     },
     {
-      "epoch": 0.21274509803921568,
-      "grad_norm": 1.9194376468658447,
-      "learning_rate": 1.8299965416565744e-05,
-      "loss": 1.0581,
+      "epoch": 0.017879212325945457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.191216834400732e-05,
+      "loss": 1.0926,
       "step": 651
     },
     {
-      "epoch": 0.2130718954248366,
-      "grad_norm": 1.7874829769134521,
-      "learning_rate": 1.829405685450202e-05,
-      "loss": 1.0482,
+      "epoch": 0.017906676553788692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1930466605672461e-05,
+      "loss": 1.1001,
       "step": 652
     },
     {
-      "epoch": 0.2133986928104575,
-      "grad_norm": 2.370438814163208,
-      "learning_rate": 1.8288138999802426e-05,
-      "loss": 1.0828,
+      "epoch": 0.017934140781631924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1948764867337604e-05,
+      "loss": 1.063,
       "step": 653
     },
     {
-      "epoch": 0.21372549019607842,
-      "grad_norm": 1.8513789176940918,
-      "learning_rate": 1.8282211859097308e-05,
-      "loss": 1.0619,
+      "epoch": 0.01796160500947516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1967063129002745e-05,
+      "loss": 1.0198,
       "step": 654
     },
     {
-      "epoch": 0.21405228758169934,
-      "grad_norm": 1.7623997926712036,
-      "learning_rate": 1.8276275439027417e-05,
-      "loss": 1.0701,
+      "epoch": 0.017989069237318394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1985361390667888e-05,
+      "loss": 1.1188,
       "step": 655
     },
     {
-      "epoch": 0.21437908496732025,
-      "grad_norm": 1.649559497833252,
-      "learning_rate": 1.8270329746243903e-05,
-      "loss": 1.0431,
+      "epoch": 0.018016533465161626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2003659652333028e-05,
+      "loss": 0.9719,
       "step": 656
     },
     {
-      "epoch": 0.21470588235294116,
-      "grad_norm": 1.6782225370407104,
-      "learning_rate": 1.82643747874083e-05,
-      "loss": 1.0505,
+      "epoch": 0.01804399769300486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202195791399817e-05,
+      "loss": 1.2876,
       "step": 657
     },
     {
-      "epoch": 0.2150326797385621,
-      "grad_norm": 2.0034573078155518,
-      "learning_rate": 1.8258410569192526e-05,
-      "loss": 1.0235,
+      "epoch": 0.018071461920848096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2040256175663314e-05,
+      "loss": 1.0763,
       "step": 658
     },
     {
-      "epoch": 0.21535947712418302,
-      "grad_norm": 1.8660086393356323,
-      "learning_rate": 1.8252437098278867e-05,
-      "loss": 1.0396,
+      "epoch": 0.018098926148691328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2058554437328456e-05,
+      "loss": 1.1041,
       "step": 659
     },
     {
-      "epoch": 0.21568627450980393,
-      "grad_norm": 1.7891356945037842,
-      "learning_rate": 1.824645438135999e-05,
-      "loss": 1.0421,
+      "epoch": 0.018126390376534563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2076852698993597e-05,
+      "loss": 1.1812,
       "step": 660
     },
     {
-      "epoch": 0.21601307189542485,
-      "grad_norm": 1.8125733137130737,
-      "learning_rate": 1.824046242513891e-05,
-      "loss": 1.0504,
+      "epoch": 0.0181538546043778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.209515096065874e-05,
+      "loss": 1.1761,
       "step": 661
     },
     {
-      "epoch": 0.21633986928104576,
-      "grad_norm": 1.9175999164581299,
-      "learning_rate": 1.8234461236328992e-05,
-      "loss": 1.0642,
+      "epoch": 0.018181318832221034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211344922232388e-05,
+      "loss": 1.1653,
       "step": 662
     },
     {
-      "epoch": 0.21666666666666667,
-      "grad_norm": 1.8372156620025635,
-      "learning_rate": 1.8228450821653956e-05,
-      "loss": 1.0979,
+      "epoch": 0.018208783060064265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2131747483989023e-05,
+      "loss": 1.1499,
       "step": 663
     },
     {
-      "epoch": 0.2169934640522876,
-      "grad_norm": 1.9329107999801636,
-      "learning_rate": 1.822243118784785e-05,
-      "loss": 1.0252,
+      "epoch": 0.0182362472879075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2150045745654164e-05,
+      "loss": 1.1956,
       "step": 664
     },
     {
-      "epoch": 0.2173202614379085,
-      "grad_norm": 1.8479384183883667,
-      "learning_rate": 1.8216402341655052e-05,
-      "loss": 1.0471,
+      "epoch": 0.018263711515750736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2168344007319305e-05,
+      "loss": 1.1268,
       "step": 665
     },
     {
-      "epoch": 0.21764705882352942,
-      "grad_norm": 1.9173794984817505,
-      "learning_rate": 1.821036428983026e-05,
-      "loss": 1.0136,
+      "epoch": 0.018291175743593967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2186642268984448e-05,
+      "loss": 1.0956,
       "step": 666
     },
     {
-      "epoch": 0.21797385620915033,
-      "grad_norm": 1.8668714761734009,
-      "learning_rate": 1.8204317039138497e-05,
-      "loss": 1.0564,
+      "epoch": 0.018318639971437203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2204940530649589e-05,
+      "loss": 1.1267,
       "step": 667
     },
     {
-      "epoch": 0.21830065359477124,
-      "grad_norm": 1.7245558500289917,
-      "learning_rate": 1.8198260596355077e-05,
-      "loss": 1.0283,
+      "epoch": 0.018346104199280438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2223238792314731e-05,
+      "loss": 1.0901,
       "step": 668
     },
     {
-      "epoch": 0.21862745098039216,
-      "grad_norm": 1.8658119440078735,
-      "learning_rate": 1.819219496826562e-05,
-      "loss": 1.0784,
+      "epoch": 0.018373568427123673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2241537053979872e-05,
+      "loss": 1.0284,
       "step": 669
     },
     {
-      "epoch": 0.21895424836601307,
-      "grad_norm": 1.8234193325042725,
-      "learning_rate": 1.818612016166604e-05,
-      "loss": 1.0595,
+      "epoch": 0.018401032654966905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2259835315645015e-05,
+      "loss": 1.0554,
       "step": 670
     },
     {
-      "epoch": 0.21928104575163399,
-      "grad_norm": 1.8511459827423096,
-      "learning_rate": 1.8180036183362535e-05,
-      "loss": 1.0874,
+      "epoch": 0.01842849688281014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2278133577310156e-05,
+      "loss": 0.9875,
       "step": 671
     },
     {
-      "epoch": 0.2196078431372549,
-      "grad_norm": 1.7114883661270142,
-      "learning_rate": 1.8173943040171567e-05,
-      "loss": 1.0246,
+      "epoch": 0.018455961110653375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2296431838975299e-05,
+      "loss": 1.0928,
       "step": 672
     },
     {
-      "epoch": 0.2199346405228758,
-      "grad_norm": 1.6927613019943237,
-      "learning_rate": 1.8167840738919883e-05,
-      "loss": 1.0333,
+      "epoch": 0.018483425338496607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231473010064044e-05,
+      "loss": 1.1957,
       "step": 673
     },
     {
-      "epoch": 0.22026143790849673,
-      "grad_norm": 1.7772390842437744,
-      "learning_rate": 1.8161729286444474e-05,
-      "loss": 1.061,
+      "epoch": 0.018510889566339842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2333028362305582e-05,
+      "loss": 1.17,
       "step": 674
     },
     {
-      "epoch": 0.22058823529411764,
-      "grad_norm": 1.6820720434188843,
-      "learning_rate": 1.8155608689592604e-05,
-      "loss": 1.0367,
+      "epoch": 0.018538353794183077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2351326623970723e-05,
+      "loss": 1.2136,
       "step": 675
     },
     {
-      "epoch": 0.22091503267973855,
-      "grad_norm": 1.6448161602020264,
-      "learning_rate": 1.814947895522176e-05,
-      "loss": 1.0154,
+      "epoch": 0.01856581802202631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2369624885635864e-05,
+      "loss": 1.1172,
       "step": 676
     },
     {
-      "epoch": 0.22124183006535947,
-      "grad_norm": 1.6766726970672607,
-      "learning_rate": 1.8143340090199687e-05,
-      "loss": 1.0474,
+      "epoch": 0.018593282249869544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2387923147301007e-05,
+      "loss": 1.1052,
       "step": 677
     },
     {
-      "epoch": 0.22156862745098038,
-      "grad_norm": 1.6982693672180176,
-      "learning_rate": 1.8137192101404343e-05,
-      "loss": 1.073,
+      "epoch": 0.01862074647771278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2406221408966147e-05,
+      "loss": 1.1858,
       "step": 678
     },
     {
-      "epoch": 0.2218954248366013,
-      "grad_norm": 1.7892935276031494,
-      "learning_rate": 1.813103499572392e-05,
-      "loss": 1.0494,
+      "epoch": 0.018648210705556015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2424519670631292e-05,
+      "loss": 1.0313,
       "step": 679
     },
     {
-      "epoch": 0.2222222222222222,
-      "grad_norm": 1.6614453792572021,
-      "learning_rate": 1.8124868780056814e-05,
-      "loss": 1.006,
+      "epoch": 0.018675674933399247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2442817932296434e-05,
+      "loss": 1.1434,
       "step": 680
     },
     {
-      "epoch": 0.22254901960784312,
-      "grad_norm": 1.6697980165481567,
-      "learning_rate": 1.8118693461311643e-05,
-      "loss": 1.051,
+      "epoch": 0.018703139161242482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2461116193961575e-05,
+      "loss": 1.0256,
       "step": 681
     },
     {
-      "epoch": 0.22287581699346407,
-      "grad_norm": 1.6075372695922852,
-      "learning_rate": 1.811250904640721e-05,
-      "loss": 1.0338,
+      "epoch": 0.018730603389085717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2479414455626718e-05,
+      "loss": 1.1349,
       "step": 682
     },
     {
-      "epoch": 0.22320261437908498,
-      "grad_norm": 1.634206771850586,
-      "learning_rate": 1.8106315542272516e-05,
-      "loss": 1.0513,
+      "epoch": 0.01875806761692895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2497712717291859e-05,
+      "loss": 1.0959,
       "step": 683
     },
     {
-      "epoch": 0.2235294117647059,
-      "grad_norm": 1.7235411405563354,
-      "learning_rate": 1.8100112955846746e-05,
-      "loss": 1.0597,
+      "epoch": 0.018785531844772184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2516010978957e-05,
+      "loss": 1.0849,
       "step": 684
     },
     {
-      "epoch": 0.2238562091503268,
-      "grad_norm": 1.7576299905776978,
-      "learning_rate": 1.8093901294079254e-05,
-      "loss": 1.0185,
+      "epoch": 0.01881299607261542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2534309240622142e-05,
+      "loss": 1.123,
       "step": 685
     },
     {
-      "epoch": 0.22418300653594772,
-      "grad_norm": 1.7927601337432861,
-      "learning_rate": 1.8087680563929574e-05,
-      "loss": 1.062,
+      "epoch": 0.018840460300458654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2552607502287283e-05,
+      "loss": 1.1606,
       "step": 686
     },
     {
-      "epoch": 0.22450980392156863,
-      "grad_norm": 1.864380955696106,
-      "learning_rate": 1.8081450772367382e-05,
-      "loss": 1.0331,
+      "epoch": 0.018867924528301886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2570905763952426e-05,
+      "loss": 1.152,
       "step": 687
     },
     {
-      "epoch": 0.22483660130718955,
-      "grad_norm": 1.8185263872146606,
-      "learning_rate": 1.807521192637253e-05,
-      "loss": 1.0396,
+      "epoch": 0.01889538875614512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2589204025617567e-05,
+      "loss": 1.0638,
       "step": 688
     },
     {
-      "epoch": 0.22516339869281046,
-      "grad_norm": 1.7746814489364624,
-      "learning_rate": 1.806896403293499e-05,
-      "loss": 1.0507,
+      "epoch": 0.018922852983988356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.260750228728271e-05,
+      "loss": 1.1698,
       "step": 689
     },
     {
-      "epoch": 0.22549019607843138,
-      "grad_norm": 1.7886286973953247,
-      "learning_rate": 1.806270709905489e-05,
-      "loss": 1.0262,
+      "epoch": 0.018950317211831588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262580054894785e-05,
+      "loss": 1.0955,
       "step": 690
     },
     {
-      "epoch": 0.2258169934640523,
-      "grad_norm": 1.5855242013931274,
-      "learning_rate": 1.8056441131742478e-05,
-      "loss": 1.0323,
+      "epoch": 0.018977781439674823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2644098810612993e-05,
+      "loss": 1.1964,
       "step": 691
     },
     {
-      "epoch": 0.2261437908496732,
-      "grad_norm": 1.7104792594909668,
-      "learning_rate": 1.805016613801813e-05,
-      "loss": 1.0451,
+      "epoch": 0.01900524566751806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2662397072278134e-05,
+      "loss": 1.0596,
       "step": 692
     },
     {
-      "epoch": 0.22647058823529412,
-      "grad_norm": 1.6900410652160645,
-      "learning_rate": 1.8043882124912317e-05,
-      "loss": 1.0178,
+      "epoch": 0.01903270989536129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2680695333943277e-05,
+      "loss": 1.215,
       "step": 693
     },
     {
-      "epoch": 0.22679738562091503,
-      "grad_norm": 1.7223681211471558,
-      "learning_rate": 1.8037589099465637e-05,
-      "loss": 1.0583,
+      "epoch": 0.019060174123204526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2698993595608417e-05,
+      "loss": 0.9673,
       "step": 694
     },
     {
-      "epoch": 0.22712418300653595,
-      "grad_norm": 1.71628737449646,
-      "learning_rate": 1.8031287068728774e-05,
-      "loss": 1.0336,
+      "epoch": 0.01908763835104776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271729185727356e-05,
+      "loss": 1.1125,
       "step": 695
     },
     {
-      "epoch": 0.22745098039215686,
-      "grad_norm": 1.7380578517913818,
-      "learning_rate": 1.8024976039762507e-05,
-      "loss": 1.0071,
+      "epoch": 0.019115102578890996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2735590118938701e-05,
+      "loss": 1.1608,
       "step": 696
     },
     {
-      "epoch": 0.22777777777777777,
-      "grad_norm": 1.7125619649887085,
-      "learning_rate": 1.801865601963769e-05,
-      "loss": 1.0614,
+      "epoch": 0.019142566806734228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2753888380603842e-05,
+      "loss": 1.1635,
       "step": 697
     },
     {
-      "epoch": 0.2281045751633987,
-      "grad_norm": 1.9443279504776,
-      "learning_rate": 1.8012327015435255e-05,
-      "loss": 1.0809,
+      "epoch": 0.019170031034577463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2772186642268985e-05,
+      "loss": 1.2012,
       "step": 698
     },
     {
-      "epoch": 0.2284313725490196,
-      "grad_norm": 1.6187223196029663,
-      "learning_rate": 1.8005989034246196e-05,
-      "loss": 1.0154,
+      "epoch": 0.019197495262420698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2790484903934129e-05,
+      "loss": 1.1119,
       "step": 699
     },
     {
-      "epoch": 0.22875816993464052,
-      "grad_norm": 1.798576831817627,
-      "learning_rate": 1.7999642083171576e-05,
-      "loss": 1.0321,
+      "epoch": 0.01922495949026393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.280878316559927e-05,
+      "loss": 1.0368,
       "step": 700
     },
     {
-      "epoch": 0.22908496732026143,
-      "grad_norm": 1.8266483545303345,
-      "learning_rate": 1.799328616932249e-05,
-      "loss": 1.0607,
+      "epoch": 0.019252423718107165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2827081427264412e-05,
+      "loss": 1.0388,
       "step": 701
     },
     {
-      "epoch": 0.22941176470588234,
-      "grad_norm": 1.789952039718628,
-      "learning_rate": 1.798692129982009e-05,
-      "loss": 1.0366,
+      "epoch": 0.0192798879459504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2845379688929553e-05,
+      "loss": 1.222,
       "step": 702
     },
     {
-      "epoch": 0.22973856209150326,
-      "grad_norm": 1.7757703065872192,
-      "learning_rate": 1.798054748179556e-05,
-      "loss": 1.0829,
+      "epoch": 0.019307352173793632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2863677950594696e-05,
+      "loss": 1.1362,
       "step": 703
     },
     {
-      "epoch": 0.23006535947712417,
-      "grad_norm": 1.8663150072097778,
-      "learning_rate": 1.7974164722390094e-05,
-      "loss": 1.0538,
+      "epoch": 0.019334816401636867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2881976212259837e-05,
+      "loss": 1.1844,
       "step": 704
     },
     {
-      "epoch": 0.23039215686274508,
-      "grad_norm": 1.8030458688735962,
-      "learning_rate": 1.7967773028754926e-05,
-      "loss": 1.045,
+      "epoch": 0.019362280629480103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2900274473924978e-05,
+      "loss": 1.1476,
       "step": 705
     },
     {
-      "epoch": 0.23071895424836603,
-      "grad_norm": 1.7396090030670166,
-      "learning_rate": 1.796137240805129e-05,
-      "loss": 1.0603,
+      "epoch": 0.019389744857323338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291857273559012e-05,
+      "loss": 1.1225,
       "step": 706
     },
     {
-      "epoch": 0.23104575163398694,
-      "grad_norm": 1.8198657035827637,
-      "learning_rate": 1.7954962867450415e-05,
-      "loss": 1.0463,
+      "epoch": 0.01941720908516657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2936870997255261e-05,
+      "loss": 1.1557,
       "step": 707
     },
     {
-      "epoch": 0.23137254901960785,
-      "grad_norm": 1.658137321472168,
-      "learning_rate": 1.7948544414133534e-05,
-      "loss": 1.0508,
+      "epoch": 0.019444673313009805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2955169258920404e-05,
+      "loss": 1.0774,
       "step": 708
     },
     {
-      "epoch": 0.23169934640522877,
-      "grad_norm": 1.7163335084915161,
-      "learning_rate": 1.7942117055291867e-05,
-      "loss": 1.0289,
+      "epoch": 0.01947213754085304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2973467520585545e-05,
+      "loss": 1.2031,
       "step": 709
     },
     {
-      "epoch": 0.23202614379084968,
-      "grad_norm": 1.7208770513534546,
-      "learning_rate": 1.7935680798126597e-05,
-      "loss": 0.9806,
+      "epoch": 0.01949960176869627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2991765782250687e-05,
+      "loss": 1.1807,
       "step": 710
     },
     {
-      "epoch": 0.2323529411764706,
-      "grad_norm": 1.8062595129013062,
-      "learning_rate": 1.79292356498489e-05,
-      "loss": 1.0239,
+      "epoch": 0.019527065996539507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3010064043915828e-05,
+      "loss": 1.1949,
       "step": 711
     },
     {
-      "epoch": 0.2326797385620915,
-      "grad_norm": 1.6323764324188232,
-      "learning_rate": 1.792278161767989e-05,
-      "loss": 1.0228,
+      "epoch": 0.019554530224382742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3028362305580971e-05,
+      "loss": 1.0384,
       "step": 712
     },
     {
-      "epoch": 0.23300653594771242,
-      "grad_norm": 1.714982509613037,
-      "learning_rate": 1.7916318708850648e-05,
-      "loss": 1.0583,
+      "epoch": 0.019581994452225977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3046660567246112e-05,
+      "loss": 1.1528,
       "step": 713
     },
     {
-      "epoch": 0.23333333333333334,
-      "grad_norm": 1.823933720588684,
-      "learning_rate": 1.79098469306022e-05,
-      "loss": 1.0515,
+      "epoch": 0.01960945868006921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3064958828911255e-05,
+      "loss": 0.9743,
       "step": 714
     },
     {
-      "epoch": 0.23366013071895425,
-      "grad_norm": 1.6771241426467896,
-      "learning_rate": 1.7903366290185498e-05,
-      "loss": 1.0117,
+      "epoch": 0.019636922907912444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3083257090576395e-05,
+      "loss": 1.0388,
       "step": 715
     },
     {
-      "epoch": 0.23398692810457516,
-      "grad_norm": 1.6607215404510498,
-      "learning_rate": 1.7896876794861443e-05,
-      "loss": 1.0978,
+      "epoch": 0.01966438713575568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3101555352241538e-05,
+      "loss": 1.1194,
       "step": 716
     },
     {
-      "epoch": 0.23431372549019608,
-      "grad_norm": 1.7140592336654663,
-      "learning_rate": 1.789037845190084e-05,
-      "loss": 1.0423,
+      "epoch": 0.01969185136359891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3119853613906679e-05,
+      "loss": 1.0681,
       "step": 717
     },
     {
-      "epoch": 0.234640522875817,
-      "grad_norm": 1.758021593093872,
-      "learning_rate": 1.788387126858441e-05,
-      "loss": 1.0445,
+      "epoch": 0.019719315591442146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313815187557182e-05,
+      "loss": 1.0046,
       "step": 718
     },
     {
-      "epoch": 0.2349673202614379,
-      "grad_norm": 1.6233147382736206,
-      "learning_rate": 1.7877355252202786e-05,
-      "loss": 1.0602,
+      "epoch": 0.01974677981928538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3156450137236964e-05,
+      "loss": 1.2114,
       "step": 719
     },
     {
-      "epoch": 0.23529411764705882,
-      "grad_norm": 1.5950539112091064,
-      "learning_rate": 1.7870830410056487e-05,
-      "loss": 1.071,
+      "epoch": 0.019774244047128613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3174748398902107e-05,
+      "loss": 0.9901,
       "step": 720
     },
     {
-      "epoch": 0.23562091503267973,
-      "grad_norm": 1.682472825050354,
-      "learning_rate": 1.7864296749455934e-05,
-      "loss": 1.0269,
+      "epoch": 0.01980170827497185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3193046660567248e-05,
+      "loss": 1.0706,
       "step": 721
     },
     {
-      "epoch": 0.23594771241830065,
-      "grad_norm": 1.7691940069198608,
-      "learning_rate": 1.7857754277721417e-05,
-      "loss": 1.0274,
+      "epoch": 0.019829172502815084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321134492223239e-05,
+      "loss": 1.1349,
       "step": 722
     },
     {
-      "epoch": 0.23627450980392156,
-      "grad_norm": 1.648249626159668,
-      "learning_rate": 1.7851203002183096e-05,
-      "loss": 1.0447,
+      "epoch": 0.01985663673065832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3229643183897531e-05,
+      "loss": 1.0236,
       "step": 723
     },
     {
-      "epoch": 0.23660130718954248,
-      "grad_norm": 1.8792474269866943,
-      "learning_rate": 1.7844642930181008e-05,
-      "loss": 1.048,
+      "epoch": 0.01988410095850155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3247941445562674e-05,
+      "loss": 1.0858,
       "step": 724
     },
     {
-      "epoch": 0.2369281045751634,
-      "grad_norm": 1.6973086595535278,
-      "learning_rate": 1.7838074069065034e-05,
-      "loss": 1.0316,
+      "epoch": 0.019911565186344786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3266239707227815e-05,
+      "loss": 1.0993,
       "step": 725
     },
     {
-      "epoch": 0.2372549019607843,
-      "grad_norm": 1.7287940979003906,
-      "learning_rate": 1.7831496426194906e-05,
-      "loss": 1.0393,
+      "epoch": 0.01993902941418802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3284537968892956e-05,
+      "loss": 1.0718,
       "step": 726
     },
     {
-      "epoch": 0.23758169934640522,
-      "grad_norm": 1.6036862134933472,
-      "learning_rate": 1.7824910008940196e-05,
-      "loss": 1.055,
+      "epoch": 0.019966493642031253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3302836230558098e-05,
+      "loss": 1.1674,
       "step": 727
     },
     {
-      "epoch": 0.23790849673202613,
-      "grad_norm": 1.7849210500717163,
-      "learning_rate": 1.78183148246803e-05,
-      "loss": 1.0095,
+      "epoch": 0.019993957869874488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332113449222324e-05,
+      "loss": 1.0833,
       "step": 728
     },
     {
-      "epoch": 0.23823529411764705,
-      "grad_norm": 1.6933237314224243,
-      "learning_rate": 1.781171088080445e-05,
-      "loss": 1.0381,
+      "epoch": 0.020021422097717723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3339432753888382e-05,
+      "loss": 1.1778,
       "step": 729
     },
     {
-      "epoch": 0.238562091503268,
-      "grad_norm": 1.748905897140503,
-      "learning_rate": 1.780509818471168e-05,
-      "loss": 1.0501,
+      "epoch": 0.02004888632556096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3357731015553523e-05,
+      "loss": 1.169,
       "step": 730
     },
     {
-      "epoch": 0.2388888888888889,
-      "grad_norm": 1.777138352394104,
-      "learning_rate": 1.7798476743810836e-05,
-      "loss": 1.0405,
+      "epoch": 0.02007635055340419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3376029277218666e-05,
+      "loss": 1.1769,
       "step": 731
     },
     {
-      "epoch": 0.23921568627450981,
-      "grad_norm": 1.5979655981063843,
-      "learning_rate": 1.779184656552056e-05,
-      "loss": 1.0294,
+      "epoch": 0.020103814781247425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3394327538883806e-05,
+      "loss": 1.0728,
       "step": 732
     },
     {
-      "epoch": 0.23954248366013073,
-      "grad_norm": 1.5690865516662598,
-      "learning_rate": 1.7785207657269282e-05,
-      "loss": 1.0329,
+      "epoch": 0.02013127900909066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3412625800548949e-05,
+      "loss": 1.1609,
       "step": 733
     },
     {
-      "epoch": 0.23986928104575164,
-      "grad_norm": 1.603372573852539,
-      "learning_rate": 1.7778560026495218e-05,
-      "loss": 1.0255,
+      "epoch": 0.020158743236933892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343092406221409e-05,
+      "loss": 1.0721,
       "step": 734
     },
     {
-      "epoch": 0.24019607843137256,
-      "grad_norm": 1.6325621604919434,
-      "learning_rate": 1.777190368064635e-05,
-      "loss": 1.0436,
+      "epoch": 0.020186207464777128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3449222323879233e-05,
+      "loss": 1.1615,
       "step": 735
     },
     {
-      "epoch": 0.24052287581699347,
-      "grad_norm": 1.6357653141021729,
-      "learning_rate": 1.7765238627180424e-05,
-      "loss": 1.046,
+      "epoch": 0.020213671692620363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3467520585544374e-05,
+      "loss": 1.1064,
       "step": 736
     },
     {
-      "epoch": 0.24084967320261438,
-      "grad_norm": 1.6516896486282349,
-      "learning_rate": 1.7758564873564955e-05,
-      "loss": 1.0219,
+      "epoch": 0.020241135920463595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3485818847209516e-05,
+      "loss": 1.1558,
       "step": 737
     },
     {
-      "epoch": 0.2411764705882353,
-      "grad_norm": 1.756772756576538,
-      "learning_rate": 1.775188242727719e-05,
-      "loss": 1.0607,
+      "epoch": 0.02026860014830683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3504117108874657e-05,
+      "loss": 1.0939,
       "step": 738
     },
     {
-      "epoch": 0.2415032679738562,
-      "grad_norm": 1.6265536546707153,
-      "learning_rate": 1.774519129580412e-05,
-      "loss": 1.0215,
+      "epoch": 0.020296064376150065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3522415370539801e-05,
+      "loss": 1.1903,
       "step": 739
     },
     {
-      "epoch": 0.24183006535947713,
-      "grad_norm": 1.7116483449935913,
-      "learning_rate": 1.773849148664247e-05,
-      "loss": 1.0445,
+      "epoch": 0.0203235286039933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3540713632204942e-05,
+      "loss": 1.1313,
       "step": 740
     },
     {
-      "epoch": 0.24215686274509804,
-      "grad_norm": 1.7070757150650024,
-      "learning_rate": 1.773178300729869e-05,
-      "loss": 1.0113,
+      "epoch": 0.020350992831836532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3559011893870085e-05,
+      "loss": 1.0938,
       "step": 741
     },
     {
-      "epoch": 0.24248366013071895,
-      "grad_norm": 1.6940354108810425,
-      "learning_rate": 1.772506586528893e-05,
-      "loss": 1.056,
+      "epoch": 0.020378457059679767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3577310155535226e-05,
+      "loss": 1.1392,
       "step": 742
     },
     {
-      "epoch": 0.24281045751633987,
-      "grad_norm": 1.8077305555343628,
-      "learning_rate": 1.7718340068139066e-05,
-      "loss": 1.0371,
+      "epoch": 0.020405921287523002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3595608417200368e-05,
+      "loss": 1.2002,
       "step": 743
     },
     {
-      "epoch": 0.24313725490196078,
-      "grad_norm": 1.7767380475997925,
-      "learning_rate": 1.7711605623384653e-05,
-      "loss": 1.0185,
+      "epoch": 0.020433385515366234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361390667886551e-05,
+      "loss": 1.0886,
       "step": 744
     },
     {
-      "epoch": 0.2434640522875817,
-      "grad_norm": 1.7848281860351562,
-      "learning_rate": 1.7704862538570948e-05,
-      "loss": 1.0508,
+      "epoch": 0.02046084974320947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.363220494053065e-05,
+      "loss": 1.1925,
       "step": 745
     },
     {
-      "epoch": 0.2437908496732026,
-      "grad_norm": 1.8287874460220337,
-      "learning_rate": 1.7698110821252873e-05,
-      "loss": 1.0709,
+      "epoch": 0.020488313971052705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3650503202195793e-05,
+      "loss": 1.0892,
       "step": 746
     },
     {
-      "epoch": 0.24411764705882352,
-      "grad_norm": 1.7005698680877686,
-      "learning_rate": 1.769135047899504e-05,
-      "loss": 0.9885,
+      "epoch": 0.020515778198895936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3668801463860934e-05,
+      "loss": 1.1697,
       "step": 747
     },
     {
-      "epoch": 0.24444444444444444,
-      "grad_norm": 1.6948485374450684,
-      "learning_rate": 1.7684581519371714e-05,
-      "loss": 0.9797,
+      "epoch": 0.02054324242673917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3687099725526076e-05,
+      "loss": 1.108,
       "step": 748
     },
     {
-      "epoch": 0.24477124183006535,
-      "grad_norm": 1.7270915508270264,
-      "learning_rate": 1.7677803949966817e-05,
-      "loss": 1.0378,
+      "epoch": 0.020570706654582407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3705397987191217e-05,
+      "loss": 1.0792,
       "step": 749
     },
     {
-      "epoch": 0.24509803921568626,
-      "grad_norm": 1.6462409496307373,
-      "learning_rate": 1.7671017778373914e-05,
-      "loss": 1.0223,
+      "epoch": 0.020598170882425642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372369624885636e-05,
+      "loss": 1.0597,
       "step": 750
     },
     {
-      "epoch": 0.24542483660130718,
-      "grad_norm": 1.623030662536621,
-      "learning_rate": 1.7664223012196213e-05,
-      "loss": 1.0567,
+      "epoch": 0.020625635110268874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3741994510521501e-05,
+      "loss": 1.1559,
       "step": 751
     },
     {
-      "epoch": 0.2457516339869281,
-      "grad_norm": 1.6456663608551025,
-      "learning_rate": 1.7657419659046555e-05,
-      "loss": 1.0156,
+      "epoch": 0.02065309933811211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3760292772186644e-05,
+      "loss": 1.0769,
       "step": 752
     },
     {
-      "epoch": 0.246078431372549,
-      "grad_norm": 1.6570730209350586,
-      "learning_rate": 1.7650607726547382e-05,
-      "loss": 1.019,
+      "epoch": 0.020680563565955344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3778591033851784e-05,
+      "loss": 1.2125,
       "step": 753
     },
     {
-      "epoch": 0.24640522875816995,
-      "grad_norm": 1.6553071737289429,
-      "learning_rate": 1.7643787222330773e-05,
-      "loss": 1.0129,
+      "epoch": 0.020708027793798576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3796889295516927e-05,
+      "loss": 1.083,
       "step": 754
     },
     {
-      "epoch": 0.24673202614379086,
-      "grad_norm": 1.8205479383468628,
-      "learning_rate": 1.7636958154038395e-05,
-      "loss": 1.0213,
+      "epoch": 0.02073549202164181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3815187557182068e-05,
+      "loss": 1.1001,
       "step": 755
     },
     {
-      "epoch": 0.24705882352941178,
-      "grad_norm": 1.6589878797531128,
-      "learning_rate": 1.7630120529321518e-05,
-      "loss": 1.0122,
+      "epoch": 0.020762956249485046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.383348581884721e-05,
+      "loss": 1.1125,
       "step": 756
     },
     {
-      "epoch": 0.2473856209150327,
-      "grad_norm": 1.6706461906433105,
-      "learning_rate": 1.7623274355841e-05,
-      "loss": 1.0316,
+      "epoch": 0.02079042047732828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3851784080512352e-05,
+      "loss": 1.1639,
       "step": 757
     },
     {
-      "epoch": 0.2477124183006536,
-      "grad_norm": 1.7780125141143799,
-      "learning_rate": 1.7616419641267267e-05,
-      "loss": 1.0246,
+      "epoch": 0.020817884705171513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3870082342177492e-05,
+      "loss": 1.1848,
       "step": 758
     },
     {
-      "epoch": 0.24803921568627452,
-      "grad_norm": 1.572438359260559,
-      "learning_rate": 1.760955639328032e-05,
-      "loss": 1.0018,
+      "epoch": 0.02084534893301475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3888380603842635e-05,
+      "loss": 1.0966,
       "step": 759
     },
     {
-      "epoch": 0.24836601307189543,
-      "grad_norm": 1.8777406215667725,
-      "learning_rate": 1.760268461956972e-05,
-      "loss": 1.0342,
+      "epoch": 0.020872813160857984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.390667886550778e-05,
+      "loss": 1.1577,
       "step": 760
     },
     {
-      "epoch": 0.24869281045751634,
-      "grad_norm": 1.6013562679290771,
-      "learning_rate": 1.759580432783458e-05,
-      "loss": 1.0187,
+      "epoch": 0.020900277388701215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.392497712717292e-05,
+      "loss": 1.0924,
       "step": 761
     },
     {
-      "epoch": 0.24901960784313726,
-      "grad_norm": 1.7960928678512573,
-      "learning_rate": 1.758891552578357e-05,
-      "loss": 1.0332,
+      "epoch": 0.02092774161654445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3943275388838063e-05,
+      "loss": 1.1098,
       "step": 762
     },
     {
-      "epoch": 0.24934640522875817,
-      "grad_norm": 1.7823951244354248,
-      "learning_rate": 1.758201822113487e-05,
-      "loss": 1.0368,
+      "epoch": 0.020955205844387686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3961573650503204e-05,
+      "loss": 1.0726,
       "step": 763
     },
     {
-      "epoch": 0.2496732026143791,
-      "grad_norm": 1.6668277978897095,
-      "learning_rate": 1.7575112421616203e-05,
-      "loss": 1.0238,
+      "epoch": 0.020982670072230918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3979871912168346e-05,
+      "loss": 1.0445,
       "step": 764
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 1.6606330871582031,
-      "learning_rate": 1.7568198134964804e-05,
-      "loss": 1.0696,
+      "epoch": 0.021010134300074153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3998170173833487e-05,
+      "loss": 1.0392,
       "step": 765
     },
     {
-      "epoch": 0.2503267973856209,
-      "grad_norm": 1.8395094871520996,
-      "learning_rate": 1.7561275368927422e-05,
-      "loss": 1.008,
+      "epoch": 0.021037598527917388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4016468435498628e-05,
+      "loss": 1.0778,
       "step": 766
     },
     {
-      "epoch": 0.2506535947712418,
-      "grad_norm": 1.7406120300292969,
-      "learning_rate": 1.75543441312603e-05,
-      "loss": 1.0424,
+      "epoch": 0.021065062755760623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4034766697163771e-05,
+      "loss": 1.1353,
       "step": 767
     },
     {
-      "epoch": 0.25098039215686274,
-      "grad_norm": 1.6293975114822388,
-      "learning_rate": 1.7547404429729174e-05,
-      "loss": 1.0634,
+      "epoch": 0.021092526983603855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4053064958828912e-05,
+      "loss": 1.0894,
       "step": 768
     },
     {
-      "epoch": 0.25130718954248366,
-      "grad_norm": 1.7319899797439575,
-      "learning_rate": 1.754045627210927e-05,
-      "loss": 1.0461,
+      "epoch": 0.02111999121144709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4071363220494054e-05,
+      "loss": 1.1312,
       "step": 769
     },
     {
-      "epoch": 0.25163398692810457,
-      "grad_norm": 1.7042118310928345,
-      "learning_rate": 1.753349966618528e-05,
-      "loss": 1.037,
+      "epoch": 0.021147455439290325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4089661482159195e-05,
+      "loss": 1.1389,
       "step": 770
     },
     {
-      "epoch": 0.2519607843137255,
-      "grad_norm": 1.6568604707717896,
-      "learning_rate": 1.752653461975136e-05,
-      "loss": 1.0101,
+      "epoch": 0.021174919667133557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4107959743824338e-05,
+      "loss": 1.1629,
       "step": 771
     },
     {
-      "epoch": 0.2522875816993464,
-      "grad_norm": 1.6463642120361328,
-      "learning_rate": 1.751956114061113e-05,
-      "loss": 1.0309,
+      "epoch": 0.021202383894976792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4126258005489479e-05,
+      "loss": 1.1049,
       "step": 772
     },
     {
-      "epoch": 0.2526143790849673,
-      "grad_norm": 1.621003270149231,
-      "learning_rate": 1.7512579236577652e-05,
-      "loss": 1.0548,
+      "epoch": 0.021229848122820028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4144556267154622e-05,
+      "loss": 1.1712,
       "step": 773
     },
     {
-      "epoch": 0.2529411764705882,
-      "grad_norm": 1.530648112297058,
-      "learning_rate": 1.7505588915473437e-05,
-      "loss": 1.0229,
+      "epoch": 0.021257312350663263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4162854528819762e-05,
+      "loss": 1.1244,
       "step": 774
     },
     {
-      "epoch": 0.25326797385620914,
-      "grad_norm": 1.717400074005127,
-      "learning_rate": 1.7498590185130413e-05,
-      "loss": 1.0334,
+      "epoch": 0.021284776578506494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4181152790484905e-05,
+      "loss": 1.0885,
       "step": 775
     },
     {
-      "epoch": 0.25359477124183005,
-      "grad_norm": 1.6760605573654175,
-      "learning_rate": 1.7491583053389937e-05,
-      "loss": 1.0329,
+      "epoch": 0.02131224080634973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4199451052150046e-05,
+      "loss": 1.1843,
       "step": 776
     },
     {
-      "epoch": 0.25392156862745097,
-      "grad_norm": 1.762571096420288,
-      "learning_rate": 1.7484567528102778e-05,
-      "loss": 1.0311,
+      "epoch": 0.021339705034192965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4217749313815189e-05,
+      "loss": 1.1932,
       "step": 777
     },
     {
-      "epoch": 0.2542483660130719,
-      "grad_norm": 1.6813228130340576,
-      "learning_rate": 1.747754361712911e-05,
-      "loss": 1.0476,
+      "epoch": 0.021367169262036197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.423604757548033e-05,
+      "loss": 1.1577,
       "step": 778
     },
     {
-      "epoch": 0.2545751633986928,
-      "grad_norm": 1.6044751405715942,
-      "learning_rate": 1.7470511328338503e-05,
-      "loss": 1.0144,
+      "epoch": 0.021394633489879432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.425434583714547e-05,
+      "loss": 1.0853,
       "step": 779
     },
     {
-      "epoch": 0.2549019607843137,
-      "grad_norm": 1.6795539855957031,
-      "learning_rate": 1.7463470669609907e-05,
-      "loss": 1.0448,
+      "epoch": 0.021422097717722667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4272644098810615e-05,
+      "loss": 1.095,
       "step": 780
     },
     {
-      "epoch": 0.2552287581699346,
-      "grad_norm": 1.6492314338684082,
-      "learning_rate": 1.7456421648831658e-05,
-      "loss": 0.9995,
+      "epoch": 0.0214495619455659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4290942360475757e-05,
+      "loss": 1.1033,
       "step": 781
     },
     {
-      "epoch": 0.25555555555555554,
-      "grad_norm": 1.4694383144378662,
-      "learning_rate": 1.7449364273901457e-05,
-      "loss": 1.054,
+      "epoch": 0.021477026173409134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4309240622140898e-05,
+      "loss": 1.2321,
       "step": 782
     },
     {
-      "epoch": 0.25588235294117645,
-      "grad_norm": 1.5415890216827393,
-      "learning_rate": 1.744229855272636e-05,
-      "loss": 1.0349,
+      "epoch": 0.02150449040125237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4327538883806041e-05,
+      "loss": 1.1648,
       "step": 783
     },
     {
-      "epoch": 0.25620915032679736,
-      "grad_norm": 1.713314175605774,
-      "learning_rate": 1.743522449322279e-05,
-      "loss": 1.0253,
+      "epoch": 0.021531954629095604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4345837145471182e-05,
+      "loss": 1.119,
       "step": 784
     },
     {
-      "epoch": 0.2565359477124183,
-      "grad_norm": 1.6568752527236938,
-      "learning_rate": 1.7428142103316493e-05,
-      "loss": 1.0232,
+      "epoch": 0.021559418856938836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4364135407136324e-05,
+      "loss": 1.1653,
       "step": 785
     },
     {
-      "epoch": 0.2568627450980392,
-      "grad_norm": 1.6191929578781128,
-      "learning_rate": 1.742105139094256e-05,
-      "loss": 1.0167,
+      "epoch": 0.02158688308478207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4382433668801465e-05,
+      "loss": 1.1479,
       "step": 786
     },
     {
-      "epoch": 0.2571895424836601,
-      "grad_norm": 1.6805230379104614,
-      "learning_rate": 1.7413952364045404e-05,
-      "loss": 0.9957,
+      "epoch": 0.021614347312625307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4400731930466606e-05,
+      "loss": 1.0477,
       "step": 787
     },
     {
-      "epoch": 0.257516339869281,
-      "grad_norm": 1.6259199380874634,
-      "learning_rate": 1.7406845030578747e-05,
-      "loss": 0.9989,
+      "epoch": 0.02164181154046854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4419030192131749e-05,
+      "loss": 1.1046,
       "step": 788
     },
     {
-      "epoch": 0.257843137254902,
-      "grad_norm": 1.6554555892944336,
-      "learning_rate": 1.7399729398505632e-05,
-      "loss": 1.0148,
+      "epoch": 0.021669275768311774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.443732845379689e-05,
+      "loss": 1.0969,
       "step": 789
     },
     {
-      "epoch": 0.2581699346405229,
-      "grad_norm": 1.705592155456543,
-      "learning_rate": 1.7392605475798387e-05,
-      "loss": 1.0408,
+      "epoch": 0.02169673999615501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455626715462032e-05,
+      "loss": 1.0845,
       "step": 790
     },
     {
-      "epoch": 0.2584967320261438,
-      "grad_norm": 1.6696909666061401,
-      "learning_rate": 1.7385473270438634e-05,
-      "loss": 1.0342,
+      "epoch": 0.021724204223998244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4473924977127173e-05,
+      "loss": 1.1139,
       "step": 791
     },
     {
-      "epoch": 0.25882352941176473,
-      "grad_norm": 1.5549250841140747,
-      "learning_rate": 1.7378332790417275e-05,
-      "loss": 1.013,
+      "epoch": 0.021751668451841476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4492223238792316e-05,
+      "loss": 1.1273,
       "step": 792
     },
     {
-      "epoch": 0.25915032679738564,
-      "grad_norm": 1.6281051635742188,
-      "learning_rate": 1.737118404373448e-05,
-      "loss": 1.0206,
+      "epoch": 0.02177913267968471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4510521500457457e-05,
+      "loss": 1.231,
       "step": 793
     },
     {
-      "epoch": 0.25947712418300656,
-      "grad_norm": 1.6301109790802002,
-      "learning_rate": 1.7364027038399682e-05,
-      "loss": 1.0341,
+      "epoch": 0.021806596907527946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.45288197621226e-05,
+      "loss": 1.1609,
       "step": 794
     },
     {
-      "epoch": 0.25980392156862747,
-      "grad_norm": 1.6722818613052368,
-      "learning_rate": 1.735686178243157e-05,
-      "loss": 1.0523,
+      "epoch": 0.021834061135371178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.454711802378774e-05,
+      "loss": 1.1396,
       "step": 795
     },
     {
-      "epoch": 0.2601307189542484,
-      "grad_norm": 1.629975438117981,
-      "learning_rate": 1.734968828385808e-05,
-      "loss": 1.0429,
+      "epoch": 0.021861525363214413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4565416285452883e-05,
+      "loss": 1.0471,
       "step": 796
     },
     {
-      "epoch": 0.2604575163398693,
-      "grad_norm": 1.7229149341583252,
-      "learning_rate": 1.7342506550716365e-05,
-      "loss": 1.0546,
+      "epoch": 0.02188898959105765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4583714547118024e-05,
+      "loss": 1.0128,
       "step": 797
     },
     {
-      "epoch": 0.2607843137254902,
-      "grad_norm": 1.6349531412124634,
-      "learning_rate": 1.7335316591052828e-05,
-      "loss": 1.0118,
+      "epoch": 0.02191645381890088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4602012808783167e-05,
+      "loss": 1.1535,
       "step": 798
     },
     {
-      "epoch": 0.2611111111111111,
-      "grad_norm": 1.606771469116211,
-      "learning_rate": 1.7328118412923073e-05,
-      "loss": 0.9997,
+      "epoch": 0.021943918046744115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4620311070448308e-05,
+      "loss": 1.1481,
       "step": 799
     },
     {
-      "epoch": 0.26143790849673204,
-      "grad_norm": 1.6385829448699951,
-      "learning_rate": 1.7320912024391915e-05,
-      "loss": 1.0574,
+      "epoch": 0.02197138227458735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4638609332113452e-05,
+      "loss": 1.1784,
       "step": 800
     },
     {
-      "epoch": 0.26176470588235295,
-      "grad_norm": 1.5211986303329468,
-      "learning_rate": 1.7313697433533375e-05,
-      "loss": 1.0019,
+      "epoch": 0.021998846502430586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4656907593778593e-05,
+      "loss": 1.1599,
       "step": 801
     },
     {
-      "epoch": 0.26209150326797387,
-      "grad_norm": 1.6180814504623413,
-      "learning_rate": 1.7306474648430653e-05,
-      "loss": 1.0077,
+      "epoch": 0.022026310730273817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4675205855443735e-05,
+      "loss": 1.1361,
       "step": 802
     },
     {
-      "epoch": 0.2624183006535948,
-      "grad_norm": 1.6715242862701416,
-      "learning_rate": 1.7299243677176142e-05,
-      "loss": 1.0264,
+      "epoch": 0.022053774958117053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4693504117108876e-05,
+      "loss": 0.9666,
       "step": 803
     },
     {
-      "epoch": 0.2627450980392157,
-      "grad_norm": 1.5747742652893066,
-      "learning_rate": 1.729200452787139e-05,
-      "loss": 1.0489,
+      "epoch": 0.022081239185960288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4711802378774019e-05,
+      "loss": 1.1938,
       "step": 804
     },
     {
-      "epoch": 0.2630718954248366,
-      "grad_norm": 1.5528409481048584,
-      "learning_rate": 1.728475720862713e-05,
-      "loss": 1.0295,
+      "epoch": 0.02210870341380352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473010064043916e-05,
+      "loss": 1.1417,
       "step": 805
     },
     {
-      "epoch": 0.2633986928104575,
-      "grad_norm": 1.5615087747573853,
-      "learning_rate": 1.7277501727563227e-05,
-      "loss": 1.0,
+      "epoch": 0.022136167641646755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4748398902104303e-05,
+      "loss": 1.142,
       "step": 806
     },
     {
-      "epoch": 0.26372549019607844,
-      "grad_norm": 1.7250020503997803,
-      "learning_rate": 1.72702380928087e-05,
-      "loss": 1.0619,
+      "epoch": 0.02216363186948999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4766697163769443e-05,
+      "loss": 1.1666,
       "step": 807
     },
     {
-      "epoch": 0.26405228758169935,
-      "grad_norm": 1.4902994632720947,
-      "learning_rate": 1.726296631250171e-05,
-      "loss": 1.0537,
+      "epoch": 0.022191096097333222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4784995425434584e-05,
+      "loss": 1.0358,
       "step": 808
     },
     {
-      "epoch": 0.26437908496732027,
-      "grad_norm": 1.6166187524795532,
-      "learning_rate": 1.725568639478953e-05,
-      "loss": 0.9839,
+      "epoch": 0.022218560325176457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4803293687099727e-05,
+      "loss": 1.0477,
       "step": 809
     },
     {
-      "epoch": 0.2647058823529412,
-      "grad_norm": 1.5854495763778687,
-      "learning_rate": 1.724839834782856e-05,
-      "loss": 0.994,
+      "epoch": 0.022246024553019692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4821591948764868e-05,
+      "loss": 1.1458,
       "step": 810
     },
     {
-      "epoch": 0.2650326797385621,
-      "grad_norm": 1.6926088333129883,
-      "learning_rate": 1.7241102179784307e-05,
-      "loss": 1.0363,
+      "epoch": 0.022273488780862927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.483989021043001e-05,
+      "loss": 1.1716,
       "step": 811
     },
     {
-      "epoch": 0.265359477124183,
-      "grad_norm": 1.6773474216461182,
-      "learning_rate": 1.7233797898831376e-05,
-      "loss": 1.054,
+      "epoch": 0.02230095300870616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4858188472095151e-05,
+      "loss": 1.184,
       "step": 812
     },
     {
-      "epoch": 0.2656862745098039,
-      "grad_norm": 1.6152057647705078,
-      "learning_rate": 1.7226485513153458e-05,
-      "loss": 1.0234,
+      "epoch": 0.022328417236549394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4876486733760294e-05,
+      "loss": 1.1531,
       "step": 813
     },
     {
-      "epoch": 0.26601307189542484,
-      "grad_norm": 1.7147127389907837,
-      "learning_rate": 1.7219165030943327e-05,
-      "loss": 0.9807,
+      "epoch": 0.02235588146439263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4894784995425435e-05,
+      "loss": 0.9835,
       "step": 814
     },
     {
-      "epoch": 0.26633986928104575,
-      "grad_norm": 1.6351487636566162,
-      "learning_rate": 1.7211836460402833e-05,
-      "loss": 1.0252,
+      "epoch": 0.02238334569223586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4913083257090578e-05,
+      "loss": 1.1466,
       "step": 815
     },
     {
-      "epoch": 0.26666666666666666,
-      "grad_norm": 1.614640235900879,
-      "learning_rate": 1.720449980974288e-05,
-      "loss": 1.0161,
+      "epoch": 0.022410809920079097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4931381518755719e-05,
+      "loss": 1.2089,
       "step": 816
     },
     {
-      "epoch": 0.2669934640522876,
-      "grad_norm": 1.6843795776367188,
-      "learning_rate": 1.719715508718343e-05,
-      "loss": 1.0294,
+      "epoch": 0.022438274147922332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4949679780420861e-05,
+      "loss": 1.1654,
       "step": 817
     },
     {
-      "epoch": 0.2673202614379085,
-      "grad_norm": 1.5865646600723267,
-      "learning_rate": 1.7189802300953488e-05,
-      "loss": 1.0456,
+      "epoch": 0.022465738375765567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4967978042086002e-05,
+      "loss": 1.0409,
       "step": 818
     },
     {
-      "epoch": 0.2676470588235294,
-      "grad_norm": 1.6357495784759521,
-      "learning_rate": 1.7182441459291095e-05,
-      "loss": 1.0312,
+      "epoch": 0.0224932026036088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4986276303751143e-05,
+      "loss": 1.1489,
       "step": 819
     },
     {
-      "epoch": 0.2679738562091503,
-      "grad_norm": 1.5449678897857666,
-      "learning_rate": 1.717507257044331e-05,
-      "loss": 1.0059,
+      "epoch": 0.022520666831452034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5004574565416287e-05,
+      "loss": 1.1263,
       "step": 820
     },
     {
-      "epoch": 0.26830065359477123,
-      "grad_norm": 1.6269075870513916,
-      "learning_rate": 1.7167695642666223e-05,
-      "loss": 1.0141,
+      "epoch": 0.02254813105929527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502287282708143e-05,
+      "loss": 1.1682,
       "step": 821
     },
     {
-      "epoch": 0.26862745098039215,
-      "grad_norm": 1.6104711294174194,
-      "learning_rate": 1.716031068422491e-05,
-      "loss": 1.0554,
+      "epoch": 0.0225755952871385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504117108874657e-05,
+      "loss": 1.2006,
       "step": 822
     },
     {
-      "epoch": 0.26895424836601306,
-      "grad_norm": 1.5809839963912964,
-      "learning_rate": 1.7152917703393462e-05,
-      "loss": 1.0014,
+      "epoch": 0.022603059514981736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5059469350411713e-05,
+      "loss": 1.1069,
       "step": 823
     },
     {
-      "epoch": 0.269281045751634,
-      "grad_norm": 1.5239191055297852,
-      "learning_rate": 1.7145516708454953e-05,
-      "loss": 1.007,
+      "epoch": 0.02263052374282497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5077767612076854e-05,
+      "loss": 1.1192,
       "step": 824
     },
     {
-      "epoch": 0.2696078431372549,
-      "grad_norm": 1.63080632686615,
-      "learning_rate": 1.7138107707701435e-05,
-      "loss": 0.9768,
+      "epoch": 0.022657987970668203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5096065873741997e-05,
+      "loss": 1.0819,
       "step": 825
     },
     {
-      "epoch": 0.2699346405228758,
-      "grad_norm": 1.5647321939468384,
-      "learning_rate": 1.7130690709433927e-05,
-      "loss": 1.0836,
+      "epoch": 0.022685452198511438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5114364135407138e-05,
+      "loss": 1.1081,
       "step": 826
     },
     {
-      "epoch": 0.2702614379084967,
-      "grad_norm": 1.548670768737793,
-      "learning_rate": 1.712326572196241e-05,
-      "loss": 1.0137,
+      "epoch": 0.022712916426354673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5132662397072279e-05,
+      "loss": 1.1073,
       "step": 827
     },
     {
-      "epoch": 0.27058823529411763,
-      "grad_norm": 1.7309716939926147,
-      "learning_rate": 1.711583275360582e-05,
-      "loss": 1.0075,
+      "epoch": 0.02274038065419791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5150960658737421e-05,
+      "loss": 1.1503,
       "step": 828
     },
     {
-      "epoch": 0.27091503267973854,
-      "grad_norm": 1.5892889499664307,
-      "learning_rate": 1.7108391812692032e-05,
-      "loss": 1.0264,
+      "epoch": 0.02276784488204114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5169258920402562e-05,
+      "loss": 1.0576,
       "step": 829
     },
     {
-      "epoch": 0.27124183006535946,
-      "grad_norm": 1.5074502229690552,
-      "learning_rate": 1.710094290755785e-05,
-      "loss": 0.9926,
+      "epoch": 0.022795309109884376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5187557182067705e-05,
+      "loss": 1.2308,
       "step": 830
     },
     {
-      "epoch": 0.27156862745098037,
-      "grad_norm": 1.5750449895858765,
-      "learning_rate": 1.709348604654901e-05,
-      "loss": 1.0232,
+      "epoch": 0.02282277333772761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5205855443732846e-05,
+      "loss": 1.1288,
       "step": 831
     },
     {
-      "epoch": 0.2718954248366013,
-      "grad_norm": 1.4483888149261475,
-      "learning_rate": 1.7086021238020154e-05,
-      "loss": 0.9978,
+      "epoch": 0.022850237565570843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5224153705397989e-05,
+      "loss": 1.1146,
       "step": 832
     },
     {
-      "epoch": 0.2722222222222222,
-      "grad_norm": 1.6392958164215088,
-      "learning_rate": 1.7078548490334824e-05,
-      "loss": 1.0316,
+      "epoch": 0.022877701793414078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524245196706313e-05,
+      "loss": 1.1626,
       "step": 833
     },
     {
-      "epoch": 0.2725490196078431,
-      "grad_norm": 1.6055431365966797,
-      "learning_rate": 1.7071067811865477e-05,
-      "loss": 1.0118,
+      "epoch": 0.022905166021257313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5260750228728272e-05,
+      "loss": 1.1212,
       "step": 834
     },
     {
-      "epoch": 0.272875816993464,
-      "grad_norm": 1.6640633344650269,
-      "learning_rate": 1.706357921099343e-05,
-      "loss": 1.03,
+      "epoch": 0.022932630249100548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5279048490393413e-05,
+      "loss": 1.1753,
       "step": 835
     },
     {
-      "epoch": 0.27320261437908494,
-      "grad_norm": 1.5221925973892212,
-      "learning_rate": 1.7056082696108896e-05,
-      "loss": 1.0113,
+      "epoch": 0.02296009447694378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5297346752058554e-05,
+      "loss": 1.0453,
       "step": 836
     },
     {
-      "epoch": 0.2735294117647059,
-      "grad_norm": 1.5559171438217163,
-      "learning_rate": 1.7048578275610943e-05,
-      "loss": 1.0233,
+      "epoch": 0.022987558704787015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5315645013723698e-05,
+      "loss": 1.155,
       "step": 837
     },
     {
-      "epoch": 0.2738562091503268,
-      "grad_norm": 1.573848843574524,
-      "learning_rate": 1.70410659579075e-05,
-      "loss": 1.0543,
+      "epoch": 0.02301502293263025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533394327538884e-05,
+      "loss": 1.1371,
       "step": 838
     },
     {
-      "epoch": 0.27418300653594774,
-      "grad_norm": 1.6698905229568481,
-      "learning_rate": 1.703354575141535e-05,
-      "loss": 1.0716,
+      "epoch": 0.023042487160473482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535224153705398e-05,
+      "loss": 1.1992,
       "step": 839
     },
     {
-      "epoch": 0.27450980392156865,
-      "grad_norm": 1.5602167844772339,
-      "learning_rate": 1.7026017664560108e-05,
-      "loss": 1.0193,
+      "epoch": 0.023069951388316717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5370539798719124e-05,
+      "loss": 0.9907,
       "step": 840
     },
     {
-      "epoch": 0.27483660130718957,
-      "grad_norm": 1.5966531038284302,
-      "learning_rate": 1.701848170577621e-05,
-      "loss": 1.022,
+      "epoch": 0.023097415616159953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5388838060384265e-05,
+      "loss": 1.019,
       "step": 841
     },
     {
-      "epoch": 0.2751633986928105,
-      "grad_norm": 1.48743736743927,
-      "learning_rate": 1.7010937883506936e-05,
-      "loss": 0.9867,
+      "epoch": 0.023124879844003184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5407136322049406e-05,
+      "loss": 1.0985,
       "step": 842
     },
     {
-      "epoch": 0.2754901960784314,
-      "grad_norm": 1.4938156604766846,
-      "learning_rate": 1.7003386206204352e-05,
-      "loss": 0.9761,
+      "epoch": 0.02315234407184642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542543458371455e-05,
+      "loss": 1.1273,
       "step": 843
     },
     {
-      "epoch": 0.2758169934640523,
-      "grad_norm": 1.5539753437042236,
-      "learning_rate": 1.699582668232934e-05,
-      "loss": 0.9902,
+      "epoch": 0.023179808299689655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544373284537969e-05,
+      "loss": 1.0858,
       "step": 844
     },
     {
-      "epoch": 0.2761437908496732,
-      "grad_norm": 1.5759625434875488,
-      "learning_rate": 1.6988259320351556e-05,
-      "loss": 1.0159,
+      "epoch": 0.02320727252753289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5462031107044832e-05,
+      "loss": 1.1279,
       "step": 845
     },
     {
-      "epoch": 0.27647058823529413,
-      "grad_norm": 1.5943830013275146,
-      "learning_rate": 1.698068412874946e-05,
-      "loss": 1.036,
+      "epoch": 0.02323473675537612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5480329368709973e-05,
+      "loss": 1.2077,
       "step": 846
     },
     {
-      "epoch": 0.27679738562091505,
-      "grad_norm": 1.5464445352554321,
-      "learning_rate": 1.697310111601027e-05,
-      "loss": 1.0348,
+      "epoch": 0.023262200983219357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5498627630375118e-05,
+      "loss": 1.1911,
       "step": 847
     },
     {
-      "epoch": 0.27712418300653596,
-      "grad_norm": 1.5485732555389404,
-      "learning_rate": 1.6965510290629973e-05,
-      "loss": 1.0245,
+      "epoch": 0.023289665211062592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551692589204026e-05,
+      "loss": 1.1707,
       "step": 848
     },
     {
-      "epoch": 0.2774509803921569,
-      "grad_norm": 1.5053657293319702,
-      "learning_rate": 1.69579116611133e-05,
-      "loss": 1.0167,
+      "epoch": 0.023317129438905824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55352241537054e-05,
+      "loss": 1.1658,
       "step": 849
     },
     {
-      "epoch": 0.2777777777777778,
-      "grad_norm": 1.6022082567214966,
-      "learning_rate": 1.695030523597374e-05,
-      "loss": 1.0272,
+      "epoch": 0.02334459366674906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.555352241537054e-05,
+      "loss": 1.1459,
       "step": 850
     },
     {
-      "epoch": 0.2781045751633987,
-      "grad_norm": 1.5617531538009644,
-      "learning_rate": 1.69426910237335e-05,
-      "loss": 1.0108,
+      "epoch": 0.023372057894592294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557182067703568e-05,
+      "loss": 1.1191,
       "step": 851
     },
     {
-      "epoch": 0.2784313725490196,
-      "grad_norm": 1.5922645330429077,
-      "learning_rate": 1.6935069032923525e-05,
-      "loss": 1.0471,
+      "epoch": 0.02339952212243553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5590118938700826e-05,
+      "loss": 1.1304,
       "step": 852
     },
     {
-      "epoch": 0.27875816993464053,
-      "grad_norm": 1.4624428749084473,
-      "learning_rate": 1.6927439272083473e-05,
-      "loss": 1.0154,
+      "epoch": 0.02342698635027876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608417200365967e-05,
+      "loss": 1.2095,
       "step": 853
     },
     {
-      "epoch": 0.27908496732026145,
-      "grad_norm": 1.5334537029266357,
-      "learning_rate": 1.6919801749761696e-05,
-      "loss": 1.0001,
+      "epoch": 0.023454450578121996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5626715462031107e-05,
+      "loss": 1.0498,
       "step": 854
     },
     {
-      "epoch": 0.27941176470588236,
-      "grad_norm": 1.5436700582504272,
-      "learning_rate": 1.6912156474515264e-05,
-      "loss": 0.9877,
+      "epoch": 0.02348191480596523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564501372369625e-05,
+      "loss": 1.2312,
       "step": 855
     },
     {
-      "epoch": 0.2797385620915033,
-      "grad_norm": 1.4679033756256104,
-      "learning_rate": 1.6904503454909905e-05,
-      "loss": 1.0313,
+      "epoch": 0.023509379033808463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5663311985361393e-05,
+      "loss": 1.0791,
       "step": 856
     },
     {
-      "epoch": 0.2800653594771242,
-      "grad_norm": 1.5296549797058105,
-      "learning_rate": 1.6896842699520047e-05,
-      "loss": 1.0455,
+      "epoch": 0.0235368432616517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5681610247026534e-05,
+      "loss": 1.1495,
       "step": 857
     },
     {
-      "epoch": 0.2803921568627451,
-      "grad_norm": 1.4714224338531494,
-      "learning_rate": 1.6889174216928776e-05,
-      "loss": 1.0295,
+      "epoch": 0.023564307489494934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5699908508691675e-05,
+      "loss": 1.1576,
       "step": 858
     },
     {
-      "epoch": 0.280718954248366,
-      "grad_norm": 1.529443383216858,
-      "learning_rate": 1.688149801572784e-05,
-      "loss": 1.0272,
+      "epoch": 0.023591771717338166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5718206770356816e-05,
+      "loss": 1.0969,
       "step": 859
     },
     {
-      "epoch": 0.28104575163398693,
-      "grad_norm": 1.5261564254760742,
-      "learning_rate": 1.6873814104517617e-05,
-      "loss": 1.03,
+      "epoch": 0.0236192359451814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5736505032021956e-05,
+      "loss": 1.1946,
       "step": 860
     },
     {
-      "epoch": 0.28137254901960784,
-      "grad_norm": 1.6551365852355957,
-      "learning_rate": 1.686612249190716e-05,
-      "loss": 1.0346,
+      "epoch": 0.023646700173024636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57548032936871e-05,
+      "loss": 1.3373,
       "step": 861
     },
     {
-      "epoch": 0.28169934640522876,
-      "grad_norm": 1.496591567993164,
-      "learning_rate": 1.6858423186514108e-05,
-      "loss": 0.9973,
+      "epoch": 0.02367416440086787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5773101555352245e-05,
+      "loss": 1.1577,
       "step": 862
     },
     {
-      "epoch": 0.28202614379084967,
-      "grad_norm": 1.6012212038040161,
-      "learning_rate": 1.6850716196964744e-05,
-      "loss": 0.9988,
+      "epoch": 0.023701628628711103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5791399817017386e-05,
+      "loss": 1.101,
       "step": 863
     },
     {
-      "epoch": 0.2823529411764706,
-      "grad_norm": 1.6853113174438477,
-      "learning_rate": 1.6843001531893967e-05,
-      "loss": 1.0706,
+      "epoch": 0.023729092856554338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5809698078682527e-05,
+      "loss": 1.1054,
       "step": 864
     },
     {
-      "epoch": 0.2826797385620915,
-      "grad_norm": 1.6495978832244873,
-      "learning_rate": 1.6835279199945248e-05,
-      "loss": 1.0362,
+      "epoch": 0.023756557084397573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5827996340347668e-05,
+      "loss": 1.1486,
       "step": 865
     },
     {
-      "epoch": 0.2830065359477124,
-      "grad_norm": 1.4526536464691162,
-      "learning_rate": 1.6827549209770676e-05,
-      "loss": 0.9687,
+      "epoch": 0.023784021312240805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5846294602012812e-05,
+      "loss": 1.165,
       "step": 866
     },
     {
-      "epoch": 0.2833333333333333,
-      "grad_norm": 1.5145591497421265,
-      "learning_rate": 1.68198115700309e-05,
-      "loss": 1.0103,
+      "epoch": 0.02381148554008404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5864592863677953e-05,
+      "loss": 1.2039,
       "step": 867
     },
     {
-      "epoch": 0.28366013071895424,
-      "grad_norm": 1.4702942371368408,
-      "learning_rate": 1.6812066289395157e-05,
-      "loss": 0.9974,
+      "epoch": 0.023838949767927275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5882891125343094e-05,
+      "loss": 1.1987,
       "step": 868
     },
     {
-      "epoch": 0.28398692810457515,
-      "grad_norm": 1.6325727701187134,
-      "learning_rate": 1.6804313376541226e-05,
-      "loss": 1.0129,
+      "epoch": 0.023866413995770507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5901189387008235e-05,
+      "loss": 1.1188,
       "step": 869
     },
     {
-      "epoch": 0.28431372549019607,
-      "grad_norm": 1.6266313791275024,
-      "learning_rate": 1.6796552840155453e-05,
-      "loss": 1.0029,
+      "epoch": 0.023893878223613742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5919487648673376e-05,
+      "loss": 1.1124,
       "step": 870
     },
     {
-      "epoch": 0.284640522875817,
-      "grad_norm": 1.4832236766815186,
-      "learning_rate": 1.6788784688932717e-05,
-      "loss": 1.0016,
+      "epoch": 0.023921342451456978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593778591033852e-05,
+      "loss": 1.0472,
       "step": 871
     },
     {
-      "epoch": 0.2849673202614379,
-      "grad_norm": 1.4989445209503174,
-      "learning_rate": 1.6781008931576433e-05,
-      "loss": 1.0429,
+      "epoch": 0.023948806679300213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595608417200366e-05,
+      "loss": 1.1035,
       "step": 872
     },
     {
-      "epoch": 0.2852941176470588,
-      "grad_norm": 1.6423624753952026,
-      "learning_rate": 1.677322557679853e-05,
-      "loss": 1.0228,
+      "epoch": 0.023976270907143445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5974382433668802e-05,
+      "loss": 1.1387,
       "step": 873
     },
     {
-      "epoch": 0.2856209150326797,
-      "grad_norm": 1.6856576204299927,
-      "learning_rate": 1.676543463331946e-05,
-      "loss": 0.9944,
+      "epoch": 0.02400373513498668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5992680695333943e-05,
+      "loss": 1.0483,
       "step": 874
     },
     {
-      "epoch": 0.28594771241830064,
-      "grad_norm": 1.4896456003189087,
-      "learning_rate": 1.675763610986817e-05,
-      "loss": 0.9839,
+      "epoch": 0.024031199362829915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6010978956999087e-05,
+      "loss": 1.0983,
       "step": 875
     },
     {
-      "epoch": 0.28627450980392155,
-      "grad_norm": 1.5448235273361206,
-      "learning_rate": 1.6749830015182106e-05,
-      "loss": 0.992,
+      "epoch": 0.024058663590673147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6029277218664228e-05,
+      "loss": 1.1874,
       "step": 876
     },
     {
-      "epoch": 0.28660130718954246,
-      "grad_norm": 1.5244494676589966,
-      "learning_rate": 1.6742016358007186e-05,
-      "loss": 0.9826,
+      "epoch": 0.024086127818516382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604757548032937e-05,
+      "loss": 1.0688,
       "step": 877
     },
     {
-      "epoch": 0.2869281045751634,
-      "grad_norm": 1.5417203903198242,
-      "learning_rate": 1.6734195147097815e-05,
-      "loss": 0.9968,
+      "epoch": 0.024113592046359617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606587374199451e-05,
+      "loss": 1.0107,
       "step": 878
     },
     {
-      "epoch": 0.2872549019607843,
-      "grad_norm": 1.5896886587142944,
-      "learning_rate": 1.6726366391216844e-05,
-      "loss": 0.9768,
+      "epoch": 0.024141056274202852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6084172003659654e-05,
+      "loss": 1.0971,
       "step": 879
     },
     {
-      "epoch": 0.2875816993464052,
-      "grad_norm": 1.568055510520935,
-      "learning_rate": 1.6718530099135596e-05,
-      "loss": 1.031,
+      "epoch": 0.024168520502046084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6102470265324795e-05,
+      "loss": 1.215,
       "step": 880
     },
     {
-      "epoch": 0.2879084967320261,
-      "grad_norm": 1.3753786087036133,
-      "learning_rate": 1.6710686279633824e-05,
-      "loss": 0.9701,
+      "epoch": 0.02419598472988932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612076852698994e-05,
+      "loss": 1.1262,
       "step": 881
     },
     {
-      "epoch": 0.28823529411764703,
-      "grad_norm": 1.4483654499053955,
-      "learning_rate": 1.6702834941499718e-05,
-      "loss": 1.0087,
+      "epoch": 0.024223448957732555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613906678865508e-05,
+      "loss": 1.121,
       "step": 882
     },
     {
-      "epoch": 0.28856209150326795,
-      "grad_norm": 1.605263590812683,
-      "learning_rate": 1.6694976093529896e-05,
-      "loss": 0.9845,
+      "epoch": 0.024250913185575786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615736505032022e-05,
+      "loss": 1.0925,
       "step": 883
     },
     {
-      "epoch": 0.28888888888888886,
-      "grad_norm": 1.4835011959075928,
-      "learning_rate": 1.6687109744529394e-05,
-      "loss": 0.978,
+      "epoch": 0.02427837741341902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6175663311985362e-05,
+      "loss": 0.9911,
       "step": 884
     },
     {
-      "epoch": 0.28921568627450983,
-      "grad_norm": 1.6153836250305176,
-      "learning_rate": 1.6679235903311633e-05,
-      "loss": 1.0149,
+      "epoch": 0.024305841641262257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6193961573650507e-05,
+      "loss": 1.1211,
       "step": 885
     },
     {
-      "epoch": 0.28954248366013075,
-      "grad_norm": 1.5385993719100952,
-      "learning_rate": 1.6671354578698444e-05,
-      "loss": 0.9969,
+      "epoch": 0.02433330586910549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6212259835315648e-05,
+      "loss": 1.1066,
       "step": 886
     },
     {
-      "epoch": 0.28986928104575166,
-      "grad_norm": 1.5651582479476929,
-      "learning_rate": 1.6663465779520042e-05,
-      "loss": 1.0124,
+      "epoch": 0.024360770096948724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623055809698079e-05,
+      "loss": 1.0241,
       "step": 887
     },
     {
-      "epoch": 0.2901960784313726,
-      "grad_norm": 1.574407696723938,
-      "learning_rate": 1.665556951461501e-05,
-      "loss": 1.0007,
+      "epoch": 0.02438823432479196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624885635864593e-05,
+      "loss": 1.1171,
       "step": 888
     },
     {
-      "epoch": 0.2905228758169935,
-      "grad_norm": 1.4966329336166382,
-      "learning_rate": 1.6647665792830303e-05,
-      "loss": 1.0255,
+      "epoch": 0.024415698552635194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626715462031107e-05,
+      "loss": 1.1105,
       "step": 889
     },
     {
-      "epoch": 0.2908496732026144,
-      "grad_norm": 1.5546950101852417,
-      "learning_rate": 1.6639754623021227e-05,
-      "loss": 0.9947,
+      "epoch": 0.024443162780478426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6285452881976215e-05,
+      "loss": 1.1872,
       "step": 890
     },
     {
-      "epoch": 0.2911764705882353,
-      "grad_norm": 1.5820887088775635,
-      "learning_rate": 1.6631836014051427e-05,
-      "loss": 1.0196,
+      "epoch": 0.02447062700832166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6303751143641356e-05,
+      "loss": 1.148,
       "step": 891
     },
     {
-      "epoch": 0.29150326797385623,
-      "grad_norm": 1.439009666442871,
-      "learning_rate": 1.6623909974792888e-05,
-      "loss": 1.0178,
+      "epoch": 0.024498091236164896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322049405306496e-05,
+      "loss": 1.0407,
       "step": 892
     },
     {
-      "epoch": 0.29183006535947714,
-      "grad_norm": 1.6108399629592896,
-      "learning_rate": 1.661597651412593e-05,
-      "loss": 1.0017,
+      "epoch": 0.024525555464008128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6340347666971637e-05,
+      "loss": 1.0945,
       "step": 893
     },
     {
-      "epoch": 0.29215686274509806,
-      "grad_norm": 1.515515685081482,
-      "learning_rate": 1.6608035640939165e-05,
-      "loss": 1.0264,
+      "epoch": 0.024553019691851363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6358645928636782e-05,
+      "loss": 1.1857,
       "step": 894
     },
     {
-      "epoch": 0.29248366013071897,
-      "grad_norm": 1.6603224277496338,
-      "learning_rate": 1.6600087364129535e-05,
-      "loss": 1.0371,
+      "epoch": 0.0245804839196946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6376944190301923e-05,
+      "loss": 1.1092,
       "step": 895
     },
     {
-      "epoch": 0.2928104575163399,
-      "grad_norm": 1.5440163612365723,
-      "learning_rate": 1.6592131692602257e-05,
-      "loss": 0.9725,
+      "epoch": 0.024607948147537834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6395242451967064e-05,
+      "loss": 1.1022,
       "step": 896
     },
     {
-      "epoch": 0.2931372549019608,
-      "grad_norm": 1.6434190273284912,
-      "learning_rate": 1.658416863527084e-05,
-      "loss": 0.9975,
+      "epoch": 0.024635412375381065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6413540713632204e-05,
+      "loss": 1.1363,
       "step": 897
     },
     {
-      "epoch": 0.2934640522875817,
-      "grad_norm": 1.5150604248046875,
-      "learning_rate": 1.657619820105708e-05,
-      "loss": 1.0306,
+      "epoch": 0.0246628766032243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643183897529735e-05,
+      "loss": 1.171,
       "step": 898
     },
     {
-      "epoch": 0.2937908496732026,
-      "grad_norm": 1.6481305360794067,
-      "learning_rate": 1.656822039889101e-05,
-      "loss": 1.0318,
+      "epoch": 0.024690340831067536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645013723696249e-05,
+      "loss": 1.1235,
       "step": 899
     },
     {
-      "epoch": 0.29411764705882354,
-      "grad_norm": 1.5463274717330933,
-      "learning_rate": 1.656023523771095e-05,
-      "loss": 1.0209,
+      "epoch": 0.024717805058910768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646843549862763e-05,
+      "loss": 1.1695,
       "step": 900
     },
     {
-      "epoch": 0.29444444444444445,
-      "grad_norm": 1.4983214139938354,
-      "learning_rate": 1.6552242726463447e-05,
-      "loss": 0.9936,
+      "epoch": 0.024745269286754003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6486733760292775e-05,
+      "loss": 1.1162,
       "step": 901
     },
     {
-      "epoch": 0.29477124183006537,
-      "grad_norm": 1.583688497543335,
-      "learning_rate": 1.6544242874103278e-05,
-      "loss": 0.9406,
+      "epoch": 0.024772733514597238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6505032021957916e-05,
+      "loss": 1.1658,
       "step": 902
     },
     {
-      "epoch": 0.2950980392156863,
-      "grad_norm": 1.572791337966919,
-      "learning_rate": 1.653623568959346e-05,
-      "loss": 1.0204,
+      "epoch": 0.02480019774244047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6523330283623057e-05,
+      "loss": 1.1012,
       "step": 903
     },
     {
-      "epoch": 0.2954248366013072,
-      "grad_norm": 1.5110663175582886,
-      "learning_rate": 1.6528221181905217e-05,
-      "loss": 1.0225,
+      "epoch": 0.024827661970283705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65416285452882e-05,
+      "loss": 1.0862,
       "step": 904
     },
     {
-      "epoch": 0.2957516339869281,
-      "grad_norm": 1.5479543209075928,
-      "learning_rate": 1.652019936001798e-05,
-      "loss": 0.9896,
+      "epoch": 0.02485512619812694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6559926806953342e-05,
+      "loss": 1.1323,
       "step": 905
     },
     {
-      "epoch": 0.296078431372549,
-      "grad_norm": 1.6475478410720825,
-      "learning_rate": 1.6512170232919372e-05,
-      "loss": 1.0338,
+      "epoch": 0.024882590425970175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6578225068618483e-05,
+      "loss": 1.1507,
       "step": 906
     },
     {
-      "epoch": 0.29640522875816994,
-      "grad_norm": 1.4979268312454224,
-      "learning_rate": 1.65041338096052e-05,
-      "loss": 0.9717,
+      "epoch": 0.024910054653813407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6596523330283624e-05,
+      "loss": 1.2062,
       "step": 907
     },
     {
-      "epoch": 0.29673202614379085,
-      "grad_norm": 1.4964370727539062,
-      "learning_rate": 1.6496090099079452e-05,
-      "loss": 1.0118,
+      "epoch": 0.024937518881656642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6614821591948768e-05,
+      "loss": 1.1064,
       "step": 908
     },
     {
-      "epoch": 0.29705882352941176,
-      "grad_norm": 1.5042089223861694,
-      "learning_rate": 1.6488039110354273e-05,
-      "loss": 1.0114,
+      "epoch": 0.024964983109499878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663311985361391e-05,
+      "loss": 1.1493,
       "step": 909
     },
     {
-      "epoch": 0.2973856209150327,
-      "grad_norm": 1.4784412384033203,
-      "learning_rate": 1.6479980852449974e-05,
-      "loss": 0.9988,
+      "epoch": 0.02499244733734311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665141811527905e-05,
+      "loss": 1.1245,
       "step": 910
     },
     {
-      "epoch": 0.2977124183006536,
-      "grad_norm": 1.568564772605896,
-      "learning_rate": 1.647191533439499e-05,
-      "loss": 0.9509,
+      "epoch": 0.025019911565186344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666971637694419e-05,
+      "loss": 1.1414,
       "step": 911
     },
     {
-      "epoch": 0.2980392156862745,
-      "grad_norm": 1.4553037881851196,
-      "learning_rate": 1.6463842565225914e-05,
-      "loss": 1.0289,
+      "epoch": 0.02504737579302958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6688014638609332e-05,
+      "loss": 1.1469,
       "step": 912
     },
     {
-      "epoch": 0.2983660130718954,
-      "grad_norm": 1.5019041299819946,
-      "learning_rate": 1.6455762553987444e-05,
-      "loss": 1.014,
+      "epoch": 0.02507484002087281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6706312900274476e-05,
+      "loss": 1.086,
       "step": 913
     },
     {
-      "epoch": 0.29869281045751633,
-      "grad_norm": 1.4880790710449219,
-      "learning_rate": 1.6447675309732403e-05,
-      "loss": 1.0103,
+      "epoch": 0.025102304248716047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6724611161939617e-05,
+      "loss": 1.0814,
       "step": 914
     },
     {
-      "epoch": 0.29901960784313725,
-      "grad_norm": 1.599459171295166,
-      "learning_rate": 1.6439580841521717e-05,
-      "loss": 1.0197,
+      "epoch": 0.025129768476559282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6742909423604758e-05,
+      "loss": 1.0671,
       "step": 915
     },
     {
-      "epoch": 0.29934640522875816,
-      "grad_norm": 1.4689435958862305,
-      "learning_rate": 1.64314791584244e-05,
-      "loss": 1.0089,
+      "epoch": 0.025157232704402517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67612076852699e-05,
+      "loss": 1.2155,
       "step": 916
     },
     {
-      "epoch": 0.2996732026143791,
-      "grad_norm": 1.5712580680847168,
-      "learning_rate": 1.6423370269517556e-05,
-      "loss": 0.9739,
+      "epoch": 0.02518469693224575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6779505946935043e-05,
+      "loss": 1.0885,
       "step": 917
     },
     {
-      "epoch": 0.3,
-      "grad_norm": 1.4598122835159302,
-      "learning_rate": 1.6415254183886356e-05,
-      "loss": 0.9701,
+      "epoch": 0.025212161160088984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6797804208600184e-05,
+      "loss": 1.1504,
       "step": 918
     },
     {
-      "epoch": 0.3003267973856209,
-      "grad_norm": 1.603790044784546,
-      "learning_rate": 1.6407130910624037e-05,
-      "loss": 0.959,
+      "epoch": 0.02523962538793222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6816102470265325e-05,
+      "loss": 1.0737,
       "step": 919
     },
     {
-      "epoch": 0.3006535947712418,
-      "grad_norm": 1.5031025409698486,
-      "learning_rate": 1.6399000458831893e-05,
-      "loss": 1.0012,
+      "epoch": 0.02526708961577545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834400731930466e-05,
+      "loss": 1.2266,
       "step": 920
     },
     {
-      "epoch": 0.30098039215686273,
-      "grad_norm": 1.4751572608947754,
-      "learning_rate": 1.6390862837619252e-05,
-      "loss": 1.0038,
+      "epoch": 0.025294553843618686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685269899359561e-05,
+      "loss": 1.1118,
       "step": 921
     },
     {
-      "epoch": 0.30130718954248364,
-      "grad_norm": 1.5720016956329346,
-      "learning_rate": 1.6382718056103486e-05,
-      "loss": 0.9772,
+      "epoch": 0.02532201807146192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687099725526075e-05,
+      "loss": 1.1451,
       "step": 922
     },
     {
-      "epoch": 0.30163398692810456,
-      "grad_norm": 1.5443570613861084,
-      "learning_rate": 1.637456612340998e-05,
-      "loss": 1.0062,
+      "epoch": 0.025349482299305157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6889295516925896e-05,
+      "loss": 1.1151,
       "step": 923
     },
     {
-      "epoch": 0.30196078431372547,
-      "grad_norm": 1.5211632251739502,
-      "learning_rate": 1.6366407048672135e-05,
-      "loss": 1.0136,
+      "epoch": 0.02537694652714839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6907593778591037e-05,
+      "loss": 1.0895,
       "step": 924
     },
     {
-      "epoch": 0.3022875816993464,
-      "grad_norm": 1.527962565422058,
-      "learning_rate": 1.635824084103135e-05,
-      "loss": 0.9473,
+      "epoch": 0.025404410754991624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6925892040256177e-05,
+      "loss": 1.1313,
       "step": 925
     },
     {
-      "epoch": 0.3026143790849673,
-      "grad_norm": 1.4427688121795654,
-      "learning_rate": 1.6350067509637028e-05,
-      "loss": 0.9841,
+      "epoch": 0.02543187498283486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694419030192132e-05,
+      "loss": 1.1009,
       "step": 926
     },
     {
-      "epoch": 0.3029411764705882,
-      "grad_norm": 1.3982694149017334,
-      "learning_rate": 1.6341887063646535e-05,
-      "loss": 1.0058,
+      "epoch": 0.02545933921067809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6962488563586463e-05,
+      "loss": 1.1416,
       "step": 927
     },
     {
-      "epoch": 0.3032679738562091,
-      "grad_norm": 1.553612232208252,
-      "learning_rate": 1.6333699512225224e-05,
-      "loss": 0.9817,
+      "epoch": 0.025486803438521326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6980786825251604e-05,
+      "loss": 1.1048,
       "step": 928
     },
     {
-      "epoch": 0.30359477124183004,
-      "grad_norm": 1.4203298091888428,
-      "learning_rate": 1.63255048645464e-05,
-      "loss": 0.986,
+      "epoch": 0.02551426766636456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6999085086916745e-05,
+      "loss": 1.097,
       "step": 929
     },
     {
-      "epoch": 0.30392156862745096,
-      "grad_norm": 1.4367247819900513,
-      "learning_rate": 1.6317303129791315e-05,
-      "loss": 0.9896,
+      "epoch": 0.025541731894207793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7017383348581885e-05,
+      "loss": 1.1074,
       "step": 930
     },
     {
-      "epoch": 0.30424836601307187,
-      "grad_norm": 1.4952465295791626,
-      "learning_rate": 1.630909431714918e-05,
-      "loss": 0.9667,
+      "epoch": 0.025569196122051028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7035681610247026e-05,
+      "loss": 1.0999,
       "step": 931
     },
     {
-      "epoch": 0.3045751633986928,
-      "grad_norm": 1.496411919593811,
-      "learning_rate": 1.6300878435817115e-05,
-      "loss": 0.9988,
+      "epoch": 0.025596660349894263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705397987191217e-05,
+      "loss": 1.2278,
       "step": 932
     },
     {
-      "epoch": 0.30490196078431375,
-      "grad_norm": 1.4844709634780884,
-      "learning_rate": 1.6292655495000163e-05,
-      "loss": 1.0049,
+      "epoch": 0.0256241245777375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707227813357731e-05,
+      "loss": 1.1425,
       "step": 933
     },
     {
-      "epoch": 0.30522875816993467,
-      "grad_norm": 1.4948958158493042,
-      "learning_rate": 1.62844255039113e-05,
-      "loss": 0.9902,
+      "epoch": 0.02565158880558073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7090576395242453e-05,
+      "loss": 1.165,
       "step": 934
     },
     {
-      "epoch": 0.3055555555555556,
-      "grad_norm": 1.5211067199707031,
-      "learning_rate": 1.6276188471771367e-05,
-      "loss": 0.9965,
+      "epoch": 0.025679053033423965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7108874656907593e-05,
+      "loss": 1.1035,
       "step": 935
     },
     {
-      "epoch": 0.3058823529411765,
-      "grad_norm": 1.4753836393356323,
-      "learning_rate": 1.626794440780911e-05,
-      "loss": 0.9719,
+      "epoch": 0.0257065172612672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7127172918572738e-05,
+      "loss": 1.1823,
       "step": 936
     },
     {
-      "epoch": 0.3062091503267974,
-      "grad_norm": 1.493467926979065,
-      "learning_rate": 1.625969332126116e-05,
-      "loss": 0.9971,
+      "epoch": 0.025733981489110432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714547118023788e-05,
+      "loss": 1.212,
       "step": 937
     },
     {
-      "epoch": 0.3065359477124183,
-      "grad_norm": 1.5081828832626343,
-      "learning_rate": 1.6251435221372007e-05,
-      "loss": 0.9971,
+      "epoch": 0.025761445716953667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716376944190302e-05,
+      "loss": 1.1843,
       "step": 938
     },
     {
-      "epoch": 0.30686274509803924,
-      "grad_norm": 1.4416155815124512,
-      "learning_rate": 1.6243170117394e-05,
-      "loss": 1.0151,
+      "epoch": 0.025788909944796903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718206770356816e-05,
+      "loss": 1.2084,
       "step": 939
     },
     {
-      "epoch": 0.30718954248366015,
-      "grad_norm": 1.542250633239746,
-      "learning_rate": 1.6234898018587336e-05,
-      "loss": 0.9841,
+      "epoch": 0.025816374172640138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7200365965233305e-05,
+      "loss": 1.1781,
       "step": 940
     },
     {
-      "epoch": 0.30751633986928106,
-      "grad_norm": 1.5334587097167969,
-      "learning_rate": 1.6226618934220048e-05,
-      "loss": 1.0037,
+      "epoch": 0.02584383840048337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7218664226898446e-05,
+      "loss": 1.0293,
       "step": 941
     },
     {
-      "epoch": 0.307843137254902,
-      "grad_norm": 1.45203697681427,
-      "learning_rate": 1.6218332873568e-05,
-      "loss": 1.0146,
+      "epoch": 0.025871302628326605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723696248856359e-05,
+      "loss": 1.1548,
       "step": 942
     },
     {
-      "epoch": 0.3081699346405229,
-      "grad_norm": 1.431759238243103,
-      "learning_rate": 1.6210039845914866e-05,
-      "loss": 0.9646,
+      "epoch": 0.02589876685616984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725526075022873e-05,
+      "loss": 1.1386,
       "step": 943
     },
     {
-      "epoch": 0.3084967320261438,
-      "grad_norm": 1.4204840660095215,
-      "learning_rate": 1.620173986055213e-05,
-      "loss": 1.0039,
+      "epoch": 0.025926231084013072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7273559011893872e-05,
+      "loss": 0.9801,
       "step": 944
     },
     {
-      "epoch": 0.3088235294117647,
-      "grad_norm": 1.5130980014801025,
-      "learning_rate": 1.619343292677907e-05,
-      "loss": 0.9599,
+      "epoch": 0.025953695311856307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7291857273559013e-05,
+      "loss": 1.1835,
       "step": 945
     },
     {
-      "epoch": 0.30915032679738563,
-      "grad_norm": 1.4743794202804565,
-      "learning_rate": 1.6185119053902748e-05,
-      "loss": 1.0177,
+      "epoch": 0.025981159539699542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7310155535224157e-05,
+      "loss": 1.1381,
       "step": 946
     },
     {
-      "epoch": 0.30947712418300655,
-      "grad_norm": 1.3975343704223633,
-      "learning_rate": 1.6176798251238004e-05,
-      "loss": 0.9664,
+      "epoch": 0.026008623767542774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7328453796889298e-05,
+      "loss": 1.0842,
       "step": 947
     },
     {
-      "epoch": 0.30980392156862746,
-      "grad_norm": 1.5124108791351318,
-      "learning_rate": 1.616847052810744e-05,
-      "loss": 0.9899,
+      "epoch": 0.02603608799538601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734675205855444e-05,
+      "loss": 1.1196,
       "step": 948
     },
     {
-      "epoch": 0.3101307189542484,
-      "grad_norm": 1.526279330253601,
-      "learning_rate": 1.6160135893841405e-05,
-      "loss": 1.001,
+      "epoch": 0.026063552223229244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736505032021958e-05,
+      "loss": 1.1144,
       "step": 949
     },
     {
-      "epoch": 0.3104575163398693,
-      "grad_norm": 1.469681739807129,
-      "learning_rate": 1.6151794357778006e-05,
-      "loss": 0.9829,
+      "epoch": 0.02609101645107248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738334858188472e-05,
+      "loss": 1.1276,
       "step": 950
     },
     {
-      "epoch": 0.3107843137254902,
-      "grad_norm": 1.4610638618469238,
-      "learning_rate": 1.614344592926307e-05,
-      "loss": 0.9833,
+      "epoch": 0.02611848067891571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7401646843549865e-05,
+      "loss": 1.1225,
       "step": 951
     },
     {
-      "epoch": 0.3111111111111111,
-      "grad_norm": 1.5406149625778198,
-      "learning_rate": 1.613509061765015e-05,
-      "loss": 0.9987,
+      "epoch": 0.026145944906758947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7419945105215006e-05,
+      "loss": 1.1017,
       "step": 952
     },
     {
-      "epoch": 0.31143790849673203,
-      "grad_norm": 1.5692520141601562,
-      "learning_rate": 1.6126728432300516e-05,
-      "loss": 0.9577,
+      "epoch": 0.026173409134602182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7438243366880147e-05,
+      "loss": 1.0299,
       "step": 953
     },
     {
-      "epoch": 0.31176470588235294,
-      "grad_norm": 1.4870448112487793,
-      "learning_rate": 1.611835938258313e-05,
-      "loss": 0.9626,
+      "epoch": 0.026200873362445413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7456541628545288e-05,
+      "loss": 1.1442,
       "step": 954
     },
     {
-      "epoch": 0.31209150326797386,
-      "grad_norm": 1.4175331592559814,
-      "learning_rate": 1.6109983477874654e-05,
-      "loss": 0.9749,
+      "epoch": 0.02622833759028865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7474839890210432e-05,
+      "loss": 1.222,
       "step": 955
     },
     {
-      "epoch": 0.31241830065359477,
-      "grad_norm": 1.5690836906433105,
-      "learning_rate": 1.6101600727559423e-05,
-      "loss": 0.9776,
+      "epoch": 0.026255801818131884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7493138151875573e-05,
+      "loss": 1.101,
       "step": 956
     },
     {
-      "epoch": 0.3127450980392157,
-      "grad_norm": 1.5043728351593018,
-      "learning_rate": 1.6093211141029445e-05,
-      "loss": 0.9654,
+      "epoch": 0.02628326604597512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7511436413540714e-05,
+      "loss": 1.0464,
       "step": 957
     },
     {
-      "epoch": 0.3130718954248366,
-      "grad_norm": 1.450974702835083,
-      "learning_rate": 1.6084814727684386e-05,
-      "loss": 0.9847,
+      "epoch": 0.02631073027381835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7529734675205855e-05,
+      "loss": 1.1752,
       "step": 958
     },
     {
-      "epoch": 0.3133986928104575,
-      "grad_norm": 1.6643537282943726,
-      "learning_rate": 1.6076411496931564e-05,
-      "loss": 1.041,
+      "epoch": 0.026338194501661586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7548032936871e-05,
+      "loss": 1.0875,
       "step": 959
     },
     {
-      "epoch": 0.3137254901960784,
-      "grad_norm": 1.4928576946258545,
-      "learning_rate": 1.6068001458185934e-05,
-      "loss": 0.9899,
+      "epoch": 0.02636565872950482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756633119853614e-05,
+      "loss": 1.0997,
       "step": 960
     },
     {
-      "epoch": 0.31405228758169934,
-      "grad_norm": 1.4915987253189087,
-      "learning_rate": 1.6059584620870076e-05,
-      "loss": 1.0012,
+      "epoch": 0.026393122957348053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758462946020128e-05,
+      "loss": 1.23,
       "step": 961
     },
     {
-      "epoch": 0.31437908496732025,
-      "grad_norm": 1.520616888999939,
-      "learning_rate": 1.6051160994414187e-05,
-      "loss": 1.0158,
+      "epoch": 0.026420587185191288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7602927721866425e-05,
+      "loss": 1.0622,
       "step": 962
     },
     {
-      "epoch": 0.31470588235294117,
-      "grad_norm": 1.463838815689087,
-      "learning_rate": 1.604273058825607e-05,
-      "loss": 0.9862,
+      "epoch": 0.026448051413034523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7621225983531566e-05,
+      "loss": 1.1304,
       "step": 963
     },
     {
-      "epoch": 0.3150326797385621,
-      "grad_norm": 1.5073494911193848,
-      "learning_rate": 1.603429341184114e-05,
-      "loss": 0.9955,
+      "epoch": 0.026475515640877755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7639524245196707e-05,
+      "loss": 1.1748,
       "step": 964
     },
     {
-      "epoch": 0.315359477124183,
-      "grad_norm": 1.58897864818573,
-      "learning_rate": 1.6025849474622365e-05,
-      "loss": 1.0019,
+      "epoch": 0.02650297986872099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765782250686185e-05,
+      "loss": 1.1423,
       "step": 965
     },
     {
-      "epoch": 0.3156862745098039,
-      "grad_norm": 1.4118837118148804,
-      "learning_rate": 1.601739878606032e-05,
-      "loss": 1.0229,
+      "epoch": 0.026530444096564226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7676120768526993e-05,
+      "loss": 1.0238,
       "step": 966
     },
     {
-      "epoch": 0.3160130718954248,
-      "grad_norm": 1.5065345764160156,
-      "learning_rate": 1.600894135562312e-05,
-      "loss": 1.0182,
+      "epoch": 0.02655790832440746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7694419030192133e-05,
+      "loss": 1.0444,
       "step": 967
     },
     {
-      "epoch": 0.31633986928104574,
-      "grad_norm": 1.456397294998169,
-      "learning_rate": 1.600047719278645e-05,
-      "loss": 0.9886,
+      "epoch": 0.026585372552250693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7712717291857274e-05,
+      "loss": 1.1864,
       "step": 968
     },
     {
-      "epoch": 0.31666666666666665,
-      "grad_norm": 1.463565468788147,
-      "learning_rate": 1.5992006307033538e-05,
-      "loss": 1.0224,
+      "epoch": 0.026612836780093928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773101555352242e-05,
+      "loss": 1.0133,
       "step": 969
     },
     {
-      "epoch": 0.31699346405228757,
-      "grad_norm": 1.4853087663650513,
-      "learning_rate": 1.5983528707855128e-05,
-      "loss": 0.9881,
+      "epoch": 0.026640301007937163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774931381518756e-05,
+      "loss": 1.1076,
       "step": 970
     },
     {
-      "epoch": 0.3173202614379085,
-      "grad_norm": 1.380228042602539,
-      "learning_rate": 1.59750444047495e-05,
-      "loss": 0.9774,
+      "epoch": 0.026667765235780395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77676120768527e-05,
+      "loss": 1.1355,
       "step": 971
     },
     {
-      "epoch": 0.3176470588235294,
-      "grad_norm": 1.6259825229644775,
-      "learning_rate": 1.596655340722244e-05,
-      "loss": 1.0338,
+      "epoch": 0.02669522946362363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778591033851784e-05,
+      "loss": 1.1296,
       "step": 972
     },
     {
-      "epoch": 0.3179738562091503,
-      "grad_norm": 1.4979356527328491,
-      "learning_rate": 1.595805572478724e-05,
-      "loss": 0.9725,
+      "epoch": 0.026722693691466865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7804208600182982e-05,
+      "loss": 1.1343,
       "step": 973
     },
     {
-      "epoch": 0.3183006535947712,
-      "grad_norm": 1.453285574913025,
-      "learning_rate": 1.5949551366964675e-05,
-      "loss": 0.9375,
+      "epoch": 0.026750157919310097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7822506861848127e-05,
+      "loss": 1.1607,
       "step": 974
     },
     {
-      "epoch": 0.31862745098039214,
-      "grad_norm": 1.4632606506347656,
-      "learning_rate": 1.5941040343282997e-05,
-      "loss": 1.0073,
+      "epoch": 0.026777622147153332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7840805123513268e-05,
+      "loss": 1.0471,
       "step": 975
     },
     {
-      "epoch": 0.31895424836601305,
-      "grad_norm": 1.4426164627075195,
-      "learning_rate": 1.593252266327794e-05,
-      "loss": 0.9868,
+      "epoch": 0.026805086374996567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785910338517841e-05,
+      "loss": 1.098,
       "step": 976
     },
     {
-      "epoch": 0.31928104575163396,
-      "grad_norm": 1.5374642610549927,
-      "learning_rate": 1.5923998336492678e-05,
-      "loss": 0.9783,
+      "epoch": 0.026832550602839803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787740164684355e-05,
+      "loss": 1.1765,
       "step": 977
     },
     {
-      "epoch": 0.3196078431372549,
-      "grad_norm": 1.4929271936416626,
-      "learning_rate": 1.5915467372477843e-05,
-      "loss": 0.9811,
+      "epoch": 0.026860014830683034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7895699908508694e-05,
+      "loss": 1.17,
       "step": 978
     },
     {
-      "epoch": 0.3199346405228758,
-      "grad_norm": 1.5067940950393677,
-      "learning_rate": 1.5906929780791502e-05,
-      "loss": 0.9738,
+      "epoch": 0.02688747905852627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7913998170173835e-05,
+      "loss": 1.1207,
       "step": 979
     },
     {
-      "epoch": 0.3202614379084967,
-      "grad_norm": 1.4917588233947754,
-      "learning_rate": 1.5898385570999146e-05,
-      "loss": 0.972,
+      "epoch": 0.026914943286369505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7932296431838976e-05,
+      "loss": 1.079,
       "step": 980
     },
     {
-      "epoch": 0.3205882352941177,
-      "grad_norm": 1.5650171041488647,
-      "learning_rate": 1.588983475267368e-05,
-      "loss": 0.9624,
+      "epoch": 0.026942407514212736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7950594693504117e-05,
+      "loss": 1.1613,
       "step": 981
     },
     {
-      "epoch": 0.3209150326797386,
-      "grad_norm": 1.433803915977478,
-      "learning_rate": 1.5881277335395425e-05,
-      "loss": 0.9656,
+      "epoch": 0.02696987174205597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796889295516926e-05,
+      "loss": 1.0463,
       "step": 982
     },
     {
-      "epoch": 0.3212418300653595,
-      "grad_norm": 1.7312647104263306,
-      "learning_rate": 1.587271332875207e-05,
-      "loss": 0.9826,
+      "epoch": 0.026997335969899207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7987191216834402e-05,
+      "loss": 1.1061,
       "step": 983
     },
     {
-      "epoch": 0.3215686274509804,
-      "grad_norm": 1.4314967393875122,
-      "learning_rate": 1.5864142742338707e-05,
-      "loss": 0.9626,
+      "epoch": 0.027024800197742442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8005489478499546e-05,
+      "loss": 1.2192,
       "step": 984
     },
     {
-      "epoch": 0.32189542483660133,
-      "grad_norm": 1.5712225437164307,
-      "learning_rate": 1.58555655857578e-05,
-      "loss": 0.9846,
+      "epoch": 0.027052264425585674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8023787740164687e-05,
+      "loss": 1.1007,
       "step": 985
     },
     {
-      "epoch": 0.32222222222222224,
-      "grad_norm": 1.4747076034545898,
-      "learning_rate": 1.584698186861916e-05,
-      "loss": 0.9938,
+      "epoch": 0.02707972865342891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042086001829828e-05,
+      "loss": 1.1652,
       "step": 986
     },
     {
-      "epoch": 0.32254901960784316,
-      "grad_norm": 1.5617436170578003,
-      "learning_rate": 1.5838391600539966e-05,
-      "loss": 0.9932,
+      "epoch": 0.027107192881272144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806038426349497e-05,
+      "loss": 1.0179,
       "step": 987
     },
     {
-      "epoch": 0.32287581699346407,
-      "grad_norm": 1.4605984687805176,
-      "learning_rate": 1.5829794791144723e-05,
-      "loss": 0.9621,
+      "epoch": 0.027134657109115376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8078682525160113e-05,
+      "loss": 1.1283,
       "step": 988
     },
     {
-      "epoch": 0.323202614379085,
-      "grad_norm": 1.4705711603164673,
-      "learning_rate": 1.582119145006527e-05,
-      "loss": 0.9729,
+      "epoch": 0.02716212133695861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8096980786825254e-05,
+      "loss": 1.1747,
       "step": 989
     },
     {
-      "epoch": 0.3235294117647059,
-      "grad_norm": 1.4788919687271118,
-      "learning_rate": 1.5812581586940767e-05,
-      "loss": 0.9723,
+      "epoch": 0.027189585564801846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8115279048490395e-05,
+      "loss": 0.9743,
       "step": 990
     },
     {
-      "epoch": 0.3238562091503268,
-      "grad_norm": 1.4782557487487793,
-      "learning_rate": 1.5803965211417676e-05,
-      "loss": 0.9926,
+      "epoch": 0.027217049792645078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8133577310155536e-05,
+      "loss": 1.0292,
       "step": 991
     },
     {
-      "epoch": 0.3241830065359477,
-      "grad_norm": 1.5275636911392212,
-      "learning_rate": 1.5795342333149757e-05,
-      "loss": 0.9753,
+      "epoch": 0.027244514020488313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8151875571820677e-05,
+      "loss": 1.1307,
       "step": 992
     },
     {
-      "epoch": 0.32450980392156864,
-      "grad_norm": 1.5167269706726074,
-      "learning_rate": 1.578671296179806e-05,
-      "loss": 1.008,
+      "epoch": 0.02727197824833155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817017383348582e-05,
+      "loss": 1.0864,
       "step": 993
     },
     {
-      "epoch": 0.32483660130718955,
-      "grad_norm": 1.6089777946472168,
-      "learning_rate": 1.57780771070309e-05,
-      "loss": 0.9869,
+      "epoch": 0.027299442476174784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8188472095150962e-05,
+      "loss": 1.0838,
       "step": 994
     },
     {
-      "epoch": 0.32516339869281047,
-      "grad_norm": 1.4116290807724,
-      "learning_rate": 1.576943477852387e-05,
-      "loss": 0.9893,
+      "epoch": 0.027326906704018016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8206770356816103e-05,
+      "loss": 1.1292,
       "step": 995
     },
     {
-      "epoch": 0.3254901960784314,
-      "grad_norm": 1.3807182312011719,
-      "learning_rate": 1.57607859859598e-05,
-      "loss": 0.9681,
+      "epoch": 0.02735437093186125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8225068618481244e-05,
+      "loss": 1.1177,
       "step": 996
     },
     {
-      "epoch": 0.3258169934640523,
-      "grad_norm": 1.3623489141464233,
-      "learning_rate": 1.575213073902878e-05,
-      "loss": 0.9689,
+      "epoch": 0.027381835159704486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8243366880146388e-05,
+      "loss": 1.1354,
       "step": 997
     },
     {
-      "epoch": 0.3261437908496732,
-      "grad_norm": 1.4567856788635254,
-      "learning_rate": 1.5743469047428113e-05,
-      "loss": 0.9517,
+      "epoch": 0.027409299387547718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826166514181153e-05,
+      "loss": 1.137,
       "step": 998
     },
     {
-      "epoch": 0.3264705882352941,
-      "grad_norm": 1.5761626958847046,
-      "learning_rate": 1.5734800920862335e-05,
-      "loss": 1.032,
+      "epoch": 0.027436763615390953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827996340347667e-05,
+      "loss": 1.1914,
       "step": 999
     },
     {
-      "epoch": 0.32679738562091504,
-      "grad_norm": 1.4622478485107422,
-      "learning_rate": 1.572612636904319e-05,
-      "loss": 0.9796,
+      "epoch": 0.027464227843234188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829826166514181e-05,
+      "loss": 1.1512,
       "step": 1000
     },
     {
-      "epoch": 0.32712418300653595,
-      "grad_norm": 1.4829388856887817,
-      "learning_rate": 1.571744540168962e-05,
-      "loss": 0.9928,
+      "epoch": 0.027491692071077423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8316559926806955e-05,
+      "loss": 1.0778,
       "step": 1001
     },
     {
-      "epoch": 0.32745098039215687,
-      "grad_norm": 1.528941035270691,
-      "learning_rate": 1.5708758028527754e-05,
-      "loss": 1.0072,
+      "epoch": 0.027519156298920655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8334858188472096e-05,
+      "loss": 1.1292,
       "step": 1002
     },
     {
-      "epoch": 0.3277777777777778,
-      "grad_norm": 1.400481104850769,
-      "learning_rate": 1.5700064259290894e-05,
-      "loss": 0.9799,
+      "epoch": 0.02754662052676389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835315645013724e-05,
+      "loss": 1.1312,
       "step": 1003
     },
     {
-      "epoch": 0.3281045751633987,
-      "grad_norm": 1.4204185009002686,
-      "learning_rate": 1.5691364103719515e-05,
-      "loss": 0.9745,
+      "epoch": 0.027574084754607125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837145471180238e-05,
+      "loss": 1.1362,
       "step": 1004
     },
     {
-      "epoch": 0.3284313725490196,
-      "grad_norm": 1.494147539138794,
-      "learning_rate": 1.568265757156125e-05,
-      "loss": 0.9843,
+      "epoch": 0.027601548982450357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8389752973467522e-05,
+      "loss": 1.1913,
       "step": 1005
     },
     {
-      "epoch": 0.3287581699346405,
-      "grad_norm": 1.4243229627609253,
-      "learning_rate": 1.5673944672570858e-05,
-      "loss": 0.9969,
+      "epoch": 0.027629013210293592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8408051235132663e-05,
+      "loss": 1.123,
       "step": 1006
     },
     {
-      "epoch": 0.32908496732026143,
-      "grad_norm": 1.470023274421692,
-      "learning_rate": 1.566522541651026e-05,
-      "loss": 0.972,
+      "epoch": 0.027656477438136828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8426349496797808e-05,
+      "loss": 1.1317,
       "step": 1007
     },
     {
-      "epoch": 0.32941176470588235,
-      "grad_norm": 1.4366976022720337,
-      "learning_rate": 1.565649981314847e-05,
-      "loss": 0.9439,
+      "epoch": 0.02768394166598006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844464775846295e-05,
+      "loss": 1.0782,
       "step": 1008
     },
     {
-      "epoch": 0.32973856209150326,
-      "grad_norm": 1.4011088609695435,
-      "learning_rate": 1.5647767872261633e-05,
-      "loss": 0.9884,
+      "epoch": 0.027711405893823295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846294602012809e-05,
+      "loss": 1.0929,
       "step": 1009
     },
     {
-      "epoch": 0.3300653594771242,
-      "grad_norm": 1.4959757328033447,
-      "learning_rate": 1.5639029603632995e-05,
-      "loss": 0.9653,
+      "epoch": 0.02773887012166653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848124428179323e-05,
+      "loss": 1.112,
       "step": 1010
     },
     {
-      "epoch": 0.3303921568627451,
-      "grad_norm": 1.4076058864593506,
-      "learning_rate": 1.5630285017052875e-05,
-      "loss": 0.989,
+      "epoch": 0.027766334349509765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8499542543458375e-05,
+      "loss": 1.1685,
       "step": 1011
     },
     {
-      "epoch": 0.330718954248366,
-      "grad_norm": 1.5262447595596313,
-      "learning_rate": 1.5621534122318682e-05,
-      "loss": 0.9865,
+      "epoch": 0.027793798577352997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8517840805123516e-05,
+      "loss": 1.1053,
       "step": 1012
     },
     {
-      "epoch": 0.3310457516339869,
-      "grad_norm": 1.3930559158325195,
-      "learning_rate": 1.56127769292349e-05,
-      "loss": 0.9593,
+      "epoch": 0.027821262805196232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8536139066788657e-05,
+      "loss": 1.1862,
       "step": 1013
     },
     {
-      "epoch": 0.33137254901960783,
-      "grad_norm": 1.4923077821731567,
-      "learning_rate": 1.5604013447613057e-05,
-      "loss": 0.9973,
+      "epoch": 0.027848727033039467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554437328453798e-05,
+      "loss": 1.2511,
       "step": 1014
     },
     {
-      "epoch": 0.33169934640522875,
-      "grad_norm": 1.431309461593628,
-      "learning_rate": 1.5595243687271726e-05,
-      "loss": 0.9742,
+      "epoch": 0.0278761912608827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857273559011894e-05,
+      "loss": 1.0842,
       "step": 1015
     },
     {
-      "epoch": 0.33202614379084966,
-      "grad_norm": 1.423667311668396,
-      "learning_rate": 1.5586467658036526e-05,
-      "loss": 0.978,
+      "epoch": 0.027903655488725934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8591033851784083e-05,
+      "loss": 1.0946,
       "step": 1016
     },
     {
-      "epoch": 0.3323529411764706,
-      "grad_norm": 1.413068175315857,
-      "learning_rate": 1.557768536974009e-05,
-      "loss": 0.9556,
+      "epoch": 0.02793111971656917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8609332113449224e-05,
+      "loss": 1.168,
       "step": 1017
     },
     {
-      "epoch": 0.3326797385620915,
-      "grad_norm": 1.3172943592071533,
-      "learning_rate": 1.556889683222207e-05,
-      "loss": 0.9646,
+      "epoch": 0.027958583944412405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8627630375114365e-05,
+      "loss": 1.1642,
       "step": 1018
     },
     {
-      "epoch": 0.3330065359477124,
-      "grad_norm": 1.4037139415740967,
-      "learning_rate": 1.5560102055329113e-05,
-      "loss": 0.9446,
+      "epoch": 0.027986048172255636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8645928636779506e-05,
+      "loss": 1.1227,
       "step": 1019
     },
     {
-      "epoch": 0.3333333333333333,
-      "grad_norm": 1.3917548656463623,
-      "learning_rate": 1.5551301048914863e-05,
-      "loss": 0.9722,
+      "epoch": 0.02801351240009887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866422689844465e-05,
+      "loss": 1.1013,
       "step": 1020
     },
     {
-      "epoch": 0.33366013071895423,
-      "grad_norm": 3.2023074626922607,
-      "learning_rate": 1.5542493822839938e-05,
-      "loss": 0.9597,
+      "epoch": 0.028040976627942107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868252516010979e-05,
+      "loss": 1.1178,
       "step": 1021
     },
     {
-      "epoch": 0.33398692810457514,
-      "grad_norm": 1.3965163230895996,
-      "learning_rate": 1.5533680386971927e-05,
-      "loss": 0.9851,
+      "epoch": 0.02806844085578534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8700823421774932e-05,
+      "loss": 1.0314,
       "step": 1022
     },
     {
-      "epoch": 0.33431372549019606,
-      "grad_norm": 1.4588124752044678,
-      "learning_rate": 1.5524860751185377e-05,
-      "loss": 0.9772,
+      "epoch": 0.028095905083628574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8719121683440076e-05,
+      "loss": 1.099,
       "step": 1023
     },
     {
-      "epoch": 0.33464052287581697,
-      "grad_norm": 1.393310785293579,
-      "learning_rate": 1.5516034925361785e-05,
-      "loss": 0.929,
+      "epoch": 0.02812336931147181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8737419945105217e-05,
+      "loss": 1.1772,
       "step": 1024
     },
     {
-      "epoch": 0.3349673202614379,
-      "grad_norm": 1.4326519966125488,
-      "learning_rate": 1.5507202919389573e-05,
-      "loss": 0.9754,
+      "epoch": 0.02815083353931504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8755718206770358e-05,
+      "loss": 1.14,
       "step": 1025
     },
     {
-      "epoch": 0.3352941176470588,
-      "grad_norm": 1.4727058410644531,
-      "learning_rate": 1.5498364743164098e-05,
-      "loss": 0.9922,
+      "epoch": 0.028178297767158276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8774016468435502e-05,
+      "loss": 1.0271,
       "step": 1026
     },
     {
-      "epoch": 0.3356209150326797,
-      "grad_norm": 1.455361008644104,
-      "learning_rate": 1.5489520406587622e-05,
-      "loss": 0.9589,
+      "epoch": 0.02820576199500151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8792314730100643e-05,
+      "loss": 1.1272,
       "step": 1027
     },
     {
-      "epoch": 0.3359477124183007,
-      "grad_norm": 1.4622852802276611,
-      "learning_rate": 1.5480669919569313e-05,
-      "loss": 0.9505,
+      "epoch": 0.028233226222844746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8810612991765784e-05,
+      "loss": 1.0765,
       "step": 1028
     },
     {
-      "epoch": 0.3362745098039216,
-      "grad_norm": 1.4417495727539062,
-      "learning_rate": 1.5471813292025232e-05,
-      "loss": 0.9878,
+      "epoch": 0.028260690450687978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8828911253430925e-05,
+      "loss": 1.1314,
       "step": 1029
     },
     {
-      "epoch": 0.3366013071895425,
-      "grad_norm": 1.50552499294281,
-      "learning_rate": 1.5462950533878318e-05,
-      "loss": 0.9646,
+      "epoch": 0.028288154678531213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884720951509607e-05,
+      "loss": 1.1808,
       "step": 1030
     },
     {
-      "epoch": 0.3369281045751634,
-      "grad_norm": 1.4367547035217285,
-      "learning_rate": 1.5454081655058378e-05,
-      "loss": 0.9638,
+      "epoch": 0.02831561890637445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886550777676121e-05,
+      "loss": 1.1776,
       "step": 1031
     },
     {
-      "epoch": 0.33725490196078434,
-      "grad_norm": 1.4587165117263794,
-      "learning_rate": 1.544520666550207e-05,
-      "loss": 0.9809,
+      "epoch": 0.02834308313421768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888380603842635e-05,
+      "loss": 1.1423,
       "step": 1032
     },
     {
-      "epoch": 0.33758169934640525,
-      "grad_norm": 1.3695908784866333,
-      "learning_rate": 1.543632557515291e-05,
-      "loss": 1.0011,
+      "epoch": 0.028370547362060915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8902104300091492e-05,
+      "loss": 1.0587,
       "step": 1033
     },
     {
-      "epoch": 0.33790849673202616,
-      "grad_norm": 1.4714150428771973,
-      "learning_rate": 1.5427438393961246e-05,
-      "loss": 0.9753,
+      "epoch": 0.02839801158990415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920402561756633e-05,
+      "loss": 1.0729,
       "step": 1034
     },
     {
-      "epoch": 0.3382352941176471,
-      "grad_norm": 1.4811055660247803,
-      "learning_rate": 1.541854513188425e-05,
-      "loss": 1.0099,
+      "epoch": 0.028425475817747382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8938700823421777e-05,
+      "loss": 1.077,
       "step": 1035
     },
     {
-      "epoch": 0.338562091503268,
-      "grad_norm": 1.468157172203064,
-      "learning_rate": 1.54096457988859e-05,
-      "loss": 0.947,
+      "epoch": 0.028452940045590618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8956999085086918e-05,
+      "loss": 1.1567,
       "step": 1036
     },
     {
-      "epoch": 0.3388888888888889,
-      "grad_norm": 1.3597301244735718,
-      "learning_rate": 1.540074040493698e-05,
-      "loss": 0.9744,
+      "epoch": 0.028480404273433853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897529734675206e-05,
+      "loss": 1.0717,
       "step": 1037
     },
     {
-      "epoch": 0.3392156862745098,
-      "grad_norm": 1.4459052085876465,
-      "learning_rate": 1.5391828960015074e-05,
-      "loss": 1.0102,
+      "epoch": 0.028507868501277088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89935956084172e-05,
+      "loss": 1.2169,
       "step": 1038
     },
     {
-      "epoch": 0.33954248366013073,
-      "grad_norm": 1.4411859512329102,
-      "learning_rate": 1.5382911474104535e-05,
-      "loss": 0.9844,
+      "epoch": 0.02853533272912032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9011893870082344e-05,
+      "loss": 1.1341,
       "step": 1039
     },
     {
-      "epoch": 0.33986928104575165,
-      "grad_norm": 1.416261076927185,
-      "learning_rate": 1.537398795719648e-05,
-      "loss": 0.9571,
+      "epoch": 0.028562796956963555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030192131747485e-05,
+      "loss": 1.0528,
       "step": 1040
     },
     {
-      "epoch": 0.34019607843137256,
-      "grad_norm": 1.3738455772399902,
-      "learning_rate": 1.53650584192888e-05,
-      "loss": 0.9495,
+      "epoch": 0.02859026118480679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9048490393412626e-05,
+      "loss": 1.1748,
       "step": 1041
     },
     {
-      "epoch": 0.3405228758169935,
-      "grad_norm": 1.5091266632080078,
-      "learning_rate": 1.5356122870386117e-05,
-      "loss": 1.0001,
+      "epoch": 0.028617725412650022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9066788655077767e-05,
+      "loss": 1.1563,
       "step": 1042
     },
     {
-      "epoch": 0.3408496732026144,
-      "grad_norm": 1.3718916177749634,
-      "learning_rate": 1.534718132049979e-05,
-      "loss": 0.9893,
+      "epoch": 0.028645189640493257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908508691674291e-05,
+      "loss": 1.0232,
       "step": 1043
     },
     {
-      "epoch": 0.3411764705882353,
-      "grad_norm": 1.522019624710083,
-      "learning_rate": 1.533823377964791e-05,
-      "loss": 1.0011,
+      "epoch": 0.028672653868336492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9103385178408052e-05,
+      "loss": 1.1147,
       "step": 1044
     },
     {
-      "epoch": 0.3415032679738562,
-      "grad_norm": 1.3760967254638672,
-      "learning_rate": 1.532928025785527e-05,
-      "loss": 0.9841,
+      "epoch": 0.028700118096179728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9121683440073197e-05,
+      "loss": 1.1955,
       "step": 1045
     },
     {
-      "epoch": 0.34183006535947713,
-      "grad_norm": 1.3594599962234497,
-      "learning_rate": 1.5320320765153367e-05,
-      "loss": 0.9813,
+      "epoch": 0.02872758232402296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9139981701738338e-05,
+      "loss": 1.0874,
       "step": 1046
     },
     {
-      "epoch": 0.34215686274509804,
-      "grad_norm": 1.4332090616226196,
-      "learning_rate": 1.5311355311580393e-05,
-      "loss": 0.9737,
+      "epoch": 0.028755046551866194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915827996340348e-05,
+      "loss": 1.0267,
       "step": 1047
     },
     {
-      "epoch": 0.34248366013071896,
-      "grad_norm": 1.4837692975997925,
-      "learning_rate": 1.5302383907181212e-05,
-      "loss": 0.9771,
+      "epoch": 0.02878251077970943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917657822506862e-05,
+      "loss": 1.0837,
       "step": 1048
     },
     {
-      "epoch": 0.3428104575163399,
-      "grad_norm": 1.4491974115371704,
-      "learning_rate": 1.5293406562007365e-05,
-      "loss": 1.0313,
+      "epoch": 0.02880997500755266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194876486733764e-05,
+      "loss": 1.1683,
       "step": 1049
     },
     {
-      "epoch": 0.3431372549019608,
-      "grad_norm": 1.3956280946731567,
-      "learning_rate": 1.5284423286117035e-05,
-      "loss": 0.9551,
+      "epoch": 0.028837439235395897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213174748398905e-05,
+      "loss": 1.0625,
       "step": 1050
     },
     {
-      "epoch": 0.3434640522875817,
-      "grad_norm": 1.4168422222137451,
-      "learning_rate": 1.5275434089575055e-05,
-      "loss": 0.9786,
+      "epoch": 0.028864903463239132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231473010064046e-05,
+      "loss": 1.1384,
       "step": 1051
     },
     {
-      "epoch": 0.3437908496732026,
-      "grad_norm": 1.467544674873352,
-      "learning_rate": 1.52664389824529e-05,
-      "loss": 0.9978,
+      "epoch": 0.028892367691082364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9249771271729187e-05,
+      "loss": 1.2318,
       "step": 1052
     },
     {
-      "epoch": 0.34411764705882353,
-      "grad_norm": 1.3388727903366089,
-      "learning_rate": 1.5257437974828652e-05,
-      "loss": 0.9803,
+      "epoch": 0.0289198319189256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268069533394327e-05,
+      "loss": 1.1513,
       "step": 1053
     },
     {
-      "epoch": 0.34444444444444444,
-      "grad_norm": 1.4426274299621582,
-      "learning_rate": 1.524843107678702e-05,
-      "loss": 1.005,
+      "epoch": 0.028947296146768834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9286367795059472e-05,
+      "loss": 1.1262,
       "step": 1054
     },
     {
-      "epoch": 0.34477124183006536,
-      "grad_norm": 1.5069611072540283,
-      "learning_rate": 1.5239418298419304e-05,
-      "loss": 0.9941,
+      "epoch": 0.02897476037461207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9304666056724613e-05,
+      "loss": 1.0817,
       "step": 1055
     },
     {
-      "epoch": 0.34509803921568627,
-      "grad_norm": 1.3334639072418213,
-      "learning_rate": 1.5230399649823389e-05,
-      "loss": 0.9653,
+      "epoch": 0.0290022246024553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322964318389754e-05,
+      "loss": 1.223,
       "step": 1056
     },
     {
-      "epoch": 0.3454248366013072,
-      "grad_norm": 1.4442048072814941,
-      "learning_rate": 1.5221375141103742e-05,
-      "loss": 0.9752,
+      "epoch": 0.029029688830298536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341262580054895e-05,
+      "loss": 1.059,
       "step": 1057
     },
     {
-      "epoch": 0.3457516339869281,
-      "grad_norm": 1.4541873931884766,
-      "learning_rate": 1.52123447823714e-05,
-      "loss": 0.9338,
+      "epoch": 0.02905715305814177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935956084172004e-05,
+      "loss": 1.2723,
       "step": 1058
     },
     {
-      "epoch": 0.346078431372549,
-      "grad_norm": 1.4234938621520996,
-      "learning_rate": 1.5203308583743942e-05,
-      "loss": 0.9671,
+      "epoch": 0.029084617285985003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937785910338518e-05,
+      "loss": 1.1831,
       "step": 1059
     },
     {
-      "epoch": 0.3464052287581699,
-      "grad_norm": 1.4356170892715454,
-      "learning_rate": 1.5194266555345505e-05,
-      "loss": 0.979,
+      "epoch": 0.02911208151382824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939615736505032e-05,
+      "loss": 1.1887,
       "step": 1060
     },
     {
-      "epoch": 0.34673202614379084,
-      "grad_norm": 1.4986679553985596,
-      "learning_rate": 1.5185218707306745e-05,
-      "loss": 0.9428,
+      "epoch": 0.029139545741671474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941445562671546e-05,
+      "loss": 1.1568,
       "step": 1061
     },
     {
-      "epoch": 0.34705882352941175,
-      "grad_norm": 1.379783272743225,
-      "learning_rate": 1.5176165049764846e-05,
-      "loss": 0.973,
+      "epoch": 0.02916700996951471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432753888380606e-05,
+      "loss": 1.0676,
       "step": 1062
     },
     {
-      "epoch": 0.34738562091503267,
-      "grad_norm": 1.3522515296936035,
-      "learning_rate": 1.516710559286349e-05,
-      "loss": 0.9891,
+      "epoch": 0.02919447419735794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451052150045747e-05,
+      "loss": 1.198,
       "step": 1063
     },
     {
-      "epoch": 0.3477124183006536,
-      "grad_norm": 1.4829210042953491,
-      "learning_rate": 1.5158040346752878e-05,
-      "loss": 0.9812,
+      "epoch": 0.029221938425201176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946935041171089e-05,
+      "loss": 1.1735,
       "step": 1064
     },
     {
-      "epoch": 0.3480392156862745,
-      "grad_norm": 1.4234751462936401,
-      "learning_rate": 1.5148969321589672e-05,
-      "loss": 0.9865,
+      "epoch": 0.02924940265304441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487648673376032e-05,
+      "loss": 1.103,
       "step": 1065
     },
     {
-      "epoch": 0.3483660130718954,
-      "grad_norm": 1.4595000743865967,
-      "learning_rate": 1.513989252753703e-05,
-      "loss": 0.9989,
+      "epoch": 0.029276866880887643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505946935041173e-05,
+      "loss": 1.1534,
       "step": 1066
     },
     {
-      "epoch": 0.3486928104575163,
-      "grad_norm": 1.395458698272705,
-      "learning_rate": 1.5130809974764557e-05,
-      "loss": 0.9348,
+      "epoch": 0.029304331108730878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524245196706314e-05,
+      "loss": 1.0709,
       "step": 1067
     },
     {
-      "epoch": 0.34901960784313724,
-      "grad_norm": 1.4494099617004395,
-      "learning_rate": 1.5121721673448319e-05,
-      "loss": 0.9499,
+      "epoch": 0.029331795336574113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9542543458371458e-05,
+      "loss": 1.1202,
       "step": 1068
     },
     {
-      "epoch": 0.34934640522875815,
-      "grad_norm": 1.4799957275390625,
-      "learning_rate": 1.5112627633770824e-05,
-      "loss": 0.9825,
+      "epoch": 0.029359259564417345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95608417200366e-05,
+      "loss": 1.1395,
       "step": 1069
     },
     {
-      "epoch": 0.34967320261437906,
-      "grad_norm": 1.453450083732605,
-      "learning_rate": 1.5103527865921005e-05,
-      "loss": 0.9716,
+      "epoch": 0.02938672379226058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957913998170174e-05,
+      "loss": 1.0221,
       "step": 1070
     },
     {
-      "epoch": 0.35,
-      "grad_norm": 1.4349570274353027,
-      "learning_rate": 1.5094422380094211e-05,
-      "loss": 0.9817,
+      "epoch": 0.029414188020103815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959743824336688e-05,
+      "loss": 1.1808,
       "step": 1071
     },
     {
-      "epoch": 0.3503267973856209,
-      "grad_norm": 1.4272875785827637,
-      "learning_rate": 1.5085311186492206e-05,
-      "loss": 1.0094,
+      "epoch": 0.02944165224794705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615736505032025e-05,
+      "loss": 1.0806,
       "step": 1072
     },
     {
-      "epoch": 0.3506535947712418,
-      "grad_norm": 1.4473227262496948,
-      "learning_rate": 1.507619429532314e-05,
-      "loss": 0.965,
+      "epoch": 0.029469116475790282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9634034766697166e-05,
+      "loss": 1.1195,
       "step": 1073
     },
     {
-      "epoch": 0.3509803921568627,
-      "grad_norm": 1.4670190811157227,
-      "learning_rate": 1.5067071716801546e-05,
-      "loss": 0.9523,
+      "epoch": 0.029496580703633517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9652333028362307e-05,
+      "loss": 1.136,
       "step": 1074
     },
     {
-      "epoch": 0.35130718954248363,
-      "grad_norm": 1.4478070735931396,
-      "learning_rate": 1.5057943461148341e-05,
-      "loss": 0.9548,
+      "epoch": 0.029524044931476753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9670631290027448e-05,
+      "loss": 1.1395,
       "step": 1075
     },
     {
-      "epoch": 0.3516339869281046,
-      "grad_norm": 1.4490259885787964,
-      "learning_rate": 1.5048809538590789e-05,
-      "loss": 1.0016,
+      "epoch": 0.029551509159319984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968892955169259e-05,
+      "loss": 1.0842,
       "step": 1076
     },
     {
-      "epoch": 0.3519607843137255,
-      "grad_norm": 1.3881380558013916,
-      "learning_rate": 1.503966995936251e-05,
-      "loss": 0.9748,
+      "epoch": 0.02957897338716322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9707227813357733e-05,
+      "loss": 1.0635,
       "step": 1077
     },
     {
-      "epoch": 0.35228758169934643,
-      "grad_norm": 1.41504967212677,
-      "learning_rate": 1.5030524733703464e-05,
-      "loss": 0.9542,
+      "epoch": 0.029606437615006455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9725526075022874e-05,
+      "loss": 1.1433,
       "step": 1078
     },
     {
-      "epoch": 0.35261437908496734,
-      "grad_norm": 1.3964641094207764,
-      "learning_rate": 1.5021373871859926e-05,
-      "loss": 0.9646,
+      "epoch": 0.029633901842849687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743824336688015e-05,
+      "loss": 1.1236,
       "step": 1079
     },
     {
-      "epoch": 0.35294117647058826,
-      "grad_norm": 1.382485032081604,
-      "learning_rate": 1.5012217384084501e-05,
-      "loss": 0.96,
+      "epoch": 0.029661366070692922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762122598353156e-05,
+      "loss": 1.1397,
       "step": 1080
     },
     {
-      "epoch": 0.35326797385620917,
-      "grad_norm": 1.4352248907089233,
-      "learning_rate": 1.5003055280636085e-05,
-      "loss": 0.9609,
+      "epoch": 0.029688830298536157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97804208600183e-05,
+      "loss": 1.0271,
       "step": 1081
     },
     {
-      "epoch": 0.3535947712418301,
-      "grad_norm": 1.3766647577285767,
-      "learning_rate": 1.4993887571779877e-05,
-      "loss": 0.9652,
+      "epoch": 0.029716294526379392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979871912168344e-05,
+      "loss": 1.1269,
       "step": 1082
     },
     {
-      "epoch": 0.353921568627451,
-      "grad_norm": 1.4267351627349854,
-      "learning_rate": 1.4984714267787344e-05,
-      "loss": 0.9996,
+      "epoch": 0.029743758754222624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817017383348586e-05,
+      "loss": 1.0535,
       "step": 1083
     },
     {
-      "epoch": 0.3542483660130719,
-      "grad_norm": 1.377723217010498,
-      "learning_rate": 1.4975535378936228e-05,
-      "loss": 0.9965,
+      "epoch": 0.02977122298206586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835315645013727e-05,
+      "loss": 1.1855,
       "step": 1084
     },
     {
-      "epoch": 0.3545751633986928,
-      "grad_norm": 1.5746901035308838,
-      "learning_rate": 1.496635091551053e-05,
-      "loss": 0.9503,
+      "epoch": 0.029798687209909094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853613906678867e-05,
+      "loss": 1.1591,
       "step": 1085
     },
     {
-      "epoch": 0.35490196078431374,
-      "grad_norm": 1.4073466062545776,
-      "learning_rate": 1.4957160887800494e-05,
-      "loss": 0.9606,
+      "epoch": 0.029826151437752326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987191216834401e-05,
+      "loss": 1.0692,
       "step": 1086
     },
     {
-      "epoch": 0.35522875816993466,
-      "grad_norm": 1.3964470624923706,
-      "learning_rate": 1.49479653061026e-05,
-      "loss": 0.9409,
+      "epoch": 0.02985361566559556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890210430009153e-05,
+      "loss": 1.1105,
       "step": 1087
     },
     {
-      "epoch": 0.35555555555555557,
-      "grad_norm": 1.3823732137680054,
-      "learning_rate": 1.4938764180719548e-05,
-      "loss": 0.991,
+      "epoch": 0.029881079893438797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908508691674294e-05,
+      "loss": 1.2206,
       "step": 1088
     },
     {
-      "epoch": 0.3558823529411765,
-      "grad_norm": 1.3708021640777588,
-      "learning_rate": 1.4929557521960251e-05,
-      "loss": 0.9796,
+      "epoch": 0.029908544121282032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926806953339435e-05,
+      "loss": 1.2359,
       "step": 1089
     },
     {
-      "epoch": 0.3562091503267974,
-      "grad_norm": 1.4518510103225708,
-      "learning_rate": 1.4920345340139819e-05,
-      "loss": 0.9531,
+      "epoch": 0.029936008349125263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945105215004575e-05,
+      "loss": 1.01,
       "step": 1090
     },
     {
-      "epoch": 0.3565359477124183,
-      "grad_norm": 1.4117237329483032,
-      "learning_rate": 1.4911127645579557e-05,
-      "loss": 0.9926,
+      "epoch": 0.0299634725769685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996340347666972e-05,
+      "loss": 1.1895,
       "step": 1091
     },
     {
-      "epoch": 0.3568627450980392,
-      "grad_norm": 1.4786967039108276,
-      "learning_rate": 1.490190444860694e-05,
-      "loss": 0.9772,
+      "epoch": 0.029990936804811734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998170173833486e-05,
+      "loss": 1.1586,
       "step": 1092
     },
     {
-      "epoch": 0.35718954248366014,
-      "grad_norm": 1.4591360092163086,
-      "learning_rate": 1.4892675759555607e-05,
-      "loss": 0.9644,
+      "epoch": 0.030018401032654966,
+      "grad_norm": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.1984,
       "step": 1093
     },
     {
-      "epoch": 0.35751633986928105,
-      "grad_norm": 1.3317713737487793,
-      "learning_rate": 1.4883441588765355e-05,
-      "loss": 0.9492,
+      "epoch": 0.0300458652604982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999960438064e-05,
+      "loss": 1.0796,
       "step": 1094
     },
     {
-      "epoch": 0.35784313725490197,
-      "grad_norm": 1.3915523290634155,
-      "learning_rate": 1.4874201946582119e-05,
-      "loss": 1.0093,
+      "epoch": 0.030073329488341436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999841752255e-05,
+      "loss": 1.1995,
       "step": 1095
     },
     {
-      "epoch": 0.3581699346405229,
-      "grad_norm": 1.4472649097442627,
-      "learning_rate": 1.4864956843357967e-05,
-      "loss": 0.9821,
+      "epoch": 0.030100793716184668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999643942574e-05,
+      "loss": 1.0811,
       "step": 1096
     },
     {
-      "epoch": 0.3584967320261438,
-      "grad_norm": 1.5317211151123047,
-      "learning_rate": 1.4855706289451088e-05,
-      "loss": 1.0163,
+      "epoch": 0.030128257944027903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999367009017e-05,
+      "loss": 1.2153,
       "step": 1097
     },
     {
-      "epoch": 0.3588235294117647,
-      "grad_norm": 1.4196431636810303,
-      "learning_rate": 1.4846450295225769e-05,
-      "loss": 0.9725,
+      "epoch": 0.030155722171871138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99999990109516e-05,
+      "loss": 1.1879,
       "step": 1098
     },
     {
-      "epoch": 0.3591503267973856,
-      "grad_norm": 1.411250352859497,
-      "learning_rate": 1.4837188871052399e-05,
-      "loss": 0.9929,
+      "epoch": 0.030183186399714373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999857577031e-05,
+      "loss": 1.0522,
       "step": 1099
     },
     {
-      "epoch": 0.35947712418300654,
-      "grad_norm": 1.4194989204406738,
-      "learning_rate": 1.482792202730745e-05,
-      "loss": 0.9745,
+      "epoch": 0.030210650627557605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999806146516e-05,
+      "loss": 1.1471,
       "step": 1100
     },
     {
-      "epoch": 0.35980392156862745,
-      "grad_norm": 1.3844318389892578,
-      "learning_rate": 1.4818649774373466e-05,
-      "loss": 0.9404,
+      "epoch": 0.03023811485540084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999746803615e-05,
+      "loss": 1.1373,
       "step": 1101
     },
     {
-      "epoch": 0.36013071895424836,
-      "grad_norm": 1.432913064956665,
-      "learning_rate": 1.4809372122639047e-05,
-      "loss": 0.9472,
+      "epoch": 0.030265579083244076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999679548329e-05,
+      "loss": 1.1516,
       "step": 1102
     },
     {
-      "epoch": 0.3604575163398693,
-      "grad_norm": 1.3780349493026733,
-      "learning_rate": 1.480008908249885e-05,
-      "loss": 0.9529,
+      "epoch": 0.030293043311087307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999604380658e-05,
+      "loss": 1.0913,
       "step": 1103
     },
     {
-      "epoch": 0.3607843137254902,
-      "grad_norm": 1.4832760095596313,
-      "learning_rate": 1.4790800664353562e-05,
-      "loss": 0.9996,
+      "epoch": 0.030320507538930543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999995213006026e-05,
+      "loss": 1.0247,
       "step": 1104
     },
     {
-      "epoch": 0.3611111111111111,
-      "grad_norm": 1.3634289503097534,
-      "learning_rate": 1.4781506878609899e-05,
-      "loss": 0.9135,
+      "epoch": 0.030347971766773778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999994303081638e-05,
+      "loss": 1.2142,
       "step": 1105
     },
     {
-      "epoch": 0.361437908496732,
-      "grad_norm": 1.3584237098693848,
-      "learning_rate": 1.477220773568059e-05,
-      "loss": 0.985,
+      "epoch": 0.030375435994617013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999993314033422e-05,
+      "loss": 1.1517,
       "step": 1106
     },
     {
-      "epoch": 0.36176470588235293,
-      "grad_norm": 1.3555259704589844,
-      "learning_rate": 1.4762903245984369e-05,
-      "loss": 0.9496,
+      "epoch": 0.030402900222460245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999992245861382e-05,
+      "loss": 1.2003,
       "step": 1107
     },
     {
-      "epoch": 0.36209150326797385,
-      "grad_norm": 1.4626364707946777,
-      "learning_rate": 1.475359341994595e-05,
-      "loss": 0.9673,
+      "epoch": 0.03043036445030348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999109856553e-05,
+      "loss": 1.1503,
       "step": 1108
     },
     {
-      "epoch": 0.36241830065359476,
-      "grad_norm": 1.338814377784729,
-      "learning_rate": 1.4744278267996044e-05,
-      "loss": 0.9605,
+      "epoch": 0.030457828678146715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999998987214588e-05,
+      "loss": 1.1548,
       "step": 1109
     },
     {
-      "epoch": 0.3627450980392157,
-      "grad_norm": 1.4388253688812256,
-      "learning_rate": 1.473495780057131e-05,
-      "loss": 0.9897,
+      "epoch": 0.030485292905989947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999988566602435e-05,
+      "loss": 1.0836,
       "step": 1110
     },
     {
-      "epoch": 0.3630718954248366,
-      "grad_norm": 1.321526050567627,
-      "learning_rate": 1.4725632028114378e-05,
-      "loss": 0.932,
+      "epoch": 0.030512757133833182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999987181935202e-05,
+      "loss": 1.0645,
       "step": 1111
     },
     {
-      "epoch": 0.3633986928104575,
-      "grad_norm": 1.41892409324646,
-      "learning_rate": 1.4716300961073808e-05,
-      "loss": 0.9441,
+      "epoch": 0.030540221361676417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99999857181442e-05,
+      "loss": 1.1333,
       "step": 1112
     },
     {
-      "epoch": 0.3637254901960784,
-      "grad_norm": 1.439854383468628,
-      "learning_rate": 1.47069646099041e-05,
-      "loss": 0.9191,
+      "epoch": 0.03056768558951965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999984175229437e-05,
+      "loss": 1.0767,
       "step": 1113
     },
     {
-      "epoch": 0.36405228758169933,
-      "grad_norm": 1.4573177099227905,
-      "learning_rate": 1.469762298506568e-05,
-      "loss": 0.9354,
+      "epoch": 0.030595149817362884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999998255319093e-05,
+      "loss": 1.0974,
       "step": 1114
     },
     {
-      "epoch": 0.36437908496732024,
-      "grad_norm": 1.3912391662597656,
-      "learning_rate": 1.468827609702487e-05,
-      "loss": 0.9751,
+      "epoch": 0.03062261404520612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999980852028683e-05,
+      "loss": 1.173,
       "step": 1115
     },
     {
-      "epoch": 0.36470588235294116,
-      "grad_norm": 1.5299745798110962,
-      "learning_rate": 1.4678923956253894e-05,
-      "loss": 0.9945,
+      "epoch": 0.030650078273049355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999979071742712e-05,
+      "loss": 1.1844,
       "step": 1116
     },
     {
-      "epoch": 0.36503267973856207,
-      "grad_norm": 1.721612572669983,
-      "learning_rate": 1.4669566573230866e-05,
-      "loss": 0.9887,
+      "epoch": 0.030677542500892586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999977212333035e-05,
+      "loss": 1.1299,
       "step": 1117
     },
     {
-      "epoch": 0.365359477124183,
-      "grad_norm": 1.374308466911316,
-      "learning_rate": 1.4660203958439768e-05,
-      "loss": 0.9782,
+      "epoch": 0.03070500672873582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999975273799666e-05,
+      "loss": 1.1294,
       "step": 1118
     },
     {
-      "epoch": 0.3656862745098039,
-      "grad_norm": 1.4233704805374146,
-      "learning_rate": 1.4650836122370442e-05,
-      "loss": 0.9517,
+      "epoch": 0.030732470956579057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999997325614262e-05,
+      "loss": 1.1804,
       "step": 1119
     },
     {
-      "epoch": 0.3660130718954248,
-      "grad_norm": 1.3755131959915161,
-      "learning_rate": 1.4641463075518587e-05,
-      "loss": 0.9706,
+      "epoch": 0.03075993518442229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999971159361906e-05,
+      "loss": 0.9846,
       "step": 1120
     },
     {
-      "epoch": 0.3663398692810457,
-      "grad_norm": 1.4304957389831543,
-      "learning_rate": 1.4632084828385732e-05,
-      "loss": 0.9685,
+      "epoch": 0.030787399412265524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999968983457552e-05,
+      "loss": 1.0946,
       "step": 1121
     },
     {
-      "epoch": 0.36666666666666664,
-      "grad_norm": 1.3257341384887695,
-      "learning_rate": 1.4622701391479237e-05,
-      "loss": 0.9858,
+      "epoch": 0.03081486364010876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999996672842957e-05,
+      "loss": 1.1655,
       "step": 1122
     },
     {
-      "epoch": 0.36699346405228755,
-      "grad_norm": 1.3386245965957642,
-      "learning_rate": 1.4613312775312276e-05,
-      "loss": 0.9566,
+      "epoch": 0.030842327867951994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999964394277974e-05,
+      "loss": 0.9561,
       "step": 1123
     },
     {
-      "epoch": 0.3673202614379085,
-      "grad_norm": 1.3333191871643066,
-      "learning_rate": 1.460391899040383e-05,
-      "loss": 0.9534,
+      "epoch": 0.030869792095795226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999996198100279e-05,
+      "loss": 1.1916,
       "step": 1124
     },
     {
-      "epoch": 0.36764705882352944,
-      "grad_norm": 1.3495841026306152,
-      "learning_rate": 1.4594520047278662e-05,
-      "loss": 1.0114,
+      "epoch": 0.03089725632363846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999995948860403e-05,
+      "loss": 1.1862,
       "step": 1125
     },
     {
-      "epoch": 0.36797385620915035,
-      "grad_norm": 1.3486690521240234,
-      "learning_rate": 1.4585115956467322e-05,
-      "loss": 0.9542,
+      "epoch": 0.030924720551481696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999956917081716e-05,
+      "loss": 1.0918,
       "step": 1126
     },
     {
-      "epoch": 0.36830065359477127,
-      "grad_norm": 1.3676090240478516,
-      "learning_rate": 1.457570672850612e-05,
-      "loss": 0.9581,
+      "epoch": 0.030952184779324928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999954266435873e-05,
+      "loss": 1.0988,
       "step": 1127
     },
     {
-      "epoch": 0.3686274509803922,
-      "grad_norm": 1.2931129932403564,
-      "learning_rate": 1.4566292373937133e-05,
-      "loss": 0.9605,
+      "epoch": 0.030979649007168163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999951536666517e-05,
+      "loss": 1.1725,
       "step": 1128
     },
     {
-      "epoch": 0.3689542483660131,
-      "grad_norm": 1.375104546546936,
-      "learning_rate": 1.4556872903308171e-05,
-      "loss": 0.973,
+      "epoch": 0.0310071132350114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999994872777367e-05,
+      "loss": 1.0359,
       "step": 1129
     },
     {
-      "epoch": 0.369281045751634,
-      "grad_norm": 1.3955081701278687,
-      "learning_rate": 1.4547448327172778e-05,
-      "loss": 0.9117,
+      "epoch": 0.03103457746285463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999945839757356e-05,
+      "loss": 1.1193,
       "step": 1130
     },
     {
-      "epoch": 0.3696078431372549,
-      "grad_norm": 1.3916276693344116,
-      "learning_rate": 1.4538018656090223e-05,
-      "loss": 0.9879,
+      "epoch": 0.031062041690697866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999942872617594e-05,
+      "loss": 1.1906,
       "step": 1131
     },
     {
-      "epoch": 0.36993464052287583,
-      "grad_norm": 1.4087636470794678,
-      "learning_rate": 1.4528583900625481e-05,
-      "loss": 0.9647,
+      "epoch": 0.0310895059185411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999939826354415e-05,
+      "loss": 1.1713,
       "step": 1132
     },
     {
-      "epoch": 0.37026143790849675,
-      "grad_norm": 1.584491491317749,
-      "learning_rate": 1.451914407134922e-05,
-      "loss": 0.9684,
+      "epoch": 0.031116970146384336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999936700967835e-05,
+      "loss": 1.1259,
       "step": 1133
     },
     {
-      "epoch": 0.37058823529411766,
-      "grad_norm": 1.4165575504302979,
-      "learning_rate": 1.45096991788378e-05,
-      "loss": 0.9789,
+      "epoch": 0.031144434374227568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999933496457884e-05,
+      "loss": 1.0923,
       "step": 1134
     },
     {
-      "epoch": 0.3709150326797386,
-      "grad_norm": 1.4099786281585693,
-      "learning_rate": 1.4500249233673247e-05,
-      "loss": 0.9716,
+      "epoch": 0.031171898602070803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999930212824587e-05,
+      "loss": 1.1053,
       "step": 1135
     },
     {
-      "epoch": 0.3712418300653595,
-      "grad_norm": 1.4616727828979492,
-      "learning_rate": 1.4490794246443249e-05,
-      "loss": 0.9576,
+      "epoch": 0.031199362829914038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999926850067967e-05,
+      "loss": 1.1274,
       "step": 1136
     },
     {
-      "epoch": 0.3715686274509804,
-      "grad_norm": 1.3645052909851074,
-      "learning_rate": 1.4481334227741153e-05,
-      "loss": 0.9757,
+      "epoch": 0.03122682705775727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999992340818805e-05,
+      "loss": 1.156,
       "step": 1137
     },
     {
-      "epoch": 0.3718954248366013,
-      "grad_norm": 1.4270594120025635,
-      "learning_rate": 1.4471869188165923e-05,
-      "loss": 0.9841,
+      "epoch": 0.031254291285600505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999919887184867e-05,
+      "loss": 1.1791,
       "step": 1138
     },
     {
-      "epoch": 0.37222222222222223,
-      "grad_norm": 1.3517379760742188,
-      "learning_rate": 1.4462399138322169e-05,
-      "loss": 0.996,
+      "epoch": 0.03128175551344374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999916287058444e-05,
+      "loss": 1.1865,
       "step": 1139
     },
     {
-      "epoch": 0.37254901960784315,
-      "grad_norm": 1.3661614656448364,
-      "learning_rate": 1.4452924088820101e-05,
-      "loss": 0.9565,
+      "epoch": 0.031309219741286975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999991260780881e-05,
+      "loss": 1.0903,
       "step": 1140
     },
     {
-      "epoch": 0.37287581699346406,
-      "grad_norm": 1.3841984272003174,
-      "learning_rate": 1.4443444050275539e-05,
-      "loss": 0.996,
+      "epoch": 0.03133668396913021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999908849435993e-05,
+      "loss": 1.1447,
       "step": 1141
     },
     {
-      "epoch": 0.373202614379085,
-      "grad_norm": 1.3503047227859497,
-      "learning_rate": 1.4433959033309888e-05,
-      "loss": 0.9616,
+      "epoch": 0.03136414819697344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999905011940026e-05,
+      "loss": 1.0643,
       "step": 1142
     },
     {
-      "epoch": 0.3735294117647059,
-      "grad_norm": 1.411984920501709,
-      "learning_rate": 1.4424469048550129e-05,
-      "loss": 0.9775,
+      "epoch": 0.031391612424816674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999901095320935e-05,
+      "loss": 1.0389,
       "step": 1143
     },
     {
-      "epoch": 0.3738562091503268,
-      "grad_norm": 1.31766676902771,
-      "learning_rate": 1.4414974106628816e-05,
-      "loss": 0.9931,
+      "epoch": 0.03141907665265991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999897099578755e-05,
+      "loss": 1.1535,
       "step": 1144
     },
     {
-      "epoch": 0.3741830065359477,
-      "grad_norm": 1.4180091619491577,
-      "learning_rate": 1.4405474218184047e-05,
-      "loss": 1.0011,
+      "epoch": 0.031446540880503145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999893024713513e-05,
+      "loss": 1.1708,
       "step": 1145
     },
     {
-      "epoch": 0.37450980392156863,
-      "grad_norm": 1.3562724590301514,
-      "learning_rate": 1.4395969393859478e-05,
-      "loss": 0.9713,
+      "epoch": 0.03147400510834638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999888870725247e-05,
+      "loss": 1.0584,
       "step": 1146
     },
     {
-      "epoch": 0.37483660130718954,
-      "grad_norm": 1.4308663606643677,
-      "learning_rate": 1.4386459644304273e-05,
-      "loss": 0.9814,
+      "epoch": 0.031501469336189615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999884637613983e-05,
+      "loss": 1.1041,
       "step": 1147
     },
     {
-      "epoch": 0.37516339869281046,
-      "grad_norm": 1.3014204502105713,
-      "learning_rate": 1.4376944980173138e-05,
-      "loss": 0.9612,
+      "epoch": 0.03152893356403285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999880325379758e-05,
+      "loss": 1.1139,
       "step": 1148
     },
     {
-      "epoch": 0.37549019607843137,
-      "grad_norm": 1.3908042907714844,
-      "learning_rate": 1.4367425412126266e-05,
-      "loss": 0.9796,
+      "epoch": 0.03155639779187608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999875934022607e-05,
+      "loss": 1.084,
       "step": 1149
     },
     {
-      "epoch": 0.3758169934640523,
-      "grad_norm": 1.435304045677185,
-      "learning_rate": 1.435790095082935e-05,
-      "loss": 0.9393,
+      "epoch": 0.031583862019719314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999871463542567e-05,
+      "loss": 1.0777,
       "step": 1150
     },
     {
-      "epoch": 0.3761437908496732,
-      "grad_norm": 1.4056068658828735,
-      "learning_rate": 1.4348371606953575e-05,
-      "loss": 0.9469,
+      "epoch": 0.03161132624756255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999866913939668e-05,
+      "loss": 1.1598,
       "step": 1151
     },
     {
-      "epoch": 0.3764705882352941,
-      "grad_norm": 1.3714854717254639,
-      "learning_rate": 1.4338837391175582e-05,
-      "loss": 0.946,
+      "epoch": 0.031638790475405784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999862285213948e-05,
+      "loss": 1.1288,
       "step": 1152
     },
     {
-      "epoch": 0.376797385620915,
-      "grad_norm": 1.3619990348815918,
-      "learning_rate": 1.432929831417748e-05,
-      "loss": 0.9415,
+      "epoch": 0.03166625470324902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999857577365443e-05,
+      "loss": 1.1953,
       "step": 1153
     },
     {
-      "epoch": 0.37712418300653594,
-      "grad_norm": 1.368129014968872,
-      "learning_rate": 1.4319754386646824e-05,
-      "loss": 0.9728,
+      "epoch": 0.031693718931092255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999852790394195e-05,
+      "loss": 1.1607,
       "step": 1154
     },
     {
-      "epoch": 0.37745098039215685,
-      "grad_norm": 1.4228788614273071,
-      "learning_rate": 1.4310205619276593e-05,
-      "loss": 0.9914,
+      "epoch": 0.03172118315893549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999847924300238e-05,
+      "loss": 1.0606,
       "step": 1155
     },
     {
-      "epoch": 0.37777777777777777,
-      "grad_norm": 1.3519084453582764,
-      "learning_rate": 1.4300652022765207e-05,
-      "loss": 0.9627,
+      "epoch": 0.03174864738677872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999842979083608e-05,
+      "loss": 1.1599,
       "step": 1156
     },
     {
-      "epoch": 0.3781045751633987,
-      "grad_norm": 1.335600733757019,
-      "learning_rate": 1.4291093607816478e-05,
-      "loss": 0.9618,
+      "epoch": 0.03177611161462195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999983795474435e-05,
+      "loss": 1.0526,
       "step": 1157
     },
     {
-      "epoch": 0.3784313725490196,
-      "grad_norm": 1.524539828300476,
-      "learning_rate": 1.4281530385139627e-05,
-      "loss": 0.9216,
+      "epoch": 0.03180357584246519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999832851282493e-05,
+      "loss": 1.1317,
       "step": 1158
     },
     {
-      "epoch": 0.3787581699346405,
-      "grad_norm": 1.492550015449524,
-      "learning_rate": 1.4271962365449266e-05,
-      "loss": 0.973,
+      "epoch": 0.031831040070308424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999827668698094e-05,
+      "loss": 1.0909,
       "step": 1159
     },
     {
-      "epoch": 0.3790849673202614,
-      "grad_norm": 1.5214204788208008,
-      "learning_rate": 1.4262389559465364e-05,
-      "loss": 0.9549,
+      "epoch": 0.03185850429815166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999982240699118e-05,
+      "loss": 1.1384,
       "step": 1160
     },
     {
-      "epoch": 0.37941176470588234,
-      "grad_norm": 1.339600920677185,
-      "learning_rate": 1.425281197791327e-05,
-      "loss": 0.9283,
+      "epoch": 0.031885968525994894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99998170661618e-05,
+      "loss": 1.2329,
       "step": 1161
     },
     {
-      "epoch": 0.37973856209150325,
-      "grad_norm": 1.339202642440796,
-      "learning_rate": 1.4243229631523678e-05,
-      "loss": 0.9555,
+      "epoch": 0.03191343275383812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999811646209992e-05,
+      "loss": 1.0784,
       "step": 1162
     },
     {
-      "epoch": 0.38006535947712417,
-      "grad_norm": 1.345339298248291,
-      "learning_rate": 1.4233642531032614e-05,
-      "loss": 0.9258,
+      "epoch": 0.03194089698168136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99998061471358e-05,
+      "loss": 1.1146,
       "step": 1163
     },
     {
-      "epoch": 0.3803921568627451,
-      "grad_norm": 1.4355287551879883,
-      "learning_rate": 1.4224050687181442e-05,
-      "loss": 0.9682,
+      "epoch": 0.03196836120952459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999980056893927e-05,
+      "loss": 1.1954,
       "step": 1164
     },
     {
-      "epoch": 0.380718954248366,
-      "grad_norm": 1.3875980377197266,
-      "learning_rate": 1.4214454110716834e-05,
-      "loss": 0.9905,
+      "epoch": 0.03199582543736783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999794911620444e-05,
+      "loss": 1.0894,
       "step": 1165
     },
     {
-      "epoch": 0.3810457516339869,
-      "grad_norm": 1.3643145561218262,
-      "learning_rate": 1.4204852812390757e-05,
-      "loss": 0.9924,
+      "epoch": 0.03202328966521106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999789175179364e-05,
+      "loss": 1.1307,
       "step": 1166
     },
     {
-      "epoch": 0.3813725490196078,
-      "grad_norm": 1.4569053649902344,
-      "learning_rate": 1.4195246802960485e-05,
-      "loss": 0.9786,
+      "epoch": 0.0320507538930543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999783359616082e-05,
+      "loss": 1.0692,
       "step": 1167
     },
     {
-      "epoch": 0.38169934640522873,
-      "grad_norm": 1.388983964920044,
-      "learning_rate": 1.4185636093188558e-05,
-      "loss": 0.9458,
+      "epoch": 0.032078218120897534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999977746493064e-05,
+      "loss": 1.1208,
       "step": 1168
     },
     {
-      "epoch": 0.38202614379084965,
-      "grad_norm": 1.3901563882827759,
-      "learning_rate": 1.4176020693842782e-05,
-      "loss": 0.9605,
+      "epoch": 0.03210568234874076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999771491123084e-05,
+      "loss": 1.1653,
       "step": 1169
     },
     {
-      "epoch": 0.38235294117647056,
-      "grad_norm": 1.4633911848068237,
-      "learning_rate": 1.4166400615696232e-05,
-      "loss": 0.9381,
+      "epoch": 0.032133146576584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999765438193463e-05,
+      "loss": 1.1486,
       "step": 1170
     },
     {
-      "epoch": 0.3826797385620915,
-      "grad_norm": 1.4735164642333984,
-      "learning_rate": 1.4156775869527204e-05,
-      "loss": 0.9582,
+      "epoch": 0.03216061080442723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999975930614183e-05,
+      "loss": 1.0872,
       "step": 1171
     },
     {
-      "epoch": 0.38300653594771245,
-      "grad_norm": 1.483892798423767,
-      "learning_rate": 1.4147146466119235e-05,
-      "loss": 0.9658,
+      "epoch": 0.03218807503227047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999975309496822e-05,
+      "loss": 1.1159,
       "step": 1172
     },
     {
-      "epoch": 0.38333333333333336,
-      "grad_norm": 1.4159458875656128,
-      "learning_rate": 1.4137512416261085e-05,
-      "loss": 0.935,
+      "epoch": 0.0322155392601137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999746804672694e-05,
+      "loss": 1.2299,
       "step": 1173
     },
     {
-      "epoch": 0.3836601307189543,
-      "grad_norm": 1.372011661529541,
-      "learning_rate": 1.4127873730746706e-05,
-      "loss": 0.9595,
+      "epoch": 0.03224300348795694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999740435255294e-05,
+      "loss": 1.1532,
       "step": 1174
     },
     {
-      "epoch": 0.3839869281045752,
-      "grad_norm": 1.3173561096191406,
-      "learning_rate": 1.4118230420375259e-05,
-      "loss": 0.9823,
+      "epoch": 0.03227046771580017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999733986716078e-05,
+      "loss": 1.1402,
       "step": 1175
     },
     {
-      "epoch": 0.3843137254901961,
-      "grad_norm": 1.4110424518585205,
-      "learning_rate": 1.4108582495951077e-05,
-      "loss": 0.9791,
+      "epoch": 0.0322979319436434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999972745905509e-05,
+      "loss": 1.1595,
       "step": 1176
     },
     {
-      "epoch": 0.384640522875817,
-      "grad_norm": 1.3587684631347656,
-      "learning_rate": 1.4098929968283661e-05,
-      "loss": 0.9828,
+      "epoch": 0.03232539617148664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999720852272387e-05,
+      "loss": 1.1029,
       "step": 1177
     },
     {
-      "epoch": 0.38496732026143793,
-      "grad_norm": 1.4578664302825928,
-      "learning_rate": 1.4089272848187678e-05,
-      "loss": 0.9771,
+      "epoch": 0.03235286039932987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999714166368017e-05,
+      "loss": 1.0836,
       "step": 1178
     },
     {
-      "epoch": 0.38529411764705884,
-      "grad_norm": 1.411005973815918,
-      "learning_rate": 1.4079611146482934e-05,
-      "loss": 0.9604,
+      "epoch": 0.03238032462717311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999707401342035e-05,
+      "loss": 1.1709,
       "step": 1179
     },
     {
-      "epoch": 0.38562091503267976,
-      "grad_norm": 1.349229097366333,
-      "learning_rate": 1.406994487399437e-05,
-      "loss": 0.9819,
+      "epoch": 0.03240778885501634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99997005571945e-05,
+      "loss": 1.2031,
       "step": 1180
     },
     {
-      "epoch": 0.38594771241830067,
-      "grad_norm": 1.3433053493499756,
-      "learning_rate": 1.4060274041552049e-05,
-      "loss": 0.9181,
+      "epoch": 0.03243525308285958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999693633925453e-05,
+      "loss": 1.0523,
       "step": 1181
     },
     {
-      "epoch": 0.3862745098039216,
-      "grad_norm": 1.401047945022583,
-      "learning_rate": 1.4050598659991144e-05,
-      "loss": 0.9841,
+      "epoch": 0.03246271731070281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999968663153496e-05,
+      "loss": 1.1217,
       "step": 1182
     },
     {
-      "epoch": 0.3866013071895425,
-      "grad_norm": 1.3352750539779663,
-      "learning_rate": 1.4040918740151917e-05,
-      "loss": 0.9776,
+      "epoch": 0.03249018153854604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999679550023074e-05,
+      "loss": 1.118,
       "step": 1183
     },
     {
-      "epoch": 0.3869281045751634,
-      "grad_norm": 1.4138652086257935,
-      "learning_rate": 1.4031234292879726e-05,
-      "loss": 0.9836,
+      "epoch": 0.032517645766389276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999967238938985e-05,
+      "loss": 1.129,
       "step": 1184
     },
     {
-      "epoch": 0.3872549019607843,
-      "grad_norm": 1.3969148397445679,
-      "learning_rate": 1.402154532902499e-05,
-      "loss": 0.9653,
+      "epoch": 0.03254510999423251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999665149635345e-05,
+      "loss": 1.1692,
       "step": 1185
     },
     {
-      "epoch": 0.38758169934640524,
-      "grad_norm": 1.3306553363800049,
-      "learning_rate": 1.40118518594432e-05,
-      "loss": 0.9774,
+      "epoch": 0.03257257422207575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999965783075961e-05,
+      "loss": 1.1975,
       "step": 1186
     },
     {
-      "epoch": 0.38790849673202615,
-      "grad_norm": 1.349782943725586,
-      "learning_rate": 1.400215389499489e-05,
-      "loss": 0.9649,
+      "epoch": 0.03260003844991898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999650432762717e-05,
+      "loss": 1.1204,
       "step": 1187
     },
     {
-      "epoch": 0.38823529411764707,
-      "grad_norm": 1.3853213787078857,
-      "learning_rate": 1.3992451446545624e-05,
-      "loss": 0.9786,
+      "epoch": 0.03262750267776222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999964295564471e-05,
+      "loss": 1.1607,
       "step": 1188
     },
     {
-      "epoch": 0.388562091503268,
-      "grad_norm": 1.40047025680542,
-      "learning_rate": 1.3982744524965999e-05,
-      "loss": 0.9822,
+      "epoch": 0.03265496690560545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999963539940566e-05,
+      "loss": 1.1918,
       "step": 1189
     },
     {
-      "epoch": 0.3888888888888889,
-      "grad_norm": 1.329954743385315,
-      "learning_rate": 1.397303314113162e-05,
-      "loss": 0.9841,
+      "epoch": 0.03268243113344868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999627764045617e-05,
+      "loss": 1.1183,
       "step": 1190
     },
     {
-      "epoch": 0.3892156862745098,
-      "grad_norm": 1.386184573173523,
-      "learning_rate": 1.3963317305923095e-05,
-      "loss": 0.9454,
+      "epoch": 0.032709895361291916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999620049564647e-05,
+      "loss": 1.0587,
       "step": 1191
     },
     {
-      "epoch": 0.3895424836601307,
-      "grad_norm": 1.3390674591064453,
-      "learning_rate": 1.3953597030226006e-05,
-      "loss": 0.9784,
+      "epoch": 0.03273735958913515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999961225596281e-05,
+      "loss": 1.0078,
       "step": 1192
     },
     {
-      "epoch": 0.38986928104575164,
-      "grad_norm": 1.3609901666641235,
-      "learning_rate": 1.3943872324930935e-05,
-      "loss": 0.9594,
+      "epoch": 0.032764823816978386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999604383240164e-05,
+      "loss": 1.1361,
       "step": 1193
     },
     {
-      "epoch": 0.39019607843137255,
-      "grad_norm": 1.394092321395874,
-      "learning_rate": 1.3934143200933398e-05,
-      "loss": 0.978,
+      "epoch": 0.03279228804482162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999596431396777e-05,
+      "loss": 1.0803,
       "step": 1194
     },
     {
-      "epoch": 0.39052287581699346,
-      "grad_norm": 1.4680426120758057,
-      "learning_rate": 1.3924409669133883e-05,
-      "loss": 0.98,
+      "epoch": 0.03281975227266486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999958840043271e-05,
+      "loss": 1.0693,
       "step": 1195
     },
     {
-      "epoch": 0.3908496732026144,
-      "grad_norm": 1.5040911436080933,
-      "learning_rate": 1.3914671740437811e-05,
-      "loss": 0.9694,
+      "epoch": 0.032847216500508085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999580290348027e-05,
+      "loss": 1.1119,
       "step": 1196
     },
     {
-      "epoch": 0.3911764705882353,
-      "grad_norm": 1.4297921657562256,
-      "learning_rate": 1.390492942575552e-05,
-      "loss": 0.9537,
+      "epoch": 0.03287468072835132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999957210114279e-05,
+      "loss": 1.1931,
       "step": 1197
     },
     {
-      "epoch": 0.3915032679738562,
-      "grad_norm": 1.349867582321167,
-      "learning_rate": 1.3895182736002276e-05,
-      "loss": 0.9724,
+      "epoch": 0.032902144956194555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999563832817067e-05,
+      "loss": 1.2068,
       "step": 1198
     },
     {
-      "epoch": 0.3918300653594771,
-      "grad_norm": 1.419834852218628,
-      "learning_rate": 1.3885431682098237e-05,
-      "loss": 0.9593,
+      "epoch": 0.03292960918403779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999955548537092e-05,
+      "loss": 1.1292,
       "step": 1199
     },
     {
-      "epoch": 0.39215686274509803,
-      "grad_norm": 1.3614758253097534,
-      "learning_rate": 1.3875676274968453e-05,
-      "loss": 0.9588,
+      "epoch": 0.032957073411881026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999547058804414e-05,
+      "loss": 1.1747,
       "step": 1200
     },
     {
-      "epoch": 0.39248366013071895,
-      "grad_norm": 1.3175439834594727,
-      "learning_rate": 1.3865916525542855e-05,
-      "loss": 0.9721,
+      "epoch": 0.03298453763972426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999953855311762e-05,
+      "loss": 1.0469,
       "step": 1201
     },
     {
-      "epoch": 0.39281045751633986,
-      "grad_norm": 1.3740148544311523,
-      "learning_rate": 1.3856152444756235e-05,
-      "loss": 0.9641,
+      "epoch": 0.033012001867567496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999529968310606e-05,
+      "loss": 1.0114,
       "step": 1202
     },
     {
-      "epoch": 0.3931372549019608,
-      "grad_norm": 1.3182121515274048,
-      "learning_rate": 1.3846384043548237e-05,
-      "loss": 0.9272,
+      "epoch": 0.033039466095410724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999521304383432e-05,
+      "loss": 1.1143,
       "step": 1203
     },
     {
-      "epoch": 0.3934640522875817,
-      "grad_norm": 1.4093552827835083,
-      "learning_rate": 1.3836611332863356e-05,
-      "loss": 0.9725,
+      "epoch": 0.03306693032325396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999512561336178e-05,
+      "loss": 1.0193,
       "step": 1204
     },
     {
-      "epoch": 0.3937908496732026,
-      "grad_norm": 1.4187290668487549,
-      "learning_rate": 1.3826834323650899e-05,
-      "loss": 0.9563,
+      "epoch": 0.033094394551097195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999503739168904e-05,
+      "loss": 1.2193,
       "step": 1205
     },
     {
-      "epoch": 0.3941176470588235,
-      "grad_norm": 1.4973227977752686,
-      "learning_rate": 1.3817053026865e-05,
-      "loss": 1.0115,
+      "epoch": 0.03312185877894043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999494837881682e-05,
+      "loss": 1.2335,
       "step": 1206
     },
     {
-      "epoch": 0.39444444444444443,
-      "grad_norm": 1.4366123676300049,
-      "learning_rate": 1.3807267453464595e-05,
-      "loss": 0.9481,
+      "epoch": 0.033149323006783665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999485857474586e-05,
+      "loss": 1.1244,
       "step": 1207
     },
     {
-      "epoch": 0.39477124183006534,
-      "grad_norm": 1.3319638967514038,
-      "learning_rate": 1.3797477614413404e-05,
-      "loss": 0.9638,
+      "epoch": 0.0331767872346269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999476797947684e-05,
+      "loss": 1.1911,
       "step": 1208
     },
     {
-      "epoch": 0.39509803921568626,
-      "grad_norm": 1.4004608392715454,
-      "learning_rate": 1.378768352067994e-05,
-      "loss": 0.978,
+      "epoch": 0.033204251462470136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999467659301048e-05,
+      "loss": 1.0219,
       "step": 1209
     },
     {
-      "epoch": 0.3954248366013072,
-      "grad_norm": 1.3906855583190918,
-      "learning_rate": 1.3777885183237473e-05,
-      "loss": 0.9641,
+      "epoch": 0.033231715690313364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999945844153475e-05,
+      "loss": 1.0452,
       "step": 1210
     },
     {
-      "epoch": 0.3957516339869281,
-      "grad_norm": 1.3628652095794678,
-      "learning_rate": 1.3768082613064031e-05,
-      "loss": 0.9768,
+      "epoch": 0.0332591799181566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999449144648864e-05,
+      "loss": 1.0916,
       "step": 1211
     },
     {
-      "epoch": 0.396078431372549,
-      "grad_norm": 1.3752762079238892,
-      "learning_rate": 1.3758275821142382e-05,
-      "loss": 0.9676,
+      "epoch": 0.033286644145999834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999439768643464e-05,
+      "loss": 1.0781,
       "step": 1212
     },
     {
-      "epoch": 0.3964052287581699,
-      "grad_norm": 1.4348105192184448,
-      "learning_rate": 1.3748464818460028e-05,
-      "loss": 0.9865,
+      "epoch": 0.03331410837384307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999943031351862e-05,
+      "loss": 1.3047,
       "step": 1213
     },
     {
-      "epoch": 0.39673202614379083,
-      "grad_norm": 1.3901528120040894,
-      "learning_rate": 1.3738649616009183e-05,
-      "loss": 0.9103,
+      "epoch": 0.033341572601686305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999420779274413e-05,
+      "loss": 1.0657,
       "step": 1214
     },
     {
-      "epoch": 0.39705882352941174,
-      "grad_norm": 1.33602774143219,
-      "learning_rate": 1.3728830224786776e-05,
-      "loss": 0.9447,
+      "epoch": 0.03336903682952954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999411165910912e-05,
+      "loss": 1.1502,
       "step": 1215
     },
     {
-      "epoch": 0.39738562091503266,
-      "grad_norm": 1.3094916343688965,
-      "learning_rate": 1.3719006655794414e-05,
-      "loss": 0.9524,
+      "epoch": 0.033396501057372775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999401473428198e-05,
+      "loss": 1.192,
       "step": 1216
     },
     {
-      "epoch": 0.39771241830065357,
-      "grad_norm": 1.5068092346191406,
-      "learning_rate": 1.3709178920038397e-05,
-      "loss": 1.0053,
+      "epoch": 0.033423965285216004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999391701826346e-05,
+      "loss": 0.9999,
       "step": 1217
     },
     {
-      "epoch": 0.3980392156862745,
-      "grad_norm": 1.4579790830612183,
-      "learning_rate": 1.3699347028529694e-05,
-      "loss": 0.9536,
+      "epoch": 0.03345142951305924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999381851105436e-05,
+      "loss": 1.0672,
       "step": 1218
     },
     {
-      "epoch": 0.3983660130718954,
-      "grad_norm": 1.272347331047058,
-      "learning_rate": 1.3689510992283923e-05,
-      "loss": 0.9442,
+      "epoch": 0.033478893740902474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999937192126554e-05,
+      "loss": 1.1297,
       "step": 1219
     },
     {
-      "epoch": 0.39869281045751637,
-      "grad_norm": 1.386574387550354,
-      "learning_rate": 1.3679670822321347e-05,
-      "loss": 0.9527,
+      "epoch": 0.03350635796874571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999361912306743e-05,
+      "loss": 1.1038,
       "step": 1220
     },
     {
-      "epoch": 0.3990196078431373,
-      "grad_norm": 1.3154230117797852,
-      "learning_rate": 1.3669826529666863e-05,
-      "loss": 1.003,
+      "epoch": 0.033533822196588944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999935182422912e-05,
+      "loss": 1.1202,
       "step": 1221
     },
     {
-      "epoch": 0.3993464052287582,
-      "grad_norm": 1.4190224409103394,
-      "learning_rate": 1.3659978125349991e-05,
-      "loss": 0.9547,
+      "epoch": 0.03356128642443218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999934165703275e-05,
+      "loss": 1.0924,
       "step": 1222
     },
     {
-      "epoch": 0.3996732026143791,
-      "grad_norm": 1.3393948078155518,
-      "learning_rate": 1.3650125620404845e-05,
-      "loss": 0.9723,
+      "epoch": 0.03358875065227541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999331410717716e-05,
+      "loss": 1.2971,
       "step": 1223
     },
     {
-      "epoch": 0.4,
-      "grad_norm": 1.3258395195007324,
-      "learning_rate": 1.3640269025870152e-05,
-      "loss": 0.9879,
+      "epoch": 0.03361621488011864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999321085284103e-05,
+      "loss": 1.0913,
       "step": 1224
     },
     {
-      "epoch": 0.40032679738562094,
-      "grad_norm": 1.3883873224258423,
-      "learning_rate": 1.3630408352789202e-05,
-      "loss": 0.9073,
+      "epoch": 0.03364367910796188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999931068073198e-05,
+      "loss": 1.191,
       "step": 1225
     },
     {
-      "epoch": 0.40065359477124185,
-      "grad_norm": 1.3437507152557373,
-      "learning_rate": 1.3620543612209861e-05,
-      "loss": 0.9673,
+      "epoch": 0.033671143335805113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999300197061443e-05,
+      "loss": 1.1277,
       "step": 1226
     },
     {
-      "epoch": 0.40098039215686276,
-      "grad_norm": 1.341525673866272,
-      "learning_rate": 1.3610674815184563e-05,
-      "loss": 1.0029,
+      "epoch": 0.03369860756364835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999928963427257e-05,
+      "loss": 1.1207,
       "step": 1227
     },
     {
-      "epoch": 0.4013071895424837,
-      "grad_norm": 1.514001727104187,
-      "learning_rate": 1.3600801972770272e-05,
-      "loss": 0.9641,
+      "epoch": 0.033726071791491584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999278992365444e-05,
+      "loss": 1.0595,
       "step": 1228
     },
     {
-      "epoch": 0.4016339869281046,
-      "grad_norm": 1.3272231817245483,
-      "learning_rate": 1.3590925096028492e-05,
-      "loss": 0.9585,
+      "epoch": 0.03375353601933482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999268271340146e-05,
+      "loss": 1.1524,
       "step": 1229
     },
     {
-      "epoch": 0.4019607843137255,
-      "grad_norm": 1.3684402704238892,
-      "learning_rate": 1.3581044196025243e-05,
-      "loss": 0.9578,
+      "epoch": 0.03378100024717805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999257471196767e-05,
+      "loss": 1.204,
       "step": 1230
     },
     {
-      "epoch": 0.4022875816993464,
-      "grad_norm": 1.5297771692276,
-      "learning_rate": 1.3571159283831057e-05,
-      "loss": 1.0088,
+      "epoch": 0.03380846447502128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999924659193539e-05,
+      "loss": 1.1821,
       "step": 1231
     },
     {
-      "epoch": 0.40261437908496733,
-      "grad_norm": 1.4102689027786255,
-      "learning_rate": 1.3561270370520957e-05,
-      "loss": 0.9396,
+      "epoch": 0.03383592870286452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999235633556096e-05,
+      "loss": 1.1135,
       "step": 1232
     },
     {
-      "epoch": 0.40294117647058825,
-      "grad_norm": 1.4257391691207886,
-      "learning_rate": 1.3551377467174455e-05,
-      "loss": 0.9611,
+      "epoch": 0.03386339293070775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999922459605898e-05,
+      "loss": 1.0543,
       "step": 1233
     },
     {
-      "epoch": 0.40326797385620916,
-      "grad_norm": 1.3843294382095337,
-      "learning_rate": 1.3541480584875531e-05,
-      "loss": 0.9453,
+      "epoch": 0.03389085715855099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999213479444127e-05,
+      "loss": 1.0289,
       "step": 1234
     },
     {
-      "epoch": 0.4035947712418301,
-      "grad_norm": 1.2900279760360718,
-      "learning_rate": 1.3531579734712616e-05,
-      "loss": 0.9753,
+      "epoch": 0.03391832138639422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999202283711624e-05,
+      "loss": 1.1027,
       "step": 1235
     },
     {
-      "epoch": 0.403921568627451,
-      "grad_norm": 1.3313140869140625,
-      "learning_rate": 1.3521674927778594e-05,
-      "loss": 0.9495,
+      "epoch": 0.03394578561423746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999191008861556e-05,
+      "loss": 1.0964,
       "step": 1236
     },
     {
-      "epoch": 0.4042483660130719,
-      "grad_norm": 1.4624375104904175,
-      "learning_rate": 1.3511766175170781e-05,
-      "loss": 0.9046,
+      "epoch": 0.03397324984208069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999179654894015e-05,
+      "loss": 1.0944,
       "step": 1237
     },
     {
-      "epoch": 0.4045751633986928,
-      "grad_norm": 1.3080902099609375,
-      "learning_rate": 1.3501853487990917e-05,
-      "loss": 0.9667,
+      "epoch": 0.03400071406992392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999168221809095e-05,
+      "loss": 1.1138,
       "step": 1238
     },
     {
-      "epoch": 0.40490196078431373,
-      "grad_norm": 1.332686185836792,
-      "learning_rate": 1.3491936877345138e-05,
-      "loss": 0.9916,
+      "epoch": 0.03402817829776716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999915670960688e-05,
+      "loss": 1.0659,
       "step": 1239
     },
     {
-      "epoch": 0.40522875816993464,
-      "grad_norm": 1.3583011627197266,
-      "learning_rate": 1.348201635434399e-05,
-      "loss": 0.9111,
+      "epoch": 0.03405564252561039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999145118287463e-05,
+      "loss": 1.0612,
       "step": 1240
     },
     {
-      "epoch": 0.40555555555555556,
-      "grad_norm": 1.3758710622787476,
-      "learning_rate": 1.3472091930102395e-05,
-      "loss": 0.9848,
+      "epoch": 0.03408310675345363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999133447850937e-05,
+      "loss": 1.1251,
       "step": 1241
     },
     {
-      "epoch": 0.40588235294117647,
-      "grad_norm": 1.367661476135254,
-      "learning_rate": 1.3462163615739647e-05,
-      "loss": 0.9748,
+      "epoch": 0.03411057098129686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999121698297398e-05,
+      "loss": 1.1481,
       "step": 1242
     },
     {
-      "epoch": 0.4062091503267974,
-      "grad_norm": 1.3344037532806396,
-      "learning_rate": 1.3452231422379401e-05,
-      "loss": 0.9512,
+      "epoch": 0.0341380352091401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999910986962693e-05,
+      "loss": 1.0667,
       "step": 1243
     },
     {
-      "epoch": 0.4065359477124183,
-      "grad_norm": 1.3805798292160034,
-      "learning_rate": 1.3442295361149651e-05,
-      "loss": 0.9763,
+      "epoch": 0.034165499436983326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999097961839632e-05,
+      "loss": 1.0521,
       "step": 1244
     },
     {
-      "epoch": 0.4068627450980392,
-      "grad_norm": 1.444422960281372,
-      "learning_rate": 1.3432355443182734e-05,
-      "loss": 0.9919,
+      "epoch": 0.03419296366482656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999085974935602e-05,
+      "loss": 1.2332,
       "step": 1245
     },
     {
-      "epoch": 0.4071895424836601,
-      "grad_norm": 1.3723170757293701,
-      "learning_rate": 1.3422411679615305e-05,
-      "loss": 0.951,
+      "epoch": 0.0342204278926698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999073908914927e-05,
+      "loss": 1.1873,
       "step": 1246
     },
     {
-      "epoch": 0.40751633986928104,
-      "grad_norm": 1.4282474517822266,
-      "learning_rate": 1.3412464081588323e-05,
-      "loss": 0.9761,
+      "epoch": 0.03424789212051303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999061763777706e-05,
+      "loss": 1.1127,
       "step": 1247
     },
     {
-      "epoch": 0.40784313725490196,
-      "grad_norm": 1.3157094717025757,
-      "learning_rate": 1.3402512660247046e-05,
-      "loss": 0.9661,
+      "epoch": 0.03427535634835627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999049539524033e-05,
+      "loss": 1.1198,
       "step": 1248
     },
     {
-      "epoch": 0.40816993464052287,
-      "grad_norm": 1.3651857376098633,
-      "learning_rate": 1.3392557426741022e-05,
-      "loss": 0.9761,
+      "epoch": 0.0343028205761995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999903723615401e-05,
+      "loss": 1.0734,
       "step": 1249
     },
     {
-      "epoch": 0.4084967320261438,
-      "grad_norm": 1.3345117568969727,
-      "learning_rate": 1.3382598392224053e-05,
-      "loss": 0.9769,
+      "epoch": 0.03433028480404274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999024853667728e-05,
+      "loss": 1.0343,
       "step": 1250
     },
     {
-      "epoch": 0.4088235294117647,
-      "grad_norm": 1.3761787414550781,
-      "learning_rate": 1.337263556785422e-05,
-      "loss": 0.9699,
+      "epoch": 0.034357749031885966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999012392065292e-05,
+      "loss": 1.1088,
       "step": 1251
     },
     {
-      "epoch": 0.4091503267973856,
-      "grad_norm": 1.3250524997711182,
-      "learning_rate": 1.336266896479384e-05,
-      "loss": 0.9898,
+      "epoch": 0.0343852132597292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998999851346795e-05,
+      "loss": 1.0565,
       "step": 1252
     },
     {
-      "epoch": 0.4094771241830065,
-      "grad_norm": 1.2792896032333374,
-      "learning_rate": 1.3352698594209458e-05,
-      "loss": 0.9859,
+      "epoch": 0.034412677487572436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998987231512337e-05,
+      "loss": 0.9549,
       "step": 1253
     },
     {
-      "epoch": 0.40980392156862744,
-      "grad_norm": 1.4234400987625122,
-      "learning_rate": 1.3342724467271852e-05,
-      "loss": 0.9393,
+      "epoch": 0.03444014171541567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999897453256202e-05,
+      "loss": 1.1564,
       "step": 1254
     },
     {
-      "epoch": 0.41013071895424835,
-      "grad_norm": 1.4791580438613892,
-      "learning_rate": 1.3332746595156002e-05,
-      "loss": 0.9932,
+      "epoch": 0.03446760594325891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998961754495942e-05,
+      "loss": 1.1839,
       "step": 1255
     },
     {
-      "epoch": 0.41045751633986927,
-      "grad_norm": 1.3307116031646729,
-      "learning_rate": 1.3322764989041086e-05,
-      "loss": 0.9783,
+      "epoch": 0.03449507017110214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998948897314206e-05,
+      "loss": 1.0466,
       "step": 1256
     },
     {
-      "epoch": 0.4107843137254902,
-      "grad_norm": 1.411434292793274,
-      "learning_rate": 1.3312779660110469e-05,
-      "loss": 0.916,
+      "epoch": 0.03452253439894537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998935961016916e-05,
+      "loss": 1.0887,
       "step": 1257
     },
     {
-      "epoch": 0.4111111111111111,
-      "grad_norm": 1.3854479789733887,
-      "learning_rate": 1.3302790619551673e-05,
-      "loss": 0.9559,
+      "epoch": 0.034549998626788606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998922945604166e-05,
+      "loss": 1.2325,
       "step": 1258
     },
     {
-      "epoch": 0.411437908496732,
-      "grad_norm": 1.2995773553848267,
-      "learning_rate": 1.3292797878556394e-05,
-      "loss": 0.963,
+      "epoch": 0.03457746285463184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999890985107607e-05,
+      "loss": 1.1257,
       "step": 1259
     },
     {
-      "epoch": 0.4117647058823529,
-      "grad_norm": 1.4266759157180786,
-      "learning_rate": 1.328280144832047e-05,
-      "loss": 0.9944,
+      "epoch": 0.034604927082475076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999889667743272e-05,
+      "loss": 1.0364,
       "step": 1260
     },
     {
-      "epoch": 0.41209150326797384,
-      "grad_norm": 1.3045318126678467,
-      "learning_rate": 1.3272801340043867e-05,
-      "loss": 0.9524,
+      "epoch": 0.03463239131031831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999888342467423e-05,
+      "loss": 1.1575,
       "step": 1261
     },
     {
-      "epoch": 0.41241830065359475,
-      "grad_norm": 1.298633098602295,
-      "learning_rate": 1.326279756493068e-05,
-      "loss": 0.979,
+      "epoch": 0.034659855538161546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998870092800703e-05,
+      "loss": 0.9541,
       "step": 1262
     },
     {
-      "epoch": 0.41274509803921566,
-      "grad_norm": 1.3313242197036743,
-      "learning_rate": 1.3252790134189106e-05,
-      "loss": 0.9543,
+      "epoch": 0.03468731976600478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999885668181224e-05,
+      "loss": 1.1764,
       "step": 1263
     },
     {
-      "epoch": 0.4130718954248366,
-      "grad_norm": 1.2922152280807495,
-      "learning_rate": 1.3242779059031436e-05,
-      "loss": 0.9359,
+      "epoch": 0.03471478399384801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999884319170895e-05,
+      "loss": 1.212,
       "step": 1264
     },
     {
-      "epoch": 0.4133986928104575,
-      "grad_norm": 1.3347480297088623,
-      "learning_rate": 1.3232764350674052e-05,
-      "loss": 0.9905,
+      "epoch": 0.034742248221691245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999882962249094e-05,
+      "loss": 1.2043,
       "step": 1265
     },
     {
-      "epoch": 0.4137254901960784,
-      "grad_norm": 1.3580937385559082,
-      "learning_rate": 1.3222746020337406e-05,
-      "loss": 0.9889,
+      "epoch": 0.03476971244953448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998815974158315e-05,
+      "loss": 1.1593,
       "step": 1266
     },
     {
-      "epoch": 0.4140522875816993,
-      "grad_norm": 1.3707002401351929,
-      "learning_rate": 1.3212724079245994e-05,
-      "loss": 0.9758,
+      "epoch": 0.034797176677377716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998802246711188e-05,
+      "loss": 1.2079,
       "step": 1267
     },
     {
-      "epoch": 0.4143790849673203,
-      "grad_norm": 1.4112038612365723,
-      "learning_rate": 1.3202698538628376e-05,
-      "loss": 0.9513,
+      "epoch": 0.03482464090522095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998788440149663e-05,
+      "loss": 1.1838,
       "step": 1268
     },
     {
-      "epoch": 0.4147058823529412,
-      "grad_norm": 1.3229252099990845,
-      "learning_rate": 1.3192669409717138e-05,
-      "loss": 0.9351,
+      "epoch": 0.034852105133064186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999877455447385e-05,
+      "loss": 1.0445,
       "step": 1269
     },
     {
-      "epoch": 0.4150326797385621,
-      "grad_norm": 1.4104151725769043,
-      "learning_rate": 1.3182636703748881e-05,
-      "loss": 0.98,
+      "epoch": 0.03487956936090742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999876058968386e-05,
+      "loss": 1.1293,
       "step": 1270
     },
     {
-      "epoch": 0.41535947712418303,
-      "grad_norm": 1.5237677097320557,
-      "learning_rate": 1.317260043196422e-05,
-      "loss": 0.9721,
+      "epoch": 0.03490703358875065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998746545779802e-05,
+      "loss": 1.1675,
       "step": 1271
     },
     {
-      "epoch": 0.41568627450980394,
-      "grad_norm": 1.4661338329315186,
-      "learning_rate": 1.3162560605607763e-05,
-      "loss": 0.9721,
+      "epoch": 0.034934497816593885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998732422761787e-05,
+      "loss": 1.0772,
       "step": 1272
     },
     {
-      "epoch": 0.41601307189542486,
-      "grad_norm": 1.4098680019378662,
-      "learning_rate": 1.3152517235928101e-05,
-      "loss": 0.9192,
+      "epoch": 0.03496196204443712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998718220629933e-05,
+      "loss": 0.9897,
       "step": 1273
     },
     {
-      "epoch": 0.41633986928104577,
-      "grad_norm": 1.3985008001327515,
-      "learning_rate": 1.3142470334177794e-05,
-      "loss": 0.9506,
+      "epoch": 0.034989426272280355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998703939384347e-05,
+      "loss": 1.0465,
       "step": 1274
     },
     {
-      "epoch": 0.4166666666666667,
-      "grad_norm": 1.3923120498657227,
-      "learning_rate": 1.3132419911613358e-05,
-      "loss": 0.9804,
+      "epoch": 0.03501689050012359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998689579025138e-05,
+      "loss": 1.1948,
       "step": 1275
     },
     {
-      "epoch": 0.4169934640522876,
-      "grad_norm": 1.327825665473938,
-      "learning_rate": 1.3122365979495259e-05,
-      "loss": 0.9566,
+      "epoch": 0.035044354727966825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998675139552426e-05,
+      "loss": 1.0717,
       "step": 1276
     },
     {
-      "epoch": 0.4173202614379085,
-      "grad_norm": 1.3447188138961792,
-      "learning_rate": 1.3112308549087888e-05,
-      "loss": 0.9073,
+      "epoch": 0.03507181895581006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999866062096632e-05,
+      "loss": 1.1771,
       "step": 1277
     },
     {
-      "epoch": 0.4176470588235294,
-      "grad_norm": 1.2850092649459839,
-      "learning_rate": 1.3102247631659564e-05,
-      "loss": 0.9766,
+      "epoch": 0.03509928318365329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998646023266942e-05,
+      "loss": 1.1096,
       "step": 1278
     },
     {
-      "epoch": 0.41797385620915034,
-      "grad_norm": 1.3578041791915894,
-      "learning_rate": 1.3092183238482506e-05,
-      "loss": 0.9534,
+      "epoch": 0.035126747411496524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998631346454402e-05,
+      "loss": 1.088,
       "step": 1279
     },
     {
-      "epoch": 0.41830065359477125,
-      "grad_norm": 1.3593051433563232,
-      "learning_rate": 1.3082115380832829e-05,
-      "loss": 0.9724,
+      "epoch": 0.03515421163933976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999861659052882e-05,
+      "loss": 1.1638,
       "step": 1280
     },
     {
-      "epoch": 0.41862745098039217,
-      "grad_norm": 1.3955191373825073,
-      "learning_rate": 1.3072044069990529e-05,
-      "loss": 0.9932,
+      "epoch": 0.035181675867182995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998601755490307e-05,
+      "loss": 1.2166,
       "step": 1281
     },
     {
-      "epoch": 0.4189542483660131,
-      "grad_norm": 1.3078430891036987,
-      "learning_rate": 1.306196931723947e-05,
-      "loss": 0.9336,
+      "epoch": 0.03520914009502623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999858684133899e-05,
+      "loss": 1.1382,
       "step": 1282
     },
     {
-      "epoch": 0.419281045751634,
-      "grad_norm": 1.4290573596954346,
-      "learning_rate": 1.3051891133867379e-05,
-      "loss": 0.9806,
+      "epoch": 0.035236604322869465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998571848074973e-05,
+      "loss": 1.1396,
       "step": 1283
     },
     {
-      "epoch": 0.4196078431372549,
-      "grad_norm": 1.3961046934127808,
-      "learning_rate": 1.3041809531165819e-05,
-      "loss": 0.9406,
+      "epoch": 0.03526406855071269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998556775698385e-05,
+      "loss": 1.1525,
       "step": 1284
     },
     {
-      "epoch": 0.4199346405228758,
-      "grad_norm": 1.3885457515716553,
-      "learning_rate": 1.3031724520430185e-05,
-      "loss": 0.9613,
+      "epoch": 0.03529153277855593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998541624209344e-05,
+      "loss": 1.0815,
       "step": 1285
     },
     {
-      "epoch": 0.42026143790849674,
-      "grad_norm": 1.378740906715393,
-      "learning_rate": 1.3021636112959696e-05,
-      "loss": 0.9562,
+      "epoch": 0.035318997006399164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999852639360797e-05,
+      "loss": 1.1761,
       "step": 1286
     },
     {
-      "epoch": 0.42058823529411765,
-      "grad_norm": 1.3788156509399414,
-      "learning_rate": 1.301154432005737e-05,
-      "loss": 0.9629,
+      "epoch": 0.0353464612342424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999851108389438e-05,
+      "loss": 1.181,
       "step": 1287
     },
     {
-      "epoch": 0.42091503267973857,
-      "grad_norm": 1.4001597166061401,
-      "learning_rate": 1.3001449153030024e-05,
-      "loss": 0.9438,
+      "epoch": 0.035373925462085634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998495695068695e-05,
+      "loss": 1.232,
       "step": 1288
     },
     {
-      "epoch": 0.4212418300653595,
-      "grad_norm": 1.303750991821289,
-      "learning_rate": 1.2991350623188245e-05,
-      "loss": 0.921,
+      "epoch": 0.03540138968992887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999848022713104e-05,
+      "loss": 1.0709,
       "step": 1289
     },
     {
-      "epoch": 0.4215686274509804,
-      "grad_norm": 1.364161491394043,
-      "learning_rate": 1.2981248741846404e-05,
-      "loss": 0.984,
+      "epoch": 0.035428853917772105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999846468008154e-05,
+      "loss": 1.1212,
       "step": 1290
     },
     {
-      "epoch": 0.4218954248366013,
-      "grad_norm": 1.383265495300293,
-      "learning_rate": 1.2971143520322613e-05,
-      "loss": 0.9175,
+      "epoch": 0.03545631814561533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998449053920314e-05,
+      "loss": 1.1989,
       "step": 1291
     },
     {
-      "epoch": 0.4222222222222222,
-      "grad_norm": 1.3645081520080566,
-      "learning_rate": 1.2961034969938732e-05,
-      "loss": 0.9745,
+      "epoch": 0.03548378237345857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998433348647484e-05,
+      "loss": 1.1813,
       "step": 1292
     },
     {
-      "epoch": 0.42254901960784313,
-      "grad_norm": 1.3608719110488892,
-      "learning_rate": 1.2950923102020348e-05,
-      "loss": 0.9374,
+      "epoch": 0.0355112466013018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998417564263176e-05,
+      "loss": 1.1766,
       "step": 1293
     },
     {
-      "epoch": 0.42287581699346405,
-      "grad_norm": 1.2957404851913452,
-      "learning_rate": 1.2940807927896773e-05,
-      "loss": 0.9282,
+      "epoch": 0.03553871082914504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998401700767516e-05,
+      "loss": 1.2605,
       "step": 1294
     },
     {
-      "epoch": 0.42320261437908496,
-      "grad_norm": 1.3601316213607788,
-      "learning_rate": 1.2930689458901007e-05,
-      "loss": 0.9804,
+      "epoch": 0.035566175056988274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998385758160628e-05,
+      "loss": 1.0739,
       "step": 1295
     },
     {
-      "epoch": 0.4235294117647059,
-      "grad_norm": 1.2380151748657227,
-      "learning_rate": 1.292056770636976e-05,
-      "loss": 0.9633,
+      "epoch": 0.03559363928483151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999836973644264e-05,
+      "loss": 1.0704,
       "step": 1296
     },
     {
-      "epoch": 0.4238562091503268,
-      "grad_norm": 1.3457262516021729,
-      "learning_rate": 1.291044268164341e-05,
-      "loss": 0.9691,
+      "epoch": 0.035621103512674744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998353635613676e-05,
+      "loss": 1.1479,
       "step": 1297
     },
     {
-      "epoch": 0.4241830065359477,
-      "grad_norm": 1.313576579093933,
-      "learning_rate": 1.2900314396066e-05,
-      "loss": 0.9358,
+      "epoch": 0.03564856774051797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999833745567387e-05,
+      "loss": 1.1611,
       "step": 1298
     },
     {
-      "epoch": 0.4245098039215686,
-      "grad_norm": 1.483548879623413,
-      "learning_rate": 1.2890182860985236e-05,
-      "loss": 0.9769,
+      "epoch": 0.03567603196836121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999832119662334e-05,
+      "loss": 1.1045,
       "step": 1299
     },
     {
-      "epoch": 0.42483660130718953,
-      "grad_norm": 1.3145712614059448,
-      "learning_rate": 1.288004808775246e-05,
-      "loss": 0.9462,
+      "epoch": 0.03570349619620444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999830485846222e-05,
+      "loss": 1.1288,
       "step": 1300
     },
     {
-      "epoch": 0.42516339869281045,
-      "grad_norm": 1.4026461839675903,
-      "learning_rate": 1.2869910087722633e-05,
-      "loss": 0.9532,
+      "epoch": 0.03573096042404768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999828844119064e-05,
+      "loss": 1.1273,
       "step": 1301
     },
     {
-      "epoch": 0.42549019607843136,
-      "grad_norm": 1.3296828269958496,
-      "learning_rate": 1.2859768872254344e-05,
-      "loss": 0.988,
+      "epoch": 0.03575842465189091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998271944808727e-05,
+      "loss": 1.2073,
       "step": 1302
     },
     {
-      "epoch": 0.4258169934640523,
-      "grad_norm": 1.3325159549713135,
-      "learning_rate": 1.284962445270978e-05,
-      "loss": 0.9321,
+      "epoch": 0.03578588887973415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998255369316616e-05,
+      "loss": 1.2317,
       "step": 1303
     },
     {
-      "epoch": 0.4261437908496732,
-      "grad_norm": 1.4755393266677856,
-      "learning_rate": 1.2839476840454714e-05,
-      "loss": 0.9395,
+      "epoch": 0.035813353107577384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998238714714436e-05,
+      "loss": 1.141,
       "step": 1304
     },
     {
-      "epoch": 0.4264705882352941,
-      "grad_norm": 1.330506443977356,
-      "learning_rate": 1.2829326046858505e-05,
-      "loss": 0.9582,
+      "epoch": 0.03584081733542061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998221981002316e-05,
+      "loss": 1.1154,
       "step": 1305
     },
     {
-      "epoch": 0.426797385620915,
-      "grad_norm": 1.3323441743850708,
-      "learning_rate": 1.2819172083294068e-05,
-      "loss": 0.9656,
+      "epoch": 0.03586828156326385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998205168180392e-05,
+      "loss": 1.0019,
       "step": 1306
     },
     {
-      "epoch": 0.42712418300653593,
-      "grad_norm": 1.359404444694519,
-      "learning_rate": 1.280901496113787e-05,
-      "loss": 0.9404,
+      "epoch": 0.03589574579110708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998188276248796e-05,
+      "loss": 1.1052,
       "step": 1307
     },
     {
-      "epoch": 0.42745098039215684,
-      "grad_norm": 1.3116964101791382,
-      "learning_rate": 1.2798854691769927e-05,
-      "loss": 0.9517,
+      "epoch": 0.03592321001895032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999817130520766e-05,
+      "loss": 1.1245,
       "step": 1308
     },
     {
-      "epoch": 0.42777777777777776,
-      "grad_norm": 1.2876540422439575,
-      "learning_rate": 1.2788691286573767e-05,
-      "loss": 0.9324,
+      "epoch": 0.03595067424679355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999815425505712e-05,
+      "loss": 1.1702,
       "step": 1309
     },
     {
-      "epoch": 0.42810457516339867,
-      "grad_norm": 1.3305782079696655,
-      "learning_rate": 1.2778524756936438e-05,
-      "loss": 0.9284,
+      "epoch": 0.03597813847463679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998137125797314e-05,
+      "loss": 1.1147,
       "step": 1310
     },
     {
-      "epoch": 0.4284313725490196,
-      "grad_norm": 1.380075216293335,
-      "learning_rate": 1.2768355114248493e-05,
-      "loss": 0.9842,
+      "epoch": 0.03600560270248002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999811991742837e-05,
+      "loss": 1.1758,
       "step": 1311
     },
     {
-      "epoch": 0.4287581699346405,
-      "grad_norm": 1.3307569026947021,
-      "learning_rate": 1.2758182369903969e-05,
-      "loss": 0.939,
+      "epoch": 0.03603306693032325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999810262995043e-05,
+      "loss": 1.1235,
       "step": 1312
     },
     {
-      "epoch": 0.4290849673202614,
-      "grad_norm": 1.3469406366348267,
-      "learning_rate": 1.2748006535300371e-05,
-      "loss": 0.9848,
+      "epoch": 0.03606053115816649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998085263363627e-05,
+      "loss": 1.0392,
       "step": 1313
     },
     {
-      "epoch": 0.4294117647058823,
-      "grad_norm": 1.306333303451538,
-      "learning_rate": 1.2737827621838678e-05,
-      "loss": 0.959,
+      "epoch": 0.03608799538600972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998067817668104e-05,
+      "loss": 1.1237,
       "step": 1314
     },
     {
-      "epoch": 0.4297385620915033,
-      "grad_norm": 1.361196517944336,
-      "learning_rate": 1.2727645640923312e-05,
-      "loss": 0.9926,
+      "epoch": 0.03611545961385296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998050292863994e-05,
+      "loss": 1.0591,
       "step": 1315
     },
     {
-      "epoch": 0.4300653594771242,
-      "grad_norm": 1.2953202724456787,
-      "learning_rate": 1.2717460603962132e-05,
-      "loss": 0.954,
+      "epoch": 0.03614292384169619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998032688951437e-05,
+      "loss": 1.1107,
       "step": 1316
     },
     {
-      "epoch": 0.4303921568627451,
-      "grad_norm": 1.3101056814193726,
-      "learning_rate": 1.2707272522366428e-05,
-      "loss": 0.9694,
+      "epoch": 0.03617038806953943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998015005930574e-05,
+      "loss": 1.105,
       "step": 1317
     },
     {
-      "epoch": 0.43071895424836604,
-      "grad_norm": 1.3079463243484497,
-      "learning_rate": 1.2697081407550886e-05,
-      "loss": 0.9273,
+      "epoch": 0.036197852297382656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999799724380154e-05,
+      "loss": 0.9619,
       "step": 1318
     },
     {
-      "epoch": 0.43104575163398695,
-      "grad_norm": 1.3118404150009155,
-      "learning_rate": 1.2686887270933605e-05,
-      "loss": 0.9785,
+      "epoch": 0.03622531652522589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997979402564483e-05,
+      "loss": 1.2092,
       "step": 1319
     },
     {
-      "epoch": 0.43137254901960786,
-      "grad_norm": 1.317522406578064,
-      "learning_rate": 1.2676690123936066e-05,
-      "loss": 0.9603,
+      "epoch": 0.036252780753069126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999796148221954e-05,
+      "loss": 1.064,
       "step": 1320
     },
     {
-      "epoch": 0.4316993464052288,
-      "grad_norm": 1.3891977071762085,
-      "learning_rate": 1.2666489977983119e-05,
-      "loss": 0.9566,
+      "epoch": 0.03628024498091236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999794348276685e-05,
+      "loss": 1.0927,
       "step": 1321
     },
     {
-      "epoch": 0.4320261437908497,
-      "grad_norm": 1.3703140020370483,
-      "learning_rate": 1.2656286844502978e-05,
-      "loss": 0.9821,
+      "epoch": 0.0363077092087556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999792540420656e-05,
+      "loss": 1.0875,
       "step": 1322
     },
     {
-      "epoch": 0.4323529411764706,
-      "grad_norm": 1.4476293325424194,
-      "learning_rate": 1.2646080734927198e-05,
-      "loss": 0.9421,
+      "epoch": 0.03633517343659883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997907246538812e-05,
+      "loss": 1.157,
       "step": 1323
     },
     {
-      "epoch": 0.4326797385620915,
-      "grad_norm": 1.3247435092926025,
-      "learning_rate": 1.2635871660690677e-05,
-      "loss": 0.9568,
+      "epoch": 0.03636263766444207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997889009763746e-05,
+      "loss": 1.1254,
       "step": 1324
     },
     {
-      "epoch": 0.43300653594771243,
-      "grad_norm": 1.3053935766220093,
-      "learning_rate": 1.2625659633231631e-05,
-      "loss": 0.9563,
+      "epoch": 0.036390101892285295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999787069388151e-05,
+      "loss": 1.1633,
       "step": 1325
     },
     {
-      "epoch": 0.43333333333333335,
-      "grad_norm": 1.3444792032241821,
-      "learning_rate": 1.261544466399158e-05,
-      "loss": 0.9436,
+      "epoch": 0.03641756612012853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997852298892254e-05,
+      "loss": 1.097,
       "step": 1326
     },
     {
-      "epoch": 0.43366013071895426,
-      "grad_norm": 1.3309049606323242,
-      "learning_rate": 1.2605226764415348e-05,
-      "loss": 0.9834,
+      "epoch": 0.036445030347971766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999783382479611e-05,
+      "loss": 1.1261,
       "step": 1327
     },
     {
-      "epoch": 0.4339869281045752,
-      "grad_norm": 1.4113153219223022,
-      "learning_rate": 1.259500594595104e-05,
-      "loss": 0.9678,
+      "epoch": 0.036472494575815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997815271593235e-05,
+      "loss": 1.1551,
       "step": 1328
     },
     {
-      "epoch": 0.4343137254901961,
-      "grad_norm": 1.4124722480773926,
-      "learning_rate": 1.2584782220050025e-05,
-      "loss": 0.9485,
+      "epoch": 0.036499958803658236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997796639283774e-05,
+      "loss": 1.1303,
       "step": 1329
     },
     {
-      "epoch": 0.434640522875817,
-      "grad_norm": 1.3799443244934082,
-      "learning_rate": 1.2574555598166932e-05,
-      "loss": 0.9402,
+      "epoch": 0.03652742303150147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997777927867874e-05,
+      "loss": 1.1335,
       "step": 1330
     },
     {
-      "epoch": 0.4349673202614379,
-      "grad_norm": 1.2639926671981812,
-      "learning_rate": 1.2564326091759647e-05,
-      "loss": 0.957,
+      "epoch": 0.03655488725934471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999775913734568e-05,
+      "loss": 1.1652,
       "step": 1331
     },
     {
-      "epoch": 0.43529411764705883,
-      "grad_norm": 1.3604811429977417,
-      "learning_rate": 1.2554093712289267e-05,
-      "loss": 0.9833,
+      "epoch": 0.036582351487187935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997740267717343e-05,
+      "loss": 1.084,
       "step": 1332
     },
     {
-      "epoch": 0.43562091503267975,
-      "grad_norm": 1.3526041507720947,
-      "learning_rate": 1.2543858471220128e-05,
-      "loss": 1.0081,
+      "epoch": 0.03660981571503117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997721318983014e-05,
+      "loss": 1.1143,
       "step": 1333
     },
     {
-      "epoch": 0.43594771241830066,
-      "grad_norm": 1.3348007202148438,
-      "learning_rate": 1.2533620380019752e-05,
-      "loss": 0.9694,
+      "epoch": 0.036637279942874405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999770229114284e-05,
+      "loss": 1.1449,
       "step": 1334
     },
     {
-      "epoch": 0.4362745098039216,
-      "grad_norm": 1.2716416120529175,
-      "learning_rate": 1.2523379450158872e-05,
-      "loss": 0.9213,
+      "epoch": 0.03666474417071764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999768318419697e-05,
+      "loss": 1.0913,
       "step": 1335
     },
     {
-      "epoch": 0.4366013071895425,
-      "grad_norm": 1.3198577165603638,
-      "learning_rate": 1.2513135693111399e-05,
-      "loss": 0.9907,
+      "epoch": 0.036692208398560876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997663998145562e-05,
+      "loss": 1.1199,
       "step": 1336
     },
     {
-      "epoch": 0.4369281045751634,
-      "grad_norm": 1.3154927492141724,
-      "learning_rate": 1.2502889120354395e-05,
-      "loss": 0.9533,
+      "epoch": 0.03671967262640411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999764473298876e-05,
+      "loss": 1.2533,
       "step": 1337
     },
     {
-      "epoch": 0.4372549019607843,
-      "grad_norm": 1.3111035823822021,
-      "learning_rate": 1.2492639743368096e-05,
-      "loss": 0.9435,
+      "epoch": 0.036747136854247346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997625388726725e-05,
+      "loss": 1.0764,
       "step": 1338
     },
     {
-      "epoch": 0.43758169934640523,
-      "grad_norm": 1.363718867301941,
-      "learning_rate": 1.2482387573635873e-05,
-      "loss": 0.9485,
+      "epoch": 0.036774601082090574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99976059653596e-05,
+      "loss": 1.1164,
       "step": 1339
     },
     {
-      "epoch": 0.43790849673202614,
-      "grad_norm": 1.4138588905334473,
-      "learning_rate": 1.2472132622644222e-05,
-      "loss": 0.966,
+      "epoch": 0.03680206530993381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999758646288755e-05,
+      "loss": 1.1017,
       "step": 1340
     },
     {
-      "epoch": 0.43823529411764706,
-      "grad_norm": 1.3649382591247559,
-      "learning_rate": 1.2461874901882763e-05,
-      "loss": 0.9488,
+      "epoch": 0.036829529537777045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999756688131072e-05,
+      "loss": 1.0451,
       "step": 1341
     },
     {
-      "epoch": 0.43856209150326797,
-      "grad_norm": 1.3368749618530273,
-      "learning_rate": 1.2451614422844208e-05,
-      "loss": 0.9443,
+      "epoch": 0.03685699376562028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999754722062927e-05,
+      "loss": 1.1865,
       "step": 1342
     },
     {
-      "epoch": 0.4388888888888889,
-      "grad_norm": 1.45650053024292,
-      "learning_rate": 1.2441351197024373e-05,
-      "loss": 0.9819,
+      "epoch": 0.036884457993463515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999752748084335e-05,
+      "loss": 1.1534,
       "step": 1343
     },
     {
-      "epoch": 0.4392156862745098,
-      "grad_norm": 1.4130500555038452,
-      "learning_rate": 1.2431085235922143e-05,
-      "loss": 0.9765,
+      "epoch": 0.03691192222130675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999750766195312e-05,
+      "loss": 1.2191,
       "step": 1344
     },
     {
-      "epoch": 0.4395424836601307,
-      "grad_norm": 1.3826642036437988,
-      "learning_rate": 1.242081655103947e-05,
-      "loss": 0.953,
+      "epoch": 0.03693938644914998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997487763958738e-05,
+      "loss": 1.1968,
       "step": 1345
     },
     {
-      "epoch": 0.4398692810457516,
-      "grad_norm": 1.4533252716064453,
-      "learning_rate": 1.2410545153881355e-05,
-      "loss": 0.9993,
+      "epoch": 0.036966850676993214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999746778686036e-05,
+      "loss": 1.1425,
       "step": 1346
     },
     {
-      "epoch": 0.44019607843137254,
-      "grad_norm": 1.3764939308166504,
-      "learning_rate": 1.2400271055955841e-05,
-      "loss": 0.9484,
+      "epoch": 0.03699431490483645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997447730658147e-05,
+      "loss": 0.9738,
       "step": 1347
     },
     {
-      "epoch": 0.44052287581699345,
-      "grad_norm": 1.3599170446395874,
-      "learning_rate": 1.2389994268773995e-05,
-      "loss": 1.0004,
+      "epoch": 0.037021779132679684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999742759535225e-05,
+      "loss": 1.0534,
       "step": 1348
     },
     {
-      "epoch": 0.44084967320261437,
-      "grad_norm": 1.3905651569366455,
-      "learning_rate": 1.2379714803849897e-05,
-      "loss": 0.9455,
+      "epoch": 0.03704924336052292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997407380942838e-05,
+      "loss": 1.1582,
       "step": 1349
     },
     {
-      "epoch": 0.4411764705882353,
-      "grad_norm": 1.502031683921814,
-      "learning_rate": 1.2369432672700634e-05,
-      "loss": 0.9799,
+      "epoch": 0.037076707588366155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997387087430064e-05,
+      "loss": 1.0776,
       "step": 1350
     },
     {
-      "epoch": 0.4415032679738562,
-      "grad_norm": 1.432819128036499,
-      "learning_rate": 1.2359147886846266e-05,
-      "loss": 0.9514,
+      "epoch": 0.03710417181620939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999736671481409e-05,
+      "loss": 1.04,
       "step": 1351
     },
     {
-      "epoch": 0.4418300653594771,
-      "grad_norm": 1.3440663814544678,
-      "learning_rate": 1.234886045780984e-05,
-      "loss": 0.9444,
+      "epoch": 0.03713163604405262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999734626309508e-05,
+      "loss": 1.1211,
       "step": 1352
     },
     {
-      "epoch": 0.442156862745098,
-      "grad_norm": 1.2929909229278564,
-      "learning_rate": 1.2338570397117357e-05,
-      "loss": 0.9405,
+      "epoch": 0.037159100271895854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999732573227319e-05,
+      "loss": 1.105,
       "step": 1353
     },
     {
-      "epoch": 0.44248366013071894,
-      "grad_norm": 1.3845266103744507,
-      "learning_rate": 1.2328277716297771e-05,
-      "loss": 0.9684,
+      "epoch": 0.03718656449973909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999730512234859e-05,
+      "loss": 1.0988,
       "step": 1354
     },
     {
-      "epoch": 0.44281045751633985,
-      "grad_norm": 1.407816767692566,
-      "learning_rate": 1.2317982426882973e-05,
-      "loss": 0.9282,
+      "epoch": 0.037214028727582324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997284433321438e-05,
+      "loss": 0.9931,
       "step": 1355
     },
     {
-      "epoch": 0.44313725490196076,
-      "grad_norm": 1.332445740699768,
-      "learning_rate": 1.2307684540407775e-05,
-      "loss": 0.9538,
+      "epoch": 0.03724149295542556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99972636651919e-05,
+      "loss": 1.0823,
       "step": 1356
     },
     {
-      "epoch": 0.4434640522875817,
-      "grad_norm": 1.3302266597747803,
-      "learning_rate": 1.2297384068409887e-05,
-      "loss": 0.9527,
+      "epoch": 0.037268957183268794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997242817960138e-05,
+      "loss": 1.075,
       "step": 1357
     },
     {
-      "epoch": 0.4437908496732026,
-      "grad_norm": 1.385046124458313,
-      "learning_rate": 1.2287081022429935e-05,
-      "loss": 0.9408,
+      "epoch": 0.03729642141111203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997221891626318e-05,
+      "loss": 1.107,
       "step": 1358
     },
     {
-      "epoch": 0.4441176470588235,
-      "grad_norm": 1.4080737829208374,
-      "learning_rate": 1.2276775414011421e-05,
-      "loss": 0.9507,
+      "epoch": 0.03732388563895526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997200886190605e-05,
+      "loss": 1.0684,
       "step": 1359
     },
     {
-      "epoch": 0.4444444444444444,
-      "grad_norm": 1.3144108057022095,
-      "learning_rate": 1.2266467254700712e-05,
-      "loss": 0.997,
+      "epoch": 0.03735134986679849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999717980165317e-05,
+      "loss": 1.1652,
       "step": 1360
     },
     {
-      "epoch": 0.44477124183006533,
-      "grad_norm": 1.3873423337936401,
-      "learning_rate": 1.2256156556047041e-05,
-      "loss": 0.9672,
+      "epoch": 0.03737881409464173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999715863801417e-05,
+      "loss": 1.1614,
       "step": 1361
     },
     {
-      "epoch": 0.44509803921568625,
-      "grad_norm": 1.3981448411941528,
-      "learning_rate": 1.224584332960248e-05,
-      "loss": 0.9414,
+      "epoch": 0.037406278322484963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997137395273783e-05,
+      "loss": 1.0645,
       "step": 1362
     },
     {
-      "epoch": 0.4454248366013072,
-      "grad_norm": 1.344169020652771,
-      "learning_rate": 1.223552758692194e-05,
-      "loss": 0.9506,
+      "epoch": 0.0374337425503282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997116073432172e-05,
+      "loss": 1.1282,
       "step": 1363
     },
     {
-      "epoch": 0.44575163398692813,
-      "grad_norm": 1.360350251197815,
-      "learning_rate": 1.2225209339563144e-05,
-      "loss": 0.9668,
+      "epoch": 0.037461206778171434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997094672489505e-05,
+      "loss": 1.1676,
       "step": 1364
     },
     {
-      "epoch": 0.44607843137254904,
-      "grad_norm": 1.3553249835968018,
-      "learning_rate": 1.221488859908663e-05,
-      "loss": 0.9774,
+      "epoch": 0.03748867100601467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999707319244595e-05,
+      "loss": 1.0147,
       "step": 1365
     },
     {
-      "epoch": 0.44640522875816996,
-      "grad_norm": 1.3439109325408936,
-      "learning_rate": 1.220456537705572e-05,
-      "loss": 0.9676,
+      "epoch": 0.0375161352338579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997051633301684e-05,
+      "loss": 1.1038,
       "step": 1366
     },
     {
-      "epoch": 0.4467320261437909,
-      "grad_norm": 1.352176547050476,
-      "learning_rate": 1.2194239685036521e-05,
-      "loss": 0.9634,
+      "epoch": 0.03754359946170113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999702999505687e-05,
+      "loss": 1.1933,
       "step": 1367
     },
     {
-      "epoch": 0.4470588235294118,
-      "grad_norm": 1.3616864681243896,
-      "learning_rate": 1.2183911534597908e-05,
-      "loss": 0.9862,
+      "epoch": 0.03757106368954437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997008277711685e-05,
+      "loss": 1.0134,
       "step": 1368
     },
     {
-      "epoch": 0.4473856209150327,
-      "grad_norm": 1.3665331602096558,
-      "learning_rate": 1.2173580937311504e-05,
-      "loss": 0.9491,
+      "epoch": 0.0375985279173876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996986481266292e-05,
+      "loss": 1.0686,
       "step": 1369
     },
     {
-      "epoch": 0.4477124183006536,
-      "grad_norm": 1.260135531425476,
-      "learning_rate": 1.2163247904751688e-05,
-      "loss": 0.9465,
+      "epoch": 0.03762599214523084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996964605720877e-05,
+      "loss": 1.0446,
       "step": 1370
     },
     {
-      "epoch": 0.44803921568627453,
-      "grad_norm": 1.361339807510376,
-      "learning_rate": 1.2152912448495548e-05,
-      "loss": 0.9485,
+      "epoch": 0.03765345637307407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99969426510756e-05,
+      "loss": 1.1151,
       "step": 1371
     },
     {
-      "epoch": 0.44836601307189544,
-      "grad_norm": 1.3791395425796509,
-      "learning_rate": 1.2142574580122903e-05,
-      "loss": 0.9783,
+      "epoch": 0.03768092060091731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996920617330646e-05,
+      "loss": 1.0977,
       "step": 1372
     },
     {
-      "epoch": 0.44869281045751636,
-      "grad_norm": 1.3638641834259033,
-      "learning_rate": 1.2132234311216271e-05,
-      "loss": 0.9548,
+      "epoch": 0.03770838482876054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999689850448618e-05,
+      "loss": 1.1223,
       "step": 1373
     },
     {
-      "epoch": 0.44901960784313727,
-      "grad_norm": 1.2861158847808838,
-      "learning_rate": 1.2121891653360851e-05,
-      "loss": 0.9764,
+      "epoch": 0.03773584905660377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999687631254238e-05,
+      "loss": 1.0515,
       "step": 1374
     },
     {
-      "epoch": 0.4493464052287582,
-      "grad_norm": 1.3022217750549316,
-      "learning_rate": 1.211154661814453e-05,
-      "loss": 0.9396,
+      "epoch": 0.03776331328444701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996854041499423e-05,
+      "loss": 1.1837,
       "step": 1375
     },
     {
-      "epoch": 0.4496732026143791,
-      "grad_norm": 1.3600519895553589,
-      "learning_rate": 1.210119921715785e-05,
-      "loss": 0.9171,
+      "epoch": 0.03779077751229024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996831691357483e-05,
+      "loss": 1.1181,
       "step": 1376
     },
     {
-      "epoch": 0.45,
-      "grad_norm": 1.3509242534637451,
-      "learning_rate": 1.2090849461994013e-05,
-      "loss": 0.9703,
+      "epoch": 0.03781824174013348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999680926211674e-05,
+      "loss": 1.2766,
       "step": 1377
     },
     {
-      "epoch": 0.4503267973856209,
-      "grad_norm": 1.3101449012756348,
-      "learning_rate": 1.2080497364248849e-05,
-      "loss": 0.9737,
+      "epoch": 0.03784570596797671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996786753777367e-05,
+      "loss": 1.1303,
       "step": 1378
     },
     {
-      "epoch": 0.45065359477124184,
-      "grad_norm": 1.3136556148529053,
-      "learning_rate": 1.2070142935520814e-05,
-      "loss": 0.9719,
+      "epoch": 0.03787317019581994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999676416633955e-05,
+      "loss": 1.0284,
       "step": 1379
     },
     {
-      "epoch": 0.45098039215686275,
-      "grad_norm": 1.2711429595947266,
-      "learning_rate": 1.2059786187410984e-05,
-      "loss": 0.9706,
+      "epoch": 0.037900634423663176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999674149980346e-05,
+      "loss": 1.172,
       "step": 1380
     },
     {
-      "epoch": 0.45130718954248367,
-      "grad_norm": 1.374591588973999,
-      "learning_rate": 1.2049427131523028e-05,
-      "loss": 0.97,
+      "epoch": 0.03792809865150641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996718754169277e-05,
+      "loss": 1.0541,
       "step": 1381
     },
     {
-      "epoch": 0.4516339869281046,
-      "grad_norm": 1.3782912492752075,
-      "learning_rate": 1.2039065779463195e-05,
-      "loss": 0.9455,
+      "epoch": 0.03795556287934965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996695929437185e-05,
+      "loss": 1.0748,
       "step": 1382
     },
     {
-      "epoch": 0.4519607843137255,
-      "grad_norm": 1.2647819519042969,
-      "learning_rate": 1.2028702142840314e-05,
-      "loss": 0.9646,
+      "epoch": 0.03798302710719288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996673025607362e-05,
+      "loss": 1.0915,
       "step": 1383
     },
     {
-      "epoch": 0.4522875816993464,
-      "grad_norm": 1.339080572128296,
-      "learning_rate": 1.2018336233265775e-05,
-      "loss": 0.977,
+      "epoch": 0.03801049133503612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999665004267999e-05,
+      "loss": 1.1509,
       "step": 1384
     },
     {
-      "epoch": 0.4526143790849673,
-      "grad_norm": 1.40694260597229,
-      "learning_rate": 1.2007968062353505e-05,
-      "loss": 0.9444,
+      "epoch": 0.03803795556287935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999662698065525e-05,
+      "loss": 1.0988,
       "step": 1385
     },
     {
-      "epoch": 0.45294117647058824,
-      "grad_norm": 1.3113360404968262,
-      "learning_rate": 1.1997597641719976e-05,
-      "loss": 0.942,
+      "epoch": 0.03806541979072258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996603839533327e-05,
+      "loss": 1.0507,
       "step": 1386
     },
     {
-      "epoch": 0.45326797385620915,
-      "grad_norm": 1.3130149841308594,
-      "learning_rate": 1.1987224982984176e-05,
-      "loss": 0.949,
+      "epoch": 0.038092884018565816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99965806193144e-05,
+      "loss": 1.0918,
       "step": 1387
     },
     {
-      "epoch": 0.45359477124183006,
-      "grad_norm": 1.3798996210098267,
-      "learning_rate": 1.1976850097767598e-05,
-      "loss": 0.9572,
+      "epoch": 0.03812034824640905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996557319998656e-05,
+      "loss": 1.1345,
       "step": 1388
     },
     {
-      "epoch": 0.453921568627451,
-      "grad_norm": 1.404359221458435,
-      "learning_rate": 1.196647299769423e-05,
-      "loss": 0.9406,
+      "epoch": 0.038147812474252286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996533941586277e-05,
+      "loss": 1.1661,
       "step": 1389
     },
     {
-      "epoch": 0.4542483660130719,
-      "grad_norm": 1.251986026763916,
-      "learning_rate": 1.1956093694390547e-05,
-      "loss": 0.9535,
+      "epoch": 0.03817527670209552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996510484077447e-05,
+      "loss": 1.1832,
       "step": 1390
     },
     {
-      "epoch": 0.4545751633986928,
-      "grad_norm": 1.381416916847229,
-      "learning_rate": 1.1945712199485486e-05,
-      "loss": 0.9599,
+      "epoch": 0.03820274092993876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996486947472358e-05,
+      "loss": 1.08,
       "step": 1391
     },
     {
-      "epoch": 0.4549019607843137,
-      "grad_norm": 1.3672189712524414,
-      "learning_rate": 1.1935328524610443e-05,
-      "loss": 0.9524,
+      "epoch": 0.03823020515778199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999646333177119e-05,
+      "loss": 1.0601,
       "step": 1392
     },
     {
-      "epoch": 0.45522875816993463,
-      "grad_norm": 1.3732043504714966,
-      "learning_rate": 1.1924942681399254e-05,
-      "loss": 0.9762,
+      "epoch": 0.03825766938562522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996439636974134e-05,
+      "loss": 1.0775,
       "step": 1393
     },
     {
-      "epoch": 0.45555555555555555,
-      "grad_norm": 1.3908774852752686,
-      "learning_rate": 1.1914554681488188e-05,
-      "loss": 0.9289,
+      "epoch": 0.038285133613468456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999641586308137e-05,
+      "loss": 1.0924,
       "step": 1394
     },
     {
-      "epoch": 0.45588235294117646,
-      "grad_norm": 1.382385492324829,
-      "learning_rate": 1.1904164536515927e-05,
-      "loss": 0.9806,
+      "epoch": 0.03831259784131169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996392010093094e-05,
+      "loss": 1.0173,
       "step": 1395
     },
     {
-      "epoch": 0.4562091503267974,
-      "grad_norm": 1.401140809059143,
-      "learning_rate": 1.1893772258123554e-05,
-      "loss": 0.9492,
+      "epoch": 0.038340062069154926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996368078009495e-05,
+      "loss": 1.0705,
       "step": 1396
     },
     {
-      "epoch": 0.4565359477124183,
-      "grad_norm": 1.2900450229644775,
-      "learning_rate": 1.1883377857954554e-05,
-      "loss": 0.9727,
+      "epoch": 0.03836752629699816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996344066830758e-05,
+      "loss": 1.1024,
       "step": 1397
     },
     {
-      "epoch": 0.4568627450980392,
-      "grad_norm": 1.2774571180343628,
-      "learning_rate": 1.1872981347654778e-05,
-      "loss": 0.9255,
+      "epoch": 0.038394990524841396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996319976557074e-05,
+      "loss": 1.095,
       "step": 1398
     },
     {
-      "epoch": 0.4571895424836601,
-      "grad_norm": 1.3976490497589111,
-      "learning_rate": 1.1862582738872438e-05,
-      "loss": 0.9413,
+      "epoch": 0.03842245475268463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996295807188636e-05,
+      "loss": 1.1134,
       "step": 1399
     },
     {
-      "epoch": 0.45751633986928103,
-      "grad_norm": 1.4456442594528198,
-      "learning_rate": 1.1852182043258112e-05,
-      "loss": 0.9465,
+      "epoch": 0.03844991898052786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999627155872563e-05,
+      "loss": 1.2131,
       "step": 1400
     },
     {
-      "epoch": 0.45784313725490194,
-      "grad_norm": 1.3840550184249878,
-      "learning_rate": 1.1841779272464703e-05,
-      "loss": 0.9527,
+      "epoch": 0.038477383208371095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996247231168253e-05,
+      "loss": 1.1008,
       "step": 1401
     },
     {
-      "epoch": 0.45816993464052286,
-      "grad_norm": 1.2869974374771118,
-      "learning_rate": 1.1831374438147445e-05,
-      "loss": 0.9176,
+      "epoch": 0.03850484743621433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996222824516694e-05,
+      "loss": 1.0741,
       "step": 1402
     },
     {
-      "epoch": 0.45849673202614377,
-      "grad_norm": 1.319387435913086,
-      "learning_rate": 1.182096755196388e-05,
-      "loss": 0.9446,
+      "epoch": 0.038532311664057566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999619833877115e-05,
+      "loss": 1.0209,
       "step": 1403
     },
     {
-      "epoch": 0.4588235294117647,
-      "grad_norm": 1.4077790975570679,
-      "learning_rate": 1.1810558625573856e-05,
-      "loss": 0.9704,
+      "epoch": 0.0385597758919008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996173773931815e-05,
+      "loss": 1.0794,
       "step": 1404
     },
     {
-      "epoch": 0.4591503267973856,
-      "grad_norm": 1.2890732288360596,
-      "learning_rate": 1.1800147670639497e-05,
-      "loss": 0.939,
+      "epoch": 0.038587240119744036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996149129998877e-05,
+      "loss": 1.109,
       "step": 1405
     },
     {
-      "epoch": 0.4594771241830065,
-      "grad_norm": 1.402363896369934,
-      "learning_rate": 1.1789734698825209e-05,
-      "loss": 0.9538,
+      "epoch": 0.038614704347587264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996124406972533e-05,
+      "loss": 1.2077,
       "step": 1406
     },
     {
-      "epoch": 0.4598039215686274,
-      "grad_norm": 1.4003798961639404,
-      "learning_rate": 1.1779319721797648e-05,
-      "loss": 0.9615,
+      "epoch": 0.0386421685754305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996099604852984e-05,
+      "loss": 1.09,
       "step": 1407
     },
     {
-      "epoch": 0.46013071895424834,
-      "grad_norm": 1.2977763414382935,
-      "learning_rate": 1.176890275122573e-05,
-      "loss": 0.9407,
+      "epoch": 0.038669632803273735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996074723640425e-05,
+      "loss": 0.9653,
       "step": 1408
     },
     {
-      "epoch": 0.46045751633986925,
-      "grad_norm": 1.276848316192627,
-      "learning_rate": 1.1758483798780593e-05,
-      "loss": 0.954,
+      "epoch": 0.03869709703111697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996049763335047e-05,
+      "loss": 1.2659,
       "step": 1409
     },
     {
-      "epoch": 0.46078431372549017,
-      "grad_norm": 1.3436230421066284,
-      "learning_rate": 1.17480628761356e-05,
-      "loss": 0.9198,
+      "epoch": 0.038724561258960205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996024723937053e-05,
+      "loss": 1.1536,
       "step": 1410
     },
     {
-      "epoch": 0.46111111111111114,
-      "grad_norm": 1.3414074182510376,
-      "learning_rate": 1.1737639994966319e-05,
-      "loss": 0.9604,
+      "epoch": 0.03875202548680344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999599960544664e-05,
+      "loss": 1.1744,
       "step": 1411
     },
     {
-      "epoch": 0.46143790849673205,
-      "grad_norm": 1.3018287420272827,
-      "learning_rate": 1.1727215166950519e-05,
-      "loss": 0.9435,
+      "epoch": 0.038779489714646675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995974407864e-05,
+      "loss": 1.1082,
       "step": 1412
     },
     {
-      "epoch": 0.46176470588235297,
-      "grad_norm": 1.3636584281921387,
-      "learning_rate": 1.1716788403768138e-05,
-      "loss": 0.9431,
+      "epoch": 0.038806953942489904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995949131189344e-05,
+      "loss": 1.1258,
       "step": 1413
     },
     {
-      "epoch": 0.4620915032679739,
-      "grad_norm": 1.3502293825149536,
-      "learning_rate": 1.1706359717101295e-05,
-      "loss": 0.9694,
+      "epoch": 0.03883441817033314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995923775422867e-05,
+      "loss": 0.9989,
       "step": 1414
     },
     {
-      "epoch": 0.4624183006535948,
-      "grad_norm": 1.2456682920455933,
-      "learning_rate": 1.1695929118634257e-05,
-      "loss": 0.9752,
+      "epoch": 0.038861882398176374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999589834056477e-05,
+      "loss": 1.0884,
       "step": 1415
     },
     {
-      "epoch": 0.4627450980392157,
-      "grad_norm": 1.3975602388381958,
-      "learning_rate": 1.1685496620053434e-05,
-      "loss": 0.9703,
+      "epoch": 0.03888934662601961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995872826615248e-05,
+      "loss": 1.1317,
       "step": 1416
     },
     {
-      "epoch": 0.4630718954248366,
-      "grad_norm": 1.2954787015914917,
-      "learning_rate": 1.1675062233047365e-05,
-      "loss": 0.9554,
+      "epoch": 0.038916810853862845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999584723357451e-05,
+      "loss": 1.1547,
       "step": 1417
     },
     {
-      "epoch": 0.46339869281045754,
-      "grad_norm": 1.297666311264038,
-      "learning_rate": 1.1664625969306707e-05,
-      "loss": 0.9456,
+      "epoch": 0.03894427508170608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999582156144276e-05,
+      "loss": 1.1069,
       "step": 1418
     },
     {
-      "epoch": 0.46372549019607845,
-      "grad_norm": 1.3304766416549683,
-      "learning_rate": 1.1654187840524216e-05,
-      "loss": 0.9573,
+      "epoch": 0.038971739309549315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995795810220193e-05,
+      "loss": 1.1379,
       "step": 1419
     },
     {
-      "epoch": 0.46405228758169936,
-      "grad_norm": 1.3945095539093018,
-      "learning_rate": 1.1643747858394743e-05,
-      "loss": 0.9842,
+      "epoch": 0.03899920353739254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999576997990702e-05,
+      "loss": 1.1479,
       "step": 1420
     },
     {
-      "epoch": 0.4643790849673203,
-      "grad_norm": 1.3466256856918335,
-      "learning_rate": 1.1633306034615208e-05,
-      "loss": 0.9324,
+      "epoch": 0.03902666776523578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999574407050344e-05,
+      "loss": 1.1394,
       "step": 1421
     },
     {
-      "epoch": 0.4647058823529412,
-      "grad_norm": 1.342419147491455,
-      "learning_rate": 1.16228623808846e-05,
-      "loss": 0.9679,
+      "epoch": 0.039054131993079014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995718082009667e-05,
+      "loss": 1.1728,
       "step": 1422
     },
     {
-      "epoch": 0.4650326797385621,
-      "grad_norm": 1.3266057968139648,
-      "learning_rate": 1.1612416908903959e-05,
-      "loss": 0.9435,
+      "epoch": 0.03908159622092225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995692014425897e-05,
+      "loss": 1.1119,
       "step": 1423
     },
     {
-      "epoch": 0.465359477124183,
-      "grad_norm": 1.3524569272994995,
-      "learning_rate": 1.1601969630376353e-05,
-      "loss": 0.9538,
+      "epoch": 0.039109060448765484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995665867752342e-05,
+      "loss": 1.0525,
       "step": 1424
     },
     {
-      "epoch": 0.46568627450980393,
-      "grad_norm": 1.3165415525436401,
-      "learning_rate": 1.1591520557006886e-05,
-      "loss": 0.9547,
+      "epoch": 0.03913652467660872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995639641989204e-05,
+      "loss": 1.1196,
       "step": 1425
     },
     {
-      "epoch": 0.46601307189542485,
-      "grad_norm": 1.3454915285110474,
-      "learning_rate": 1.158106970050267e-05,
-      "loss": 0.9673,
+      "epoch": 0.039163988904451955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999561333713669e-05,
+      "loss": 1.2214,
       "step": 1426
     },
     {
-      "epoch": 0.46633986928104576,
-      "grad_norm": 1.3289947509765625,
-      "learning_rate": 1.1570617072572806e-05,
-      "loss": 0.9674,
+      "epoch": 0.03919145313229518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999558695319502e-05,
+      "loss": 1.1845,
       "step": 1427
     },
     {
-      "epoch": 0.4666666666666667,
-      "grad_norm": 1.429916501045227,
-      "learning_rate": 1.156016268492839e-05,
-      "loss": 0.9647,
+      "epoch": 0.03921891736013842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995560490164386e-05,
+      "loss": 1.0981,
       "step": 1428
     },
     {
-      "epoch": 0.4669934640522876,
-      "grad_norm": 1.2956998348236084,
-      "learning_rate": 1.1549706549282486e-05,
-      "loss": 0.9565,
+      "epoch": 0.03924638158798165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995533948045013e-05,
+      "loss": 0.9366,
       "step": 1429
     },
     {
-      "epoch": 0.4673202614379085,
-      "grad_norm": 1.4843283891677856,
-      "learning_rate": 1.1539248677350114e-05,
-      "loss": 0.9575,
+      "epoch": 0.03927384581582489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99955073268371e-05,
+      "loss": 1.044,
       "step": 1430
     },
     {
-      "epoch": 0.4676470588235294,
-      "grad_norm": 1.3585447072982788,
-      "learning_rate": 1.1528789080848247e-05,
-      "loss": 0.9332,
+      "epoch": 0.039301310043668124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999548062654086e-05,
+      "loss": 1.1995,
       "step": 1431
     },
     {
-      "epoch": 0.46797385620915033,
-      "grad_norm": 1.327380895614624,
-      "learning_rate": 1.151832777149578e-05,
-      "loss": 0.9447,
+      "epoch": 0.03932877427151136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995453847156506e-05,
+      "loss": 1.158,
       "step": 1432
     },
     {
-      "epoch": 0.46830065359477124,
-      "grad_norm": 1.324302077293396,
-      "learning_rate": 1.1507864761013531e-05,
-      "loss": 0.9369,
+      "epoch": 0.03935623849935459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999542698868425e-05,
+      "loss": 1.066,
       "step": 1433
     },
     {
-      "epoch": 0.46862745098039216,
-      "grad_norm": 1.3640440702438354,
-      "learning_rate": 1.1497400061124231e-05,
-      "loss": 0.9576,
+      "epoch": 0.03938370272719782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99954000511243e-05,
+      "loss": 1.1367,
       "step": 1434
     },
     {
-      "epoch": 0.46895424836601307,
-      "grad_norm": 1.3630412817001343,
-      "learning_rate": 1.1486933683552492e-05,
-      "loss": 0.9325,
+      "epoch": 0.03941116695504106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995373034476878e-05,
+      "loss": 1.2473,
       "step": 1435
     },
     {
-      "epoch": 0.469281045751634,
-      "grad_norm": 1.2723362445831299,
-      "learning_rate": 1.1476465640024814e-05,
-      "loss": 0.9746,
+      "epoch": 0.03943863118288429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999534593874219e-05,
+      "loss": 1.131,
       "step": 1436
     },
     {
-      "epoch": 0.4696078431372549,
-      "grad_norm": 1.4808999300003052,
-      "learning_rate": 1.146599594226956e-05,
-      "loss": 0.9665,
+      "epoch": 0.03946609541072753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995318763920456e-05,
+      "loss": 1.049,
       "step": 1437
     },
     {
-      "epoch": 0.4699346405228758,
-      "grad_norm": 1.478496789932251,
-      "learning_rate": 1.1455524602016946e-05,
-      "loss": 0.9613,
+      "epoch": 0.03949355963857076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995291510011884e-05,
+      "loss": 1.1998,
       "step": 1438
     },
     {
-      "epoch": 0.4702614379084967,
-      "grad_norm": 1.3673841953277588,
-      "learning_rate": 1.1445051630999031e-05,
-      "loss": 0.9896,
+      "epoch": 0.039521023866414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995264177016696e-05,
+      "loss": 1.0563,
       "step": 1439
     },
     {
-      "epoch": 0.47058823529411764,
-      "grad_norm": 1.42827570438385,
-      "learning_rate": 1.1434577040949699e-05,
-      "loss": 0.9682,
+      "epoch": 0.03954848809425723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995236764935105e-05,
+      "loss": 1.031,
       "step": 1440
     },
     {
-      "epoch": 0.47091503267973855,
-      "grad_norm": 1.3216973543167114,
-      "learning_rate": 1.1424100843604649e-05,
-      "loss": 0.9953,
+      "epoch": 0.03957595232210046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999520927376733e-05,
+      "loss": 1.0489,
       "step": 1441
     },
     {
-      "epoch": 0.47124183006535947,
-      "grad_norm": 1.332844614982605,
-      "learning_rate": 1.1413623050701381e-05,
-      "loss": 0.9792,
+      "epoch": 0.0396034165499437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995181703513588e-05,
+      "loss": 1.1566,
       "step": 1442
     },
     {
-      "epoch": 0.4715686274509804,
-      "grad_norm": 1.2975749969482422,
-      "learning_rate": 1.1403143673979182e-05,
-      "loss": 0.9492,
+      "epoch": 0.03963088077778693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995154054174095e-05,
+      "loss": 0.9637,
       "step": 1443
     },
     {
-      "epoch": 0.4718954248366013,
-      "grad_norm": 1.3647875785827637,
-      "learning_rate": 1.1392662725179114e-05,
-      "loss": 0.9144,
+      "epoch": 0.03965834500563017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995126325749073e-05,
+      "loss": 1.1254,
       "step": 1444
     },
     {
-      "epoch": 0.4722222222222222,
-      "grad_norm": 1.262791633605957,
-      "learning_rate": 1.1382180216043994e-05,
-      "loss": 0.9577,
+      "epoch": 0.0396858092334734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999509851823874e-05,
+      "loss": 1.0787,
       "step": 1445
     },
     {
-      "epoch": 0.4725490196078431,
-      "grad_norm": 1.3134890794754028,
-      "learning_rate": 1.1371696158318404e-05,
-      "loss": 0.9496,
+      "epoch": 0.03971327346131664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995070631643313e-05,
+      "loss": 1.08,
       "step": 1446
     },
     {
-      "epoch": 0.47287581699346404,
-      "grad_norm": 1.3186434507369995,
-      "learning_rate": 1.136121056374864e-05,
-      "loss": 0.9337,
+      "epoch": 0.039740737689159866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995042665963018e-05,
+      "loss": 1.0806,
       "step": 1447
     },
     {
-      "epoch": 0.47320261437908495,
-      "grad_norm": 1.3771674633026123,
-      "learning_rate": 1.1350723444082736e-05,
-      "loss": 0.9825,
+      "epoch": 0.0397682019170031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995014621198072e-05,
+      "loss": 1.1971,
       "step": 1448
     },
     {
-      "epoch": 0.47352941176470587,
-      "grad_norm": 1.4197664260864258,
-      "learning_rate": 1.1340234811070427e-05,
-      "loss": 0.9647,
+      "epoch": 0.03979566614484634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99949864973487e-05,
+      "loss": 1.0364,
       "step": 1449
     },
     {
-      "epoch": 0.4738562091503268,
-      "grad_norm": 1.3049182891845703,
-      "learning_rate": 1.1329744676463144e-05,
-      "loss": 0.9657,
+      "epoch": 0.03982313037268957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994958294415123e-05,
+      "loss": 1.1105,
       "step": 1450
     },
     {
-      "epoch": 0.4741830065359477,
-      "grad_norm": 1.374398112297058,
-      "learning_rate": 1.1319253052014003e-05,
-      "loss": 0.9719,
+      "epoch": 0.03985059460053281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994930012397563e-05,
+      "loss": 1.0645,
       "step": 1451
     },
     {
-      "epoch": 0.4745098039215686,
-      "grad_norm": 1.3778860569000244,
-      "learning_rate": 1.1308759949477786e-05,
-      "loss": 0.9708,
+      "epoch": 0.03987805882837604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994901651296247e-05,
+      "loss": 1.0642,
       "step": 1452
     },
     {
-      "epoch": 0.4748366013071895,
-      "grad_norm": 1.3674229383468628,
-      "learning_rate": 1.1298265380610935e-05,
-      "loss": 0.9274,
+      "epoch": 0.03990552305621928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994873211111395e-05,
+      "loss": 1.0126,
       "step": 1453
     },
     {
-      "epoch": 0.47516339869281043,
-      "grad_norm": 1.3554354906082153,
-      "learning_rate": 1.1287769357171536e-05,
-      "loss": 0.9703,
+      "epoch": 0.039932987284062506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994844691843237e-05,
+      "loss": 1.148,
       "step": 1454
     },
     {
-      "epoch": 0.47549019607843135,
-      "grad_norm": 1.2360559701919556,
-      "learning_rate": 1.1277271890919295e-05,
-      "loss": 0.9183,
+      "epoch": 0.03996045151190574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994816093491994e-05,
+      "loss": 1.0907,
       "step": 1455
     },
     {
-      "epoch": 0.47581699346405226,
-      "grad_norm": 1.2722046375274658,
-      "learning_rate": 1.1266772993615543e-05,
-      "loss": 0.9381,
+      "epoch": 0.039987915739748976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994787416057896e-05,
+      "loss": 1.112,
       "step": 1456
     },
     {
-      "epoch": 0.4761437908496732,
-      "grad_norm": 1.3531367778778076,
-      "learning_rate": 1.1256272677023214e-05,
-      "loss": 0.9482,
+      "epoch": 0.04001537996759221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999475865954117e-05,
+      "loss": 1.1577,
       "step": 1457
     },
     {
-      "epoch": 0.4764705882352941,
-      "grad_norm": 1.308112621307373,
-      "learning_rate": 1.1245770952906829e-05,
-      "loss": 0.9079,
+      "epoch": 0.04004284419543545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999472982394204e-05,
+      "loss": 1.1306,
       "step": 1458
     },
     {
-      "epoch": 0.47679738562091506,
-      "grad_norm": 1.4019052982330322,
-      "learning_rate": 1.1235267833032487e-05,
-      "loss": 1.0002,
+      "epoch": 0.04007030842327868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994700909260735e-05,
+      "loss": 1.1169,
       "step": 1459
     },
     {
-      "epoch": 0.477124183006536,
-      "grad_norm": 1.3358079195022583,
-      "learning_rate": 1.1224763329167859e-05,
-      "loss": 0.9805,
+      "epoch": 0.04009777265112192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994671915497486e-05,
+      "loss": 1.0703,
       "step": 1460
     },
     {
-      "epoch": 0.4774509803921569,
-      "grad_norm": 1.357197642326355,
-      "learning_rate": 1.1214257453082144e-05,
-      "loss": 0.9736,
+      "epoch": 0.040125236878965145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999464284265252e-05,
+      "loss": 1.0417,
       "step": 1461
     },
     {
-      "epoch": 0.4777777777777778,
-      "grad_norm": 1.3121967315673828,
-      "learning_rate": 1.1203750216546105e-05,
-      "loss": 0.9574,
+      "epoch": 0.04015270110680838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999461369072607e-05,
+      "loss": 1.1478,
       "step": 1462
     },
     {
-      "epoch": 0.4781045751633987,
-      "grad_norm": 1.390708565711975,
-      "learning_rate": 1.1193241631332015e-05,
-      "loss": 0.9449,
+      "epoch": 0.040180165334651616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994584459718363e-05,
+      "loss": 1.1831,
       "step": 1463
     },
     {
-      "epoch": 0.47843137254901963,
-      "grad_norm": 1.296386480331421,
-      "learning_rate": 1.1182731709213658e-05,
-      "loss": 0.9034,
+      "epoch": 0.04020762956249485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994555149629634e-05,
+      "loss": 1.1418,
       "step": 1464
     },
     {
-      "epoch": 0.47875816993464054,
-      "grad_norm": 1.4047290086746216,
-      "learning_rate": 1.1172220461966324e-05,
-      "loss": 0.9283,
+      "epoch": 0.040235093790338086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994525760460113e-05,
+      "loss": 1.0913,
       "step": 1465
     },
     {
-      "epoch": 0.47908496732026146,
-      "grad_norm": 1.289014458656311,
-      "learning_rate": 1.1161707901366776e-05,
-      "loss": 0.9637,
+      "epoch": 0.04026255801818132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999449629221003e-05,
+      "loss": 1.0508,
       "step": 1466
     },
     {
-      "epoch": 0.47941176470588237,
-      "grad_norm": 1.2450826168060303,
-      "learning_rate": 1.1151194039193256e-05,
-      "loss": 0.953,
+      "epoch": 0.04029002224602455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994466744879625e-05,
+      "loss": 1.0169,
       "step": 1467
     },
     {
-      "epoch": 0.4797385620915033,
-      "grad_norm": 1.3335509300231934,
-      "learning_rate": 1.1140678887225468e-05,
-      "loss": 0.9729,
+      "epoch": 0.040317486473867785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994437118469126e-05,
+      "loss": 1.0698,
       "step": 1468
     },
     {
-      "epoch": 0.4800653594771242,
-      "grad_norm": 1.2920773029327393,
-      "learning_rate": 1.1130162457244551e-05,
-      "loss": 0.9393,
+      "epoch": 0.04034495070171102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999440741297877e-05,
+      "loss": 1.1407,
       "step": 1469
     },
     {
-      "epoch": 0.4803921568627451,
-      "grad_norm": 1.3004705905914307,
-      "learning_rate": 1.1119644761033079e-05,
-      "loss": 0.9601,
+      "epoch": 0.040372414929554255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999437762840879e-05,
+      "loss": 1.1255,
       "step": 1470
     },
     {
-      "epoch": 0.480718954248366,
-      "grad_norm": 1.2358909845352173,
-      "learning_rate": 1.1109125810375054e-05,
-      "loss": 0.9787,
+      "epoch": 0.04039987915739749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994347764759422e-05,
+      "loss": 1.1318,
       "step": 1471
     },
     {
-      "epoch": 0.48104575163398694,
-      "grad_norm": 1.3743312358856201,
-      "learning_rate": 1.1098605617055871e-05,
-      "loss": 0.9306,
+      "epoch": 0.040427343385240726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994317822030905e-05,
+      "loss": 1.144,
       "step": 1472
     },
     {
-      "epoch": 0.48137254901960785,
-      "grad_norm": 1.3764363527297974,
-      "learning_rate": 1.108808419286232e-05,
-      "loss": 0.9785,
+      "epoch": 0.04045480761308396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999428780022347e-05,
+      "loss": 1.0975,
       "step": 1473
     },
     {
-      "epoch": 0.48169934640522877,
-      "grad_norm": 1.3222495317459106,
-      "learning_rate": 1.1077561549582578e-05,
-      "loss": 0.9606,
+      "epoch": 0.04048227184092719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999425769933736e-05,
+      "loss": 1.1222,
       "step": 1474
     },
     {
-      "epoch": 0.4820261437908497,
-      "grad_norm": 1.2947359085083008,
-      "learning_rate": 1.106703769900618e-05,
-      "loss": 0.9588,
+      "epoch": 0.040509736068770424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994227519372817e-05,
+      "loss": 1.1429,
       "step": 1475
     },
     {
-      "epoch": 0.4823529411764706,
-      "grad_norm": 1.3065094947814941,
-      "learning_rate": 1.1056512652924014e-05,
-      "loss": 0.9613,
+      "epoch": 0.04053720029661366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994197260330068e-05,
+      "loss": 1.1343,
       "step": 1476
     },
     {
-      "epoch": 0.4826797385620915,
-      "grad_norm": 1.3212991952896118,
-      "learning_rate": 1.1045986423128315e-05,
-      "loss": 0.9587,
+      "epoch": 0.040564664524456895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999416692220936e-05,
+      "loss": 1.0792,
       "step": 1477
     },
     {
-      "epoch": 0.4830065359477124,
-      "grad_norm": 1.2825583219528198,
-      "learning_rate": 1.1035459021412631e-05,
-      "loss": 0.9703,
+      "epoch": 0.04059212875230013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994136505010934e-05,
+      "loss": 1.1547,
       "step": 1478
     },
     {
-      "epoch": 0.48333333333333334,
-      "grad_norm": 1.3545998334884644,
-      "learning_rate": 1.1024930459571834e-05,
-      "loss": 0.9703,
+      "epoch": 0.040619592980143365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999410600873503e-05,
+      "loss": 1.1038,
       "step": 1479
     },
     {
-      "epoch": 0.48366013071895425,
-      "grad_norm": 1.285112977027893,
-      "learning_rate": 1.1014400749402096e-05,
-      "loss": 0.9914,
+      "epoch": 0.0406470572079866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994075433381884e-05,
+      "loss": 1.1655,
       "step": 1480
     },
     {
-      "epoch": 0.48398692810457516,
-      "grad_norm": 1.3975061178207397,
-      "learning_rate": 1.1003869902700866e-05,
-      "loss": 0.9504,
+      "epoch": 0.04067452143582983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994044778951745e-05,
+      "loss": 1.124,
       "step": 1481
     },
     {
-      "epoch": 0.4843137254901961,
-      "grad_norm": 1.3113744258880615,
-      "learning_rate": 1.0993337931266879e-05,
-      "loss": 0.9152,
+      "epoch": 0.040701985663673064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999401404544485e-05,
+      "loss": 1.0949,
       "step": 1482
     },
     {
-      "epoch": 0.484640522875817,
-      "grad_norm": 1.3781546354293823,
-      "learning_rate": 1.0982804846900114e-05,
-      "loss": 0.9458,
+      "epoch": 0.0407294498915163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993983232861445e-05,
+      "loss": 1.2471,
       "step": 1483
     },
     {
-      "epoch": 0.4849673202614379,
-      "grad_norm": 1.30940842628479,
-      "learning_rate": 1.0972270661401812e-05,
-      "loss": 0.9629,
+      "epoch": 0.040756914119359534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993952341201775e-05,
+      "loss": 1.1098,
       "step": 1484
     },
     {
-      "epoch": 0.4852941176470588,
-      "grad_norm": 1.3795424699783325,
-      "learning_rate": 1.0961735386574447e-05,
-      "loss": 0.9323,
+      "epoch": 0.04078437834720277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993921370466078e-05,
+      "loss": 1.1487,
       "step": 1485
     },
     {
-      "epoch": 0.48562091503267973,
-      "grad_norm": 1.2878742218017578,
-      "learning_rate": 1.09511990342217e-05,
-      "loss": 0.9388,
+      "epoch": 0.040811842575046005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999389032065461e-05,
+      "loss": 1.1359,
       "step": 1486
     },
     {
-      "epoch": 0.48594771241830065,
-      "grad_norm": 1.3650363683700562,
-      "learning_rate": 1.0940661616148473e-05,
-      "loss": 0.9917,
+      "epoch": 0.04083930680288924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999385919176761e-05,
+      "loss": 1.0837,
       "step": 1487
     },
     {
-      "epoch": 0.48627450980392156,
-      "grad_norm": 1.321696400642395,
-      "learning_rate": 1.0930123144160857e-05,
-      "loss": 0.9757,
+      "epoch": 0.04086677103073247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993827983805318e-05,
+      "loss": 1.0532,
       "step": 1488
     },
     {
-      "epoch": 0.4866013071895425,
-      "grad_norm": 1.6134222745895386,
-      "learning_rate": 1.091958363006612e-05,
-      "loss": 0.932,
+      "epoch": 0.040894235258575704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993796696767992e-05,
+      "loss": 1.0772,
       "step": 1489
     },
     {
-      "epoch": 0.4869281045751634,
-      "grad_norm": 1.3480255603790283,
-      "learning_rate": 1.0909043085672703e-05,
-      "loss": 0.9835,
+      "epoch": 0.04092169948641894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993765330655878e-05,
+      "loss": 1.1079,
       "step": 1490
     },
     {
-      "epoch": 0.4872549019607843,
-      "grad_norm": 1.406970739364624,
-      "learning_rate": 1.0898501522790204e-05,
-      "loss": 0.9135,
+      "epoch": 0.040949163714262174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993733885469217e-05,
+      "loss": 1.1014,
       "step": 1491
     },
     {
-      "epoch": 0.4875816993464052,
-      "grad_norm": 1.301840901374817,
-      "learning_rate": 1.0887958953229349e-05,
-      "loss": 0.9404,
+      "epoch": 0.04097662794210541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993702361208266e-05,
+      "loss": 1.0337,
       "step": 1492
     },
     {
-      "epoch": 0.48790849673202613,
-      "grad_norm": 1.2699828147888184,
-      "learning_rate": 1.0877415388802009e-05,
-      "loss": 0.9424,
+      "epoch": 0.041004092169948644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993670757873273e-05,
+      "loss": 1.1542,
       "step": 1493
     },
     {
-      "epoch": 0.48823529411764705,
-      "grad_norm": 1.3617887496948242,
-      "learning_rate": 1.0866870841321156e-05,
-      "loss": 0.8972,
+      "epoch": 0.04103155639779187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999363907546448e-05,
+      "loss": 1.0563,
       "step": 1494
     },
     {
-      "epoch": 0.48856209150326796,
-      "grad_norm": 1.439496636390686,
-      "learning_rate": 1.085632532260087e-05,
-      "loss": 0.9418,
+      "epoch": 0.04105902062563511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993607313982144e-05,
+      "loss": 1.1481,
       "step": 1495
     },
     {
-      "epoch": 0.4888888888888889,
-      "grad_norm": 1.3464581966400146,
-      "learning_rate": 1.0845778844456319e-05,
-      "loss": 0.9747,
+      "epoch": 0.04108648485347834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993575473426517e-05,
+      "loss": 1.0743,
       "step": 1496
     },
     {
-      "epoch": 0.4892156862745098,
-      "grad_norm": 1.2794173955917358,
-      "learning_rate": 1.0835231418703742e-05,
-      "loss": 0.9438,
+      "epoch": 0.04111394908132158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999354355379785e-05,
+      "loss": 1.0218,
       "step": 1497
     },
     {
-      "epoch": 0.4895424836601307,
-      "grad_norm": 1.3043569326400757,
-      "learning_rate": 1.0824683057160445e-05,
-      "loss": 0.9676,
+      "epoch": 0.041141413309164813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993511555096395e-05,
+      "loss": 1.1301,
       "step": 1498
     },
     {
-      "epoch": 0.4898692810457516,
-      "grad_norm": 1.3247284889221191,
-      "learning_rate": 1.0814133771644783e-05,
-      "loss": 0.9551,
+      "epoch": 0.04116887753700805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993479477322405e-05,
+      "loss": 1.123,
       "step": 1499
     },
     {
-      "epoch": 0.49019607843137253,
-      "grad_norm": 1.4013457298278809,
-      "learning_rate": 1.0803583573976137e-05,
-      "loss": 0.9275,
+      "epoch": 0.041196341764851284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993447320476135e-05,
+      "loss": 1.0616,
       "step": 1500
     },
     {
-      "epoch": 0.49052287581699344,
-      "grad_norm": 1.3295892477035522,
-      "learning_rate": 1.0793032475974919e-05,
-      "loss": 0.9543,
+      "epoch": 0.04122380599269451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999341508455784e-05,
+      "loss": 1.1098,
       "step": 1501
     },
     {
-      "epoch": 0.49084967320261436,
-      "grad_norm": 1.298079013824463,
-      "learning_rate": 1.0782480489462546e-05,
-      "loss": 0.9335,
+      "epoch": 0.04125127022053775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993382769567774e-05,
+      "loss": 1.2339,
       "step": 1502
     },
     {
-      "epoch": 0.49117647058823527,
-      "grad_norm": 1.4506229162216187,
-      "learning_rate": 1.0771927626261431e-05,
-      "loss": 0.965,
+      "epoch": 0.04127873444838098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999335037550619e-05,
+      "loss": 1.1478,
       "step": 1503
     },
     {
-      "epoch": 0.4915032679738562,
-      "grad_norm": 1.3715262413024902,
-      "learning_rate": 1.0761373898194972e-05,
-      "loss": 0.9688,
+      "epoch": 0.04130619867622422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993317902373347e-05,
+      "loss": 1.0292,
       "step": 1504
     },
     {
-      "epoch": 0.4918300653594771,
-      "grad_norm": 1.2949753999710083,
-      "learning_rate": 1.075081931708753e-05,
-      "loss": 0.9975,
+      "epoch": 0.04133366290406745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993285350169505e-05,
+      "loss": 1.0732,
       "step": 1505
     },
     {
-      "epoch": 0.492156862745098,
-      "grad_norm": 1.3576465845108032,
-      "learning_rate": 1.074026389476443e-05,
-      "loss": 0.9545,
+      "epoch": 0.04136112713191069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993252718894913e-05,
+      "loss": 1.1031,
       "step": 1506
     },
     {
-      "epoch": 0.492483660130719,
-      "grad_norm": 1.281836986541748,
-      "learning_rate": 1.072970764305193e-05,
-      "loss": 0.9248,
+      "epoch": 0.04138859135975392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999322000854984e-05,
+      "loss": 0.9936,
       "step": 1507
     },
     {
-      "epoch": 0.4928104575163399,
-      "grad_norm": 1.2643693685531616,
-      "learning_rate": 1.0719150573777226e-05,
-      "loss": 0.9523,
+      "epoch": 0.04141605558759715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993187219134537e-05,
+      "loss": 1.1103,
       "step": 1508
     },
     {
-      "epoch": 0.4931372549019608,
-      "grad_norm": 1.2795814275741577,
-      "learning_rate": 1.0708592698768423e-05,
-      "loss": 0.9843,
+      "epoch": 0.04144351981544039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993154350649263e-05,
+      "loss": 1.1171,
       "step": 1509
     },
     {
-      "epoch": 0.4934640522875817,
-      "grad_norm": 1.308885097503662,
-      "learning_rate": 1.0698034029854533e-05,
-      "loss": 0.9344,
+      "epoch": 0.04147098404328362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993121403094283e-05,
+      "loss": 1.1772,
       "step": 1510
     },
     {
-      "epoch": 0.49379084967320264,
-      "grad_norm": 1.3478662967681885,
-      "learning_rate": 1.0687474578865455e-05,
-      "loss": 0.9677,
+      "epoch": 0.04149844827112686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999308837646986e-05,
+      "loss": 1.0696,
       "step": 1511
     },
     {
-      "epoch": 0.49411764705882355,
-      "grad_norm": 1.268430471420288,
-      "learning_rate": 1.0676914357631968e-05,
-      "loss": 0.9212,
+      "epoch": 0.04152591249897009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993055270776244e-05,
+      "loss": 1.1703,
       "step": 1512
     },
     {
-      "epoch": 0.49444444444444446,
-      "grad_norm": 1.3500936031341553,
-      "learning_rate": 1.0666353377985712e-05,
-      "loss": 0.8934,
+      "epoch": 0.04155337672681333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993022086013706e-05,
+      "loss": 1.0565,
       "step": 1513
     },
     {
-      "epoch": 0.4947712418300654,
-      "grad_norm": 1.26417875289917,
-      "learning_rate": 1.0655791651759174e-05,
-      "loss": 0.9574,
+      "epoch": 0.04158084095465656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992988822182505e-05,
+      "loss": 1.1143,
       "step": 1514
     },
     {
-      "epoch": 0.4950980392156863,
-      "grad_norm": 1.363433599472046,
-      "learning_rate": 1.0645229190785678e-05,
-      "loss": 0.9429,
+      "epoch": 0.04160830518249979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992955479282907e-05,
+      "loss": 1.0247,
       "step": 1515
     },
     {
-      "epoch": 0.4954248366013072,
-      "grad_norm": 1.3637996912002563,
-      "learning_rate": 1.0634666006899375e-05,
-      "loss": 0.9567,
+      "epoch": 0.041635769410343026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992922057315175e-05,
+      "loss": 1.0713,
       "step": 1516
     },
     {
-      "epoch": 0.4957516339869281,
-      "grad_norm": 1.395114779472351,
-      "learning_rate": 1.0624102111935224e-05,
-      "loss": 0.9636,
+      "epoch": 0.04166323363818626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999288855627957e-05,
+      "loss": 1.1756,
       "step": 1517
     },
     {
-      "epoch": 0.49607843137254903,
-      "grad_norm": 1.339880347251892,
-      "learning_rate": 1.061353751772898e-05,
-      "loss": 0.9317,
+      "epoch": 0.0416906978660295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992854976176365e-05,
+      "loss": 1.0892,
       "step": 1518
     },
     {
-      "epoch": 0.49640522875816995,
-      "grad_norm": 1.2920236587524414,
-      "learning_rate": 1.0602972236117187e-05,
-      "loss": 0.9685,
+      "epoch": 0.04171816209387273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992821317005818e-05,
+      "loss": 1.1765,
       "step": 1519
     },
     {
-      "epoch": 0.49673202614379086,
-      "grad_norm": 1.3776479959487915,
-      "learning_rate": 1.0592406278937143e-05,
-      "loss": 0.9507,
+      "epoch": 0.04174562632171597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992787578768194e-05,
+      "loss": 1.1036,
       "step": 1520
     },
     {
-      "epoch": 0.4970588235294118,
-      "grad_norm": 1.3718396425247192,
-      "learning_rate": 1.0581839658026924e-05,
-      "loss": 0.9425,
+      "epoch": 0.0417730905495592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999275376146377e-05,
+      "loss": 1.1964,
       "step": 1521
     },
     {
-      "epoch": 0.4973856209150327,
-      "grad_norm": 1.3950423002243042,
-      "learning_rate": 1.0571272385225338e-05,
-      "loss": 0.935,
+      "epoch": 0.04180055477740243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992719865092806e-05,
+      "loss": 1.1903,
       "step": 1522
     },
     {
-      "epoch": 0.4977124183006536,
-      "grad_norm": 1.3125123977661133,
-      "learning_rate": 1.0560704472371919e-05,
-      "loss": 0.9145,
+      "epoch": 0.041828019005245666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999268588965557e-05,
+      "loss": 1.1965,
       "step": 1523
     },
     {
-      "epoch": 0.4980392156862745,
-      "grad_norm": 1.3265632390975952,
-      "learning_rate": 1.055013593130693e-05,
-      "loss": 0.9524,
+      "epoch": 0.0418554832330889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992651835152333e-05,
+      "loss": 1.0888,
       "step": 1524
     },
     {
-      "epoch": 0.49836601307189543,
-      "grad_norm": 1.3799357414245605,
-      "learning_rate": 1.0539566773871332e-05,
-      "loss": 0.9859,
+      "epoch": 0.041882947460932136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992617701583362e-05,
+      "loss": 1.1324,
       "step": 1525
     },
     {
-      "epoch": 0.49869281045751634,
-      "grad_norm": 1.3639473915100098,
-      "learning_rate": 1.0528997011906775e-05,
-      "loss": 0.9629,
+      "epoch": 0.04191041168877537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992583488948928e-05,
+      "loss": 1.0461,
       "step": 1526
     },
     {
-      "epoch": 0.49901960784313726,
-      "grad_norm": 1.2557101249694824,
-      "learning_rate": 1.0518426657255585e-05,
-      "loss": 0.9359,
+      "epoch": 0.04193787591661861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992549197249307e-05,
+      "loss": 1.1664,
       "step": 1527
     },
     {
-      "epoch": 0.4993464052287582,
-      "grad_norm": 1.3006701469421387,
-      "learning_rate": 1.0507855721760757e-05,
-      "loss": 0.9357,
+      "epoch": 0.041965340144461835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992514826484765e-05,
+      "loss": 1.1328,
       "step": 1528
     },
     {
-      "epoch": 0.4996732026143791,
-      "grad_norm": 1.3332628011703491,
-      "learning_rate": 1.0497284217265938e-05,
-      "loss": 0.9212,
+      "epoch": 0.04199280437230507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992480376655573e-05,
+      "loss": 1.0967,
       "step": 1529
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 1.4079618453979492,
-      "learning_rate": 1.0486712155615408e-05,
-      "loss": 0.9645,
+      "epoch": 0.042020268600148306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999244584776201e-05,
+      "loss": 1.0629,
       "step": 1530
     },
     {
-      "epoch": 0.5003267973856209,
-      "grad_norm": 1.3262722492218018,
-      "learning_rate": 1.047613954865407e-05,
-      "loss": 0.9791,
+      "epoch": 0.04204773282799154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999241123980434e-05,
+      "loss": 1.0956,
       "step": 1531
     },
     {
-      "epoch": 0.5006535947712418,
-      "grad_norm": 1.412726879119873,
-      "learning_rate": 1.046556640822744e-05,
-      "loss": 0.9452,
+      "epoch": 0.042075197055834776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992376552782842e-05,
+      "loss": 1.1458,
       "step": 1532
     },
     {
-      "epoch": 0.5009803921568627,
-      "grad_norm": 1.3216849565505981,
-      "learning_rate": 1.0454992746181636e-05,
-      "loss": 0.9813,
+      "epoch": 0.04210266128367801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999234178669779e-05,
+      "loss": 1.0731,
       "step": 1533
     },
     {
-      "epoch": 0.5013071895424837,
-      "grad_norm": 1.3122748136520386,
-      "learning_rate": 1.0444418574363354e-05,
-      "loss": 1.0072,
+      "epoch": 0.042130125511521246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992306941549466e-05,
+      "loss": 1.0903,
       "step": 1534
     },
     {
-      "epoch": 0.5016339869281046,
-      "grad_norm": 1.3326642513275146,
-      "learning_rate": 1.0433843904619862e-05,
-      "loss": 0.9708,
+      "epoch": 0.042157589739364475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999227201733813e-05,
+      "loss": 1.1249,
       "step": 1535
     },
     {
-      "epoch": 0.5019607843137255,
-      "grad_norm": 1.3189812898635864,
-      "learning_rate": 1.0423268748798992e-05,
-      "loss": 0.9495,
+      "epoch": 0.04218505396720771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999223701406407e-05,
+      "loss": 1.1137,
       "step": 1536
     },
     {
-      "epoch": 0.5022875816993464,
-      "grad_norm": 1.2931123971939087,
-      "learning_rate": 1.041269311874911e-05,
-      "loss": 0.9793,
+      "epoch": 0.042212518195050945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999220193172756e-05,
+      "loss": 1.2112,
       "step": 1537
     },
     {
-      "epoch": 0.5026143790849673,
-      "grad_norm": 1.2677775621414185,
-      "learning_rate": 1.0402117026319128e-05,
-      "loss": 0.9391,
+      "epoch": 0.04223998242289418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992166770328878e-05,
+      "loss": 1.1382,
       "step": 1538
     },
     {
-      "epoch": 0.5029411764705882,
-      "grad_norm": 1.2882936000823975,
-      "learning_rate": 1.0391540483358458e-05,
-      "loss": 0.9077,
+      "epoch": 0.042267446650737416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99921315298683e-05,
+      "loss": 1.1875,
       "step": 1539
     },
     {
-      "epoch": 0.5032679738562091,
-      "grad_norm": 1.3218257427215576,
-      "learning_rate": 1.0380963501717034e-05,
-      "loss": 0.9438,
+      "epoch": 0.04229491087858065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999209621034611e-05,
+      "loss": 1.1365,
       "step": 1540
     },
     {
-      "epoch": 0.50359477124183,
-      "grad_norm": 1.2979398965835571,
-      "learning_rate": 1.0370386093245271e-05,
-      "loss": 0.9358,
+      "epoch": 0.042322375106423886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992060811762585e-05,
+      "loss": 1.0805,
       "step": 1541
     },
     {
-      "epoch": 0.503921568627451,
-      "grad_norm": 1.3852407932281494,
-      "learning_rate": 1.0359808269794063e-05,
-      "loss": 0.9306,
+      "epoch": 0.042349839334267114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992025334118e-05,
+      "loss": 1.0887,
       "step": 1542
     },
     {
-      "epoch": 0.5042483660130719,
-      "grad_norm": 1.3797996044158936,
-      "learning_rate": 1.0349230043214773e-05,
-      "loss": 0.9346,
+      "epoch": 0.04237730356211035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991989777412644e-05,
+      "loss": 1.0458,
       "step": 1543
     },
     {
-      "epoch": 0.5045751633986928,
-      "grad_norm": 1.2847553491592407,
-      "learning_rate": 1.0338651425359217e-05,
-      "loss": 0.9341,
+      "epoch": 0.042404767789953585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991954141646794e-05,
+      "loss": 1.1426,
       "step": 1544
     },
     {
-      "epoch": 0.5049019607843137,
-      "grad_norm": 1.2948590517044067,
-      "learning_rate": 1.0328072428079637e-05,
-      "loss": 0.9363,
+      "epoch": 0.04243223201779682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999191842682073e-05,
+      "loss": 1.1859,
       "step": 1545
     },
     {
-      "epoch": 0.5052287581699346,
-      "grad_norm": 1.3046973943710327,
-      "learning_rate": 1.0317493063228716e-05,
-      "loss": 0.9378,
+      "epoch": 0.042459696245640055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999188263293474e-05,
+      "loss": 1.1261,
       "step": 1546
     },
     {
-      "epoch": 0.5055555555555555,
-      "grad_norm": 1.368383526802063,
-      "learning_rate": 1.0306913342659544e-05,
-      "loss": 0.9416,
+      "epoch": 0.04248716047348329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99918467599891e-05,
+      "loss": 1.127,
       "step": 1547
     },
     {
-      "epoch": 0.5058823529411764,
-      "grad_norm": 1.308611512184143,
-      "learning_rate": 1.0296333278225599e-05,
-      "loss": 0.9606,
+      "epoch": 0.042514624701326525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99918108079841e-05,
+      "loss": 1.0364,
       "step": 1548
     },
     {
-      "epoch": 0.5062091503267974,
-      "grad_norm": 1.27964448928833,
-      "learning_rate": 1.0285752881780759e-05,
-      "loss": 0.9679,
+      "epoch": 0.042542088929169754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991774776920025e-05,
+      "loss": 1.0835,
       "step": 1549
     },
     {
-      "epoch": 0.5065359477124183,
-      "grad_norm": 1.285178303718567,
-      "learning_rate": 1.0275172165179268e-05,
-      "loss": 0.9543,
+      "epoch": 0.04256955315701299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999173866679715e-05,
+      "loss": 1.0703,
       "step": 1550
     },
     {
-      "epoch": 0.5068627450980392,
-      "grad_norm": 1.2803425788879395,
-      "learning_rate": 1.0264591140275726e-05,
-      "loss": 0.9615,
+      "epoch": 0.042597017384856224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991702477615776e-05,
+      "loss": 1.019,
       "step": 1551
     },
     {
-      "epoch": 0.5071895424836601,
-      "grad_norm": 1.3136039972305298,
-      "learning_rate": 1.0254009818925082e-05,
-      "loss": 0.9494,
+      "epoch": 0.04262448161269946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999166620937618e-05,
+      "loss": 1.1271,
       "step": 1552
     },
     {
-      "epoch": 0.507516339869281,
-      "grad_norm": 1.279526948928833,
-      "learning_rate": 1.0243428212982615e-05,
-      "loss": 0.9602,
+      "epoch": 0.042651945840542695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999162986207865e-05,
+      "loss": 1.072,
       "step": 1553
     },
     {
-      "epoch": 0.5078431372549019,
-      "grad_norm": 1.2501652240753174,
-      "learning_rate": 1.0232846334303925e-05,
-      "loss": 0.9528,
+      "epoch": 0.04267941006838593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991593435723473e-05,
+      "loss": 1.17,
       "step": 1554
     },
     {
-      "epoch": 0.5081699346405228,
-      "grad_norm": 1.2541040182113647,
-      "learning_rate": 1.0222264194744918e-05,
-      "loss": 0.9788,
+      "epoch": 0.04270687429622916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999155693031094e-05,
+      "loss": 1.0809,
       "step": 1555
     },
     {
-      "epoch": 0.5084967320261438,
-      "grad_norm": 1.3344426155090332,
-      "learning_rate": 1.0211681806161787e-05,
-      "loss": 0.9518,
+      "epoch": 0.04273433852407239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991520345841334e-05,
+      "loss": 1.2201,
       "step": 1556
     },
     {
-      "epoch": 0.5088235294117647,
-      "grad_norm": 1.3039205074310303,
-      "learning_rate": 1.0201099180411008e-05,
-      "loss": 0.9143,
+      "epoch": 0.04276180275191563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991483682314953e-05,
+      "loss": 1.0883,
       "step": 1557
     },
     {
-      "epoch": 0.5091503267973856,
-      "grad_norm": 1.279098391532898,
-      "learning_rate": 1.0190516329349328e-05,
-      "loss": 0.9302,
+      "epoch": 0.042789266979758864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991446939732083e-05,
+      "loss": 1.1515,
       "step": 1558
     },
     {
-      "epoch": 0.5094771241830065,
-      "grad_norm": 1.316380500793457,
-      "learning_rate": 1.0179933264833732e-05,
-      "loss": 0.9655,
+      "epoch": 0.0428167312076021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991410118093012e-05,
+      "loss": 1.1153,
       "step": 1559
     },
     {
-      "epoch": 0.5098039215686274,
-      "grad_norm": 1.317154049873352,
-      "learning_rate": 1.0169349998721459e-05,
-      "loss": 0.9461,
+      "epoch": 0.042844195435445334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991373217398032e-05,
+      "loss": 1.1674,
       "step": 1560
     },
     {
-      "epoch": 0.5101307189542483,
-      "grad_norm": 1.2698218822479248,
-      "learning_rate": 1.0158766542869967e-05,
-      "loss": 0.9471,
+      "epoch": 0.04287165966328857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999133623764744e-05,
+      "loss": 1.0005,
       "step": 1561
     },
     {
-      "epoch": 0.5104575163398692,
-      "grad_norm": 1.338861346244812,
-      "learning_rate": 1.0148182909136928e-05,
-      "loss": 0.9437,
+      "epoch": 0.0428991238911318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991299178841527e-05,
+      "loss": 1.0767,
       "step": 1562
     },
     {
-      "epoch": 0.5107843137254902,
-      "grad_norm": 1.2621362209320068,
-      "learning_rate": 1.0137599109380209e-05,
-      "loss": 0.9528,
+      "epoch": 0.04292658811897503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999126204098058e-05,
+      "loss": 1.0087,
       "step": 1563
     },
     {
-      "epoch": 0.5111111111111111,
-      "grad_norm": 1.3028907775878906,
-      "learning_rate": 1.0127015155457875e-05,
-      "loss": 0.8833,
+      "epoch": 0.04295405234681827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991224824064902e-05,
+      "loss": 1.1348,
       "step": 1564
     },
     {
-      "epoch": 0.511437908496732,
-      "grad_norm": 1.3314871788024902,
-      "learning_rate": 1.0116431059228147e-05,
-      "loss": 0.9507,
+      "epoch": 0.0429815165746615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991187528094784e-05,
+      "loss": 1.1924,
       "step": 1565
     },
     {
-      "epoch": 0.5117647058823529,
-      "grad_norm": 1.2624768018722534,
-      "learning_rate": 1.0105846832549417e-05,
-      "loss": 0.942,
+      "epoch": 0.04300898080250474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991150153070515e-05,
+      "loss": 1.1496,
       "step": 1566
     },
     {
-      "epoch": 0.5120915032679738,
-      "grad_norm": 1.2908921241760254,
-      "learning_rate": 1.0095262487280222e-05,
-      "loss": 1.0005,
+      "epoch": 0.043036445030347974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991112698992397e-05,
+      "loss": 1.1633,
       "step": 1567
     },
     {
-      "epoch": 0.5124183006535947,
-      "grad_norm": 1.298193335533142,
-      "learning_rate": 1.0084678035279227e-05,
-      "loss": 0.9449,
+      "epoch": 0.04306390925819121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991075165860727e-05,
+      "loss": 1.2357,
       "step": 1568
     },
     {
-      "epoch": 0.5127450980392156,
-      "grad_norm": 1.3066232204437256,
-      "learning_rate": 1.0074093488405223e-05,
-      "loss": 0.9372,
+      "epoch": 0.04309137348603444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991037553675802e-05,
+      "loss": 1.0214,
       "step": 1569
     },
     {
-      "epoch": 0.5130718954248366,
-      "grad_norm": 1.3465402126312256,
-      "learning_rate": 1.00635088585171e-05,
-      "loss": 0.9625,
+      "epoch": 0.04311883771387767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999099986243791e-05,
+      "loss": 1.0533,
       "step": 1570
     },
     {
-      "epoch": 0.5133986928104575,
-      "grad_norm": 1.2949174642562866,
-      "learning_rate": 1.0052924157473847e-05,
-      "loss": 0.9428,
+      "epoch": 0.04314630194172091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990962092147364e-05,
+      "loss": 1.1349,
       "step": 1571
     },
     {
-      "epoch": 0.5137254901960784,
-      "grad_norm": 1.3134602308273315,
-      "learning_rate": 1.0042339397134528e-05,
-      "loss": 0.9574,
+      "epoch": 0.04317376616956414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990924242804452e-05,
+      "loss": 1.0977,
       "step": 1572
     },
     {
-      "epoch": 0.5140522875816993,
-      "grad_norm": 1.2252238988876343,
-      "learning_rate": 1.003175458935828e-05,
-      "loss": 0.9594,
+      "epoch": 0.04320123039740738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990886314409482e-05,
+      "loss": 1.186,
       "step": 1573
     },
     {
-      "epoch": 0.5143790849673202,
-      "grad_norm": 1.2998151779174805,
-      "learning_rate": 1.0021169746004287e-05,
-      "loss": 0.922,
+      "epoch": 0.04322869462525061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990848306962744e-05,
+      "loss": 1.1252,
       "step": 1574
     },
     {
-      "epoch": 0.5147058823529411,
-      "grad_norm": 1.298860788345337,
-      "learning_rate": 1.001058487893178e-05,
-      "loss": 0.9389,
+      "epoch": 0.04325615885309385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999081022046455e-05,
+      "loss": 1.144,
       "step": 1575
     },
     {
-      "epoch": 0.515032679738562,
-      "grad_norm": 1.3734941482543945,
-      "learning_rate": 1e-05,
-      "loss": 0.9457,
+      "epoch": 0.04328362308093708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990772054915187e-05,
+      "loss": 1.1851,
       "step": 1576
     },
     {
-      "epoch": 0.515359477124183,
-      "grad_norm": 1.2623703479766846,
-      "learning_rate": 9.989415121068224e-06,
-      "loss": 1.0068,
+      "epoch": 0.04331108730878031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990733810314972e-05,
+      "loss": 1.1704,
       "step": 1577
     },
     {
-      "epoch": 0.515686274509804,
-      "grad_norm": 1.370967149734497,
-      "learning_rate": 9.978830253995713e-06,
-      "loss": 0.9503,
+      "epoch": 0.04333855153662355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99906954866642e-05,
+      "loss": 1.1309,
       "step": 1578
     },
     {
-      "epoch": 0.5160130718954249,
-      "grad_norm": 1.2963651418685913,
-      "learning_rate": 9.968245410641726e-06,
-      "loss": 0.9045,
+      "epoch": 0.04336601576446678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990657083963175e-05,
+      "loss": 1.1971,
       "step": 1579
     },
     {
-      "epoch": 0.5163398692810458,
-      "grad_norm": 1.3179384469985962,
-      "learning_rate": 9.957660602865477e-06,
-      "loss": 0.9466,
+      "epoch": 0.04339347999231002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990618602212203e-05,
+      "loss": 1.0319,
       "step": 1580
     },
     {
-      "epoch": 0.5166666666666667,
-      "grad_norm": 1.3017592430114746,
-      "learning_rate": 9.947075842526156e-06,
-      "loss": 0.9501,
+      "epoch": 0.04342094422015325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990580041411583e-05,
+      "loss": 1.189,
       "step": 1581
     },
     {
-      "epoch": 0.5169934640522876,
-      "grad_norm": 1.2774056196212769,
-      "learning_rate": 9.936491141482902e-06,
-      "loss": 0.9503,
+      "epoch": 0.04344840844799649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990541401561624e-05,
+      "loss": 1.1293,
       "step": 1582
     },
     {
-      "epoch": 0.5173202614379085,
-      "grad_norm": 1.382008671760559,
-      "learning_rate": 9.92590651159478e-06,
-      "loss": 0.9648,
+      "epoch": 0.043475872675839716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990502682662636e-05,
+      "loss": 1.0504,
       "step": 1583
     },
     {
-      "epoch": 0.5176470588235295,
-      "grad_norm": 1.3609120845794678,
-      "learning_rate": 9.915321964720773e-06,
-      "loss": 0.9533,
+      "epoch": 0.04350333690368295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990463884714915e-05,
+      "loss": 1.1084,
       "step": 1584
     },
     {
-      "epoch": 0.5179738562091504,
-      "grad_norm": 1.3207111358642578,
-      "learning_rate": 9.904737512719781e-06,
-      "loss": 0.9394,
+      "epoch": 0.04353080113152619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990425007718777e-05,
+      "loss": 1.2344,
       "step": 1585
     },
     {
-      "epoch": 0.5183006535947713,
-      "grad_norm": 1.3128529787063599,
-      "learning_rate": 9.894153167450586e-06,
-      "loss": 0.9571,
+      "epoch": 0.04355826535936942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990386051674525e-05,
+      "loss": 1.0804,
       "step": 1586
     },
     {
-      "epoch": 0.5186274509803922,
-      "grad_norm": 1.305372714996338,
-      "learning_rate": 9.883568940771856e-06,
-      "loss": 0.9693,
+      "epoch": 0.04358572958721266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990347016582468e-05,
+      "loss": 1.1159,
       "step": 1587
     },
     {
-      "epoch": 0.5189542483660131,
-      "grad_norm": 1.3641544580459595,
-      "learning_rate": 9.872984844542128e-06,
-      "loss": 0.9668,
+      "epoch": 0.04361319381505589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999030790244292e-05,
+      "loss": 1.1407,
       "step": 1588
     },
     {
-      "epoch": 0.519281045751634,
-      "grad_norm": 1.4035249948501587,
-      "learning_rate": 9.862400890619791e-06,
-      "loss": 0.9384,
+      "epoch": 0.04364065804289912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999026870925618e-05,
+      "loss": 1.0469,
       "step": 1589
     },
     {
-      "epoch": 0.5196078431372549,
-      "grad_norm": 1.2797895669937134,
-      "learning_rate": 9.851817090863073e-06,
-      "loss": 0.9725,
+      "epoch": 0.043668122270742356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999022943702257e-05,
+      "loss": 1.0571,
       "step": 1590
     },
     {
-      "epoch": 0.5199346405228759,
-      "grad_norm": 1.2850505113601685,
-      "learning_rate": 9.841233457130036e-06,
-      "loss": 0.9466,
+      "epoch": 0.04369558649858559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999019008574239e-05,
+      "loss": 1.1315,
       "step": 1591
     },
     {
-      "epoch": 0.5202614379084968,
-      "grad_norm": 1.2995980978012085,
-      "learning_rate": 9.830650001278543e-06,
-      "loss": 0.9307,
+      "epoch": 0.043723050726428826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990150655415958e-05,
+      "loss": 1.2513,
       "step": 1592
     },
     {
-      "epoch": 0.5205882352941177,
-      "grad_norm": 1.3166271448135376,
-      "learning_rate": 9.820066735166271e-06,
-      "loss": 0.9437,
+      "epoch": 0.04375051495427206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990111146043585e-05,
+      "loss": 1.0316,
       "step": 1593
     },
     {
-      "epoch": 0.5209150326797386,
-      "grad_norm": 1.2951669692993164,
-      "learning_rate": 9.809483670650676e-06,
-      "loss": 0.9641,
+      "epoch": 0.0437779791821153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999007155762558e-05,
+      "loss": 1.1522,
       "step": 1594
     },
     {
-      "epoch": 0.5212418300653595,
-      "grad_norm": 1.3487752676010132,
-      "learning_rate": 9.798900819588992e-06,
-      "loss": 0.9706,
+      "epoch": 0.04380544340995853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990031890162264e-05,
+      "loss": 1.0114,
       "step": 1595
     },
     {
-      "epoch": 0.5215686274509804,
-      "grad_norm": 1.3355836868286133,
-      "learning_rate": 9.788318193838218e-06,
-      "loss": 0.9518,
+      "epoch": 0.04383290763780176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989992143653943e-05,
+      "loss": 1.1215,
       "step": 1596
     },
     {
-      "epoch": 0.5218954248366013,
-      "grad_norm": 1.3281069993972778,
-      "learning_rate": 9.777735805255087e-06,
-      "loss": 0.9498,
+      "epoch": 0.043860371865644995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989952318100935e-05,
+      "loss": 0.9905,
       "step": 1597
     },
     {
-      "epoch": 0.5222222222222223,
-      "grad_norm": 1.3336400985717773,
-      "learning_rate": 9.767153665696077e-06,
-      "loss": 0.9512,
+      "epoch": 0.04388783609348823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989912413503554e-05,
+      "loss": 1.0385,
       "step": 1598
     },
     {
-      "epoch": 0.5225490196078432,
-      "grad_norm": 1.3495811223983765,
-      "learning_rate": 9.756571787017386e-06,
-      "loss": 0.9339,
+      "epoch": 0.043915300321331466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989872429862116e-05,
+      "loss": 1.0811,
       "step": 1599
     },
     {
-      "epoch": 0.5228758169934641,
-      "grad_norm": 1.317995548248291,
-      "learning_rate": 9.74599018107492e-06,
-      "loss": 0.9958,
+      "epoch": 0.0439427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989832367176938e-05,
+      "loss": 1.1783,
       "step": 1600
     },
     {
-      "epoch": 0.523202614379085,
-      "grad_norm": 1.3135356903076172,
-      "learning_rate": 9.735408859724276e-06,
-      "loss": 0.9391,
+      "epoch": 0.043970228777017936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989792225448337e-05,
+      "loss": 0.9966,
       "step": 1601
     },
     {
-      "epoch": 0.5235294117647059,
-      "grad_norm": 1.2901458740234375,
-      "learning_rate": 9.724827834820736e-06,
-      "loss": 0.9404,
+      "epoch": 0.04399769300486117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989752004676626e-05,
+      "loss": 1.073,
       "step": 1602
     },
     {
-      "epoch": 0.5238562091503268,
-      "grad_norm": 1.3350518941879272,
-      "learning_rate": 9.714247118219243e-06,
-      "loss": 0.9418,
+      "epoch": 0.0440251572327044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989711704862132e-05,
+      "loss": 1.0443,
       "step": 1603
     },
     {
-      "epoch": 0.5241830065359477,
-      "grad_norm": 1.375985860824585,
-      "learning_rate": 9.703666721774403e-06,
-      "loss": 0.9479,
+      "epoch": 0.044052621460547635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998967132600517e-05,
+      "loss": 1.069,
       "step": 1604
     },
     {
-      "epoch": 0.5245098039215687,
-      "grad_norm": 1.3404607772827148,
-      "learning_rate": 9.69308665734046e-06,
-      "loss": 0.9353,
+      "epoch": 0.04408008568839087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989630868106058e-05,
+      "loss": 1.0896,
       "step": 1605
     },
     {
-      "epoch": 0.5248366013071896,
-      "grad_norm": 1.322981595993042,
-      "learning_rate": 9.682506936771284e-06,
-      "loss": 0.9336,
+      "epoch": 0.044107549916234105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989590331165117e-05,
+      "loss": 1.0698,
       "step": 1606
     },
     {
-      "epoch": 0.5251633986928105,
-      "grad_norm": 1.2817202806472778,
-      "learning_rate": 9.671927571920368e-06,
-      "loss": 0.9423,
+      "epoch": 0.04413501414407734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989549715182667e-05,
+      "loss": 1.0467,
       "step": 1607
     },
     {
-      "epoch": 0.5254901960784314,
-      "grad_norm": 1.3941251039505005,
-      "learning_rate": 9.66134857464079e-06,
-      "loss": 0.9571,
+      "epoch": 0.044162478371920576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998950902015903e-05,
+      "loss": 1.1206,
       "step": 1608
     },
     {
-      "epoch": 0.5258169934640523,
-      "grad_norm": 1.24065363407135,
-      "learning_rate": 9.65076995678523e-06,
-      "loss": 0.9698,
+      "epoch": 0.04418994259976381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998946824609453e-05,
+      "loss": 1.1543,
       "step": 1609
     },
     {
-      "epoch": 0.5261437908496732,
-      "grad_norm": 1.2351130247116089,
-      "learning_rate": 9.64019173020594e-06,
-      "loss": 0.9882,
+      "epoch": 0.04421740682760704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989427392989483e-05,
+      "loss": 1.1306,
       "step": 1610
     },
     {
-      "epoch": 0.5264705882352941,
-      "grad_norm": 1.2933920621871948,
-      "learning_rate": 9.629613906754732e-06,
-      "loss": 0.9625,
+      "epoch": 0.044244871055450274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998938646084422e-05,
+      "loss": 1.0474,
       "step": 1611
     },
     {
-      "epoch": 0.526797385620915,
-      "grad_norm": 1.2806869745254517,
-      "learning_rate": 9.619036498282968e-06,
-      "loss": 0.9703,
+      "epoch": 0.04427233528329351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989345449659063e-05,
+      "loss": 1.0363,
       "step": 1612
     },
     {
-      "epoch": 0.527124183006536,
-      "grad_norm": 1.3265235424041748,
-      "learning_rate": 9.608459516641543e-06,
-      "loss": 0.9423,
+      "epoch": 0.044299799511136745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989304359434334e-05,
+      "loss": 1.1586,
       "step": 1613
     },
     {
-      "epoch": 0.5274509803921569,
-      "grad_norm": 1.2592096328735352,
-      "learning_rate": 9.597882973680877e-06,
-      "loss": 0.9577,
+      "epoch": 0.04432726373897998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998926319017036e-05,
+      "loss": 1.1279,
       "step": 1614
     },
     {
-      "epoch": 0.5277777777777778,
-      "grad_norm": 1.2904943227767944,
-      "learning_rate": 9.587306881250892e-06,
-      "loss": 0.9471,
+      "epoch": 0.044354727966823215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989221941867463e-05,
+      "loss": 1.1363,
       "step": 1615
     },
     {
-      "epoch": 0.5281045751633987,
-      "grad_norm": 1.3368679285049438,
-      "learning_rate": 9.57673125120101e-06,
-      "loss": 0.9424,
+      "epoch": 0.044382192194666444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989180614525975e-05,
+      "loss": 1.077,
       "step": 1616
     },
     {
-      "epoch": 0.5284313725490196,
-      "grad_norm": 1.3245139122009277,
-      "learning_rate": 9.56615609538014e-06,
-      "loss": 0.9496,
+      "epoch": 0.04440965642250968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989139208146222e-05,
+      "loss": 1.1896,
       "step": 1617
     },
     {
-      "epoch": 0.5287581699346405,
-      "grad_norm": 1.256258487701416,
-      "learning_rate": 9.555581425636648e-06,
-      "loss": 0.95,
+      "epoch": 0.044437120650352914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989097722728528e-05,
+      "loss": 1.0579,
       "step": 1618
     },
     {
-      "epoch": 0.5290849673202614,
-      "grad_norm": 1.3639835119247437,
-      "learning_rate": 9.54500725381837e-06,
-      "loss": 0.9366,
+      "epoch": 0.04446458487819615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989056158273227e-05,
+      "loss": 1.2067,
       "step": 1619
     },
     {
-      "epoch": 0.5294117647058824,
-      "grad_norm": 1.3078570365905762,
-      "learning_rate": 9.534433591772562e-06,
-      "loss": 0.9437,
+      "epoch": 0.044492049106039384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998901451478064e-05,
+      "loss": 0.9919,
       "step": 1620
     },
     {
-      "epoch": 0.5297385620915033,
-      "grad_norm": 1.3476070165634155,
-      "learning_rate": 9.523860451345934e-06,
-      "loss": 0.9492,
+      "epoch": 0.04451951333388262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988972792251103e-05,
+      "loss": 1.0942,
       "step": 1621
     },
     {
-      "epoch": 0.5300653594771242,
-      "grad_norm": 1.3051077127456665,
-      "learning_rate": 9.513287844384594e-06,
-      "loss": 0.9855,
+      "epoch": 0.044546977561725855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988930990684944e-05,
+      "loss": 1.1451,
       "step": 1622
     },
     {
-      "epoch": 0.5303921568627451,
-      "grad_norm": 1.3136478662490845,
-      "learning_rate": 9.502715782734064e-06,
-      "loss": 0.9245,
+      "epoch": 0.04457444178956908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998888911008249e-05,
+      "loss": 1.1305,
       "step": 1623
     },
     {
-      "epoch": 0.530718954248366,
-      "grad_norm": 1.2764275074005127,
-      "learning_rate": 9.492144278239244e-06,
-      "loss": 0.9092,
+      "epoch": 0.04460190601741232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988847150444083e-05,
+      "loss": 1.0267,
       "step": 1624
     },
     {
-      "epoch": 0.5310457516339869,
-      "grad_norm": 1.3181718587875366,
-      "learning_rate": 9.481573342744418e-06,
-      "loss": 0.9641,
+      "epoch": 0.044629370245255554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998880511177004e-05,
+      "loss": 1.024,
       "step": 1625
     },
     {
-      "epoch": 0.5313725490196078,
-      "grad_norm": 1.3150854110717773,
-      "learning_rate": 9.47100298809323e-06,
-      "loss": 0.9587,
+      "epoch": 0.04465683447309879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988762994060706e-05,
+      "loss": 1.092,
       "step": 1626
     },
     {
-      "epoch": 0.5316993464052288,
-      "grad_norm": 1.391109824180603,
-      "learning_rate": 9.46043322612867e-06,
-      "loss": 0.9284,
+      "epoch": 0.044684298700942024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998872079731641e-05,
+      "loss": 1.0772,
       "step": 1627
     },
     {
-      "epoch": 0.5320261437908497,
-      "grad_norm": 1.2786201238632202,
-      "learning_rate": 9.449864068693072e-06,
-      "loss": 0.9287,
+      "epoch": 0.04471176292878526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988678521537485e-05,
+      "loss": 1.0069,
       "step": 1628
     },
     {
-      "epoch": 0.5323529411764706,
-      "grad_norm": 1.2938239574432373,
-      "learning_rate": 9.439295527628083e-06,
-      "loss": 0.9544,
+      "epoch": 0.044739227156628494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988636166724265e-05,
+      "loss": 1.0704,
       "step": 1629
     },
     {
-      "epoch": 0.5326797385620915,
-      "grad_norm": 1.348069667816162,
-      "learning_rate": 9.428727614774667e-06,
-      "loss": 0.945,
+      "epoch": 0.04476669138447172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988593732877088e-05,
+      "loss": 1.1602,
       "step": 1630
     },
     {
-      "epoch": 0.5330065359477124,
-      "grad_norm": 1.2734588384628296,
-      "learning_rate": 9.418160341973079e-06,
-      "loss": 0.9696,
+      "epoch": 0.04479415561231496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998855121999629e-05,
+      "loss": 1.0738,
       "step": 1631
     },
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 1.3421164751052856,
-      "learning_rate": 9.407593721062858e-06,
-      "loss": 0.9565,
+      "epoch": 0.04482161984015819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988508628082203e-05,
+      "loss": 1.1185,
       "step": 1632
     },
     {
-      "epoch": 0.5336601307189542,
-      "grad_norm": 1.3137770891189575,
-      "learning_rate": 9.397027763882818e-06,
-      "loss": 0.9332,
+      "epoch": 0.04484908406800143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988465957135166e-05,
+      "loss": 1.1825,
       "step": 1633
     },
     {
-      "epoch": 0.5339869281045752,
-      "grad_norm": 1.2879812717437744,
-      "learning_rate": 9.38646248227102e-06,
-      "loss": 0.9604,
+      "epoch": 0.044876548295844663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998842320715552e-05,
+      "loss": 1.0959,
       "step": 1634
     },
     {
-      "epoch": 0.5343137254901961,
-      "grad_norm": 1.3488954305648804,
-      "learning_rate": 9.375897888064776e-06,
-      "loss": 0.9507,
+      "epoch": 0.0449040125236879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99883803781436e-05,
+      "loss": 1.1181,
       "step": 1635
     },
     {
-      "epoch": 0.534640522875817,
-      "grad_norm": 1.275715708732605,
-      "learning_rate": 9.365333993100628e-06,
-      "loss": 0.9397,
+      "epoch": 0.044931476751531134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988337470099743e-05,
+      "loss": 1.1384,
       "step": 1636
     },
     {
-      "epoch": 0.5349673202614379,
-      "grad_norm": 1.2202703952789307,
-      "learning_rate": 9.354770809214326e-06,
-      "loss": 0.9437,
+      "epoch": 0.04495894097937436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988294483024292e-05,
+      "loss": 1.0325,
       "step": 1637
     },
     {
-      "epoch": 0.5352941176470588,
-      "grad_norm": 1.3167457580566406,
-      "learning_rate": 9.34420834824083e-06,
-      "loss": 0.9245,
+      "epoch": 0.0449864052072176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988251416917585e-05,
+      "loss": 1.124,
       "step": 1638
     },
     {
-      "epoch": 0.5356209150326797,
-      "grad_norm": 1.4090172052383423,
-      "learning_rate": 9.33364662201429e-06,
-      "loss": 0.933,
+      "epoch": 0.04501386943506083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998820827177997e-05,
+      "loss": 1.0251,
       "step": 1639
     },
     {
-      "epoch": 0.5359477124183006,
-      "grad_norm": 1.296369194984436,
-      "learning_rate": 9.323085642368032e-06,
-      "loss": 0.9461,
+      "epoch": 0.04504133366290407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988165047611773e-05,
+      "loss": 1.1958,
       "step": 1640
     },
     {
-      "epoch": 0.5362745098039216,
-      "grad_norm": 1.3061758279800415,
-      "learning_rate": 9.312525421134549e-06,
-      "loss": 0.9193,
+      "epoch": 0.0450687978907473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998812174441335e-05,
+      "loss": 1.0209,
       "step": 1641
     },
     {
-      "epoch": 0.5366013071895425,
-      "grad_norm": 1.328555703163147,
-      "learning_rate": 9.301965970145472e-06,
-      "loss": 0.9406,
+      "epoch": 0.04509626211859054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988078362185036e-05,
+      "loss": 1.0861,
       "step": 1642
     },
     {
-      "epoch": 0.5369281045751634,
-      "grad_norm": 1.2992769479751587,
-      "learning_rate": 9.29140730123158e-06,
-      "loss": 0.9235,
+      "epoch": 0.04512372634643377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998803490092718e-05,
+      "loss": 1.1608,
       "step": 1643
     },
     {
-      "epoch": 0.5372549019607843,
-      "grad_norm": 1.3379188776016235,
-      "learning_rate": 9.280849426222778e-06,
-      "loss": 0.9183,
+      "epoch": 0.045151190574277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998799136064012e-05,
+      "loss": 1.2289,
       "step": 1644
     },
     {
-      "epoch": 0.5375816993464052,
-      "grad_norm": 1.311373233795166,
-      "learning_rate": 9.270292356948071e-06,
-      "loss": 0.922,
+      "epoch": 0.04517865480212024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998794774132421e-05,
+      "loss": 1.0323,
       "step": 1645
     },
     {
-      "epoch": 0.5379084967320261,
-      "grad_norm": 1.3128135204315186,
-      "learning_rate": 9.259736105235572e-06,
-      "loss": 0.9645,
+      "epoch": 0.04520611902996347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998790404297978e-05,
+      "loss": 1.1471,
       "step": 1646
     },
     {
-      "epoch": 0.538235294117647,
-      "grad_norm": 1.361523985862732,
-      "learning_rate": 9.249180682912472e-06,
-      "loss": 0.9482,
+      "epoch": 0.04523358325780671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987860265607187e-05,
+      "loss": 1.1214,
       "step": 1647
     },
     {
-      "epoch": 0.538562091503268,
-      "grad_norm": 1.300734519958496,
-      "learning_rate": 9.23862610180503e-06,
-      "loss": 0.9318,
+      "epoch": 0.04526104748564994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987816409206776e-05,
+      "loss": 1.2086,
       "step": 1648
     },
     {
-      "epoch": 0.5388888888888889,
-      "grad_norm": 1.2874706983566284,
-      "learning_rate": 9.22807237373857e-06,
-      "loss": 0.9292,
+      "epoch": 0.04528851171349318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998777247377889e-05,
+      "loss": 1.1169,
       "step": 1649
     },
     {
-      "epoch": 0.5392156862745098,
-      "grad_norm": 1.2916414737701416,
-      "learning_rate": 9.217519510537455e-06,
-      "loss": 0.9607,
+      "epoch": 0.045315975941336406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987728459323882e-05,
+      "loss": 1.1726,
       "step": 1650
     },
     {
-      "epoch": 0.5395424836601307,
-      "grad_norm": 1.3067023754119873,
-      "learning_rate": 9.206967524025083e-06,
-      "loss": 0.95,
+      "epoch": 0.04534344016917964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987684365842096e-05,
+      "loss": 1.0911,
       "step": 1651
     },
     {
-      "epoch": 0.5398692810457516,
-      "grad_norm": 1.3494466543197632,
-      "learning_rate": 9.196416426023868e-06,
-      "loss": 0.9292,
+      "epoch": 0.045370904397022876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987640193333882e-05,
+      "loss": 1.1061,
       "step": 1652
     },
     {
-      "epoch": 0.5401960784313725,
-      "grad_norm": 1.323143720626831,
-      "learning_rate": 9.18586622835522e-06,
-      "loss": 0.9871,
+      "epoch": 0.04539836862486611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987595941799594e-05,
+      "loss": 1.0684,
       "step": 1653
     },
     {
-      "epoch": 0.5405228758169934,
-      "grad_norm": 1.2681559324264526,
-      "learning_rate": 9.175316942839557e-06,
-      "loss": 0.9633,
+      "epoch": 0.04542583285270935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987551611239575e-05,
+      "loss": 1.2385,
       "step": 1654
     },
     {
-      "epoch": 0.5408496732026143,
-      "grad_norm": 1.354444980621338,
-      "learning_rate": 9.16476858129626e-06,
-      "loss": 0.9498,
+      "epoch": 0.04545329708055258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987507201654178e-05,
+      "loss": 1.2244,
       "step": 1655
     },
     {
-      "epoch": 0.5411764705882353,
-      "grad_norm": 1.255180835723877,
-      "learning_rate": 9.154221155543684e-06,
-      "loss": 0.9555,
+      "epoch": 0.04548076130839582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987462713043757e-05,
+      "loss": 1.1581,
       "step": 1656
     },
     {
-      "epoch": 0.5415032679738562,
-      "grad_norm": 1.3421871662139893,
-      "learning_rate": 9.143674677399132e-06,
-      "loss": 0.9602,
+      "epoch": 0.045508225536239046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987418145408662e-05,
+      "loss": 1.0394,
       "step": 1657
     },
     {
-      "epoch": 0.5418300653594771,
-      "grad_norm": 1.3424783945083618,
-      "learning_rate": 9.133129158678848e-06,
-      "loss": 0.9575,
+      "epoch": 0.04553568976408228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987373498749246e-05,
+      "loss": 1.1108,
       "step": 1658
     },
     {
-      "epoch": 0.542156862745098,
-      "grad_norm": 1.2601569890975952,
-      "learning_rate": 9.122584611197995e-06,
-      "loss": 0.964,
+      "epoch": 0.045563153991925516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998732877306586e-05,
+      "loss": 1.1799,
       "step": 1659
     },
     {
-      "epoch": 0.5424836601307189,
-      "grad_norm": 1.2274696826934814,
-      "learning_rate": 9.112041046770653e-06,
-      "loss": 0.9667,
+      "epoch": 0.04559061821976875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987283968358864e-05,
+      "loss": 1.1287,
       "step": 1660
     },
     {
-      "epoch": 0.5428104575163398,
-      "grad_norm": 1.354696273803711,
-      "learning_rate": 9.1014984772098e-06,
-      "loss": 0.9513,
+      "epoch": 0.045618082447611986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987239084628607e-05,
+      "loss": 1.0519,
       "step": 1661
     },
     {
-      "epoch": 0.5431372549019607,
-      "grad_norm": 1.3198022842407227,
-      "learning_rate": 9.090956914327297e-06,
-      "loss": 0.9278,
+      "epoch": 0.04564554667545522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998719412187544e-05,
+      "loss": 1.1774,
       "step": 1662
     },
     {
-      "epoch": 0.5434640522875817,
-      "grad_norm": 1.2905367612838745,
-      "learning_rate": 9.080416369933882e-06,
-      "loss": 0.9436,
+      "epoch": 0.04567301090329846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987149080099733e-05,
+      "loss": 1.005,
       "step": 1663
     },
     {
-      "epoch": 0.5437908496732026,
-      "grad_norm": 1.2591315507888794,
-      "learning_rate": 9.069876855839148e-06,
-      "loss": 0.9378,
+      "epoch": 0.045700475131141685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998710395930183e-05,
+      "loss": 1.1598,
       "step": 1664
     },
     {
-      "epoch": 0.5441176470588235,
-      "grad_norm": 1.3195788860321045,
-      "learning_rate": 9.05933838385153e-06,
-      "loss": 0.9229,
+      "epoch": 0.04572793935898492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987058759482093e-05,
+      "loss": 1.2667,
       "step": 1665
     },
     {
-      "epoch": 0.5444444444444444,
-      "grad_norm": 1.1943581104278564,
-      "learning_rate": 9.048800965778304e-06,
-      "loss": 0.9705,
+      "epoch": 0.045755403586828156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987013480640876e-05,
+      "loss": 1.0586,
       "step": 1666
     },
     {
-      "epoch": 0.5447712418300653,
-      "grad_norm": 1.34080171585083,
-      "learning_rate": 9.038264613425556e-06,
-      "loss": 0.9357,
+      "epoch": 0.04578286781467139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998696812277854e-05,
+      "loss": 1.1747,
       "step": 1667
     },
     {
-      "epoch": 0.5450980392156862,
-      "grad_norm": 1.255176305770874,
-      "learning_rate": 9.027729338598188e-06,
-      "loss": 0.9089,
+      "epoch": 0.045810332042514626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986922685895446e-05,
+      "loss": 1.0144,
       "step": 1668
     },
     {
-      "epoch": 0.5454248366013071,
-      "grad_norm": 1.4001544713974,
-      "learning_rate": 9.017195153099891e-06,
-      "loss": 0.9328,
+      "epoch": 0.04583779627035786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998687716999195e-05,
+      "loss": 1.1742,
       "step": 1669
     },
     {
-      "epoch": 0.545751633986928,
-      "grad_norm": 1.3079835176467896,
-      "learning_rate": 9.006662068733128e-06,
-      "loss": 0.9522,
+      "epoch": 0.045865260498201096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986831575068413e-05,
+      "loss": 0.9866,
       "step": 1670
     },
     {
-      "epoch": 0.546078431372549,
-      "grad_norm": 1.293014407157898,
-      "learning_rate": 8.996130097299136e-06,
-      "loss": 0.9443,
+      "epoch": 0.045892724726044325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986785901125195e-05,
+      "loss": 1.0557,
       "step": 1671
     },
     {
-      "epoch": 0.5464052287581699,
-      "grad_norm": 1.278877854347229,
-      "learning_rate": 8.985599250597906e-06,
-      "loss": 0.9371,
+      "epoch": 0.04592018895388756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998674014816266e-05,
+      "loss": 1.2347,
       "step": 1672
     },
     {
-      "epoch": 0.5467320261437908,
-      "grad_norm": 1.318183183670044,
-      "learning_rate": 8.975069540428166e-06,
-      "loss": 0.9253,
+      "epoch": 0.045947653181730795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998669431618117e-05,
+      "loss": 1.1135,
       "step": 1673
     },
     {
-      "epoch": 0.5470588235294118,
-      "grad_norm": 1.311918020248413,
-      "learning_rate": 8.96454097858737e-06,
-      "loss": 0.9468,
+      "epoch": 0.04597511740957403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986648405181084e-05,
+      "loss": 1.1283,
       "step": 1674
     },
     {
-      "epoch": 0.5473856209150327,
-      "grad_norm": 1.2638307809829712,
-      "learning_rate": 8.95401357687169e-06,
-      "loss": 0.9558,
+      "epoch": 0.046002581637417266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986602415162765e-05,
+      "loss": 1.0427,
       "step": 1675
     },
     {
-      "epoch": 0.5477124183006536,
-      "grad_norm": 1.3056546449661255,
-      "learning_rate": 8.943487347075988e-06,
-      "loss": 0.9998,
+      "epoch": 0.0460300458652605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986556346126586e-05,
+      "loss": 1.1172,
       "step": 1676
     },
     {
-      "epoch": 0.5480392156862746,
-      "grad_norm": 1.286923885345459,
-      "learning_rate": 8.932962300993824e-06,
-      "loss": 0.9726,
+      "epoch": 0.04605751009310373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986510198072898e-05,
+      "loss": 1.0533,
       "step": 1677
     },
     {
-      "epoch": 0.5483660130718955,
-      "grad_norm": 1.3575732707977295,
-      "learning_rate": 8.922438450417424e-06,
-      "loss": 0.9542,
+      "epoch": 0.046084974320946964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998646397100208e-05,
+      "loss": 1.2004,
       "step": 1678
     },
     {
-      "epoch": 0.5486928104575164,
-      "grad_norm": 1.315389633178711,
-      "learning_rate": 8.911915807137681e-06,
-      "loss": 0.9589,
+      "epoch": 0.0461124385487902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986417664914482e-05,
+      "loss": 1.1533,
       "step": 1679
     },
     {
-      "epoch": 0.5490196078431373,
-      "grad_norm": 1.3121126890182495,
-      "learning_rate": 8.901394382944135e-06,
-      "loss": 0.9634,
+      "epoch": 0.046139902776633435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986371279810483e-05,
+      "loss": 1.0764,
       "step": 1680
     },
     {
-      "epoch": 0.5493464052287582,
-      "grad_norm": 1.2542829513549805,
-      "learning_rate": 8.890874189624951e-06,
-      "loss": 0.9085,
+      "epoch": 0.04616736700447667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998632481569045e-05,
+      "loss": 1.148,
       "step": 1681
     },
     {
-      "epoch": 0.5496732026143791,
-      "grad_norm": 1.260268211364746,
-      "learning_rate": 8.880355238966923e-06,
-      "loss": 0.9333,
+      "epoch": 0.046194831232319905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998627827255474e-05,
+      "loss": 0.9512,
       "step": 1682
     },
     {
-      "epoch": 0.55,
-      "grad_norm": 1.3131664991378784,
-      "learning_rate": 8.869837542755454e-06,
-      "loss": 0.9598,
+      "epoch": 0.04622229546016314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986231650403732e-05,
+      "loss": 1.1142,
       "step": 1683
     },
     {
-      "epoch": 0.550326797385621,
-      "grad_norm": 1.311390995979309,
-      "learning_rate": 8.859321112774535e-06,
-      "loss": 0.9379,
+      "epoch": 0.04624975968800637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998618494923779e-05,
+      "loss": 1.1177,
       "step": 1684
     },
     {
-      "epoch": 0.5506535947712419,
-      "grad_norm": 1.2927840948104858,
-      "learning_rate": 8.848805960806744e-06,
-      "loss": 0.9375,
+      "epoch": 0.046277223915849604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986138169057282e-05,
+      "loss": 1.0459,
       "step": 1685
     },
     {
-      "epoch": 0.5509803921568628,
-      "grad_norm": 1.3297604322433472,
-      "learning_rate": 8.838292098633229e-06,
-      "loss": 0.9293,
+      "epoch": 0.04630468814369284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986091309862586e-05,
+      "loss": 1.117,
       "step": 1686
     },
     {
-      "epoch": 0.5513071895424837,
-      "grad_norm": 1.3215869665145874,
-      "learning_rate": 8.827779538033681e-06,
-      "loss": 0.9564,
+      "epoch": 0.046332152371536074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986044371654064e-05,
+      "loss": 1.0585,
       "step": 1687
     },
     {
-      "epoch": 0.5516339869281046,
-      "grad_norm": 1.316103219985962,
-      "learning_rate": 8.817268290786343e-06,
-      "loss": 0.9203,
+      "epoch": 0.04635961659937931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998599735443209e-05,
+      "loss": 1.1556,
       "step": 1688
     },
     {
-      "epoch": 0.5519607843137255,
-      "grad_norm": 1.385805368423462,
-      "learning_rate": 8.806758368667989e-06,
-      "loss": 0.9741,
+      "epoch": 0.046387080827222545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985950258197036e-05,
+      "loss": 0.9615,
       "step": 1689
     },
     {
-      "epoch": 0.5522875816993464,
-      "grad_norm": 1.2502243518829346,
-      "learning_rate": 8.796249783453897e-06,
-      "loss": 0.9886,
+      "epoch": 0.04641454505506578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985903082949282e-05,
+      "loss": 1.1682,
       "step": 1690
     },
     {
-      "epoch": 0.5526143790849674,
-      "grad_norm": 1.2364052534103394,
-      "learning_rate": 8.785742546917857e-06,
-      "loss": 0.9563,
+      "epoch": 0.04644200928290901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985855828689186e-05,
+      "loss": 1.0351,
       "step": 1691
     },
     {
-      "epoch": 0.5529411764705883,
-      "grad_norm": 1.33072829246521,
-      "learning_rate": 8.775236670832146e-06,
-      "loss": 0.9515,
+      "epoch": 0.04646947351075224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985808495417136e-05,
+      "loss": 1.1722,
       "step": 1692
     },
     {
-      "epoch": 0.5532679738562092,
-      "grad_norm": 1.2890251874923706,
-      "learning_rate": 8.764732166967514e-06,
-      "loss": 0.9587,
+      "epoch": 0.04649693773859548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99857610831335e-05,
+      "loss": 1.0854,
       "step": 1693
     },
     {
-      "epoch": 0.5535947712418301,
-      "grad_norm": 1.2830970287322998,
-      "learning_rate": 8.754229047093173e-06,
-      "loss": 0.9522,
+      "epoch": 0.046524401966438714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998571359183865e-05,
+      "loss": 1.1688,
       "step": 1694
     },
     {
-      "epoch": 0.553921568627451,
-      "grad_norm": 1.2740323543548584,
-      "learning_rate": 8.743727322976788e-06,
-      "loss": 0.9688,
+      "epoch": 0.04655186619428195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998566602153297e-05,
+      "loss": 1.0742,
       "step": 1695
     },
     {
-      "epoch": 0.5542483660130719,
-      "grad_norm": 1.291920781135559,
-      "learning_rate": 8.733227006384459e-06,
-      "loss": 0.9316,
+      "epoch": 0.046579330422125184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985618372216834e-05,
+      "loss": 1.123,
       "step": 1696
     },
     {
-      "epoch": 0.5545751633986928,
-      "grad_norm": 1.2420905828475952,
-      "learning_rate": 8.72272810908071e-06,
-      "loss": 0.9395,
+      "epoch": 0.04660679464996842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985570643890613e-05,
+      "loss": 1.1326,
       "step": 1697
     },
     {
-      "epoch": 0.5549019607843138,
-      "grad_norm": 1.2543922662734985,
-      "learning_rate": 8.71223064282847e-06,
-      "loss": 0.9083,
+      "epoch": 0.04663425887781165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985522836554693e-05,
+      "loss": 0.9834,
       "step": 1698
     },
     {
-      "epoch": 0.5552287581699347,
-      "grad_norm": 1.302888035774231,
-      "learning_rate": 8.701734619389068e-06,
-      "loss": 0.9311,
+      "epoch": 0.04666172310565488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985474950209447e-05,
+      "loss": 1.1126,
       "step": 1699
     },
     {
-      "epoch": 0.5555555555555556,
-      "grad_norm": 1.325583577156067,
-      "learning_rate": 8.691240050522215e-06,
-      "loss": 0.9591,
+      "epoch": 0.04668918733349812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985426984855255e-05,
+      "loss": 1.0847,
       "step": 1700
     },
     {
-      "epoch": 0.5558823529411765,
-      "grad_norm": 1.2668852806091309,
-      "learning_rate": 8.680746947986e-06,
-      "loss": 0.9165,
+      "epoch": 0.04671665156134135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985378940492496e-05,
+      "loss": 1.1101,
       "step": 1701
     },
     {
-      "epoch": 0.5562091503267974,
-      "grad_norm": 1.3429179191589355,
-      "learning_rate": 8.670255323536858e-06,
-      "loss": 0.9711,
+      "epoch": 0.04674411578918459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998533081712155e-05,
+      "loss": 1.155,
       "step": 1702
     },
     {
-      "epoch": 0.5565359477124183,
-      "grad_norm": 1.3458527326583862,
-      "learning_rate": 8.659765188929578e-06,
-      "loss": 0.9568,
+      "epoch": 0.046771580017027824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985282614742802e-05,
+      "loss": 1.145,
       "step": 1703
     },
     {
-      "epoch": 0.5568627450980392,
-      "grad_norm": 1.2896274328231812,
-      "learning_rate": 8.649276555917267e-06,
-      "loss": 0.934,
+      "epoch": 0.04679904424487106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985234333356626e-05,
+      "loss": 1.0054,
       "step": 1704
     },
     {
-      "epoch": 0.5571895424836601,
-      "grad_norm": 1.3474249839782715,
-      "learning_rate": 8.638789436251363e-06,
-      "loss": 0.9691,
+      "epoch": 0.04682650847271429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998518597296341e-05,
+      "loss": 1.0073,
       "step": 1705
     },
     {
-      "epoch": 0.5575163398692811,
-      "grad_norm": 1.4122213125228882,
-      "learning_rate": 8.6283038416816e-06,
-      "loss": 0.9224,
+      "epoch": 0.04685397270055752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985137533563534e-05,
+      "loss": 1.1435,
       "step": 1706
     },
     {
-      "epoch": 0.557843137254902,
-      "grad_norm": 1.3908412456512451,
-      "learning_rate": 8.617819783956006e-06,
-      "loss": 0.9468,
+      "epoch": 0.04688143692840076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985089015157383e-05,
+      "loss": 1.0096,
       "step": 1707
     },
     {
-      "epoch": 0.5581699346405229,
-      "grad_norm": 1.3300267457962036,
-      "learning_rate": 8.607337274820888e-06,
-      "loss": 0.9493,
+      "epoch": 0.04690890115624399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998504041774534e-05,
+      "loss": 1.1222,
       "step": 1708
     },
     {
-      "epoch": 0.5584967320261438,
-      "grad_norm": 1.3099459409713745,
-      "learning_rate": 8.596856326020822e-06,
-      "loss": 0.9313,
+      "epoch": 0.04693636538408723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984991741327787e-05,
+      "loss": 1.1042,
       "step": 1709
     },
     {
-      "epoch": 0.5588235294117647,
-      "grad_norm": 1.441552996635437,
-      "learning_rate": 8.586376949298622e-06,
-      "loss": 0.9646,
+      "epoch": 0.04696382961193046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984942985905114e-05,
+      "loss": 1.0204,
       "step": 1710
     },
     {
-      "epoch": 0.5591503267973856,
-      "grad_norm": 1.2971066236495972,
-      "learning_rate": 8.575899156395353e-06,
-      "loss": 0.9479,
+      "epoch": 0.04699129383977369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984894151477707e-05,
+      "loss": 1.1312,
       "step": 1711
     },
     {
-      "epoch": 0.5594771241830065,
-      "grad_norm": 1.3888888359069824,
-      "learning_rate": 8.565422959050303e-06,
-      "loss": 0.9408,
+      "epoch": 0.04701875806761693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984845238045944e-05,
+      "loss": 1.0644,
       "step": 1712
     },
     {
-      "epoch": 0.5598039215686275,
-      "grad_norm": 1.298083782196045,
-      "learning_rate": 8.554948369000972e-06,
-      "loss": 0.9672,
+      "epoch": 0.04704622229546016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998479624561022e-05,
+      "loss": 1.1714,
       "step": 1713
     },
     {
-      "epoch": 0.5601307189542484,
-      "grad_norm": 1.2731727361679077,
-      "learning_rate": 8.54447539798306e-06,
-      "loss": 0.9223,
+      "epoch": 0.0470736865233034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998474717417092e-05,
+      "loss": 1.1518,
       "step": 1714
     },
     {
-      "epoch": 0.5604575163398693,
-      "grad_norm": 1.2675257921218872,
-      "learning_rate": 8.534004057730445e-06,
-      "loss": 0.9229,
+      "epoch": 0.04710115075114663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984698023728435e-05,
+      "loss": 1.2137,
       "step": 1715
     },
     {
-      "epoch": 0.5607843137254902,
-      "grad_norm": 1.2248355150222778,
-      "learning_rate": 8.52353435997519e-06,
-      "loss": 0.9473,
+      "epoch": 0.04712861497898987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998464879428315e-05,
+      "loss": 1.1224,
       "step": 1716
     },
     {
-      "epoch": 0.5611111111111111,
-      "grad_norm": 1.2996864318847656,
-      "learning_rate": 8.51306631644751e-06,
-      "loss": 0.9448,
+      "epoch": 0.0471560792068331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984599485835455e-05,
+      "loss": 1.1053,
       "step": 1717
     },
     {
-      "epoch": 0.561437908496732,
-      "grad_norm": 1.456566572189331,
-      "learning_rate": 8.50259993887577e-06,
-      "loss": 0.9504,
+      "epoch": 0.04718354343467633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998455009838574e-05,
+      "loss": 1.0828,
       "step": 1718
     },
     {
-      "epoch": 0.5617647058823529,
-      "grad_norm": 1.2593883275985718,
-      "learning_rate": 8.492135238986467e-06,
-      "loss": 0.9047,
+      "epoch": 0.047211007662519566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99845006319344e-05,
+      "loss": 1.09,
       "step": 1719
     },
     {
-      "epoch": 0.5620915032679739,
-      "grad_norm": 1.292035698890686,
-      "learning_rate": 8.481672228504224e-06,
-      "loss": 0.9841,
+      "epoch": 0.0472384718903628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998445108648182e-05,
+      "loss": 1.0825,
       "step": 1720
     },
     {
-      "epoch": 0.5624183006535948,
-      "grad_norm": 1.308863639831543,
-      "learning_rate": 8.471210919151757e-06,
-      "loss": 0.9473,
+      "epoch": 0.04726593611820604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99844014620284e-05,
+      "loss": 1.0556,
       "step": 1721
     },
     {
-      "epoch": 0.5627450980392157,
-      "grad_norm": 1.226660966873169,
-      "learning_rate": 8.460751322649887e-06,
-      "loss": 0.9206,
+      "epoch": 0.04729340034604927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984351758574526e-05,
+      "loss": 1.1493,
       "step": 1722
     },
     {
-      "epoch": 0.5630718954248366,
-      "grad_norm": 1.3895764350891113,
-      "learning_rate": 8.450293450717517e-06,
-      "loss": 0.9885,
+      "epoch": 0.04732086457389251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984301976120593e-05,
+      "loss": 1.0899,
       "step": 1723
     },
     {
-      "epoch": 0.5633986928104575,
-      "grad_norm": 1.3398152589797974,
-      "learning_rate": 8.439837315071612e-06,
-      "loss": 0.9314,
+      "epoch": 0.04734832880173574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984252114666997e-05,
+      "loss": 1.1938,
       "step": 1724
     },
     {
-      "epoch": 0.5637254901960784,
-      "grad_norm": 1.2985576391220093,
-      "learning_rate": 8.429382927427199e-06,
-      "loss": 0.9543,
+      "epoch": 0.04737579302957897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984202174214128e-05,
+      "loss": 1.1201,
       "step": 1725
     },
     {
-      "epoch": 0.5640522875816993,
-      "grad_norm": 1.2682744264602661,
-      "learning_rate": 8.418930299497335e-06,
-      "loss": 0.9742,
+      "epoch": 0.047403257257422206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984152154762385e-05,
+      "loss": 1.0835,
       "step": 1726
     },
     {
-      "epoch": 0.5643790849673203,
-      "grad_norm": 1.2885727882385254,
-      "learning_rate": 8.408479442993115e-06,
-      "loss": 0.9432,
+      "epoch": 0.04743072148526544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998410205631216e-05,
+      "loss": 1.1157,
       "step": 1727
     },
     {
-      "epoch": 0.5647058823529412,
-      "grad_norm": 1.3005753755569458,
-      "learning_rate": 8.39803036962365e-06,
-      "loss": 0.9352,
+      "epoch": 0.047458185713108676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984051878863856e-05,
+      "loss": 1.0888,
       "step": 1728
     },
     {
-      "epoch": 0.5650326797385621,
-      "grad_norm": 1.3040106296539307,
-      "learning_rate": 8.387583091096045e-06,
-      "loss": 0.9361,
+      "epoch": 0.04748564994095191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984001622417865e-05,
+      "loss": 1.1666,
       "step": 1729
     },
     {
-      "epoch": 0.565359477124183,
-      "grad_norm": 1.249686360359192,
-      "learning_rate": 8.377137619115401e-06,
-      "loss": 0.9124,
+      "epoch": 0.04751311416879515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983951286974586e-05,
+      "loss": 1.1549,
       "step": 1730
     },
     {
-      "epoch": 0.5656862745098039,
-      "grad_norm": 1.3115404844284058,
-      "learning_rate": 8.366693965384795e-06,
-      "loss": 0.9547,
+      "epoch": 0.04754057839663838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983900872534416e-05,
+      "loss": 0.9919,
       "step": 1731
     },
     {
-      "epoch": 0.5660130718954248,
-      "grad_norm": 1.2941303253173828,
-      "learning_rate": 8.35625214160526e-06,
-      "loss": 0.9258,
+      "epoch": 0.04756804262448161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983850379097754e-05,
+      "loss": 1.2313,
       "step": 1732
     },
     {
-      "epoch": 0.5663398692810457,
-      "grad_norm": 1.2151806354522705,
-      "learning_rate": 8.345812159475786e-06,
-      "loss": 0.9248,
+      "epoch": 0.047595506852324845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983799806665e-05,
+      "loss": 1.0303,
       "step": 1733
     },
     {
-      "epoch": 0.5666666666666667,
-      "grad_norm": 1.3290979862213135,
-      "learning_rate": 8.335374030693295e-06,
-      "loss": 0.9805,
+      "epoch": 0.04762297108016808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998374915523656e-05,
+      "loss": 1.1075,
       "step": 1734
     },
     {
-      "epoch": 0.5669934640522876,
-      "grad_norm": 1.3190641403198242,
-      "learning_rate": 8.324937766952638e-06,
-      "loss": 0.9638,
+      "epoch": 0.047650435308011316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983698424812825e-05,
+      "loss": 1.1476,
       "step": 1735
     },
     {
-      "epoch": 0.5673202614379085,
-      "grad_norm": 1.3452814817428589,
-      "learning_rate": 8.314503379946569e-06,
-      "loss": 0.9717,
+      "epoch": 0.04767789953585455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99836476153942e-05,
+      "loss": 1.1154,
       "step": 1736
     },
     {
-      "epoch": 0.5676470588235294,
-      "grad_norm": 1.2460103034973145,
-      "learning_rate": 8.304070881365748e-06,
-      "loss": 0.8927,
+      "epoch": 0.047705363763697786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998359672698109e-05,
+      "loss": 1.0756,
       "step": 1737
     },
     {
-      "epoch": 0.5679738562091503,
-      "grad_norm": 1.2788786888122559,
-      "learning_rate": 8.293640282898709e-06,
-      "loss": 0.9312,
+      "epoch": 0.047732827991541014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983545759573894e-05,
+      "loss": 1.0745,
       "step": 1738
     },
     {
-      "epoch": 0.5683006535947712,
-      "grad_norm": 1.371330738067627,
-      "learning_rate": 8.283211596231864e-06,
-      "loss": 0.9244,
+      "epoch": 0.04776029221938425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983494713173017e-05,
+      "loss": 1.0967,
       "step": 1739
     },
     {
-      "epoch": 0.5686274509803921,
-      "grad_norm": 1.2956725358963013,
-      "learning_rate": 8.272784833049485e-06,
-      "loss": 0.9232,
+      "epoch": 0.047787756447227485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983443587778863e-05,
+      "loss": 1.1851,
       "step": 1740
     },
     {
-      "epoch": 0.568954248366013,
-      "grad_norm": 1.2829607725143433,
-      "learning_rate": 8.262360005033681e-06,
-      "loss": 0.9581,
+      "epoch": 0.04781522067507072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998339238339184e-05,
+      "loss": 1.0267,
       "step": 1741
     },
     {
-      "epoch": 0.569281045751634,
-      "grad_norm": 1.4109143018722534,
-      "learning_rate": 8.251937123864405e-06,
-      "loss": 0.9568,
+      "epoch": 0.047842684902913955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983341100012342e-05,
+      "loss": 1.0895,
       "step": 1742
     },
     {
-      "epoch": 0.5696078431372549,
-      "grad_norm": 1.3069216012954712,
-      "learning_rate": 8.24151620121941e-06,
-      "loss": 0.9688,
+      "epoch": 0.04787014913075719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983289737640786e-05,
+      "loss": 1.0855,
       "step": 1743
     },
     {
-      "epoch": 0.5699346405228758,
-      "grad_norm": 1.2894386053085327,
-      "learning_rate": 8.231097248774273e-06,
-      "loss": 0.9715,
+      "epoch": 0.047897613358600426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983238296277576e-05,
+      "loss": 0.9915,
       "step": 1744
     },
     {
-      "epoch": 0.5702614379084967,
-      "grad_norm": 1.311722755432129,
-      "learning_rate": 8.220680278202355e-06,
-      "loss": 0.9462,
+      "epoch": 0.047925077586443654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983186775923112e-05,
+      "loss": 1.1456,
       "step": 1745
     },
     {
-      "epoch": 0.5705882352941176,
-      "grad_norm": 1.234694004058838,
-      "learning_rate": 8.210265301174796e-06,
-      "loss": 0.9779,
+      "epoch": 0.04795254181428689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983135176577812e-05,
+      "loss": 1.0811,
       "step": 1746
     },
     {
-      "epoch": 0.5709150326797385,
-      "grad_norm": 1.2654495239257812,
-      "learning_rate": 8.199852329360506e-06,
-      "loss": 0.9434,
+      "epoch": 0.047980006042130124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983083498242078e-05,
+      "loss": 1.0261,
       "step": 1747
     },
     {
-      "epoch": 0.5712418300653594,
-      "grad_norm": 1.2887895107269287,
-      "learning_rate": 8.18944137442615e-06,
-      "loss": 0.9498,
+      "epoch": 0.04800747026997336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983031740916317e-05,
+      "loss": 1.1087,
       "step": 1748
     },
     {
-      "epoch": 0.5715686274509804,
-      "grad_norm": 1.3159551620483398,
-      "learning_rate": 8.179032448036122e-06,
-      "loss": 0.9389,
+      "epoch": 0.048034934497816595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982979904600946e-05,
+      "loss": 1.0824,
       "step": 1749
     },
     {
-      "epoch": 0.5718954248366013,
-      "grad_norm": 1.3015402555465698,
-      "learning_rate": 8.16862556185256e-06,
-      "loss": 0.9587,
+      "epoch": 0.04806239872565983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998292798929637e-05,
+      "loss": 1.1148,
       "step": 1750
     },
     {
-      "epoch": 0.5722222222222222,
-      "grad_norm": 1.263421654701233,
-      "learning_rate": 8.158220727535298e-06,
-      "loss": 0.9396,
+      "epoch": 0.048089862953503065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982875995003003e-05,
+      "loss": 1.1648,
       "step": 1751
     },
     {
-      "epoch": 0.5725490196078431,
-      "grad_norm": 1.2772361040115356,
-      "learning_rate": 8.14781795674189e-06,
-      "loss": 0.9524,
+      "epoch": 0.048117327181346294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982823921721247e-05,
+      "loss": 1.1505,
       "step": 1752
     },
     {
-      "epoch": 0.572875816993464,
-      "grad_norm": 1.3076915740966797,
-      "learning_rate": 8.137417261127562e-06,
-      "loss": 0.9738,
+      "epoch": 0.04814479140918953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982771769451528e-05,
+      "loss": 0.9576,
       "step": 1753
     },
     {
-      "epoch": 0.5732026143790849,
-      "grad_norm": 1.3718947172164917,
-      "learning_rate": 8.12701865234523e-06,
-      "loss": 0.9564,
+      "epoch": 0.048172255637032764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982719538194248e-05,
+      "loss": 1.1726,
       "step": 1754
     },
     {
-      "epoch": 0.5735294117647058,
-      "grad_norm": 1.4951456785202026,
-      "learning_rate": 8.11662214204545e-06,
-      "loss": 0.9523,
+      "epoch": 0.048199719864876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982667227949828e-05,
+      "loss": 1.0727,
       "step": 1755
     },
     {
-      "epoch": 0.5738562091503268,
-      "grad_norm": 1.2847355604171753,
-      "learning_rate": 8.106227741876447e-06,
-      "loss": 0.9268,
+      "epoch": 0.048227184092719234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982614838718674e-05,
+      "loss": 1.1664,
       "step": 1756
     },
     {
-      "epoch": 0.5741830065359477,
-      "grad_norm": 1.4003204107284546,
-      "learning_rate": 8.095835463484076e-06,
-      "loss": 0.9336,
+      "epoch": 0.04825464832056247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982562370501207e-05,
+      "loss": 0.9953,
       "step": 1757
     },
     {
-      "epoch": 0.5745098039215686,
-      "grad_norm": 1.3730101585388184,
-      "learning_rate": 8.085445318511814e-06,
-      "loss": 0.9507,
+      "epoch": 0.048282112548405705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998250982329784e-05,
+      "loss": 1.2021,
       "step": 1758
     },
     {
-      "epoch": 0.5748366013071895,
-      "grad_norm": 1.3485863208770752,
-      "learning_rate": 8.075057318600751e-06,
-      "loss": 0.9939,
+      "epoch": 0.04830957677624893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982457197108985e-05,
+      "loss": 1.0302,
       "step": 1759
     },
     {
-      "epoch": 0.5751633986928104,
-      "grad_norm": 1.3372048139572144,
-      "learning_rate": 8.064671475389562e-06,
-      "loss": 0.9371,
+      "epoch": 0.04833704100409217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982404491935067e-05,
+      "loss": 1.0033,
       "step": 1760
     },
     {
-      "epoch": 0.5754901960784313,
-      "grad_norm": 1.2840646505355835,
-      "learning_rate": 8.054287800514517e-06,
-      "loss": 0.9218,
+      "epoch": 0.048364505231935404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982351707776492e-05,
+      "loss": 1.0586,
       "step": 1761
     },
     {
-      "epoch": 0.5758169934640522,
-      "grad_norm": 1.2206528186798096,
-      "learning_rate": 8.043906305609454e-06,
-      "loss": 0.9685,
+      "epoch": 0.04839196945977864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982298844633687e-05,
+      "loss": 1.0868,
       "step": 1762
     },
     {
-      "epoch": 0.5761437908496732,
-      "grad_norm": 1.3098160028457642,
-      "learning_rate": 8.03352700230577e-06,
-      "loss": 0.9443,
+      "epoch": 0.048419433687621874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982245902507066e-05,
+      "loss": 1.0891,
       "step": 1763
     },
     {
-      "epoch": 0.5764705882352941,
-      "grad_norm": 1.2614731788635254,
-      "learning_rate": 8.023149902232404e-06,
-      "loss": 0.9348,
+      "epoch": 0.04844689791546511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982192881397048e-05,
+      "loss": 1.1758,
       "step": 1764
     },
     {
-      "epoch": 0.576797385620915,
-      "grad_norm": 1.385893702507019,
-      "learning_rate": 8.012775017015829e-06,
-      "loss": 0.982,
+      "epoch": 0.04847436214330834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982139781304054e-05,
+      "loss": 1.0431,
       "step": 1765
     },
     {
-      "epoch": 0.5771241830065359,
-      "grad_norm": 1.317423701286316,
-      "learning_rate": 8.002402358280028e-06,
-      "loss": 0.9465,
+      "epoch": 0.04850182637115157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982086602228504e-05,
+      "loss": 1.1748,
       "step": 1766
     },
     {
-      "epoch": 0.5774509803921568,
-      "grad_norm": 1.3293441534042358,
-      "learning_rate": 7.992031937646496e-06,
-      "loss": 0.8953,
+      "epoch": 0.04852929059899481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982033344170818e-05,
+      "loss": 1.1643,
       "step": 1767
     },
     {
-      "epoch": 0.5777777777777777,
-      "grad_norm": 1.2714464664459229,
-      "learning_rate": 7.981663766734228e-06,
-      "loss": 0.9459,
+      "epoch": 0.04855675482683804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981980007131416e-05,
+      "loss": 1.1262,
       "step": 1768
     },
     {
-      "epoch": 0.5781045751633986,
-      "grad_norm": 1.269906997680664,
-      "learning_rate": 7.971297857159686e-06,
-      "loss": 0.9366,
+      "epoch": 0.04858421905468128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981926591110725e-05,
+      "loss": 1.0764,
       "step": 1769
     },
     {
-      "epoch": 0.5784313725490197,
-      "grad_norm": 1.2781094312667847,
-      "learning_rate": 7.960934220536812e-06,
-      "loss": 0.9362,
+      "epoch": 0.048611683282524513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981873096109158e-05,
+      "loss": 1.1094,
       "step": 1770
     },
     {
-      "epoch": 0.5787581699346406,
-      "grad_norm": 1.3664278984069824,
-      "learning_rate": 7.950572868476979e-06,
-      "loss": 0.9596,
+      "epoch": 0.04863914751036775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998181952212715e-05,
+      "loss": 1.1259,
       "step": 1771
     },
     {
-      "epoch": 0.5790849673202615,
-      "grad_norm": 1.3123564720153809,
-      "learning_rate": 7.940213812589018e-06,
-      "loss": 0.9641,
+      "epoch": 0.04866661173821098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998176586916512e-05,
+      "loss": 1.1649,
       "step": 1772
     },
     {
-      "epoch": 0.5794117647058824,
-      "grad_norm": 1.2991548776626587,
-      "learning_rate": 7.929857064479188e-06,
-      "loss": 0.9469,
+      "epoch": 0.04869407596605421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981712137223486e-05,
+      "loss": 1.0959,
       "step": 1773
     },
     {
-      "epoch": 0.5797385620915033,
-      "grad_norm": 1.3414918184280396,
-      "learning_rate": 7.919502635751154e-06,
-      "loss": 0.9813,
+      "epoch": 0.04872154019389745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981658326302682e-05,
+      "loss": 1.1734,
       "step": 1774
     },
     {
-      "epoch": 0.5800653594771242,
-      "grad_norm": 1.287213683128357,
-      "learning_rate": 7.909150538005989e-06,
-      "loss": 0.913,
+      "epoch": 0.04874900442174068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981604436403134e-05,
+      "loss": 1.2246,
       "step": 1775
     },
     {
-      "epoch": 0.5803921568627451,
-      "grad_norm": 1.2575128078460693,
-      "learning_rate": 7.898800782842153e-06,
-      "loss": 0.8914,
+      "epoch": 0.04877646864958392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998155046752526e-05,
+      "loss": 1.134,
       "step": 1776
     },
     {
-      "epoch": 0.5807189542483661,
-      "grad_norm": 1.2723140716552734,
-      "learning_rate": 7.888453381855474e-06,
-      "loss": 0.9731,
+      "epoch": 0.04880393287742715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981496419669495e-05,
+      "loss": 1.21,
       "step": 1777
     },
     {
-      "epoch": 0.581045751633987,
-      "grad_norm": 1.3260555267333984,
-      "learning_rate": 7.878108346639152e-06,
-      "loss": 0.9288,
+      "epoch": 0.04883139710527039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981442292836264e-05,
+      "loss": 1.1331,
       "step": 1778
     },
     {
-      "epoch": 0.5813725490196079,
-      "grad_norm": 1.2974891662597656,
-      "learning_rate": 7.867765688783732e-06,
-      "loss": 0.9397,
+      "epoch": 0.048858861333113617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981388087025993e-05,
+      "loss": 1.186,
       "step": 1779
     },
     {
-      "epoch": 0.5816993464052288,
-      "grad_norm": 1.2711511850357056,
-      "learning_rate": 7.857425419877097e-06,
-      "loss": 0.9872,
+      "epoch": 0.04888632556095685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981333802239116e-05,
+      "loss": 1.1276,
       "step": 1780
     },
     {
-      "epoch": 0.5820261437908497,
-      "grad_norm": 1.2716833353042603,
-      "learning_rate": 7.847087551504452e-06,
-      "loss": 0.9646,
+      "epoch": 0.04891378978880009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998127943847606e-05,
+      "loss": 0.9937,
       "step": 1781
     },
     {
-      "epoch": 0.5823529411764706,
-      "grad_norm": 1.2697283029556274,
-      "learning_rate": 7.836752095248318e-06,
-      "loss": 0.9536,
+      "epoch": 0.04894125401664332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981224995737252e-05,
+      "loss": 1.0339,
       "step": 1782
     },
     {
-      "epoch": 0.5826797385620915,
-      "grad_norm": 1.3035037517547607,
-      "learning_rate": 7.826419062688497e-06,
-      "loss": 0.9627,
+      "epoch": 0.04896871824448656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981170474023126e-05,
+      "loss": 1.1602,
       "step": 1783
     },
     {
-      "epoch": 0.5830065359477125,
-      "grad_norm": 1.2804960012435913,
-      "learning_rate": 7.816088465402097e-06,
-      "loss": 0.9659,
+      "epoch": 0.04899618247232979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998111587333411e-05,
+      "loss": 1.0262,
       "step": 1784
     },
     {
-      "epoch": 0.5833333333333334,
-      "grad_norm": 1.2710847854614258,
-      "learning_rate": 7.80576031496348e-06,
-      "loss": 0.9047,
+      "epoch": 0.04902364670017303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981061193670642e-05,
+      "loss": 1.0777,
       "step": 1785
     },
     {
-      "epoch": 0.5836601307189543,
-      "grad_norm": 1.2580628395080566,
-      "learning_rate": 7.795434622944282e-06,
-      "loss": 0.9188,
+      "epoch": 0.049051110928016256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981006435033154e-05,
+      "loss": 1.0043,
       "step": 1786
     },
     {
-      "epoch": 0.5839869281045752,
-      "grad_norm": 1.2942194938659668,
-      "learning_rate": 7.785111400913372e-06,
-      "loss": 0.9375,
+      "epoch": 0.04907857515585949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980951597422075e-05,
+      "loss": 1.093,
       "step": 1787
     },
     {
-      "epoch": 0.5843137254901961,
-      "grad_norm": 1.2111730575561523,
-      "learning_rate": 7.774790660436857e-06,
-      "loss": 0.9407,
+      "epoch": 0.049106039383702726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998089668083784e-05,
+      "loss": 1.072,
       "step": 1788
     },
     {
-      "epoch": 0.584640522875817,
-      "grad_norm": 1.2613211870193481,
-      "learning_rate": 7.764472413078063e-06,
-      "loss": 0.9117,
+      "epoch": 0.04913350361154596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980841685280885e-05,
+      "loss": 1.1698,
       "step": 1789
     },
     {
-      "epoch": 0.5849673202614379,
-      "grad_norm": 1.2823266983032227,
-      "learning_rate": 7.754156670397522e-06,
-      "loss": 0.9697,
+      "epoch": 0.0491609678393892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980786610751645e-05,
+      "loss": 1.1804,
       "step": 1790
     },
     {
-      "epoch": 0.5852941176470589,
-      "grad_norm": 1.4463274478912354,
-      "learning_rate": 7.743843443952962e-06,
-      "loss": 0.9943,
+      "epoch": 0.04918843206723243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998073145725055e-05,
+      "loss": 1.1175,
       "step": 1791
     },
     {
-      "epoch": 0.5856209150326798,
-      "grad_norm": 1.2582892179489136,
-      "learning_rate": 7.73353274529929e-06,
-      "loss": 0.9345,
+      "epoch": 0.04921589629507567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998067622477805e-05,
+      "loss": 1.1038,
       "step": 1792
     },
     {
-      "epoch": 0.5859477124183007,
-      "grad_norm": 1.2807694673538208,
-      "learning_rate": 7.723224585988585e-06,
-      "loss": 0.933,
+      "epoch": 0.049243360522918896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980620913334564e-05,
+      "loss": 1.1707,
       "step": 1793
     },
     {
-      "epoch": 0.5862745098039216,
-      "grad_norm": 1.377860426902771,
-      "learning_rate": 7.712918977570067e-06,
-      "loss": 0.9498,
+      "epoch": 0.04927082475076213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980565522920548e-05,
+      "loss": 1.0996,
       "step": 1794
     },
     {
-      "epoch": 0.5866013071895425,
-      "grad_norm": 1.3099807500839233,
-      "learning_rate": 7.702615931590116e-06,
-      "loss": 0.9399,
+      "epoch": 0.049298288978605366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998051005353643e-05,
+      "loss": 1.0688,
       "step": 1795
     },
     {
-      "epoch": 0.5869281045751634,
-      "grad_norm": 1.2924737930297852,
-      "learning_rate": 7.69231545959223e-06,
-      "loss": 0.9271,
+      "epoch": 0.0493257532064486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980454505182644e-05,
+      "loss": 1.1835,
       "step": 1796
     },
     {
-      "epoch": 0.5872549019607843,
-      "grad_norm": 1.3693186044692993,
-      "learning_rate": 7.682017573117027e-06,
-      "loss": 0.8828,
+      "epoch": 0.049353217434291836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998039887785964e-05,
+      "loss": 1.0141,
       "step": 1797
     },
     {
-      "epoch": 0.5875816993464053,
-      "grad_norm": 1.252362608909607,
-      "learning_rate": 7.671722283702227e-06,
-      "loss": 0.9317,
+      "epoch": 0.04938068166213507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980343171567855e-05,
+      "loss": 1.0136,
       "step": 1798
     },
     {
-      "epoch": 0.5879084967320262,
-      "grad_norm": 1.2799091339111328,
-      "learning_rate": 7.661429602882646e-06,
-      "loss": 0.9063,
+      "epoch": 0.0494081458899783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998028738630773e-05,
+      "loss": 1.23,
       "step": 1799
     },
     {
-      "epoch": 0.5882352941176471,
-      "grad_norm": 1.299665927886963,
-      "learning_rate": 7.651139542190164e-06,
-      "loss": 0.9611,
+      "epoch": 0.049435610117821535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99802315220797e-05,
+      "loss": 1.1166,
       "step": 1800
     },
     {
-      "epoch": 0.588562091503268,
-      "grad_norm": 1.2916074991226196,
-      "learning_rate": 7.640852113153738e-06,
-      "loss": 0.9513,
+      "epoch": 0.04946307434566477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980175578884218e-05,
+      "loss": 1.1113,
       "step": 1801
     },
     {
-      "epoch": 0.5888888888888889,
-      "grad_norm": 1.2568702697753906,
-      "learning_rate": 7.63056732729937e-06,
-      "loss": 0.9146,
+      "epoch": 0.049490538573508006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998011955672172e-05,
+      "loss": 1.0511,
       "step": 1802
     },
     {
-      "epoch": 0.5892156862745098,
-      "grad_norm": 1.2732142210006714,
-      "learning_rate": 7.6202851961501025e-06,
-      "loss": 0.9351,
+      "epoch": 0.04951800280135124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980063455592647e-05,
+      "loss": 1.1025,
       "step": 1803
     },
     {
-      "epoch": 0.5895424836601307,
-      "grad_norm": 1.3110524415969849,
-      "learning_rate": 7.610005731226009e-06,
-      "loss": 0.9342,
+      "epoch": 0.049545467029194476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980007275497443e-05,
+      "loss": 1.2063,
       "step": 1804
     },
     {
-      "epoch": 0.5898692810457516,
-      "grad_norm": 1.2713409662246704,
-      "learning_rate": 7.599728944044162e-06,
-      "loss": 0.9767,
+      "epoch": 0.04957293125703771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979951016436564e-05,
+      "loss": 1.1231,
       "step": 1805
     },
     {
-      "epoch": 0.5901960784313726,
-      "grad_norm": 1.2243762016296387,
-      "learning_rate": 7.589454846118648e-06,
-      "loss": 0.9877,
+      "epoch": 0.04960039548488094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997989467841044e-05,
+      "loss": 0.9869,
       "step": 1806
     },
     {
-      "epoch": 0.5905228758169935,
-      "grad_norm": 1.2715017795562744,
-      "learning_rate": 7.579183448960532e-06,
-      "loss": 0.9442,
+      "epoch": 0.049627859712724175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997983826141953e-05,
+      "loss": 1.127,
       "step": 1807
     },
     {
-      "epoch": 0.5908496732026144,
-      "grad_norm": 1.328645944595337,
-      "learning_rate": 7.568914764077857e-06,
-      "loss": 0.9588,
+      "epoch": 0.04965532394056741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979781765464267e-05,
+      "loss": 1.1299,
       "step": 1808
     },
     {
-      "epoch": 0.5911764705882353,
-      "grad_norm": 1.2790685892105103,
-      "learning_rate": 7.558648802975627e-06,
-      "loss": 0.9433,
+      "epoch": 0.049682788168410645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979725190545106e-05,
+      "loss": 1.0416,
       "step": 1809
     },
     {
-      "epoch": 0.5915032679738562,
-      "grad_norm": 1.2579392194747925,
-      "learning_rate": 7.548385577155795e-06,
-      "loss": 0.9578,
+      "epoch": 0.04971025239625388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979668536662498e-05,
+      "loss": 1.036,
       "step": 1810
     },
     {
-      "epoch": 0.5918300653594771,
-      "grad_norm": 1.2528767585754395,
-      "learning_rate": 7.538125098117241e-06,
-      "loss": 0.9578,
+      "epoch": 0.049737716624097116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979611803816883e-05,
+      "loss": 1.1056,
       "step": 1811
     },
     {
-      "epoch": 0.592156862745098,
-      "grad_norm": 1.2944865226745605,
-      "learning_rate": 7.52786737735578e-06,
-      "loss": 0.9567,
+      "epoch": 0.04976518085194035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979554992008716e-05,
+      "loss": 1.1059,
       "step": 1812
     },
     {
-      "epoch": 0.592483660130719,
-      "grad_norm": 1.2380448579788208,
-      "learning_rate": 7.517612426364129e-06,
-      "loss": 0.9507,
+      "epoch": 0.04979264507978358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997949810123844e-05,
+      "loss": 1.0726,
       "step": 1813
     },
     {
-      "epoch": 0.5928104575163399,
-      "grad_norm": 1.2937313318252563,
-      "learning_rate": 7.5073602566319045e-06,
-      "loss": 0.9367,
+      "epoch": 0.049820109307626814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979441131506513e-05,
+      "loss": 0.9709,
       "step": 1814
     },
     {
-      "epoch": 0.5931372549019608,
-      "grad_norm": 1.321370244026184,
-      "learning_rate": 7.49711087964561e-06,
-      "loss": 0.9726,
+      "epoch": 0.04984757353547005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997938408281338e-05,
+      "loss": 1.0541,
       "step": 1815
     },
     {
-      "epoch": 0.5934640522875817,
-      "grad_norm": 1.3093054294586182,
-      "learning_rate": 7.486864306888608e-06,
-      "loss": 0.9677,
+      "epoch": 0.049875037763313285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99793269551595e-05,
+      "loss": 1.0665,
       "step": 1816
     },
     {
-      "epoch": 0.5937908496732026,
-      "grad_norm": 1.2691104412078857,
-      "learning_rate": 7.4766205498411296e-06,
-      "loss": 0.9208,
+      "epoch": 0.04990250199115652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979269748545315e-05,
+      "loss": 1.0222,
       "step": 1817
     },
     {
-      "epoch": 0.5941176470588235,
-      "grad_norm": 1.3374755382537842,
-      "learning_rate": 7.466379619980251e-06,
-      "loss": 0.9379,
+      "epoch": 0.049929966218999755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997921246297128e-05,
+      "loss": 1.1407,
       "step": 1818
     },
     {
-      "epoch": 0.5944444444444444,
-      "grad_norm": 1.2850759029388428,
-      "learning_rate": 7.456141528779876e-06,
-      "loss": 0.9376,
+      "epoch": 0.04995743044684299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979155098437853e-05,
+      "loss": 1.0754,
       "step": 1819
     },
     {
-      "epoch": 0.5947712418300654,
-      "grad_norm": 1.2573597431182861,
-      "learning_rate": 7.445906287710733e-06,
-      "loss": 0.9424,
+      "epoch": 0.04998489467468622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979097654945487e-05,
+      "loss": 1.0852,
       "step": 1820
     },
     {
-      "epoch": 0.5950980392156863,
-      "grad_norm": 1.2702178955078125,
-      "learning_rate": 7.435673908240357e-06,
-      "loss": 0.9496,
+      "epoch": 0.050012358902529454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997904013249463e-05,
+      "loss": 1.0934,
       "step": 1821
     },
     {
-      "epoch": 0.5954248366013072,
-      "grad_norm": 1.2464447021484375,
-      "learning_rate": 7.4254444018330685e-06,
-      "loss": 0.9451,
+      "epoch": 0.05003982313037269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997898253108575e-05,
+      "loss": 1.0656,
       "step": 1822
     },
     {
-      "epoch": 0.5957516339869281,
-      "grad_norm": 1.2257614135742188,
-      "learning_rate": 7.415217779949979e-06,
-      "loss": 0.9297,
+      "epoch": 0.050067287358215924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978924850719287e-05,
+      "loss": 1.1223,
       "step": 1823
     },
     {
-      "epoch": 0.596078431372549,
-      "grad_norm": 1.334830403327942,
-      "learning_rate": 7.404994054048962e-06,
-      "loss": 0.92,
+      "epoch": 0.05009475158605916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978867091395708e-05,
+      "loss": 1.1384,
       "step": 1824
     },
     {
-      "epoch": 0.5964052287581699,
-      "grad_norm": 1.3287346363067627,
-      "learning_rate": 7.394773235584651e-06,
-      "loss": 0.9076,
+      "epoch": 0.050122215813902395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978809253115468e-05,
+      "loss": 1.2151,
       "step": 1825
     },
     {
-      "epoch": 0.5967320261437908,
-      "grad_norm": 1.317328929901123,
-      "learning_rate": 7.384555336008419e-06,
-      "loss": 0.8931,
+      "epoch": 0.05014968004174562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997875133587902e-05,
+      "loss": 1.0601,
       "step": 1826
     },
     {
-      "epoch": 0.5970588235294118,
-      "grad_norm": 1.350876808166504,
-      "learning_rate": 7.374340366768373e-06,
-      "loss": 0.9354,
+      "epoch": 0.05017714426958886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978693339686832e-05,
+      "loss": 1.1395,
       "step": 1827
     },
     {
-      "epoch": 0.5973856209150327,
-      "grad_norm": 1.2896265983581543,
-      "learning_rate": 7.364128339309326e-06,
-      "loss": 0.9022,
+      "epoch": 0.05020460849743209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978635264539355e-05,
+      "loss": 1.1347,
       "step": 1828
     },
     {
-      "epoch": 0.5977124183006536,
-      "grad_norm": 1.3443280458450317,
-      "learning_rate": 7.353919265072804e-06,
-      "loss": 0.95,
+      "epoch": 0.05023207272527533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978577110437046e-05,
+      "loss": 1.0446,
       "step": 1829
     },
     {
-      "epoch": 0.5980392156862745,
-      "grad_norm": 1.3198127746582031,
-      "learning_rate": 7.3437131554970254e-06,
-      "loss": 0.9619,
+      "epoch": 0.050259536953118564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978518877380374e-05,
+      "loss": 1.071,
       "step": 1830
     },
     {
-      "epoch": 0.5983660130718954,
-      "grad_norm": 1.304330587387085,
-      "learning_rate": 7.333510022016883e-06,
-      "loss": 0.9724,
+      "epoch": 0.0502870011809618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978460565369792e-05,
+      "loss": 1.2056,
       "step": 1831
     },
     {
-      "epoch": 0.5986928104575163,
-      "grad_norm": 1.3617892265319824,
-      "learning_rate": 7.323309876063936e-06,
-      "loss": 0.9579,
+      "epoch": 0.050314465408805034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978402174405768e-05,
+      "loss": 1.1087,
       "step": 1832
     },
     {
-      "epoch": 0.5990196078431372,
-      "grad_norm": 1.383641242980957,
-      "learning_rate": 7.313112729066396e-06,
-      "loss": 0.9457,
+      "epoch": 0.05034192963664826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997834370448876e-05,
+      "loss": 1.1219,
       "step": 1833
     },
     {
-      "epoch": 0.5993464052287582,
-      "grad_norm": 1.2400273084640503,
-      "learning_rate": 7.302918592449116e-06,
-      "loss": 0.9587,
+      "epoch": 0.0503693938644915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997828515561923e-05,
+      "loss": 1.1836,
       "step": 1834
     },
     {
-      "epoch": 0.5996732026143791,
-      "grad_norm": 1.344286322593689,
-      "learning_rate": 7.292727477633576e-06,
-      "loss": 0.907,
+      "epoch": 0.05039685809233473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978226527797643e-05,
+      "loss": 1.0547,
       "step": 1835
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 1.2989883422851562,
-      "learning_rate": 7.282539396037868e-06,
-      "loss": 0.944,
+      "epoch": 0.05042432232017797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978167821024463e-05,
+      "loss": 1.111,
       "step": 1836
     },
     {
-      "epoch": 0.6003267973856209,
-      "grad_norm": 1.270693063735962,
-      "learning_rate": 7.272354359076689e-06,
-      "loss": 0.9099,
+      "epoch": 0.0504517865480212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978109035300158e-05,
+      "loss": 1.261,
       "step": 1837
     },
     {
-      "epoch": 0.6006535947712418,
-      "grad_norm": 1.2538220882415771,
-      "learning_rate": 7.262172378161326e-06,
-      "loss": 0.9632,
+      "epoch": 0.05047925077586444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978050170625183e-05,
+      "loss": 1.0371,
       "step": 1838
     },
     {
-      "epoch": 0.6009803921568627,
-      "grad_norm": 1.3532196283340454,
-      "learning_rate": 7.251993464699632e-06,
-      "loss": 0.969,
+      "epoch": 0.050506715003707674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997799122700001e-05,
+      "loss": 1.0367,
       "step": 1839
     },
     {
-      "epoch": 0.6013071895424836,
-      "grad_norm": 1.2595572471618652,
-      "learning_rate": 7.241817630096035e-06,
-      "loss": 0.9495,
+      "epoch": 0.0505341792315509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997793220442511e-05,
+      "loss": 1.1274,
       "step": 1840
     },
     {
-      "epoch": 0.6016339869281045,
-      "grad_norm": 1.3065401315689087,
-      "learning_rate": 7.2316448857515076e-06,
-      "loss": 0.9271,
+      "epoch": 0.05056164345939414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977873102900944e-05,
+      "loss": 1.0847,
       "step": 1841
     },
     {
-      "epoch": 0.6019607843137255,
-      "grad_norm": 1.2753115892410278,
-      "learning_rate": 7.2214752430635625e-06,
-      "loss": 0.9271,
+      "epoch": 0.05058910768723737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997781392242798e-05,
+      "loss": 1.2308,
       "step": 1842
     },
     {
-      "epoch": 0.6022875816993464,
-      "grad_norm": 1.3620848655700684,
-      "learning_rate": 7.211308713426235e-06,
-      "loss": 0.9199,
+      "epoch": 0.05061657191508061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977754663006686e-05,
+      "loss": 1.0503,
       "step": 1843
     },
     {
-      "epoch": 0.6026143790849673,
-      "grad_norm": 1.369003415107727,
-      "learning_rate": 7.201145308230075e-06,
-      "loss": 0.9725,
+      "epoch": 0.05064403614292384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997769532463753e-05,
+      "loss": 1.1591,
       "step": 1844
     },
     {
-      "epoch": 0.6029411764705882,
-      "grad_norm": 1.3323901891708374,
-      "learning_rate": 7.190985038862131e-06,
-      "loss": 0.9658,
+      "epoch": 0.05067150037076708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997763590732099e-05,
+      "loss": 1.0695,
       "step": 1845
     },
     {
-      "epoch": 0.6032679738562091,
-      "grad_norm": 1.385352373123169,
-      "learning_rate": 7.180827916705935e-06,
-      "loss": 0.9482,
+      "epoch": 0.05069896459861031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977576411057528e-05,
+      "loss": 1.067,
       "step": 1846
     },
     {
-      "epoch": 0.60359477124183,
-      "grad_norm": 1.3072097301483154,
-      "learning_rate": 7.170673953141496e-06,
-      "loss": 0.9387,
+      "epoch": 0.05072642882645354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977516835847613e-05,
+      "loss": 1.1255,
       "step": 1847
     },
     {
-      "epoch": 0.6039215686274509,
-      "grad_norm": 1.2783094644546509,
-      "learning_rate": 7.160523159545285e-06,
-      "loss": 0.9525,
+      "epoch": 0.05075389305429678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997745718169172e-05,
+      "loss": 1.1036,
       "step": 1848
     },
     {
-      "epoch": 0.6042483660130719,
-      "grad_norm": 1.3002747297286987,
-      "learning_rate": 7.1503755472902245e-06,
-      "loss": 0.9527,
+      "epoch": 0.05078135728214001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977397448590324e-05,
+      "loss": 1.0379,
       "step": 1849
     },
     {
-      "epoch": 0.6045751633986928,
-      "grad_norm": 1.2587488889694214,
-      "learning_rate": 7.140231127745661e-06,
-      "loss": 0.9284,
+      "epoch": 0.05080882150998325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977337636543895e-05,
+      "loss": 1.1605,
       "step": 1850
     },
     {
-      "epoch": 0.6049019607843137,
-      "grad_norm": 1.292233943939209,
-      "learning_rate": 7.1300899122773715e-06,
-      "loss": 0.922,
+      "epoch": 0.05083628573782648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99772777455529e-05,
+      "loss": 1.1621,
       "step": 1851
     },
     {
-      "epoch": 0.6052287581699346,
-      "grad_norm": 1.2650859355926514,
-      "learning_rate": 7.119951912247545e-06,
-      "loss": 0.9244,
+      "epoch": 0.05086374996566972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977217775617826e-05,
+      "loss": 1.1233,
       "step": 1852
     },
     {
-      "epoch": 0.6055555555555555,
-      "grad_norm": 1.2286444902420044,
-      "learning_rate": 7.109817139014764e-06,
-      "loss": 0.9637,
+      "epoch": 0.05089121419351295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977157726739135e-05,
+      "loss": 1.1412,
       "step": 1853
     },
     {
-      "epoch": 0.6058823529411764,
-      "grad_norm": 1.325933814048767,
-      "learning_rate": 7.099685603934e-06,
-      "loss": 0.9286,
+      "epoch": 0.05091867842135618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997709759891731e-05,
+      "loss": 1.1341,
       "step": 1854
     },
     {
-      "epoch": 0.6062091503267973,
-      "grad_norm": 1.2955578565597534,
-      "learning_rate": 7.089557318356594e-06,
-      "loss": 0.9493,
+      "epoch": 0.050946142649199416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977037392152822e-05,
+      "loss": 1.1243,
       "step": 1855
     },
     {
-      "epoch": 0.6065359477124183,
-      "grad_norm": 1.3022781610488892,
-      "learning_rate": 7.079432293630244e-06,
-      "loss": 0.9384,
+      "epoch": 0.05097360687704265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997697710644615e-05,
+      "loss": 1.2134,
       "step": 1856
     },
     {
-      "epoch": 0.6068627450980392,
-      "grad_norm": 1.2156546115875244,
-      "learning_rate": 7.069310541098995e-06,
-      "loss": 0.9605,
+      "epoch": 0.05100107110488589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976916741797773e-05,
+      "loss": 1.0567,
       "step": 1857
     },
     {
-      "epoch": 0.6071895424836601,
-      "grad_norm": 1.2894723415374756,
-      "learning_rate": 7.0591920721032314e-06,
-      "loss": 0.9574,
+      "epoch": 0.05102853533272912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976856298208162e-05,
+      "loss": 1.189,
       "step": 1858
     },
     {
-      "epoch": 0.607516339869281,
-      "grad_norm": 1.3411719799041748,
-      "learning_rate": 7.049076897979652e-06,
-      "loss": 0.9656,
+      "epoch": 0.05105599956057236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976795775677802e-05,
+      "loss": 1.0287,
       "step": 1859
     },
     {
-      "epoch": 0.6078431372549019,
-      "grad_norm": 1.2647956609725952,
-      "learning_rate": 7.038965030061273e-06,
-      "loss": 0.9366,
+      "epoch": 0.051083463788415585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997673517420717e-05,
+      "loss": 1.1181,
       "step": 1860
     },
     {
-      "epoch": 0.6081699346405228,
-      "grad_norm": 1.3523671627044678,
-      "learning_rate": 7.028856479677391e-06,
-      "loss": 0.9663,
+      "epoch": 0.05111092801625882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976674493796742e-05,
+      "loss": 1.1156,
       "step": 1861
     },
     {
-      "epoch": 0.6084967320261437,
-      "grad_norm": 1.2523976564407349,
-      "learning_rate": 7.018751258153599e-06,
-      "loss": 0.9189,
+      "epoch": 0.051138392244102056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976613734447e-05,
+      "loss": 1.0669,
       "step": 1862
     },
     {
-      "epoch": 0.6088235294117647,
-      "grad_norm": 1.2521415948867798,
-      "learning_rate": 7.008649376811756e-06,
-      "loss": 0.9273,
+      "epoch": 0.05116585647194529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997655289615843e-05,
+      "loss": 1.0004,
       "step": 1863
     },
     {
-      "epoch": 0.6091503267973856,
-      "grad_norm": 1.2997777462005615,
-      "learning_rate": 6.998550846969981e-06,
-      "loss": 0.9018,
+      "epoch": 0.051193320699788526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976491978931505e-05,
+      "loss": 1.0598,
       "step": 1864
     },
     {
-      "epoch": 0.6094771241830066,
-      "grad_norm": 1.3328218460083008,
-      "learning_rate": 6.9884556799426316e-06,
-      "loss": 0.9705,
+      "epoch": 0.05122078492763176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976430982766714e-05,
+      "loss": 1.1857,
       "step": 1865
     },
     {
-      "epoch": 0.6098039215686275,
-      "grad_norm": 1.3121150732040405,
-      "learning_rate": 6.978363887040308e-06,
-      "loss": 0.9143,
+      "epoch": 0.051248249155475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976369907664534e-05,
+      "loss": 1.0616,
       "step": 1866
     },
     {
-      "epoch": 0.6101307189542484,
-      "grad_norm": 1.2431137561798096,
-      "learning_rate": 6.968275479569817e-06,
-      "loss": 0.9231,
+      "epoch": 0.051275713383318225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976308753625453e-05,
+      "loss": 1.127,
       "step": 1867
     },
     {
-      "epoch": 0.6104575163398693,
-      "grad_norm": 1.2959319353103638,
-      "learning_rate": 6.9581904688341854e-06,
-      "loss": 0.9388,
+      "epoch": 0.05130317761116146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997624752064995e-05,
+      "loss": 1.2144,
       "step": 1868
     },
     {
-      "epoch": 0.6107843137254902,
-      "grad_norm": 1.297324299812317,
-      "learning_rate": 6.9481088661326235e-06,
-      "loss": 0.8957,
+      "epoch": 0.051330641839004695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976186208738512e-05,
+      "loss": 1.051,
       "step": 1869
     },
     {
-      "epoch": 0.6111111111111112,
-      "grad_norm": 1.298065185546875,
-      "learning_rate": 6.938030682760532e-06,
-      "loss": 0.9252,
+      "epoch": 0.05135810606684793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976124817891628e-05,
+      "loss": 1.0743,
       "step": 1870
     },
     {
-      "epoch": 0.6114379084967321,
-      "grad_norm": 1.3203293085098267,
-      "learning_rate": 6.927955930009474e-06,
-      "loss": 0.9451,
+      "epoch": 0.051385570294691166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976063348109778e-05,
+      "loss": 0.9536,
       "step": 1871
     },
     {
-      "epoch": 0.611764705882353,
-      "grad_norm": 1.275282621383667,
-      "learning_rate": 6.917884619167177e-06,
-      "loss": 0.949,
+      "epoch": 0.0514130345225344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997600179939345e-05,
+      "loss": 1.1194,
       "step": 1872
     },
     {
-      "epoch": 0.6120915032679739,
-      "grad_norm": 1.2524340152740479,
-      "learning_rate": 6.9078167615174985e-06,
-      "loss": 0.9538,
+      "epoch": 0.051440498750377636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975940171743132e-05,
+      "loss": 1.0635,
       "step": 1873
     },
     {
-      "epoch": 0.6124183006535948,
-      "grad_norm": 1.295903205871582,
-      "learning_rate": 6.897752368340438e-06,
-      "loss": 0.9543,
+      "epoch": 0.051467962978220864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997587846515931e-05,
+      "loss": 1.1398,
       "step": 1874
     },
     {
-      "epoch": 0.6127450980392157,
-      "grad_norm": 1.3161076307296753,
-      "learning_rate": 6.887691450912113e-06,
-      "loss": 0.9513,
+      "epoch": 0.0514954272060641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975816679642476e-05,
+      "loss": 1.0529,
       "step": 1875
     },
     {
-      "epoch": 0.6130718954248366,
-      "grad_norm": 1.3265007734298706,
-      "learning_rate": 6.8776340205047446e-06,
-      "loss": 0.9096,
+      "epoch": 0.051522891433907335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975754815193115e-05,
+      "loss": 1.1368,
       "step": 1876
     },
     {
-      "epoch": 0.6133986928104576,
-      "grad_norm": 1.3182584047317505,
-      "learning_rate": 6.867580088386646e-06,
-      "loss": 0.9302,
+      "epoch": 0.05155035566175057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975692871811714e-05,
+      "loss": 1.1238,
       "step": 1877
     },
     {
-      "epoch": 0.6137254901960785,
-      "grad_norm": 1.3227144479751587,
-      "learning_rate": 6.857529665822211e-06,
-      "loss": 0.9283,
+      "epoch": 0.051577819889593805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997563084949877e-05,
+      "loss": 1.0302,
       "step": 1878
     },
     {
-      "epoch": 0.6140522875816994,
-      "grad_norm": 1.2266169786453247,
-      "learning_rate": 6.847482764071902e-06,
-      "loss": 0.9555,
+      "epoch": 0.05160528411743704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997556874825477e-05,
+      "loss": 1.1223,
       "step": 1879
     },
     {
-      "epoch": 0.6143790849673203,
-      "grad_norm": 1.3152949810028076,
-      "learning_rate": 6.837439394392238e-06,
-      "loss": 0.9446,
+      "epoch": 0.051632748345280276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997550656808021e-05,
+      "loss": 1.1836,
       "step": 1880
     },
     {
-      "epoch": 0.6147058823529412,
-      "grad_norm": 1.3192017078399658,
-      "learning_rate": 6.8273995680357805e-06,
-      "loss": 0.9542,
+      "epoch": 0.051660212573123504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975444308975575e-05,
+      "loss": 1.1198,
       "step": 1881
     },
     {
-      "epoch": 0.6150326797385621,
-      "grad_norm": 1.3173855543136597,
-      "learning_rate": 6.817363296251119e-06,
-      "loss": 0.9708,
+      "epoch": 0.05168767680096674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975381970941355e-05,
+      "loss": 0.988,
       "step": 1882
     },
     {
-      "epoch": 0.615359477124183,
-      "grad_norm": 1.3220322132110596,
-      "learning_rate": 6.807330590282866e-06,
-      "loss": 0.9542,
+      "epoch": 0.051715141028809974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975319553978054e-05,
+      "loss": 1.101,
       "step": 1883
     },
     {
-      "epoch": 0.615686274509804,
-      "grad_norm": 1.3128364086151123,
-      "learning_rate": 6.797301461371626e-06,
-      "loss": 0.9177,
+      "epoch": 0.05174260525665321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975257058086163e-05,
+      "loss": 1.1357,
       "step": 1884
     },
     {
-      "epoch": 0.6160130718954249,
-      "grad_norm": 1.3310080766677856,
-      "learning_rate": 6.7872759207540085e-06,
-      "loss": 0.9402,
+      "epoch": 0.051770069484496445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997519448326617e-05,
+      "loss": 1.0499,
       "step": 1885
     },
     {
-      "epoch": 0.6163398692810458,
-      "grad_norm": 1.3009871244430542,
-      "learning_rate": 6.7772539796625995e-06,
-      "loss": 0.942,
+      "epoch": 0.05179753371233968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975131829518575e-05,
+      "loss": 0.9369,
       "step": 1886
     },
     {
-      "epoch": 0.6166666666666667,
-      "grad_norm": 1.2685109376907349,
-      "learning_rate": 6.7672356493259495e-06,
-      "loss": 0.9008,
+      "epoch": 0.05182499794018291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975069096843872e-05,
+      "loss": 1.1661,
       "step": 1887
     },
     {
-      "epoch": 0.6169934640522876,
-      "grad_norm": 1.3362884521484375,
-      "learning_rate": 6.757220940968567e-06,
-      "loss": 0.9809,
+      "epoch": 0.051852462168026144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997500628524256e-05,
+      "loss": 1.1891,
       "step": 1888
     },
     {
-      "epoch": 0.6173202614379085,
-      "grad_norm": 1.3671109676361084,
-      "learning_rate": 6.7472098658108994e-06,
-      "loss": 0.9438,
+      "epoch": 0.05187992639586938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974943394715133e-05,
+      "loss": 1.1174,
       "step": 1889
     },
     {
-      "epoch": 0.6176470588235294,
-      "grad_norm": 1.2992867231369019,
-      "learning_rate": 6.737202435069323e-06,
-      "loss": 0.942,
+      "epoch": 0.051907390623712614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997488042526209e-05,
+      "loss": 1.0058,
       "step": 1890
     },
     {
-      "epoch": 0.6179738562091504,
-      "grad_norm": 1.3224472999572754,
-      "learning_rate": 6.727198659956133e-06,
-      "loss": 0.9117,
+      "epoch": 0.05193485485155585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974817376883932e-05,
+      "loss": 1.0635,
       "step": 1891
     },
     {
-      "epoch": 0.6183006535947713,
-      "grad_norm": 1.271720051765442,
-      "learning_rate": 6.7171985516795315e-06,
-      "loss": 0.9557,
+      "epoch": 0.051962319079399084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974754249581153e-05,
+      "loss": 1.0108,
       "step": 1892
     },
     {
-      "epoch": 0.6186274509803922,
-      "grad_norm": 1.265630841255188,
-      "learning_rate": 6.707202121443607e-06,
-      "loss": 0.9524,
+      "epoch": 0.05198978330724232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974691043354256e-05,
+      "loss": 1.0352,
       "step": 1893
     },
     {
-      "epoch": 0.6189542483660131,
-      "grad_norm": 1.2504501342773438,
-      "learning_rate": 6.697209380448333e-06,
-      "loss": 0.911,
+      "epoch": 0.05201724753508555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997462775820374e-05,
+      "loss": 0.9921,
       "step": 1894
     },
     {
-      "epoch": 0.619281045751634,
-      "grad_norm": 1.3188557624816895,
-      "learning_rate": 6.687220339889537e-06,
-      "loss": 0.9229,
+      "epoch": 0.05204471176292878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974564394130104e-05,
+      "loss": 0.9554,
       "step": 1895
     },
     {
-      "epoch": 0.6196078431372549,
-      "grad_norm": 1.363122582435608,
-      "learning_rate": 6.677235010958916e-06,
-      "loss": 0.9418,
+      "epoch": 0.05207217599077202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974500951133845e-05,
+      "loss": 1.0767,
       "step": 1896
     },
     {
-      "epoch": 0.6199346405228758,
-      "grad_norm": 1.2452425956726074,
-      "learning_rate": 6.667253404843999e-06,
-      "loss": 0.9674,
+      "epoch": 0.052099640218615254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974437429215478e-05,
+      "loss": 1.0846,
       "step": 1897
     },
     {
-      "epoch": 0.6202614379084967,
-      "grad_norm": 1.3406429290771484,
-      "learning_rate": 6.6572755327281506e-06,
-      "loss": 0.9588,
+      "epoch": 0.05212710444645849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974373828375497e-05,
+      "loss": 1.1307,
       "step": 1898
     },
     {
-      "epoch": 0.6205882352941177,
-      "grad_norm": 1.2450374364852905,
-      "learning_rate": 6.647301405790545e-06,
-      "loss": 0.9688,
+      "epoch": 0.052154568674301724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974310148614408e-05,
+      "loss": 1.1655,
       "step": 1899
     },
     {
-      "epoch": 0.6209150326797386,
-      "grad_norm": 1.2432974576950073,
-      "learning_rate": 6.637331035206166e-06,
-      "loss": 0.9566,
+      "epoch": 0.05218203290214496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997424638993271e-05,
+      "loss": 1.0916,
       "step": 1900
     },
     {
-      "epoch": 0.6212418300653595,
-      "grad_norm": 1.3671040534973145,
-      "learning_rate": 6.627364432145782e-06,
-      "loss": 0.9064,
+      "epoch": 0.05220949712998819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974182552330913e-05,
+      "loss": 1.1318,
       "step": 1901
     },
     {
-      "epoch": 0.6215686274509804,
-      "grad_norm": 1.321744441986084,
-      "learning_rate": 6.617401607775948e-06,
-      "loss": 0.8982,
+      "epoch": 0.05223696135783142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997411863580952e-05,
+      "loss": 1.0681,
       "step": 1902
     },
     {
-      "epoch": 0.6218954248366013,
-      "grad_norm": 1.4005368947982788,
-      "learning_rate": 6.607442573258982e-06,
-      "loss": 0.9819,
+      "epoch": 0.05226442558567466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974054640369034e-05,
+      "loss": 1.1635,
       "step": 1903
     },
     {
-      "epoch": 0.6222222222222222,
-      "grad_norm": 1.3378090858459473,
-      "learning_rate": 6.597487339752953e-06,
-      "loss": 0.9367,
+      "epoch": 0.05229188981351789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973990566009965e-05,
+      "loss": 1.1179,
       "step": 1904
     },
     {
-      "epoch": 0.6225490196078431,
-      "grad_norm": 1.2628504037857056,
-      "learning_rate": 6.58753591841168e-06,
-      "loss": 0.962,
+      "epoch": 0.05231935404136113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973926412732822e-05,
+      "loss": 1.1581,
       "step": 1905
     },
     {
-      "epoch": 0.6228758169934641,
-      "grad_norm": 1.2778384685516357,
-      "learning_rate": 6.577588320384698e-06,
-      "loss": 0.9253,
+      "epoch": 0.052346818269204363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973862180538108e-05,
+      "loss": 1.2034,
       "step": 1906
     },
     {
-      "epoch": 0.623202614379085,
-      "grad_norm": 1.30204176902771,
-      "learning_rate": 6.567644556817267e-06,
-      "loss": 0.939,
+      "epoch": 0.0523742824970476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973797869426334e-05,
+      "loss": 1.0673,
       "step": 1907
     },
     {
-      "epoch": 0.6235294117647059,
-      "grad_norm": 1.266796588897705,
-      "learning_rate": 6.557704638850352e-06,
-      "loss": 0.9306,
+      "epoch": 0.05240174672489083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973733479398007e-05,
+      "loss": 1.0087,
       "step": 1908
     },
     {
-      "epoch": 0.6238562091503268,
-      "grad_norm": 1.32464599609375,
-      "learning_rate": 6.547768577620603e-06,
-      "loss": 0.9118,
+      "epoch": 0.05242921095273406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973669010453637e-05,
+      "loss": 1.0422,
       "step": 1909
     },
     {
-      "epoch": 0.6241830065359477,
-      "grad_norm": 1.3334121704101562,
-      "learning_rate": 6.537836384260356e-06,
-      "loss": 0.9711,
+      "epoch": 0.0524566751805773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997360446259373e-05,
+      "loss": 1.0634,
       "step": 1910
     },
     {
-      "epoch": 0.6245098039215686,
-      "grad_norm": 1.2726194858551025,
-      "learning_rate": 6.5279080698976094e-06,
-      "loss": 0.9584,
+      "epoch": 0.05248413940842053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973539835818808e-05,
+      "loss": 1.0543,
       "step": 1911
     },
     {
-      "epoch": 0.6248366013071895,
-      "grad_norm": 1.2086796760559082,
-      "learning_rate": 6.517983645656014e-06,
-      "loss": 0.9521,
+      "epoch": 0.05251160363626377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973475130129373e-05,
+      "loss": 1.1231,
       "step": 1912
     },
     {
-      "epoch": 0.6251633986928105,
-      "grad_norm": 1.2000842094421387,
-      "learning_rate": 6.508063122654865e-06,
-      "loss": 0.9327,
+      "epoch": 0.052539067864107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997341034552594e-05,
+      "loss": 1.0197,
       "step": 1913
     },
     {
-      "epoch": 0.6254901960784314,
-      "grad_norm": 1.310484528541565,
-      "learning_rate": 6.498146512009087e-06,
-      "loss": 0.9619,
+      "epoch": 0.05256653209195024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973345482009022e-05,
+      "loss": 0.9733,
       "step": 1914
     },
     {
-      "epoch": 0.6258169934640523,
-      "grad_norm": 1.2892115116119385,
-      "learning_rate": 6.48823382482922e-06,
-      "loss": 0.9781,
+      "epoch": 0.052593996319793467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973280539579126e-05,
+      "loss": 1.0772,
       "step": 1915
     },
     {
-      "epoch": 0.6261437908496732,
-      "grad_norm": 1.3448349237442017,
-      "learning_rate": 6.4783250722214066e-06,
-      "loss": 0.9501,
+      "epoch": 0.0526214605476367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973215518236775e-05,
+      "loss": 1.0345,
       "step": 1916
     },
     {
-      "epoch": 0.6264705882352941,
-      "grad_norm": 1.2603986263275146,
-      "learning_rate": 6.46842026528739e-06,
-      "loss": 0.945,
+      "epoch": 0.05264892477547994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973150417982482e-05,
+      "loss": 1.0753,
       "step": 1917
     },
     {
-      "epoch": 0.626797385620915,
-      "grad_norm": 1.2588222026824951,
-      "learning_rate": 6.458519415124473e-06,
-      "loss": 0.9369,
+      "epoch": 0.05267638900332317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973085238816754e-05,
+      "loss": 1.075,
       "step": 1918
     },
     {
-      "epoch": 0.6271241830065359,
-      "grad_norm": 1.329470157623291,
-      "learning_rate": 6.448622532825546e-06,
-      "loss": 0.9343,
+      "epoch": 0.05270385323116641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997301998074012e-05,
+      "loss": 1.0295,
       "step": 1919
     },
     {
-      "epoch": 0.6274509803921569,
-      "grad_norm": 1.3524428606033325,
-      "learning_rate": 6.438729629479045e-06,
-      "loss": 0.9093,
+      "epoch": 0.05273131745900964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997295464375308e-05,
+      "loss": 1.1008,
       "step": 1920
     },
     {
-      "epoch": 0.6277777777777778,
-      "grad_norm": 1.3444948196411133,
-      "learning_rate": 6.428840716168947e-06,
-      "loss": 0.9476,
+      "epoch": 0.05275878168685287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972889227856165e-05,
+      "loss": 1.1131,
       "step": 1921
     },
     {
-      "epoch": 0.6281045751633987,
-      "grad_norm": 1.3245820999145508,
-      "learning_rate": 6.418955803974763e-06,
-      "loss": 0.9361,
+      "epoch": 0.052786245914696106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972823733049888e-05,
+      "loss": 1.0971,
       "step": 1922
     },
     {
-      "epoch": 0.6284313725490196,
-      "grad_norm": 1.229536771774292,
-      "learning_rate": 6.409074903971512e-06,
-      "loss": 0.9331,
+      "epoch": 0.05281371014253934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972758159334763e-05,
+      "loss": 1.0712,
       "step": 1923
     },
     {
-      "epoch": 0.6287581699346405,
-      "grad_norm": 1.2253104448318481,
-      "learning_rate": 6.399198027229732e-06,
-      "loss": 0.926,
+      "epoch": 0.052841174370382576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972692506711312e-05,
+      "loss": 1.1279,
       "step": 1924
     },
     {
-      "epoch": 0.6290849673202614,
-      "grad_norm": 1.3077677488327026,
-      "learning_rate": 6.3893251848154385e-06,
-      "loss": 0.9254,
+      "epoch": 0.05286863859822581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972626775180057e-05,
+      "loss": 1.0433,
       "step": 1925
     },
     {
-      "epoch": 0.6294117647058823,
-      "grad_norm": 1.3274539709091187,
-      "learning_rate": 6.3794563877901385e-06,
-      "loss": 0.9542,
+      "epoch": 0.05289610282606905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972560964741517e-05,
+      "loss": 1.1733,
       "step": 1926
     },
     {
-      "epoch": 0.6297385620915033,
-      "grad_norm": 1.3351601362228394,
-      "learning_rate": 6.3695916472108e-06,
-      "loss": 0.9118,
+      "epoch": 0.05292356705391228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997249507539621e-05,
+      "loss": 1.0798,
       "step": 1927
     },
     {
-      "epoch": 0.6300653594771242,
-      "grad_norm": 1.321487307548523,
-      "learning_rate": 6.359730974129852e-06,
-      "loss": 0.9411,
+      "epoch": 0.05295103128175551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972429107144658e-05,
+      "loss": 0.9866,
       "step": 1928
     },
     {
-      "epoch": 0.6303921568627451,
-      "grad_norm": 1.2582429647445679,
-      "learning_rate": 6.349874379595155e-06,
-      "loss": 0.8965,
+      "epoch": 0.052978495509598746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972363059987386e-05,
+      "loss": 1.053,
       "step": 1929
     },
     {
-      "epoch": 0.630718954248366,
-      "grad_norm": 1.2316243648529053,
-      "learning_rate": 6.340021874650013e-06,
-      "loss": 0.9451,
+      "epoch": 0.05300595973744198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972296933924915e-05,
+      "loss": 1.1227,
       "step": 1930
     },
     {
-      "epoch": 0.6310457516339869,
-      "grad_norm": 1.2679141759872437,
-      "learning_rate": 6.3301734703331386e-06,
-      "loss": 0.966,
+      "epoch": 0.053033423965285216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972230728957768e-05,
+      "loss": 1.1154,
       "step": 1931
     },
     {
-      "epoch": 0.6313725490196078,
-      "grad_norm": 1.3112471103668213,
-      "learning_rate": 6.320329177678656e-06,
-      "loss": 0.9254,
+      "epoch": 0.05306088819312845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972164445086462e-05,
+      "loss": 1.1192,
       "step": 1932
     },
     {
-      "epoch": 0.6316993464052287,
-      "grad_norm": 1.2813457250595093,
-      "learning_rate": 6.310489007716083e-06,
-      "loss": 0.9475,
+      "epoch": 0.053088352420971686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972098082311534e-05,
+      "loss": 1.2255,
       "step": 1933
     },
     {
-      "epoch": 0.6320261437908496,
-      "grad_norm": 1.2821322679519653,
-      "learning_rate": 6.300652971470309e-06,
-      "loss": 0.9324,
+      "epoch": 0.05311581664881492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972031640633504e-05,
+      "loss": 1.1224,
       "step": 1934
     },
     {
-      "epoch": 0.6323529411764706,
-      "grad_norm": 1.2638098001480103,
-      "learning_rate": 6.290821079961605e-06,
-      "loss": 0.9667,
+      "epoch": 0.05314328087665815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971965120052896e-05,
+      "loss": 1.076,
       "step": 1935
     },
     {
-      "epoch": 0.6326797385620915,
-      "grad_norm": 1.2944446802139282,
-      "learning_rate": 6.28099334420559e-06,
-      "loss": 0.9525,
+      "epoch": 0.053170745104501385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971898520570233e-05,
+      "loss": 1.072,
       "step": 1936
     },
     {
-      "epoch": 0.6330065359477124,
-      "grad_norm": 1.2825645208358765,
-      "learning_rate": 6.271169775213228e-06,
-      "loss": 0.9518,
+      "epoch": 0.05319820933234462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997183184218605e-05,
+      "loss": 1.1373,
       "step": 1937
     },
     {
-      "epoch": 0.6333333333333333,
-      "grad_norm": 1.300518274307251,
-      "learning_rate": 6.261350383990816e-06,
-      "loss": 0.9215,
+      "epoch": 0.053225673560187856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971765084900872e-05,
+      "loss": 1.0603,
       "step": 1938
     },
     {
-      "epoch": 0.6336601307189542,
-      "grad_norm": 1.2976813316345215,
-      "learning_rate": 6.251535181539976e-06,
-      "loss": 0.9351,
+      "epoch": 0.05325313778803109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971698248715225e-05,
+      "loss": 1.0974,
       "step": 1939
     },
     {
-      "epoch": 0.6339869281045751,
-      "grad_norm": 1.3761931657791138,
-      "learning_rate": 6.241724178857621e-06,
-      "loss": 0.9451,
+      "epoch": 0.053280602015874326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971631333629637e-05,
+      "loss": 1.0856,
       "step": 1940
     },
     {
-      "epoch": 0.634313725490196,
-      "grad_norm": 1.3003026247024536,
-      "learning_rate": 6.231917386935973e-06,
-      "loss": 0.9293,
+      "epoch": 0.05330806624371756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997156433964464e-05,
+      "loss": 1.1251,
       "step": 1941
     },
     {
-      "epoch": 0.634640522875817,
-      "grad_norm": 1.260115623474121,
-      "learning_rate": 6.22211481676253e-06,
-      "loss": 0.9628,
+      "epoch": 0.05333553047156079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971497266760764e-05,
+      "loss": 1.1489,
       "step": 1942
     },
     {
-      "epoch": 0.6349673202614379,
-      "grad_norm": 1.318030834197998,
-      "learning_rate": 6.21231647932006e-06,
-      "loss": 0.9509,
+      "epoch": 0.053362994699404025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971430114978538e-05,
+      "loss": 1.2977,
       "step": 1943
     },
     {
-      "epoch": 0.6352941176470588,
-      "grad_norm": 1.235325813293457,
-      "learning_rate": 6.202522385586595e-06,
-      "loss": 0.9595,
+      "epoch": 0.05339045892724726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971362884298496e-05,
+      "loss": 1.0414,
       "step": 1944
     },
     {
-      "epoch": 0.6356209150326797,
-      "grad_norm": 1.3156641721725464,
-      "learning_rate": 6.192732546535411e-06,
-      "loss": 0.9698,
+      "epoch": 0.053417923155090495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971295574721168e-05,
+      "loss": 1.0656,
       "step": 1945
     },
     {
-      "epoch": 0.6359477124183006,
-      "grad_norm": 1.283003568649292,
-      "learning_rate": 6.182946973135004e-06,
-      "loss": 0.9218,
+      "epoch": 0.05344538738293373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971228186247088e-05,
+      "loss": 1.1924,
       "step": 1946
     },
     {
-      "epoch": 0.6362745098039215,
-      "grad_norm": 1.2664074897766113,
-      "learning_rate": 6.173165676349103e-06,
-      "loss": 0.9351,
+      "epoch": 0.053472851610776966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971160718876784e-05,
+      "loss": 1.1474,
       "step": 1947
     },
     {
-      "epoch": 0.6366013071895424,
-      "grad_norm": 1.3313953876495361,
-      "learning_rate": 6.163388667136646e-06,
-      "loss": 0.9438,
+      "epoch": 0.053500315838620194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971093172610797e-05,
+      "loss": 0.9936,
       "step": 1948
     },
     {
-      "epoch": 0.6369281045751634,
-      "grad_norm": 1.261316180229187,
-      "learning_rate": 6.153615956451761e-06,
-      "loss": 0.967,
+      "epoch": 0.05352778006646343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971025547449656e-05,
+      "loss": 1.1666,
       "step": 1949
     },
     {
-      "epoch": 0.6372549019607843,
-      "grad_norm": 1.3480398654937744,
-      "learning_rate": 6.14384755524377e-06,
-      "loss": 0.9152,
+      "epoch": 0.053555244294306664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970957843393902e-05,
+      "loss": 1.0975,
       "step": 1950
     },
     {
-      "epoch": 0.6375816993464052,
-      "grad_norm": 1.3015077114105225,
-      "learning_rate": 6.134083474457151e-06,
-      "loss": 0.9617,
+      "epoch": 0.0535827085221499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970890060444063e-05,
+      "loss": 1.1689,
       "step": 1951
     },
     {
-      "epoch": 0.6379084967320261,
-      "grad_norm": 1.2585914134979248,
-      "learning_rate": 6.12432372503155e-06,
-      "loss": 0.9364,
+      "epoch": 0.053610172749993135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970822198600683e-05,
+      "loss": 1.1205,
       "step": 1952
     },
     {
-      "epoch": 0.638235294117647,
-      "grad_norm": 1.2697882652282715,
-      "learning_rate": 6.114568317901768e-06,
-      "loss": 0.9765,
+      "epoch": 0.05363763697783637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970754257864296e-05,
+      "loss": 0.9596,
       "step": 1953
     },
     {
-      "epoch": 0.6385620915032679,
-      "grad_norm": 1.2860521078109741,
-      "learning_rate": 6.104817263997727e-06,
-      "loss": 0.9182,
+      "epoch": 0.053665101205679605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970686238235437e-05,
+      "loss": 1.1124,
       "step": 1954
     },
     {
-      "epoch": 0.6388888888888888,
-      "grad_norm": 1.2592734098434448,
-      "learning_rate": 6.095070574244481e-06,
-      "loss": 0.9294,
+      "epoch": 0.05369256543352283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970618139714646e-05,
+      "loss": 1.1366,
       "step": 1955
     },
     {
-      "epoch": 0.6392156862745098,
-      "grad_norm": 1.275815725326538,
-      "learning_rate": 6.085328259562195e-06,
-      "loss": 0.9275,
+      "epoch": 0.05372002966136607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970549962302465e-05,
+      "loss": 1.0269,
       "step": 1956
     },
     {
-      "epoch": 0.6395424836601307,
-      "grad_norm": 1.3299214839935303,
-      "learning_rate": 6.075590330866119e-06,
-      "loss": 0.9531,
+      "epoch": 0.053747493889209304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970481705999427e-05,
+      "loss": 1.031,
       "step": 1957
     },
     {
-      "epoch": 0.6398692810457516,
-      "grad_norm": 1.3658503293991089,
-      "learning_rate": 6.065856799066604e-06,
-      "loss": 0.9549,
+      "epoch": 0.05377495811705254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970413370806074e-05,
+      "loss": 1.0837,
       "step": 1958
     },
     {
-      "epoch": 0.6401960784313725,
-      "grad_norm": 1.253770112991333,
-      "learning_rate": 6.056127675069068e-06,
-      "loss": 0.9806,
+      "epoch": 0.053802422344895774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970344956722953e-05,
+      "loss": 1.1679,
       "step": 1959
     },
     {
-      "epoch": 0.6405228758169934,
-      "grad_norm": 1.2686128616333008,
-      "learning_rate": 6.046402969773992e-06,
-      "loss": 0.964,
+      "epoch": 0.05382988657273901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970276463750598e-05,
+      "loss": 1.1783,
       "step": 1960
     },
     {
-      "epoch": 0.6408496732026144,
-      "grad_norm": 1.3286341428756714,
-      "learning_rate": 6.0366826940769075e-06,
-      "loss": 0.9627,
+      "epoch": 0.053857350800582245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970207891889552e-05,
+      "loss": 1.1729,
       "step": 1961
     },
     {
-      "epoch": 0.6411764705882353,
-      "grad_norm": 1.3399343490600586,
-      "learning_rate": 6.0269668588683815e-06,
-      "loss": 0.9549,
+      "epoch": 0.05388481502842547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997013924114036e-05,
+      "loss": 1.1409,
       "step": 1962
     },
     {
-      "epoch": 0.6415032679738563,
-      "grad_norm": 1.2503085136413574,
-      "learning_rate": 6.017255475034003e-06,
-      "loss": 0.9412,
+      "epoch": 0.05391227925626871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970070511503564e-05,
+      "loss": 1.132,
       "step": 1963
     },
     {
-      "epoch": 0.6418300653594772,
-      "grad_norm": 1.2332344055175781,
-      "learning_rate": 6.007548553454379e-06,
-      "loss": 0.9189,
+      "epoch": 0.05393974348411194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970001702979714e-05,
+      "loss": 1.082,
       "step": 1964
     },
     {
-      "epoch": 0.6421568627450981,
-      "grad_norm": 1.2479175329208374,
-      "learning_rate": 5.997846105005114e-06,
-      "loss": 0.9299,
+      "epoch": 0.05396720771195518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996993281556934e-05,
+      "loss": 1.0412,
       "step": 1965
     },
     {
-      "epoch": 0.642483660130719,
-      "grad_norm": 1.272969365119934,
-      "learning_rate": 5.9881481405568005e-06,
-      "loss": 0.9207,
+      "epoch": 0.053994671939798414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969863849273003e-05,
+      "loss": 1.0487,
       "step": 1966
     },
     {
-      "epoch": 0.6428104575163399,
-      "grad_norm": 1.255947232246399,
-      "learning_rate": 5.978454670975014e-06,
-      "loss": 0.9208,
+      "epoch": 0.05402213616764165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969794804091236e-05,
+      "loss": 1.0347,
       "step": 1967
     },
     {
-      "epoch": 0.6431372549019608,
-      "grad_norm": 1.2407008409500122,
-      "learning_rate": 5.96876570712028e-06,
-      "loss": 0.9599,
+      "epoch": 0.054049600395484884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996972568002459e-05,
+      "loss": 1.1183,
       "step": 1968
     },
     {
-      "epoch": 0.6434640522875817,
-      "grad_norm": 1.2795063257217407,
-      "learning_rate": 5.959081259848088e-06,
-      "loss": 0.9443,
+      "epoch": 0.05407706462332811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969656477073617e-05,
+      "loss": 1.0965,
       "step": 1969
     },
     {
-      "epoch": 0.6437908496732027,
-      "grad_norm": 1.3054927587509155,
-      "learning_rate": 5.94940134000886e-06,
-      "loss": 0.9609,
+      "epoch": 0.05410452885117135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996958719523886e-05,
+      "loss": 1.0128,
       "step": 1970
     },
     {
-      "epoch": 0.6441176470588236,
-      "grad_norm": 1.3237295150756836,
-      "learning_rate": 5.939725958447951e-06,
-      "loss": 0.9595,
+      "epoch": 0.05413199307901458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996951783452086e-05,
+      "loss": 1.0424,
       "step": 1971
     },
     {
-      "epoch": 0.6444444444444445,
-      "grad_norm": 1.3166065216064453,
-      "learning_rate": 5.93005512600563e-06,
-      "loss": 0.9412,
+      "epoch": 0.05415945730685782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996944839492018e-05,
+      "loss": 1.11,
       "step": 1972
     },
     {
-      "epoch": 0.6447712418300654,
-      "grad_norm": 1.3048123121261597,
-      "learning_rate": 5.920388853517068e-06,
-      "loss": 0.9311,
+      "epoch": 0.05418692153470105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996937887643736e-05,
+      "loss": 1.0423,
       "step": 1973
     },
     {
-      "epoch": 0.6450980392156863,
-      "grad_norm": 1.321239709854126,
-      "learning_rate": 5.910727151812324e-06,
-      "loss": 0.9548,
+      "epoch": 0.05421438576254429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969309279072955e-05,
+      "loss": 1.1295,
       "step": 1974
     },
     {
-      "epoch": 0.6454248366013072,
-      "grad_norm": 1.2679786682128906,
-      "learning_rate": 5.90107003171634e-06,
-      "loss": 0.9604,
+      "epoch": 0.054241849990387524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996923960282751e-05,
+      "loss": 1.1369,
       "step": 1975
     },
     {
-      "epoch": 0.6457516339869281,
-      "grad_norm": 1.2374626398086548,
-      "learning_rate": 5.891417504048926e-06,
-      "loss": 0.9399,
+      "epoch": 0.05426931421823075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969169847701582e-05,
+      "loss": 1.1417,
       "step": 1976
     },
     {
-      "epoch": 0.6460784313725491,
-      "grad_norm": 1.3411341905593872,
-      "learning_rate": 5.881769579624741e-06,
-      "loss": 0.9363,
+      "epoch": 0.05429677844607399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969100013695718e-05,
+      "loss": 1.0543,
       "step": 1977
     },
     {
-      "epoch": 0.64640522875817,
-      "grad_norm": 1.3259007930755615,
-      "learning_rate": 5.872126269253298e-06,
-      "loss": 0.9466,
+      "epoch": 0.05432424267391722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969030100810477e-05,
+      "loss": 1.2327,
       "step": 1978
     },
     {
-      "epoch": 0.6467320261437909,
-      "grad_norm": 1.3261858224868774,
-      "learning_rate": 5.862487583738921e-06,
-      "loss": 0.9352,
+      "epoch": 0.05435170690176046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968960109046404e-05,
+      "loss": 1.1285,
       "step": 1979
     },
     {
-      "epoch": 0.6470588235294118,
-      "grad_norm": 1.2464438676834106,
-      "learning_rate": 5.852853533880768e-06,
-      "loss": 0.9628,
+      "epoch": 0.05437917112960369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968890038404058e-05,
+      "loss": 1.0437,
       "step": 1980
     },
     {
-      "epoch": 0.6473856209150327,
-      "grad_norm": 1.2368334531784058,
-      "learning_rate": 5.8432241304728e-06,
-      "loss": 0.9334,
+      "epoch": 0.05440663535744693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968819888883994e-05,
+      "loss": 1.0317,
       "step": 1981
     },
     {
-      "epoch": 0.6477124183006536,
-      "grad_norm": 1.2783849239349365,
-      "learning_rate": 5.8335993843037695e-06,
-      "loss": 0.966,
+      "epoch": 0.054434099585290156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968749660486763e-05,
+      "loss": 1.093,
       "step": 1982
     },
     {
-      "epoch": 0.6480392156862745,
-      "grad_norm": 1.2768876552581787,
-      "learning_rate": 5.823979306157215e-06,
-      "loss": 0.9449,
+      "epoch": 0.05446156381313339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968679353212925e-05,
+      "loss": 1.0808,
       "step": 1983
     },
     {
-      "epoch": 0.6483660130718955,
-      "grad_norm": 1.2539700269699097,
-      "learning_rate": 5.8143639068114465e-06,
-      "loss": 0.929,
+      "epoch": 0.05448902804097663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968608967063034e-05,
+      "loss": 1.1489,
       "step": 1984
     },
     {
-      "epoch": 0.6486928104575164,
-      "grad_norm": 1.306246042251587,
-      "learning_rate": 5.804753197039519e-06,
-      "loss": 0.9823,
+      "epoch": 0.05451649226881986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968538502037652e-05,
+      "loss": 1.0541,
       "step": 1985
     },
     {
-      "epoch": 0.6490196078431373,
-      "grad_norm": 1.2558245658874512,
-      "learning_rate": 5.7951471876092445e-06,
-      "loss": 0.9222,
+      "epoch": 0.0545439564966631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968467958137324e-05,
+      "loss": 1.1199,
       "step": 1986
     },
     {
-      "epoch": 0.6493464052287582,
-      "grad_norm": 1.2939560413360596,
-      "learning_rate": 5.7855458892831706e-06,
-      "loss": 0.9624,
+      "epoch": 0.05457142072450633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968397335362623e-05,
+      "loss": 1.154,
       "step": 1987
     },
     {
-      "epoch": 0.6496732026143791,
-      "grad_norm": 1.2461154460906982,
-      "learning_rate": 5.7759493128185584e-06,
-      "loss": 0.9224,
+      "epoch": 0.05459888495234957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968326633714094e-05,
+      "loss": 1.0039,
       "step": 1988
     },
     {
-      "epoch": 0.65,
-      "grad_norm": 1.2774848937988281,
-      "learning_rate": 5.766357468967385e-06,
-      "loss": 0.9502,
+      "epoch": 0.054626349180192796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996825585319231e-05,
+      "loss": 0.9897,
       "step": 1989
     },
     {
-      "epoch": 0.6503267973856209,
-      "grad_norm": 1.2813912630081177,
-      "learning_rate": 5.756770368476326e-06,
-      "loss": 0.9172,
+      "epoch": 0.05465381340803603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996818499379782e-05,
+      "loss": 1.1892,
       "step": 1990
     },
     {
-      "epoch": 0.6506535947712418,
-      "grad_norm": 1.3252019882202148,
-      "learning_rate": 5.747188022086732e-06,
-      "loss": 0.928,
+      "epoch": 0.054681277635879266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996811405553119e-05,
+      "loss": 1.0601,
       "step": 1991
     },
     {
-      "epoch": 0.6509803921568628,
-      "grad_norm": 1.2251453399658203,
-      "learning_rate": 5.7376104405346376e-06,
-      "loss": 0.9572,
+      "epoch": 0.0547087418637225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968043038392978e-05,
+      "loss": 1.1167,
       "step": 1992
     },
     {
-      "epoch": 0.6513071895424837,
-      "grad_norm": 1.297636866569519,
-      "learning_rate": 5.728037634550738e-06,
-      "loss": 0.9151,
+      "epoch": 0.05473620609156574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996797194238375e-05,
+      "loss": 1.1521,
       "step": 1993
     },
     {
-      "epoch": 0.6516339869281046,
-      "grad_norm": 1.277099370956421,
-      "learning_rate": 5.7184696148603725e-06,
-      "loss": 0.9361,
+      "epoch": 0.05476367031940897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996790076750407e-05,
+      "loss": 1.0674,
       "step": 1994
     },
     {
-      "epoch": 0.6519607843137255,
-      "grad_norm": 1.2842209339141846,
-      "learning_rate": 5.708906392183527e-06,
-      "loss": 0.9416,
+      "epoch": 0.05479113454725221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967829513754497e-05,
+      "loss": 1.0182,
       "step": 1995
     },
     {
-      "epoch": 0.6522875816993464,
-      "grad_norm": 1.3014440536499023,
-      "learning_rate": 5.699347977234799e-06,
-      "loss": 0.9492,
+      "epoch": 0.054818598775095435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967758181135594e-05,
+      "loss": 1.0782,
       "step": 1996
     },
     {
-      "epoch": 0.6526143790849673,
-      "grad_norm": 1.2981011867523193,
-      "learning_rate": 5.68979438072341e-06,
-      "loss": 0.9181,
+      "epoch": 0.05484606300293867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967686769647926e-05,
+      "loss": 1.0376,
       "step": 1997
     },
     {
-      "epoch": 0.6529411764705882,
-      "grad_norm": 1.2902876138687134,
-      "learning_rate": 5.680245613353181e-06,
-      "loss": 0.9171,
+      "epoch": 0.054873527230781906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996761527929206e-05,
+      "loss": 1.1552,
       "step": 1998
     },
     {
-      "epoch": 0.6532679738562092,
-      "grad_norm": 1.34819495677948,
-      "learning_rate": 5.670701685822521e-06,
-      "loss": 0.9256,
+      "epoch": 0.05490099145862514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967543710068562e-05,
+      "loss": 1.1494,
       "step": 1999
     },
     {
-      "epoch": 0.6535947712418301,
-      "grad_norm": 1.249009132385254,
-      "learning_rate": 5.66116260882442e-06,
-      "loss": 0.9484,
+      "epoch": 0.054928455686468376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967472061978e-05,
+      "loss": 1.0996,
       "step": 2000
     },
     {
-      "epoch": 0.653921568627451,
-      "grad_norm": 1.2547657489776611,
-      "learning_rate": 5.651628393046427e-06,
-      "loss": 0.955,
+      "epoch": 0.05495591991431161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967400335020932e-05,
+      "loss": 1.1067,
       "step": 2001
     },
     {
-      "epoch": 0.6542483660130719,
-      "grad_norm": 1.1922270059585571,
-      "learning_rate": 5.642099049170651e-06,
-      "loss": 0.9366,
+      "epoch": 0.05498338414215485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967328529197933e-05,
+      "loss": 1.1536,
       "step": 2002
     },
     {
-      "epoch": 0.6545751633986928,
-      "grad_norm": 1.2661941051483154,
-      "learning_rate": 5.632574587873736e-06,
-      "loss": 0.9378,
+      "epoch": 0.055010848369998075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967256644509573e-05,
+      "loss": 1.1129,
       "step": 2003
     },
     {
-      "epoch": 0.6549019607843137,
-      "grad_norm": 1.2265642881393433,
-      "learning_rate": 5.623055019826862e-06,
-      "loss": 0.9234,
+      "epoch": 0.05503831259784131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967184680956414e-05,
+      "loss": 1.1011,
       "step": 2004
     },
     {
-      "epoch": 0.6552287581699346,
-      "grad_norm": 1.2926719188690186,
-      "learning_rate": 5.613540355695724e-06,
-      "loss": 0.9278,
+      "epoch": 0.055065776825684545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996711263853903e-05,
+      "loss": 1.1088,
       "step": 2005
     },
     {
-      "epoch": 0.6555555555555556,
-      "grad_norm": 1.2733089923858643,
-      "learning_rate": 5.604030606140522e-06,
-      "loss": 0.9506,
+      "epoch": 0.05509324105352778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996704051725799e-05,
+      "loss": 1.1873,
       "step": 2006
     },
     {
-      "epoch": 0.6558823529411765,
-      "grad_norm": 1.3303554058074951,
-      "learning_rate": 5.594525781815953e-06,
-      "loss": 0.9317,
+      "epoch": 0.055120705281371016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966968317113865e-05,
+      "loss": 1.0958,
       "step": 2007
     },
     {
-      "epoch": 0.6562091503267974,
-      "grad_norm": 1.2537962198257446,
-      "learning_rate": 5.585025893371187e-06,
-      "loss": 0.9416,
+      "epoch": 0.05514816950921425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966896038107226e-05,
+      "loss": 0.9865,
       "step": 2008
     },
     {
-      "epoch": 0.6565359477124183,
-      "grad_norm": 1.2984381914138794,
-      "learning_rate": 5.575530951449872e-06,
-      "loss": 0.9439,
+      "epoch": 0.05517563373705748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966823680238644e-05,
+      "loss": 1.036,
       "step": 2009
     },
     {
-      "epoch": 0.6568627450980392,
-      "grad_norm": 1.3022828102111816,
-      "learning_rate": 5.566040966690115e-06,
-      "loss": 0.9408,
+      "epoch": 0.055203097964900714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996675124350869e-05,
+      "loss": 1.0966,
       "step": 2010
     },
     {
-      "epoch": 0.6571895424836601,
-      "grad_norm": 1.2518284320831299,
-      "learning_rate": 5.556555949724463e-06,
-      "loss": 0.9405,
+      "epoch": 0.05523056219274395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966678727917943e-05,
+      "loss": 1.0232,
       "step": 2011
     },
     {
-      "epoch": 0.657516339869281,
-      "grad_norm": 1.2200355529785156,
-      "learning_rate": 5.547075911179902e-06,
-      "loss": 0.9277,
+      "epoch": 0.055258026420587185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966606133466974e-05,
+      "loss": 1.0867,
       "step": 2012
     },
     {
-      "epoch": 0.657843137254902,
-      "grad_norm": 1.2811434268951416,
-      "learning_rate": 5.537600861677835e-06,
-      "loss": 0.9438,
+      "epoch": 0.05528549064843042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996653346015635e-05,
+      "loss": 1.1513,
       "step": 2013
     },
     {
-      "epoch": 0.6581699346405229,
-      "grad_norm": 1.2770450115203857,
-      "learning_rate": 5.528130811834081e-06,
-      "loss": 0.9389,
+      "epoch": 0.055312954876273655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966460707986658e-05,
+      "loss": 1.1193,
       "step": 2014
     },
     {
-      "epoch": 0.6584967320261438,
-      "grad_norm": 1.3234399557113647,
-      "learning_rate": 5.518665772258851e-06,
-      "loss": 0.92,
+      "epoch": 0.05534041910411689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996638787695847e-05,
+      "loss": 1.0476,
       "step": 2015
     },
     {
-      "epoch": 0.6588235294117647,
-      "grad_norm": 1.3321138620376587,
-      "learning_rate": 5.50920575355675e-06,
-      "loss": 0.9536,
+      "epoch": 0.05536788333196012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966314967072356e-05,
+      "loss": 1.1219,
       "step": 2016
     },
     {
-      "epoch": 0.6591503267973856,
-      "grad_norm": 1.2285000085830688,
-      "learning_rate": 5.4997507663267546e-06,
-      "loss": 0.9427,
+      "epoch": 0.055395347559803354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99662419783289e-05,
+      "loss": 1.1198,
       "step": 2017
     },
     {
-      "epoch": 0.6594771241830065,
-      "grad_norm": 1.270706057548523,
-      "learning_rate": 5.490300821162204e-06,
-      "loss": 0.9597,
+      "epoch": 0.05542281178764659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966168910728677e-05,
+      "loss": 1.0192,
       "step": 2018
     },
     {
-      "epoch": 0.6598039215686274,
-      "grad_norm": 1.29932701587677,
-      "learning_rate": 5.480855928650784e-06,
-      "loss": 0.9372,
+      "epoch": 0.055450276015489824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966095764272263e-05,
+      "loss": 1.1472,
       "step": 2019
     },
     {
-      "epoch": 0.6601307189542484,
-      "grad_norm": 1.310821294784546,
-      "learning_rate": 5.471416099374525e-06,
-      "loss": 0.9531,
+      "epoch": 0.05547774024333306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966022538960242e-05,
+      "loss": 1.0216,
       "step": 2020
     },
     {
-      "epoch": 0.6604575163398693,
-      "grad_norm": 1.2434585094451904,
-      "learning_rate": 5.461981343909781e-06,
-      "loss": 0.9193,
+      "epoch": 0.055505204471176295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965949234793188e-05,
+      "loss": 1.0855,
       "step": 2021
     },
     {
-      "epoch": 0.6607843137254902,
-      "grad_norm": 1.2814429998397827,
-      "learning_rate": 5.452551672827227e-06,
-      "loss": 0.9369,
+      "epoch": 0.05553266869901953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965875851771683e-05,
+      "loss": 1.0442,
       "step": 2022
     },
     {
-      "epoch": 0.6611111111111111,
-      "grad_norm": 1.2687089443206787,
-      "learning_rate": 5.443127096691833e-06,
-      "loss": 0.9217,
+      "epoch": 0.05556013292686276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965802389896306e-05,
+      "loss": 1.1287,
       "step": 2023
     },
     {
-      "epoch": 0.661437908496732,
-      "grad_norm": 1.2956420183181763,
-      "learning_rate": 5.43370762606287e-06,
-      "loss": 0.9422,
+      "epoch": 0.055587597154705994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965728849167644e-05,
+      "loss": 1.0944,
       "step": 2024
     },
     {
-      "epoch": 0.6617647058823529,
-      "grad_norm": 1.3218458890914917,
-      "learning_rate": 5.424293271493881e-06,
-      "loss": 0.9119,
+      "epoch": 0.05561506138254923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965655229586272e-05,
+      "loss": 1.11,
       "step": 2025
     },
     {
-      "epoch": 0.6620915032679738,
-      "grad_norm": 1.301421046257019,
-      "learning_rate": 5.41488404353268e-06,
-      "loss": 0.9482,
+      "epoch": 0.055642525610392464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965581531152778e-05,
+      "loss": 1.0048,
       "step": 2026
     },
     {
-      "epoch": 0.6624183006535947,
-      "grad_norm": 1.2251275777816772,
-      "learning_rate": 5.405479952721339e-06,
-      "loss": 0.9528,
+      "epoch": 0.0556699898382357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965507753867742e-05,
+      "loss": 1.0468,
       "step": 2027
     },
     {
-      "epoch": 0.6627450980392157,
-      "grad_norm": 1.2208938598632812,
-      "learning_rate": 5.3960810095961705e-06,
-      "loss": 0.9586,
+      "epoch": 0.055697454066078934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996543389773175e-05,
+      "loss": 1.0591,
       "step": 2028
     },
     {
-      "epoch": 0.6630718954248366,
-      "grad_norm": 1.3645243644714355,
-      "learning_rate": 5.386687224687726e-06,
-      "loss": 0.9347,
+      "epoch": 0.05572491829392217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965359962745384e-05,
+      "loss": 1.1725,
       "step": 2029
     },
     {
-      "epoch": 0.6633986928104575,
-      "grad_norm": 1.2502869367599487,
-      "learning_rate": 5.377298608520767e-06,
-      "loss": 0.9306,
+      "epoch": 0.0557523825217654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965285948909224e-05,
+      "loss": 0.9329,
       "step": 2030
     },
     {
-      "epoch": 0.6637254901960784,
-      "grad_norm": 1.2618389129638672,
-      "learning_rate": 5.367915171614273e-06,
-      "loss": 0.9773,
+      "epoch": 0.05577984674960863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965211856223868e-05,
+      "loss": 1.0685,
       "step": 2031
     },
     {
-      "epoch": 0.6640522875816993,
-      "grad_norm": 1.3442938327789307,
-      "learning_rate": 5.358536924481416e-06,
-      "loss": 0.9236,
+      "epoch": 0.05580731097745187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965137684689892e-05,
+      "loss": 1.0846,
       "step": 2032
     },
     {
-      "epoch": 0.6643790849673202,
-      "grad_norm": 1.3153409957885742,
-      "learning_rate": 5.3491638776295595e-06,
-      "loss": 0.9759,
+      "epoch": 0.055834775205295104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996506343430789e-05,
+      "loss": 1.1438,
       "step": 2033
     },
     {
-      "epoch": 0.6647058823529411,
-      "grad_norm": 1.3087702989578247,
-      "learning_rate": 5.339796041560233e-06,
-      "loss": 0.9398,
+      "epoch": 0.05586223943313834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996498910507844e-05,
+      "loss": 0.9934,
       "step": 2034
     },
     {
-      "epoch": 0.6650326797385621,
-      "grad_norm": 1.2601157426834106,
-      "learning_rate": 5.330433426769137e-06,
-      "loss": 0.9175,
+      "epoch": 0.055889703660981574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964914697002142e-05,
+      "loss": 1.1197,
       "step": 2035
     },
     {
-      "epoch": 0.665359477124183,
-      "grad_norm": 1.3070228099822998,
-      "learning_rate": 5.321076043746108e-06,
-      "loss": 0.9095,
+      "epoch": 0.05591716788882481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996484021007958e-05,
+      "loss": 1.1024,
       "step": 2036
     },
     {
-      "epoch": 0.6656862745098039,
-      "grad_norm": 1.2797455787658691,
-      "learning_rate": 5.311723902975133e-06,
-      "loss": 0.9059,
+      "epoch": 0.05594463211666804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996476564431134e-05,
+      "loss": 1.123,
       "step": 2037
     },
     {
-      "epoch": 0.6660130718954248,
-      "grad_norm": 1.2125977277755737,
-      "learning_rate": 5.302377014934322e-06,
-      "loss": 0.9147,
+      "epoch": 0.05597209634451127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964690999698013e-05,
+      "loss": 1.1594,
       "step": 2038
     },
     {
-      "epoch": 0.6663398692810457,
-      "grad_norm": 1.321542739868164,
-      "learning_rate": 5.2930353900959e-06,
-      "loss": 0.9493,
+      "epoch": 0.05599956057235451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996461627624019e-05,
+      "loss": 1.1131,
       "step": 2039
     },
     {
-      "epoch": 0.6666666666666666,
-      "grad_norm": 1.280273675918579,
-      "learning_rate": 5.283699038926199e-06,
-      "loss": 0.9344,
+      "epoch": 0.05602702480019774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964541473938465e-05,
+      "loss": 1.0194,
       "step": 2040
     },
     {
-      "epoch": 0.6669934640522875,
-      "grad_norm": 1.3468877077102661,
-      "learning_rate": 5.274367971885629e-06,
-      "loss": 0.9393,
+      "epoch": 0.05605448902804098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964466592793425e-05,
+      "loss": 1.152,
       "step": 2041
     },
     {
-      "epoch": 0.6673202614379085,
-      "grad_norm": 1.3337552547454834,
-      "learning_rate": 5.265042199428693e-06,
-      "loss": 0.9444,
+      "epoch": 0.056081953255884213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996439163280567e-05,
+      "loss": 1.0764,
       "step": 2042
     },
     {
-      "epoch": 0.6676470588235294,
-      "grad_norm": 1.2955427169799805,
-      "learning_rate": 5.25572173200396e-06,
-      "loss": 0.9306,
+      "epoch": 0.05610941748372744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964316593975788e-05,
+      "loss": 1.1235,
       "step": 2043
     },
     {
-      "epoch": 0.6679738562091503,
-      "grad_norm": 1.256926417350769,
-      "learning_rate": 5.246406580054051e-06,
-      "loss": 0.9557,
+      "epoch": 0.05613688171157068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996424147630437e-05,
+      "loss": 1.0213,
       "step": 2044
     },
     {
-      "epoch": 0.6683006535947712,
-      "grad_norm": 1.2732112407684326,
-      "learning_rate": 5.237096754015635e-06,
-      "loss": 0.9258,
+      "epoch": 0.05616434593941391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964166279792016e-05,
+      "loss": 1.0699,
       "step": 2045
     },
     {
-      "epoch": 0.6686274509803921,
-      "grad_norm": 1.2502480745315552,
-      "learning_rate": 5.227792264319411e-06,
-      "loss": 0.9434,
+      "epoch": 0.05619181016725715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996409100443932e-05,
+      "loss": 1.1033,
       "step": 2046
     },
     {
-      "epoch": 0.668954248366013,
-      "grad_norm": 1.291792869567871,
-      "learning_rate": 5.218493121390103e-06,
-      "loss": 0.9444,
+      "epoch": 0.05621927439510038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964015650246873e-05,
+      "loss": 1.0896,
       "step": 2047
     },
     {
-      "epoch": 0.6692810457516339,
-      "grad_norm": 1.2806559801101685,
-      "learning_rate": 5.20919933564644e-06,
-      "loss": 0.969,
+      "epoch": 0.05624673862294362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963940217215273e-05,
+      "loss": 1.1648,
       "step": 2048
     },
     {
-      "epoch": 0.6696078431372549,
-      "grad_norm": 1.251795768737793,
-      "learning_rate": 5.199910917501151e-06,
-      "loss": 0.9387,
+      "epoch": 0.05627420285078685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963864705345123e-05,
+      "loss": 1.0358,
       "step": 2049
     },
     {
-      "epoch": 0.6699346405228758,
-      "grad_norm": 1.2040321826934814,
-      "learning_rate": 5.190627877360953e-06,
-      "loss": 0.9172,
+      "epoch": 0.05630166707863008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963789114637016e-05,
+      "loss": 1.0991,
       "step": 2050
     },
     {
-      "epoch": 0.6702614379084967,
-      "grad_norm": 1.332192063331604,
-      "learning_rate": 5.181350225626538e-06,
-      "loss": 0.9411,
+      "epoch": 0.056329131306473317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963713445091547e-05,
+      "loss": 1.1445,
       "step": 2051
     },
     {
-      "epoch": 0.6705882352941176,
-      "grad_norm": 1.340325951576233,
-      "learning_rate": 5.172077972692553e-06,
-      "loss": 0.9726,
+      "epoch": 0.05635659553431655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963637696709317e-05,
+      "loss": 1.1793,
       "step": 2052
     },
     {
-      "epoch": 0.6709150326797385,
-      "grad_norm": 1.3411011695861816,
-      "learning_rate": 5.1628111289476025e-06,
-      "loss": 0.9639,
+      "epoch": 0.05638405976215979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963561869490928e-05,
+      "loss": 1.1923,
       "step": 2053
     },
     {
-      "epoch": 0.6712418300653594,
-      "grad_norm": 1.2233967781066895,
-      "learning_rate": 5.153549704774232e-06,
-      "loss": 0.9253,
+      "epoch": 0.05641152399000302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996348596343698e-05,
+      "loss": 1.1155,
       "step": 2054
     },
     {
-      "epoch": 0.6715686274509803,
-      "grad_norm": 1.2887128591537476,
-      "learning_rate": 5.144293710548913e-06,
-      "loss": 0.9146,
+      "epoch": 0.05643898821784626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963409978548067e-05,
+      "loss": 1.1576,
       "step": 2055
     },
     {
-      "epoch": 0.6718954248366014,
-      "grad_norm": 1.2739453315734863,
-      "learning_rate": 5.1350431566420326e-06,
-      "loss": 0.9544,
+      "epoch": 0.05646645244568949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963333914824798e-05,
+      "loss": 1.2032,
       "step": 2056
     },
     {
-      "epoch": 0.6722222222222223,
-      "grad_norm": 1.2394267320632935,
-      "learning_rate": 5.125798053417885e-06,
-      "loss": 0.9348,
+      "epoch": 0.05649391667353272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996325777226777e-05,
+      "loss": 1.1024,
       "step": 2057
     },
     {
-      "epoch": 0.6725490196078432,
-      "grad_norm": 1.2548054456710815,
-      "learning_rate": 5.116558411234651e-06,
-      "loss": 0.9328,
+      "epoch": 0.056521380901375956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963181550877588e-05,
+      "loss": 1.1315,
       "step": 2058
     },
     {
-      "epoch": 0.6728758169934641,
-      "grad_norm": 1.29739248752594,
-      "learning_rate": 5.107324240444398e-06,
-      "loss": 0.9479,
+      "epoch": 0.05654884512921919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963105250654853e-05,
+      "loss": 1.0534,
       "step": 2059
     },
     {
-      "epoch": 0.673202614379085,
-      "grad_norm": 1.2666230201721191,
-      "learning_rate": 5.098095551393066e-06,
-      "loss": 0.9317,
+      "epoch": 0.056576309357062426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996302887160017e-05,
+      "loss": 1.0356,
       "step": 2060
     },
     {
-      "epoch": 0.6735294117647059,
-      "grad_norm": 1.2312803268432617,
-      "learning_rate": 5.088872354420447e-06,
-      "loss": 0.9168,
+      "epoch": 0.05660377358490566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962952413714145e-05,
+      "loss": 0.9467,
       "step": 2061
     },
     {
-      "epoch": 0.6738562091503268,
-      "grad_norm": 1.2918168306350708,
-      "learning_rate": 5.079654659860183e-06,
-      "loss": 0.9156,
+      "epoch": 0.0566312378127489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996287587699738e-05,
+      "loss": 1.0452,
       "step": 2062
     },
     {
-      "epoch": 0.6741830065359478,
-      "grad_norm": 1.310848355293274,
-      "learning_rate": 5.070442478039755e-06,
-      "loss": 0.9512,
+      "epoch": 0.05665870204059213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962799261450484e-05,
+      "loss": 1.0857,
       "step": 2063
     },
     {
-      "epoch": 0.6745098039215687,
-      "grad_norm": 1.3425322771072388,
-      "learning_rate": 5.061235819280456e-06,
-      "loss": 0.9348,
+      "epoch": 0.05668616626843536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996272256707406e-05,
+      "loss": 1.059,
       "step": 2064
     },
     {
-      "epoch": 0.6748366013071896,
-      "grad_norm": 1.2287566661834717,
-      "learning_rate": 5.052034693897404e-06,
-      "loss": 0.9146,
+      "epoch": 0.056713630496278596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996264579386871e-05,
+      "loss": 1.1174,
       "step": 2065
     },
     {
-      "epoch": 0.6751633986928105,
-      "grad_norm": 1.3022691011428833,
-      "learning_rate": 5.042839112199509e-06,
-      "loss": 0.9032,
+      "epoch": 0.05674109472412183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962568941835053e-05,
+      "loss": 1.1178,
       "step": 2066
     },
     {
-      "epoch": 0.6754901960784314,
-      "grad_norm": 1.2883180379867554,
-      "learning_rate": 5.0336490844894725e-06,
-      "loss": 0.9331,
+      "epoch": 0.056768558951965066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962492010973694e-05,
+      "loss": 1.1415,
       "step": 2067
     },
     {
-      "epoch": 0.6758169934640523,
-      "grad_norm": 1.2419108152389526,
-      "learning_rate": 5.024464621063773e-06,
-      "loss": 0.9339,
+      "epoch": 0.0567960231798083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962415001285235e-05,
+      "loss": 1.0,
       "step": 2068
     },
     {
-      "epoch": 0.6761437908496732,
-      "grad_norm": 1.2996271848678589,
-      "learning_rate": 5.015285732212658e-06,
-      "loss": 0.9354,
+      "epoch": 0.056823487407651536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996233791277029e-05,
+      "loss": 1.1223,
       "step": 2069
     },
     {
-      "epoch": 0.6764705882352942,
-      "grad_norm": 1.2840467691421509,
-      "learning_rate": 5.006112428220124e-06,
-      "loss": 0.9724,
+      "epoch": 0.056850951635494765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962260745429466e-05,
+      "loss": 1.097,
       "step": 2070
     },
     {
-      "epoch": 0.6767973856209151,
-      "grad_norm": 1.2778745889663696,
-      "learning_rate": 4.996944719363914e-06,
-      "loss": 0.9182,
+      "epoch": 0.056878415863338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996218349926338e-05,
+      "loss": 1.0066,
       "step": 2071
     },
     {
-      "epoch": 0.677124183006536,
-      "grad_norm": 1.233568549156189,
-      "learning_rate": 4.9877826159154995e-06,
-      "loss": 0.9151,
+      "epoch": 0.056905880091181235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962106174272637e-05,
+      "loss": 1.1251,
       "step": 2072
     },
     {
-      "epoch": 0.6774509803921569,
-      "grad_norm": 1.2906852960586548,
-      "learning_rate": 4.978626128140074e-06,
-      "loss": 0.9393,
+      "epoch": 0.05693334431902447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996202877045785e-05,
+      "loss": 1.0867,
       "step": 2073
     },
     {
-      "epoch": 0.6777777777777778,
-      "grad_norm": 1.261096477508545,
-      "learning_rate": 4.9694752662965415e-06,
-      "loss": 0.9266,
+      "epoch": 0.056960808546867706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961951287819636e-05,
+      "loss": 1.0814,
       "step": 2074
     },
     {
-      "epoch": 0.6781045751633987,
-      "grad_norm": 1.3285647630691528,
-      "learning_rate": 4.960330040637493e-06,
-      "loss": 0.9285,
+      "epoch": 0.05698827277471094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961873726358605e-05,
+      "loss": 1.0876,
       "step": 2075
     },
     {
-      "epoch": 0.6784313725490196,
-      "grad_norm": 1.3131134510040283,
-      "learning_rate": 4.951190461409214e-06,
-      "loss": 0.9506,
+      "epoch": 0.057015737002554176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961796086075368e-05,
+      "loss": 1.1146,
       "step": 2076
     },
     {
-      "epoch": 0.6787581699346406,
-      "grad_norm": 1.2469526529312134,
-      "learning_rate": 4.942056538851661e-06,
-      "loss": 0.8988,
+      "epoch": 0.057043201230397404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996171836697054e-05,
+      "loss": 1.0302,
       "step": 2077
     },
     {
-      "epoch": 0.6790849673202615,
-      "grad_norm": 1.3948842287063599,
-      "learning_rate": 4.932928283198454e-06,
-      "loss": 0.9204,
+      "epoch": 0.05707066545824064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996164056904474e-05,
+      "loss": 1.0464,
       "step": 2078
     },
     {
-      "epoch": 0.6794117647058824,
-      "grad_norm": 1.4249918460845947,
-      "learning_rate": 4.923805704676862e-06,
-      "loss": 0.9449,
+      "epoch": 0.057098129686083875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996156269229858e-05,
+      "loss": 1.1027,
       "step": 2079
     },
     {
-      "epoch": 0.6797385620915033,
-      "grad_norm": 1.274198055267334,
-      "learning_rate": 4.914688813507798e-06,
-      "loss": 0.9275,
+      "epoch": 0.05712559391392711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961484736732677e-05,
+      "loss": 1.1761,
       "step": 2080
     },
     {
-      "epoch": 0.6800653594771242,
-      "grad_norm": 1.3410102128982544,
-      "learning_rate": 4.90557761990579e-06,
-      "loss": 0.9108,
+      "epoch": 0.057153058141770345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961406702347646e-05,
+      "loss": 1.1399,
       "step": 2081
     },
     {
-      "epoch": 0.6803921568627451,
-      "grad_norm": 1.3644161224365234,
-      "learning_rate": 4.896472134078998e-06,
-      "loss": 0.94,
+      "epoch": 0.05718052236961358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961328589144112e-05,
+      "loss": 1.0562,
       "step": 2082
     },
     {
-      "epoch": 0.680718954248366,
-      "grad_norm": 1.2575759887695312,
-      "learning_rate": 4.887372366229177e-06,
-      "loss": 0.9682,
+      "epoch": 0.057207986597456816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961250397122684e-05,
+      "loss": 1.1214,
       "step": 2083
     },
     {
-      "epoch": 0.681045751633987,
-      "grad_norm": 1.3028315305709839,
-      "learning_rate": 4.878278326551682e-06,
-      "loss": 0.9447,
+      "epoch": 0.057235450825300044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961172126283984e-05,
+      "loss": 1.0846,
       "step": 2084
     },
     {
-      "epoch": 0.6813725490196079,
-      "grad_norm": 1.2630622386932373,
-      "learning_rate": 4.8691900252354485e-06,
-      "loss": 0.9141,
+      "epoch": 0.05726291505314328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996109377662863e-05,
+      "loss": 1.1302,
       "step": 2085
     },
     {
-      "epoch": 0.6816993464052288,
-      "grad_norm": 1.2724767923355103,
-      "learning_rate": 4.860107472462976e-06,
-      "loss": 0.927,
+      "epoch": 0.057290379280986514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961015348157245e-05,
+      "loss": 1.1644,
       "step": 2086
     },
     {
-      "epoch": 0.6820261437908497,
-      "grad_norm": 1.2285809516906738,
-      "learning_rate": 4.851030678410331e-06,
-      "loss": 0.8983,
+      "epoch": 0.05731784350882975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996093684087045e-05,
+      "loss": 1.1212,
       "step": 2087
     },
     {
-      "epoch": 0.6823529411764706,
-      "grad_norm": 1.4442353248596191,
-      "learning_rate": 4.8419596532471266e-06,
-      "loss": 0.9152,
+      "epoch": 0.057345307736672985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960858254768865e-05,
+      "loss": 1.0677,
       "step": 2088
     },
     {
-      "epoch": 0.6826797385620915,
-      "grad_norm": 1.2724831104278564,
-      "learning_rate": 4.83289440713651e-06,
-      "loss": 0.9147,
+      "epoch": 0.05737277196451622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996077958985311e-05,
+      "loss": 1.1179,
       "step": 2089
     },
     {
-      "epoch": 0.6830065359477124,
-      "grad_norm": 1.3054497241973877,
-      "learning_rate": 4.823834950235158e-06,
-      "loss": 0.9234,
+      "epoch": 0.057400236192359455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960700846123803e-05,
+      "loss": 1.0602,
       "step": 2090
     },
     {
-      "epoch": 0.6833333333333333,
-      "grad_norm": 1.3336055278778076,
-      "learning_rate": 4.814781292693256e-06,
-      "loss": 0.9433,
+      "epoch": 0.05742770042020268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996062202358158e-05,
+      "loss": 1.092,
       "step": 2091
     },
     {
-      "epoch": 0.6836601307189543,
-      "grad_norm": 1.246656894683838,
-      "learning_rate": 4.805733444654496e-06,
-      "loss": 0.9842,
+      "epoch": 0.05745516464804592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996054312222705e-05,
+      "loss": 1.1054,
       "step": 2092
     },
     {
-      "epoch": 0.6839869281045752,
-      "grad_norm": 1.2881683111190796,
-      "learning_rate": 4.796691416256056e-06,
-      "loss": 0.8957,
+      "epoch": 0.057482628875889154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996046414206085e-05,
+      "loss": 1.1282,
       "step": 2093
     },
     {
-      "epoch": 0.6843137254901961,
-      "grad_norm": 1.296237826347351,
-      "learning_rate": 4.7876552176286005e-06,
-      "loss": 0.9254,
+      "epoch": 0.05751009310373239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99603850830836e-05,
+      "loss": 1.111,
       "step": 2094
     },
     {
-      "epoch": 0.684640522875817,
-      "grad_norm": 1.2583386898040771,
-      "learning_rate": 4.778624858896257e-06,
-      "loss": 0.9498,
+      "epoch": 0.057537557331575624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996030594529592e-05,
+      "loss": 1.1931,
       "step": 2095
     },
     {
-      "epoch": 0.6849673202614379,
-      "grad_norm": 1.2192224264144897,
-      "learning_rate": 4.7696003501766155e-06,
-      "loss": 0.9705,
+      "epoch": 0.05756502155941886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960226728698447e-05,
+      "loss": 1.1327,
       "step": 2096
     },
     {
-      "epoch": 0.6852941176470588,
-      "grad_norm": 1.2473993301391602,
-      "learning_rate": 4.7605817015807e-06,
-      "loss": 0.9447,
+      "epoch": 0.05759248578726209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99601474332918e-05,
+      "loss": 1.1391,
       "step": 2097
     },
     {
-      "epoch": 0.6856209150326797,
-      "grad_norm": 1.3113734722137451,
-      "learning_rate": 4.751568923212983e-06,
-      "loss": 0.9316,
+      "epoch": 0.05761995001510532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960068059076606e-05,
+      "loss": 1.0922,
       "step": 2098
     },
     {
-      "epoch": 0.6859477124183007,
-      "grad_norm": 1.409050464630127,
-      "learning_rate": 4.74256202517135e-06,
-      "loss": 0.9574,
+      "epoch": 0.05764741424294856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959988606053496e-05,
+      "loss": 1.1633,
       "step": 2099
     },
     {
-      "epoch": 0.6862745098039216,
-      "grad_norm": 1.292858362197876,
-      "learning_rate": 4.733561017547104e-06,
-      "loss": 0.902,
+      "epoch": 0.05767487847079179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959909074223098e-05,
+      "loss": 1.194,
       "step": 2100
     },
     {
-      "epoch": 0.6866013071895425,
-      "grad_norm": 1.2944626808166504,
-      "learning_rate": 4.724565910424946e-06,
-      "loss": 0.9462,
+      "epoch": 0.05770234269863503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959829463586042e-05,
+      "loss": 1.171,
       "step": 2101
     },
     {
-      "epoch": 0.6869281045751634,
-      "grad_norm": 1.269584059715271,
-      "learning_rate": 4.715576713882971e-06,
-      "loss": 0.9493,
+      "epoch": 0.057729806926478264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995974977414296e-05,
+      "loss": 1.1057,
       "step": 2102
     },
     {
-      "epoch": 0.6872549019607843,
-      "grad_norm": 1.28876793384552,
-      "learning_rate": 4.706593437992639e-06,
-      "loss": 0.9104,
+      "epoch": 0.0577572711543215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959670005894474e-05,
+      "loss": 1.1126,
       "step": 2103
     },
     {
-      "epoch": 0.6875816993464052,
-      "grad_norm": 1.3064723014831543,
-      "learning_rate": 4.697616092818789e-06,
-      "loss": 0.9524,
+      "epoch": 0.05778473538216473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959590158841223e-05,
+      "loss": 1.1894,
       "step": 2104
     },
     {
-      "epoch": 0.6879084967320261,
-      "grad_norm": 1.2883776426315308,
-      "learning_rate": 4.688644688419609e-06,
-      "loss": 0.9105,
+      "epoch": 0.05781219961000796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959510232983837e-05,
+      "loss": 1.0307,
       "step": 2105
     },
     {
-      "epoch": 0.6882352941176471,
-      "grad_norm": 1.2635353803634644,
-      "learning_rate": 4.679679234846636e-06,
-      "loss": 0.9066,
+      "epoch": 0.0578396638378512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959430228322948e-05,
+      "loss": 1.1527,
       "step": 2106
     },
     {
-      "epoch": 0.688562091503268,
-      "grad_norm": 1.251063585281372,
-      "learning_rate": 4.670719742144734e-06,
-      "loss": 0.9298,
+      "epoch": 0.05786712806569443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959350144859192e-05,
+      "loss": 1.1478,
       "step": 2107
     },
     {
-      "epoch": 0.6888888888888889,
-      "grad_norm": 1.274909496307373,
-      "learning_rate": 4.661766220352098e-06,
-      "loss": 0.9397,
+      "epoch": 0.05789459229353767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959269982593197e-05,
+      "loss": 1.0979,
       "step": 2108
     },
     {
-      "epoch": 0.6892156862745098,
-      "grad_norm": 1.3359507322311401,
-      "learning_rate": 4.652818679500215e-06,
-      "loss": 0.9323,
+      "epoch": 0.0579220565213809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99591897415256e-05,
+      "loss": 1.166,
       "step": 2109
     },
     {
-      "epoch": 0.6895424836601307,
-      "grad_norm": 1.350987434387207,
-      "learning_rate": 4.643877129613889e-06,
-      "loss": 0.9779,
+      "epoch": 0.05794952074922414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959109421657038e-05,
+      "loss": 1.1562,
       "step": 2110
     },
     {
-      "epoch": 0.6898692810457516,
-      "grad_norm": 1.265624761581421,
-      "learning_rate": 4.634941580711204e-06,
-      "loss": 0.9384,
+      "epoch": 0.05797698497706737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995902902298814e-05,
+      "loss": 1.1491,
       "step": 2111
     },
     {
-      "epoch": 0.6901960784313725,
-      "grad_norm": 1.2385672330856323,
-      "learning_rate": 4.626012042803521e-06,
-      "loss": 0.9616,
+      "epoch": 0.0580044492049106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995894854551955e-05,
+      "loss": 1.1004,
       "step": 2112
     },
     {
-      "epoch": 0.6905228758169935,
-      "grad_norm": 1.255768895149231,
-      "learning_rate": 4.61708852589547e-06,
-      "loss": 0.9091,
+      "epoch": 0.05803191343275384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958867989251903e-05,
+      "loss": 1.08,
       "step": 2113
     },
     {
-      "epoch": 0.6908496732026144,
-      "grad_norm": 1.2002631425857544,
-      "learning_rate": 4.608171039984928e-06,
-      "loss": 0.9533,
+      "epoch": 0.05805937766059707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995878735418583e-05,
+      "loss": 1.126,
       "step": 2114
     },
     {
-      "epoch": 0.6911764705882353,
-      "grad_norm": 1.2909184694290161,
-      "learning_rate": 4.59925959506302e-06,
-      "loss": 0.9463,
+      "epoch": 0.05808684188844031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995870664032198e-05,
+      "loss": 1.1005,
       "step": 2115
     },
     {
-      "epoch": 0.6915032679738562,
-      "grad_norm": 1.3141326904296875,
-      "learning_rate": 4.590354201114103e-06,
-      "loss": 0.9754,
+      "epoch": 0.05811430611628354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995862584766098e-05,
+      "loss": 1.0972,
       "step": 2116
     },
     {
-      "epoch": 0.6918300653594771,
-      "grad_norm": 1.2583376169204712,
-      "learning_rate": 4.581454868115752e-06,
-      "loss": 0.9229,
+      "epoch": 0.05814177034412678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958544976203475e-05,
+      "loss": 0.9771,
       "step": 2117
     },
     {
-      "epoch": 0.692156862745098,
-      "grad_norm": 1.2764437198638916,
-      "learning_rate": 4.5725616060387525e-06,
-      "loss": 0.9687,
+      "epoch": 0.058169234571970006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958464025950105e-05,
+      "loss": 1.0648,
       "step": 2118
     },
     {
-      "epoch": 0.6924836601307189,
-      "grad_norm": 1.2724894285202026,
-      "learning_rate": 4.563674424847092e-06,
-      "loss": 0.9553,
+      "epoch": 0.05819669879981324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995838299690151e-05,
+      "loss": 1.0837,
       "step": 2119
     },
     {
-      "epoch": 0.6928104575163399,
-      "grad_norm": 1.2915507555007935,
-      "learning_rate": 4.554793334497933e-06,
-      "loss": 0.9513,
+      "epoch": 0.05822416302765648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958301889058336e-05,
+      "loss": 1.0826,
       "step": 2120
     },
     {
-      "epoch": 0.6931372549019608,
-      "grad_norm": 1.2594969272613525,
-      "learning_rate": 4.545918344941627e-06,
-      "loss": 0.9461,
+      "epoch": 0.05825162725549971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958220702421217e-05,
+      "loss": 1.0681,
       "step": 2121
     },
     {
-      "epoch": 0.6934640522875817,
-      "grad_norm": 1.2576066255569458,
-      "learning_rate": 4.5370494661216835e-06,
-      "loss": 0.9273,
+      "epoch": 0.05827909148334295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958139436990794e-05,
+      "loss": 1.0468,
       "step": 2122
     },
     {
-      "epoch": 0.6937908496732026,
-      "grad_norm": 1.3197733163833618,
-      "learning_rate": 4.5281867079747675e-06,
-      "loss": 0.9654,
+      "epoch": 0.05830655571118618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995805809276772e-05,
+      "loss": 1.1191,
       "step": 2123
     },
     {
-      "epoch": 0.6941176470588235,
-      "grad_norm": 1.2651792764663696,
-      "learning_rate": 4.519330080430687e-06,
-      "loss": 0.9604,
+      "epoch": 0.05833401993902942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957976669752633e-05,
+      "loss": 1.081,
       "step": 2124
     },
     {
-      "epoch": 0.6944444444444444,
-      "grad_norm": 1.23296320438385,
-      "learning_rate": 4.510479593412383e-06,
-      "loss": 0.9494,
+      "epoch": 0.058361484166872646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957895167946173e-05,
+      "loss": 1.102,
       "step": 2125
     },
     {
-      "epoch": 0.6947712418300653,
-      "grad_norm": 1.2053828239440918,
-      "learning_rate": 4.501635256835908e-06,
-      "loss": 0.9467,
+      "epoch": 0.05838894839471588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957813587348993e-05,
+      "loss": 1.1228,
       "step": 2126
     },
     {
-      "epoch": 0.6950980392156862,
-      "grad_norm": 1.2530354261398315,
-      "learning_rate": 4.492797080610431e-06,
-      "loss": 0.9617,
+      "epoch": 0.058416412622559116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957731927961733e-05,
+      "loss": 1.0013,
       "step": 2127
     },
     {
-      "epoch": 0.6954248366013072,
-      "grad_norm": 2.485301971435547,
-      "learning_rate": 4.483965074638219e-06,
-      "loss": 0.9375,
+      "epoch": 0.05844387685040235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957650189785037e-05,
+      "loss": 1.1611,
       "step": 2128
     },
     {
-      "epoch": 0.6957516339869281,
-      "grad_norm": 1.2692816257476807,
-      "learning_rate": 4.475139248814625e-06,
-      "loss": 0.9647,
+      "epoch": 0.05847134107824559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957568372819562e-05,
+      "loss": 1.1725,
       "step": 2129
     },
     {
-      "epoch": 0.696078431372549,
-      "grad_norm": 1.2615432739257812,
-      "learning_rate": 4.46631961302808e-06,
-      "loss": 0.9551,
+      "epoch": 0.05849880530608882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957486477065946e-05,
+      "loss": 1.0687,
       "step": 2130
     },
     {
-      "epoch": 0.6964052287581699,
-      "grad_norm": 1.3874883651733398,
-      "learning_rate": 4.457506177160069e-06,
-      "loss": 0.9959,
+      "epoch": 0.05852626953393205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995740450252484e-05,
+      "loss": 1.1572,
       "step": 2131
     },
     {
-      "epoch": 0.6967320261437908,
-      "grad_norm": 1.249599575996399,
-      "learning_rate": 4.448698951085143e-06,
-      "loss": 0.9598,
+      "epoch": 0.058553733761775285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957322449196893e-05,
+      "loss": 1.0721,
       "step": 2132
     },
     {
-      "epoch": 0.6970588235294117,
-      "grad_norm": 1.27445387840271,
-      "learning_rate": 4.439897944670888e-06,
-      "loss": 0.9535,
+      "epoch": 0.05858119798961852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957240317082754e-05,
+      "loss": 1.0245,
       "step": 2133
     },
     {
-      "epoch": 0.6973856209150326,
-      "grad_norm": 1.2119786739349365,
-      "learning_rate": 4.43110316777793e-06,
-      "loss": 0.9521,
+      "epoch": 0.058608662217461756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995715810618307e-05,
+      "loss": 0.9688,
       "step": 2134
     },
     {
-      "epoch": 0.6977124183006536,
-      "grad_norm": 1.1981520652770996,
-      "learning_rate": 4.422314630259909e-06,
-      "loss": 0.9254,
+      "epoch": 0.05863612644530499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957075816498498e-05,
+      "loss": 1.1148,
       "step": 2135
     },
     {
-      "epoch": 0.6980392156862745,
-      "grad_norm": 1.2601613998413086,
-      "learning_rate": 4.413532341963477e-06,
-      "loss": 0.9367,
+      "epoch": 0.058663590673148226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995699344802968e-05,
+      "loss": 1.1845,
       "step": 2136
     },
     {
-      "epoch": 0.6983660130718954,
-      "grad_norm": 1.3326412439346313,
-      "learning_rate": 4.404756312728275e-06,
-      "loss": 0.961,
+      "epoch": 0.05869105490099146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956911000777273e-05,
+      "loss": 1.1538,
       "step": 2137
     },
     {
-      "epoch": 0.6986928104575163,
-      "grad_norm": 1.2742072343826294,
-      "learning_rate": 4.395986552386947e-06,
-      "loss": 0.936,
+      "epoch": 0.05871851912883469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956828474741935e-05,
+      "loss": 0.9809,
       "step": 2138
     },
     {
-      "epoch": 0.6990196078431372,
-      "grad_norm": 1.2486897706985474,
-      "learning_rate": 4.387223070765101e-06,
-      "loss": 0.9619,
+      "epoch": 0.058745983356677925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995674586992431e-05,
+      "loss": 1.0445,
       "step": 2139
     },
     {
-      "epoch": 0.6993464052287581,
-      "grad_norm": 1.3668092489242554,
-      "learning_rate": 4.378465877681317e-06,
-      "loss": 0.9209,
+      "epoch": 0.05877344758452116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956663186325053e-05,
+      "loss": 1.1552,
       "step": 2140
     },
     {
-      "epoch": 0.699673202614379,
-      "grad_norm": 1.3340882062911987,
-      "learning_rate": 4.36971498294713e-06,
-      "loss": 0.9748,
+      "epoch": 0.058800911812364395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956580423944825e-05,
+      "loss": 1.1414,
       "step": 2141
     },
     {
-      "epoch": 0.7,
-      "grad_norm": 1.314589500427246,
-      "learning_rate": 4.360970396367012e-06,
-      "loss": 0.9625,
+      "epoch": 0.05882837604020763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995649758278427e-05,
+      "loss": 1.0768,
       "step": 2142
     },
     {
-      "epoch": 0.7003267973856209,
-      "grad_norm": 1.335323691368103,
-      "learning_rate": 4.3522321277383686e-06,
-      "loss": 0.9458,
+      "epoch": 0.058855840268050866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956414662844053e-05,
+      "loss": 1.0857,
       "step": 2143
     },
     {
-      "epoch": 0.7006535947712418,
-      "grad_norm": 1.3030949831008911,
-      "learning_rate": 4.343500186851534e-06,
-      "loss": 0.9442,
+      "epoch": 0.0588833044958941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956331664124824e-05,
+      "loss": 1.0384,
       "step": 2144
     },
     {
-      "epoch": 0.7009803921568627,
-      "grad_norm": 1.3078614473342896,
-      "learning_rate": 4.3347745834897445e-06,
-      "loss": 0.9399,
+      "epoch": 0.05891076872373733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956248586627245e-05,
+      "loss": 1.2621,
       "step": 2145
     },
     {
-      "epoch": 0.7013071895424836,
-      "grad_norm": 1.303547978401184,
-      "learning_rate": 4.326055327429143e-06,
-      "loss": 0.9766,
+      "epoch": 0.058938232951580564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995616543035197e-05,
+      "loss": 1.1174,
       "step": 2146
     },
     {
-      "epoch": 0.7016339869281045,
-      "grad_norm": 1.3770030736923218,
-      "learning_rate": 4.3173424284387575e-06,
-      "loss": 0.9476,
+      "epoch": 0.0589656971794238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956082195299656e-05,
+      "loss": 1.0029,
       "step": 2147
     },
     {
-      "epoch": 0.7019607843137254,
-      "grad_norm": 1.2833129167556763,
-      "learning_rate": 4.3086358962804885e-06,
-      "loss": 0.9164,
+      "epoch": 0.058993161407267035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955998881470964e-05,
+      "loss": 1.0051,
       "step": 2148
     },
     {
-      "epoch": 0.7022875816993464,
-      "grad_norm": 1.2867724895477295,
-      "learning_rate": 4.29993574070911e-06,
-      "loss": 0.905,
+      "epoch": 0.05902062563511027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995591548886655e-05,
+      "loss": 1.0331,
       "step": 2149
     },
     {
-      "epoch": 0.7026143790849673,
-      "grad_norm": 1.2636572122573853,
-      "learning_rate": 4.29124197147225e-06,
-      "loss": 0.9073,
+      "epoch": 0.059048089862953505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995583201748708e-05,
+      "loss": 1.1311,
       "step": 2150
     },
     {
-      "epoch": 0.7029411764705882,
-      "grad_norm": 1.26131010055542,
-      "learning_rate": 4.282554598310382e-06,
-      "loss": 0.9392,
+      "epoch": 0.05907555409079674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955748467333207e-05,
+      "loss": 1.1152,
       "step": 2151
     },
     {
-      "epoch": 0.7032679738562092,
-      "grad_norm": 1.270751953125,
-      "learning_rate": 4.273873630956811e-06,
-      "loss": 0.9312,
+      "epoch": 0.05910301831863997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99556648384056e-05,
+      "loss": 1.0098,
       "step": 2152
     },
     {
-      "epoch": 0.7035947712418301,
-      "grad_norm": 1.3046401739120483,
-      "learning_rate": 4.2651990791376666e-06,
-      "loss": 0.9527,
+      "epoch": 0.059130482546483204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955581130704913e-05,
+      "loss": 1.0768,
       "step": 2153
     },
     {
-      "epoch": 0.703921568627451,
-      "grad_norm": 1.2530076503753662,
-      "learning_rate": 4.256530952571889e-06,
-      "loss": 0.908,
+      "epoch": 0.05915794677432644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955497344231812e-05,
+      "loss": 1.1131,
       "step": 2154
     },
     {
-      "epoch": 0.704248366013072,
-      "grad_norm": 1.232655644416809,
-      "learning_rate": 4.247869260971223e-06,
-      "loss": 0.9172,
+      "epoch": 0.059185411002169674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995541347898696e-05,
+      "loss": 1.0433,
       "step": 2155
     },
     {
-      "epoch": 0.7045751633986929,
-      "grad_norm": 1.2453532218933105,
-      "learning_rate": 4.2392140140401996e-06,
-      "loss": 0.9416,
+      "epoch": 0.05921287523001291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955329534971018e-05,
+      "loss": 1.0222,
       "step": 2156
     },
     {
-      "epoch": 0.7049019607843138,
-      "grad_norm": 1.338639497756958,
-      "learning_rate": 4.2305652214761305e-06,
-      "loss": 0.9359,
+      "epoch": 0.059240339457856145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955245512184656e-05,
+      "loss": 1.1523,
       "step": 2157
     },
     {
-      "epoch": 0.7052287581699347,
-      "grad_norm": 1.3959466218948364,
-      "learning_rate": 4.221922892969103e-06,
-      "loss": 0.9544,
+      "epoch": 0.05926780368569937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955161410628534e-05,
+      "loss": 1.0733,
       "step": 2158
     },
     {
-      "epoch": 0.7055555555555556,
-      "grad_norm": 1.2707918882369995,
-      "learning_rate": 4.213287038201943e-06,
-      "loss": 0.9517,
+      "epoch": 0.05929526791354261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955077230303317e-05,
+      "loss": 1.1375,
       "step": 2159
     },
     {
-      "epoch": 0.7058823529411765,
-      "grad_norm": 1.220543622970581,
-      "learning_rate": 4.204657666850245e-06,
-      "loss": 0.9458,
+      "epoch": 0.059322732141385844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954992971209674e-05,
+      "loss": 1.1806,
       "step": 2160
     },
     {
-      "epoch": 0.7062091503267974,
-      "grad_norm": 1.4205716848373413,
-      "learning_rate": 4.196034788582326e-06,
-      "loss": 0.9401,
+      "epoch": 0.05935019636922908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954908633348272e-05,
+      "loss": 0.9215,
       "step": 2161
     },
     {
-      "epoch": 0.7065359477124183,
-      "grad_norm": 1.2461148500442505,
-      "learning_rate": 4.187418413059234e-06,
-      "loss": 0.8925,
+      "epoch": 0.059377660597072314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954824216719776e-05,
+      "loss": 1.1516,
       "step": 2162
     },
     {
-      "epoch": 0.7068627450980393,
-      "grad_norm": 1.2306709289550781,
-      "learning_rate": 4.17880854993473e-06,
-      "loss": 0.9469,
+      "epoch": 0.05940512482491555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954739721324853e-05,
+      "loss": 1.1408,
       "step": 2163
     },
     {
-      "epoch": 0.7071895424836602,
-      "grad_norm": 1.3214656114578247,
-      "learning_rate": 4.170205208855281e-06,
-      "loss": 0.9076,
+      "epoch": 0.059432589052758784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954655147164176e-05,
+      "loss": 1.075,
       "step": 2164
     },
     {
-      "epoch": 0.7075163398692811,
-      "grad_norm": 1.2420142889022827,
-      "learning_rate": 4.161608399460038e-06,
-      "loss": 0.9293,
+      "epoch": 0.05946005328060201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954570494238407e-05,
+      "loss": 1.1329,
       "step": 2165
     },
     {
-      "epoch": 0.707843137254902,
-      "grad_norm": 1.270175576210022,
-      "learning_rate": 4.153018131380843e-06,
-      "loss": 0.9138,
+      "epoch": 0.05948751750844525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954485762548223e-05,
+      "loss": 1.1376,
       "step": 2166
     },
     {
-      "epoch": 0.7081699346405229,
-      "grad_norm": 1.350549578666687,
-      "learning_rate": 4.144434414242204e-06,
-      "loss": 0.8958,
+      "epoch": 0.05951498173628848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954400952094293e-05,
+      "loss": 1.069,
       "step": 2167
     },
     {
-      "epoch": 0.7084967320261438,
-      "grad_norm": 1.2426217794418335,
-      "learning_rate": 4.135857257661295e-06,
-      "loss": 0.9481,
+      "epoch": 0.05954244596413172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954316062877285e-05,
+      "loss": 1.1691,
       "step": 2168
     },
     {
-      "epoch": 0.7088235294117647,
-      "grad_norm": 1.246176838874817,
-      "learning_rate": 4.127286671247933e-06,
-      "loss": 0.9386,
+      "epoch": 0.059569910191974954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954231094897873e-05,
+      "loss": 1.0702,
       "step": 2169
     },
     {
-      "epoch": 0.7091503267973857,
-      "grad_norm": 1.2749322652816772,
-      "learning_rate": 4.118722664604582e-06,
-      "loss": 0.9286,
+      "epoch": 0.05959737441981819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954146048156725e-05,
+      "loss": 1.0491,
       "step": 2170
     },
     {
-      "epoch": 0.7094771241830066,
-      "grad_norm": 1.24730384349823,
-      "learning_rate": 4.11016524732632e-06,
-      "loss": 0.9381,
+      "epoch": 0.059624838647661424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995406092265452e-05,
+      "loss": 1.1378,
       "step": 2171
     },
     {
-      "epoch": 0.7098039215686275,
-      "grad_norm": 1.2631783485412598,
-      "learning_rate": 4.101614429000857e-06,
-      "loss": 0.9178,
+      "epoch": 0.05965230287550465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995397571839193e-05,
+      "loss": 1.1063,
       "step": 2172
     },
     {
-      "epoch": 0.7101307189542484,
-      "grad_norm": 1.2931915521621704,
-      "learning_rate": 4.0930702192085e-06,
-      "loss": 0.941,
+      "epoch": 0.05967976710334789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995389043536963e-05,
+      "loss": 1.0781,
       "step": 2173
     },
     {
-      "epoch": 0.7104575163398693,
-      "grad_norm": 1.2745211124420166,
-      "learning_rate": 4.08453262752216e-06,
-      "loss": 0.9296,
+      "epoch": 0.05970723133119112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953805073588294e-05,
+      "loss": 1.1424,
       "step": 2174
     },
     {
-      "epoch": 0.7107843137254902,
-      "grad_norm": 1.3035430908203125,
-      "learning_rate": 4.0760016635073255e-06,
-      "loss": 0.9454,
+      "epoch": 0.05973469555903436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953719633048593e-05,
+      "loss": 1.0725,
       "step": 2175
     },
     {
-      "epoch": 0.7111111111111111,
-      "grad_norm": 1.3056156635284424,
-      "learning_rate": 4.067477336722063e-06,
-      "loss": 0.9079,
+      "epoch": 0.05976215978687759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953634113751207e-05,
+      "loss": 1.192,
       "step": 2176
     },
     {
-      "epoch": 0.711437908496732,
-      "grad_norm": 1.258553385734558,
-      "learning_rate": 4.058959656717002e-06,
-      "loss": 0.9244,
+      "epoch": 0.05978962401472083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995354851569681e-05,
+      "loss": 1.0372,
       "step": 2177
     },
     {
-      "epoch": 0.711764705882353,
-      "grad_norm": 1.3076857328414917,
-      "learning_rate": 4.050448633035326e-06,
-      "loss": 0.9203,
+      "epoch": 0.059817088242564063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953462838886083e-05,
+      "loss": 1.0217,
       "step": 2178
     },
     {
-      "epoch": 0.7120915032679739,
-      "grad_norm": 1.2771118879318237,
-      "learning_rate": 4.041944275212759e-06,
-      "loss": 0.9419,
+      "epoch": 0.05984455247040729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995337708331971e-05,
+      "loss": 1.0807,
       "step": 2179
     },
     {
-      "epoch": 0.7124183006535948,
-      "grad_norm": 1.3147903680801392,
-      "learning_rate": 4.033446592777558e-06,
-      "loss": 0.9453,
+      "epoch": 0.05987201669825053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953291248998348e-05,
+      "loss": 1.0516,
       "step": 2180
     },
     {
-      "epoch": 0.7127450980392157,
-      "grad_norm": 1.2800912857055664,
-      "learning_rate": 4.024955595250503e-06,
-      "loss": 0.9531,
+      "epoch": 0.05989948092609376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99532053359227e-05,
+      "loss": 1.1505,
       "step": 2181
     },
     {
-      "epoch": 0.7130718954248366,
-      "grad_norm": 1.3125413656234741,
-      "learning_rate": 4.016471292144875e-06,
-      "loss": 0.9271,
+      "epoch": 0.059926945153937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953119344093436e-05,
+      "loss": 1.0631,
       "step": 2182
     },
     {
-      "epoch": 0.7133986928104575,
-      "grad_norm": 1.279025912284851,
-      "learning_rate": 4.0079936929664654e-06,
-      "loss": 0.9489,
+      "epoch": 0.05995440938178023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953033273511232e-05,
+      "loss": 1.0006,
       "step": 2183
     },
     {
-      "epoch": 0.7137254901960784,
-      "grad_norm": 1.3816391229629517,
-      "learning_rate": 3.999522807213549e-06,
-      "loss": 0.9303,
+      "epoch": 0.05998187360962347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952947124176774e-05,
+      "loss": 1.1206,
       "step": 2184
     },
     {
-      "epoch": 0.7140522875816994,
-      "grad_norm": 1.253772497177124,
-      "learning_rate": 3.991058644376881e-06,
-      "loss": 0.9339,
+      "epoch": 0.0600093378374667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952860896090748e-05,
+      "loss": 1.0253,
       "step": 2185
     },
     {
-      "epoch": 0.7143790849673203,
-      "grad_norm": 1.3472222089767456,
-      "learning_rate": 3.982601213939686e-06,
-      "loss": 0.9433,
+      "epoch": 0.06003680206530993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952774589253823e-05,
+      "loss": 1.0913,
       "step": 2186
     },
     {
-      "epoch": 0.7147058823529412,
-      "grad_norm": 1.2160375118255615,
-      "learning_rate": 3.974150525377638e-06,
-      "loss": 0.9359,
+      "epoch": 0.060064266293153167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952688203666697e-05,
+      "loss": 1.2063,
       "step": 2187
     },
     {
-      "epoch": 0.7150326797385621,
-      "grad_norm": 1.2808998823165894,
-      "learning_rate": 3.965706588158865e-06,
-      "loss": 0.9176,
+      "epoch": 0.0600917305209964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995260173933004e-05,
+      "loss": 1.0977,
       "step": 2188
     },
     {
-      "epoch": 0.715359477124183,
-      "grad_norm": 1.2269909381866455,
-      "learning_rate": 3.957269411743929e-06,
-      "loss": 0.9463,
+      "epoch": 0.06011919474883964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952515196244547e-05,
+      "loss": 1.0015,
       "step": 2189
     },
     {
-      "epoch": 0.7156862745098039,
-      "grad_norm": 1.3262627124786377,
-      "learning_rate": 3.948839005585815e-06,
-      "loss": 0.9161,
+      "epoch": 0.06014665897668287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99524285744109e-05,
+      "loss": 1.0919,
       "step": 2190
     },
     {
-      "epoch": 0.7160130718954248,
-      "grad_norm": 1.2593330144882202,
-      "learning_rate": 3.940415379129926e-06,
-      "loss": 0.947,
+      "epoch": 0.06017412320452611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995234187382978e-05,
+      "loss": 1.0539,
       "step": 2191
     },
     {
-      "epoch": 0.7163398692810458,
-      "grad_norm": 1.2663335800170898,
-      "learning_rate": 3.931998541814069e-06,
-      "loss": 0.9273,
+      "epoch": 0.060201587432369336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995225509450188e-05,
+      "loss": 1.1963,
       "step": 2192
     },
     {
-      "epoch": 0.7166666666666667,
-      "grad_norm": 1.2774310111999512,
-      "learning_rate": 3.923588503068438e-06,
-      "loss": 0.9232,
+      "epoch": 0.06022905166021257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995216823642788e-05,
+      "loss": 1.098,
       "step": 2193
     },
     {
-      "epoch": 0.7169934640522876,
-      "grad_norm": 1.2557518482208252,
-      "learning_rate": 3.915185272315616e-06,
-      "loss": 0.8876,
+      "epoch": 0.060256515888055806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995208129960847e-05,
+      "loss": 1.0349,
       "step": 2194
     },
     {
-      "epoch": 0.7173202614379085,
-      "grad_norm": 1.262776494026184,
-      "learning_rate": 3.906788858970558e-06,
-      "loss": 0.9338,
+      "epoch": 0.06028398011589904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951994284044335e-05,
+      "loss": 1.0462,
       "step": 2195
     },
     {
-      "epoch": 0.7176470588235294,
-      "grad_norm": 1.2615878582000732,
-      "learning_rate": 3.89839927244058e-06,
-      "loss": 0.9042,
+      "epoch": 0.060311444343742276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951907189736167e-05,
+      "loss": 1.0574,
       "step": 2196
     },
     {
-      "epoch": 0.7179738562091503,
-      "grad_norm": 1.2954107522964478,
-      "learning_rate": 3.890016522125347e-06,
-      "loss": 0.9413,
+      "epoch": 0.06033890857158551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951820016684657e-05,
+      "loss": 1.0485,
       "step": 2197
     },
     {
-      "epoch": 0.7183006535947712,
-      "grad_norm": 1.2931324243545532,
-      "learning_rate": 3.8816406174168695e-06,
-      "loss": 0.9232,
+      "epoch": 0.06036637279942875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995173276489049e-05,
+      "loss": 1.1301,
       "step": 2198
     },
     {
-      "epoch": 0.7186274509803922,
-      "grad_norm": 1.3032928705215454,
-      "learning_rate": 3.873271567699485e-06,
-      "loss": 0.9236,
+      "epoch": 0.060393837027271975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995164543435436e-05,
+      "loss": 1.1986,
       "step": 2199
     },
     {
-      "epoch": 0.7189542483660131,
-      "grad_norm": 1.2376676797866821,
-      "learning_rate": 3.86490938234985e-06,
-      "loss": 0.9539,
+      "epoch": 0.06042130125511521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951558025076955e-05,
+      "loss": 1.0254,
       "step": 2200
     },
     {
-      "epoch": 0.719281045751634,
-      "grad_norm": 1.2765297889709473,
-      "learning_rate": 3.856554070736931e-06,
-      "loss": 0.9111,
+      "epoch": 0.060448765482958446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951470537058972e-05,
+      "loss": 1.1257,
       "step": 2201
     },
     {
-      "epoch": 0.7196078431372549,
-      "grad_norm": 1.378954529762268,
-      "learning_rate": 3.848205642221995e-06,
-      "loss": 0.9345,
+      "epoch": 0.06047622971080168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951382970301094e-05,
+      "loss": 0.9713,
       "step": 2202
     },
     {
-      "epoch": 0.7199346405228758,
-      "grad_norm": 1.242738127708435,
-      "learning_rate": 3.839864106158599e-06,
-      "loss": 0.9301,
+      "epoch": 0.060503693938644916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995129532480402e-05,
+      "loss": 0.9919,
       "step": 2203
     },
     {
-      "epoch": 0.7202614379084967,
-      "grad_norm": 1.2957960367202759,
-      "learning_rate": 3.8315294718925656e-06,
-      "loss": 0.92,
+      "epoch": 0.06053115816648815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951207600568442e-05,
+      "loss": 1.1957,
       "step": 2204
     },
     {
-      "epoch": 0.7205882352941176,
-      "grad_norm": 1.2394717931747437,
-      "learning_rate": 3.823201748761998e-06,
-      "loss": 0.9302,
+      "epoch": 0.060558622394331386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951119797595058e-05,
+      "loss": 1.1597,
       "step": 2205
     },
     {
-      "epoch": 0.7209150326797386,
-      "grad_norm": 1.2651242017745972,
-      "learning_rate": 3.814880946097253e-06,
-      "loss": 0.948,
+      "epoch": 0.060586086622174615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951031915884556e-05,
+      "loss": 1.1558,
       "step": 2206
     },
     {
-      "epoch": 0.7212418300653595,
-      "grad_norm": 1.2907092571258545,
-      "learning_rate": 3.8065670732209316e-06,
-      "loss": 0.9219,
+      "epoch": 0.06061355085001785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995094395543764e-05,
+      "loss": 1.0923,
       "step": 2207
     },
     {
-      "epoch": 0.7215686274509804,
-      "grad_norm": 1.2839285135269165,
-      "learning_rate": 3.79826013944787e-06,
-      "loss": 0.9123,
+      "epoch": 0.060641015077861085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950855916254996e-05,
+      "loss": 1.1626,
       "step": 2208
     },
     {
-      "epoch": 0.7218954248366013,
-      "grad_norm": 1.257580280303955,
-      "learning_rate": 3.789960154085137e-06,
-      "loss": 0.9305,
+      "epoch": 0.06066847930570432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950767798337327e-05,
+      "loss": 1.0943,
       "step": 2209
     },
     {
-      "epoch": 0.7222222222222222,
-      "grad_norm": 1.2792307138442993,
-      "learning_rate": 3.7816671264320026e-06,
-      "loss": 0.9452,
+      "epoch": 0.060695943533547556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995067960168533e-05,
+      "loss": 1.0827,
       "step": 2210
     },
     {
-      "epoch": 0.7225490196078431,
-      "grad_norm": 1.2887842655181885,
-      "learning_rate": 3.773381065779953e-06,
-      "loss": 0.9406,
+      "epoch": 0.06072340776139079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950591326299697e-05,
+      "loss": 1.1484,
       "step": 2211
     },
     {
-      "epoch": 0.722875816993464,
-      "grad_norm": 1.243760585784912,
-      "learning_rate": 3.7651019814126656e-06,
-      "loss": 0.9621,
+      "epoch": 0.060750871989234026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950502972181137e-05,
+      "loss": 1.1119,
       "step": 2212
     },
     {
-      "epoch": 0.723202614379085,
-      "grad_norm": 1.2308412790298462,
-      "learning_rate": 3.756829882606001e-06,
-      "loss": 0.9506,
+      "epoch": 0.060778336217077254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995041453933034e-05,
+      "loss": 1.0041,
       "step": 2213
     },
     {
-      "epoch": 0.7235294117647059,
-      "grad_norm": 1.2629681825637817,
-      "learning_rate": 3.748564778627993e-06,
-      "loss": 0.9184,
+      "epoch": 0.06080580044492049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950326027748005e-05,
+      "loss": 1.0297,
       "step": 2214
     },
     {
-      "epoch": 0.7238562091503268,
-      "grad_norm": 1.2908703088760376,
-      "learning_rate": 3.740306678738843e-06,
-      "loss": 0.9035,
+      "epoch": 0.060833264672763725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995023743743484e-05,
+      "loss": 1.0578,
       "step": 2215
     },
     {
-      "epoch": 0.7241830065359477,
-      "grad_norm": 1.2861124277114868,
-      "learning_rate": 3.732055592190893e-06,
-      "loss": 0.9195,
+      "epoch": 0.06086072890060696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950148768391544e-05,
+      "loss": 1.069,
       "step": 2216
     },
     {
-      "epoch": 0.7245098039215686,
-      "grad_norm": 1.2668068408966064,
-      "learning_rate": 3.7238115282286393e-06,
-      "loss": 0.9476,
+      "epoch": 0.060888193128450195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950060020618816e-05,
+      "loss": 1.1979,
       "step": 2217
     },
     {
-      "epoch": 0.7248366013071895,
-      "grad_norm": 1.2540007829666138,
-      "learning_rate": 3.7155744960887054e-06,
-      "loss": 0.9258,
+      "epoch": 0.06091565735629343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949971194117354e-05,
+      "loss": 1.0042,
       "step": 2218
     },
     {
-      "epoch": 0.7251633986928104,
-      "grad_norm": 1.3234748840332031,
-      "learning_rate": 3.707344504999837e-06,
-      "loss": 0.9322,
+      "epoch": 0.06094312158413666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994988228888787e-05,
+      "loss": 1.0232,
       "step": 2219
     },
     {
-      "epoch": 0.7254901960784313,
-      "grad_norm": 1.2365049123764038,
-      "learning_rate": 3.6991215641828903e-06,
-      "loss": 0.9215,
+      "epoch": 0.060970585811979894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949793304931062e-05,
+      "loss": 1.0377,
       "step": 2220
     },
     {
-      "epoch": 0.7258169934640523,
-      "grad_norm": 1.2423840761184692,
-      "learning_rate": 3.690905682850824e-06,
-      "loss": 0.9746,
+      "epoch": 0.06099805003982313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949704242247636e-05,
+      "loss": 1.0602,
       "step": 2221
     },
     {
-      "epoch": 0.7261437908496732,
-      "grad_norm": 1.3728364706039429,
-      "learning_rate": 3.682696870208685e-06,
-      "loss": 0.9565,
+      "epoch": 0.061025514267666364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949615100838294e-05,
+      "loss": 1.1431,
       "step": 2222
     },
     {
-      "epoch": 0.7264705882352941,
-      "grad_norm": 1.3115557432174683,
-      "learning_rate": 3.674495135453603e-06,
-      "loss": 0.9816,
+      "epoch": 0.0610529784955096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949525880703745e-05,
+      "loss": 1.0693,
       "step": 2223
     },
     {
-      "epoch": 0.726797385620915,
-      "grad_norm": 1.2645386457443237,
-      "learning_rate": 3.6663004877747765e-06,
-      "loss": 0.9622,
+      "epoch": 0.061080442723352835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949436581844693e-05,
+      "loss": 1.019,
       "step": 2224
     },
     {
-      "epoch": 0.7271241830065359,
-      "grad_norm": 1.2133152484893799,
-      "learning_rate": 3.658112936353464e-06,
-      "loss": 0.9358,
+      "epoch": 0.06110790695119607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949347204261844e-05,
+      "loss": 1.0875,
       "step": 2225
     },
     {
-      "epoch": 0.7274509803921568,
-      "grad_norm": 1.3602253198623657,
-      "learning_rate": 3.649932490362974e-06,
-      "loss": 0.916,
+      "epoch": 0.0611353711790393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949257747955905e-05,
+      "loss": 1.0856,
       "step": 2226
     },
     {
-      "epoch": 0.7277777777777777,
-      "grad_norm": 1.2741917371749878,
-      "learning_rate": 3.64175915896865e-06,
-      "loss": 0.9293,
+      "epoch": 0.06116283540688253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949168212927587e-05,
+      "loss": 1.0367,
       "step": 2227
     },
     {
-      "epoch": 0.7281045751633987,
-      "grad_norm": 1.228398323059082,
-      "learning_rate": 3.6335929513278667e-06,
-      "loss": 0.9397,
+      "epoch": 0.06119029963472577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949078599177593e-05,
+      "loss": 1.2577,
       "step": 2228
     },
     {
-      "epoch": 0.7284313725490196,
-      "grad_norm": 1.2594088315963745,
-      "learning_rate": 3.625433876590021e-06,
-      "loss": 0.8993,
+      "epoch": 0.061217763862569004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948988906706637e-05,
+      "loss": 1.0112,
       "step": 2229
     },
     {
-      "epoch": 0.7287581699346405,
-      "grad_norm": 1.271207571029663,
-      "learning_rate": 3.6172819438965157e-06,
-      "loss": 0.9121,
+      "epoch": 0.06124522809041224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948899135515428e-05,
+      "loss": 0.9849,
       "step": 2230
     },
     {
-      "epoch": 0.7290849673202614,
-      "grad_norm": 1.198533296585083,
-      "learning_rate": 3.6091371623807513e-06,
-      "loss": 0.954,
+      "epoch": 0.061272692318255474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948809285604676e-05,
+      "loss": 1.0949,
       "step": 2231
     },
     {
-      "epoch": 0.7294117647058823,
-      "grad_norm": 1.2795460224151611,
-      "learning_rate": 3.600999541168113e-06,
-      "loss": 0.9192,
+      "epoch": 0.06130015654609871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994871935697509e-05,
+      "loss": 1.1158,
       "step": 2232
     },
     {
-      "epoch": 0.7297385620915032,
-      "grad_norm": 1.29898202419281,
-      "learning_rate": 3.592869089375968e-06,
-      "loss": 0.9153,
+      "epoch": 0.06132762077394194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994862934962738e-05,
+      "loss": 1.0233,
       "step": 2233
     },
     {
-      "epoch": 0.7300653594771241,
-      "grad_norm": 1.2823206186294556,
-      "learning_rate": 3.5847458161136485e-06,
-      "loss": 0.9542,
+      "epoch": 0.06135508500178517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948539263562264e-05,
+      "loss": 1.0212,
       "step": 2234
     },
     {
-      "epoch": 0.7303921568627451,
-      "grad_norm": 1.3131471872329712,
-      "learning_rate": 3.576629730482447e-06,
-      "loss": 0.9492,
+      "epoch": 0.06138254922962841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994844909878045e-05,
+      "loss": 1.0556,
       "step": 2235
     },
     {
-      "epoch": 0.730718954248366,
-      "grad_norm": 1.26218843460083,
-      "learning_rate": 3.568520841575601e-06,
-      "loss": 0.9478,
+      "epoch": 0.06141001345747164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948358855282654e-05,
+      "loss": 1.0523,
       "step": 2236
     },
     {
-      "epoch": 0.7310457516339869,
-      "grad_norm": 1.302687644958496,
-      "learning_rate": 3.560419158478288e-06,
-      "loss": 0.9532,
+      "epoch": 0.06143747768531488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948268533069586e-05,
+      "loss": 1.1239,
       "step": 2237
     },
     {
-      "epoch": 0.7313725490196078,
-      "grad_norm": 1.2744784355163574,
-      "learning_rate": 3.5523246902676e-06,
-      "loss": 0.9156,
+      "epoch": 0.061464941913158114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948178132141968e-05,
+      "loss": 1.1212,
       "step": 2238
     },
     {
-      "epoch": 0.7316993464052287,
-      "grad_norm": 1.2645971775054932,
-      "learning_rate": 3.5442374460125594e-06,
-      "loss": 0.9527,
+      "epoch": 0.06149240614100135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948087652500512e-05,
+      "loss": 1.0367,
       "step": 2239
     },
     {
-      "epoch": 0.7320261437908496,
-      "grad_norm": 1.2463462352752686,
-      "learning_rate": 3.5361574347740912e-06,
-      "loss": 0.9016,
+      "epoch": 0.06151987036884458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994799709414593e-05,
+      "loss": 1.1675,
       "step": 2240
     },
     {
-      "epoch": 0.7323529411764705,
-      "grad_norm": 1.27919340133667,
-      "learning_rate": 3.5280846656050127e-06,
-      "loss": 0.9475,
+      "epoch": 0.06154733459668781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947906457078944e-05,
+      "loss": 1.0439,
       "step": 2241
     },
     {
-      "epoch": 0.7326797385620915,
-      "grad_norm": 1.2300465106964111,
-      "learning_rate": 3.5200191475500313e-06,
-      "loss": 0.9483,
+      "epoch": 0.06157479882453105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947815741300264e-05,
+      "loss": 1.0801,
       "step": 2242
     },
     {
-      "epoch": 0.7330065359477124,
-      "grad_norm": 1.2382996082305908,
-      "learning_rate": 3.511960889645728e-06,
-      "loss": 0.928,
+      "epoch": 0.06160226305237428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947724946810617e-05,
+      "loss": 1.0773,
       "step": 2243
     },
     {
-      "epoch": 0.7333333333333333,
-      "grad_norm": 1.24155855178833,
-      "learning_rate": 3.5039099009205503e-06,
-      "loss": 0.9115,
+      "epoch": 0.06162972728021752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947634073610715e-05,
+      "loss": 1.0301,
       "step": 2244
     },
     {
-      "epoch": 0.7336601307189542,
-      "grad_norm": 1.2844840288162231,
-      "learning_rate": 3.4958661903948023e-06,
-      "loss": 0.9115,
+      "epoch": 0.06165719150806075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994754312170128e-05,
+      "loss": 1.0658,
       "step": 2245
     },
     {
-      "epoch": 0.7339869281045751,
-      "grad_norm": 1.2614479064941406,
-      "learning_rate": 3.4878297670806303e-06,
-      "loss": 0.914,
+      "epoch": 0.06168465573590399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994745209108303e-05,
+      "loss": 1.0983,
       "step": 2246
     },
     {
-      "epoch": 0.734313725490196,
-      "grad_norm": 1.2739602327346802,
-      "learning_rate": 3.4798006399820204e-06,
-      "loss": 0.9471,
+      "epoch": 0.06171211996374722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947360981756684e-05,
+      "loss": 1.1582,
       "step": 2247
     },
     {
-      "epoch": 0.734640522875817,
-      "grad_norm": 1.2271217107772827,
-      "learning_rate": 3.4717788180947855e-06,
-      "loss": 0.8843,
+      "epoch": 0.06173958419159045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947269793722967e-05,
+      "loss": 1.1279,
       "step": 2248
     },
     {
-      "epoch": 0.734967320261438,
-      "grad_norm": 1.2279889583587646,
-      "learning_rate": 3.463764310406542e-06,
-      "loss": 0.9416,
+      "epoch": 0.06176704841943369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947178526982596e-05,
+      "loss": 1.0559,
       "step": 2249
     },
     {
-      "epoch": 0.7352941176470589,
-      "grad_norm": 1.2678091526031494,
-      "learning_rate": 3.455757125896725e-06,
-      "loss": 0.9195,
+      "epoch": 0.06179451264727692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947087181536296e-05,
+      "loss": 1.1351,
       "step": 2250
     },
     {
-      "epoch": 0.7356209150326798,
-      "grad_norm": 1.2275513410568237,
-      "learning_rate": 3.447757273536557e-06,
-      "loss": 0.9246,
+      "epoch": 0.06182197687512016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994699575738479e-05,
+      "loss": 1.1606,
       "step": 2251
     },
     {
-      "epoch": 0.7359477124183007,
-      "grad_norm": 1.2553077936172485,
-      "learning_rate": 3.439764762289051e-06,
-      "loss": 0.969,
+      "epoch": 0.06184944110296339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99469042545288e-05,
+      "loss": 1.1398,
       "step": 2252
     },
     {
-      "epoch": 0.7362745098039216,
-      "grad_norm": 1.3087111711502075,
-      "learning_rate": 3.431779601108989e-06,
-      "loss": 0.9715,
+      "epoch": 0.06187690533080662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946812672969052e-05,
+      "loss": 1.0501,
       "step": 2253
     },
     {
-      "epoch": 0.7366013071895425,
-      "grad_norm": 1.2428115606307983,
-      "learning_rate": 3.423801798942926e-06,
-      "loss": 0.9084,
+      "epoch": 0.061904369558649856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946721012706264e-05,
+      "loss": 1.0285,
       "step": 2254
     },
     {
-      "epoch": 0.7369281045751634,
-      "grad_norm": 1.2260981798171997,
-      "learning_rate": 3.415831364729161e-06,
-      "loss": 0.9333,
+      "epoch": 0.06193183378649309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946629273741173e-05,
+      "loss": 1.1141,
       "step": 2255
     },
     {
-      "epoch": 0.7372549019607844,
-      "grad_norm": 1.3168964385986328,
-      "learning_rate": 3.407868307397747e-06,
-      "loss": 0.909,
+      "epoch": 0.06195929801433633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946537456074493e-05,
+      "loss": 1.0414,
       "step": 2256
     },
     {
-      "epoch": 0.7375816993464053,
-      "grad_norm": 1.234456181526184,
-      "learning_rate": 3.3999126358704683e-06,
-      "loss": 0.9192,
+      "epoch": 0.06198676224217956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994644555970696e-05,
+      "loss": 1.0877,
       "step": 2257
     },
     {
-      "epoch": 0.7379084967320262,
-      "grad_norm": 1.3223031759262085,
-      "learning_rate": 3.3919643590608355e-06,
-      "loss": 0.9483,
+      "epoch": 0.0620142264700228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994635358463929e-05,
+      "loss": 1.0086,
       "step": 2258
     },
     {
-      "epoch": 0.7382352941176471,
-      "grad_norm": 1.3415145874023438,
-      "learning_rate": 3.3840234858740763e-06,
-      "loss": 0.934,
+      "epoch": 0.06204169069786603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946261530872224e-05,
+      "loss": 1.0765,
       "step": 2259
     },
     {
-      "epoch": 0.738562091503268,
-      "grad_norm": 1.2171876430511475,
-      "learning_rate": 3.376090025207115e-06,
-      "loss": 0.9549,
+      "epoch": 0.06206915492570926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946169398406484e-05,
+      "loss": 1.1381,
       "step": 2260
     },
     {
-      "epoch": 0.7388888888888889,
-      "grad_norm": 1.2916749715805054,
-      "learning_rate": 3.3681639859485794e-06,
-      "loss": 0.9315,
+      "epoch": 0.062096619153552496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946077187242795e-05,
+      "loss": 1.0488,
       "step": 2261
     },
     {
-      "epoch": 0.7392156862745098,
-      "grad_norm": 1.289625644683838,
-      "learning_rate": 3.360245376978779e-06,
-      "loss": 0.9081,
+      "epoch": 0.06212408338139573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945984897381895e-05,
+      "loss": 1.0624,
       "step": 2262
     },
     {
-      "epoch": 0.7395424836601308,
-      "grad_norm": 1.2830357551574707,
-      "learning_rate": 3.3523342071696994e-06,
-      "loss": 0.9498,
+      "epoch": 0.062151547609238966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945892528824505e-05,
+      "loss": 1.0989,
       "step": 2263
     },
     {
-      "epoch": 0.7398692810457517,
-      "grad_norm": 1.3245270252227783,
-      "learning_rate": 3.3444304853849918e-06,
-      "loss": 0.9376,
+      "epoch": 0.0621790118370822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945800081571365e-05,
+      "loss": 1.1377,
       "step": 2264
     },
     {
-      "epoch": 0.7401960784313726,
-      "grad_norm": 1.233991026878357,
-      "learning_rate": 3.3365342204799613e-06,
-      "loss": 0.9214,
+      "epoch": 0.06220647606492544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99457075556232e-05,
+      "loss": 1.0483,
       "step": 2265
     },
     {
-      "epoch": 0.7405228758169935,
-      "grad_norm": 1.3004252910614014,
-      "learning_rate": 3.3286454213015582e-06,
-      "loss": 0.915,
+      "epoch": 0.06223394029276867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945614950980745e-05,
+      "loss": 1.0127,
       "step": 2266
     },
     {
-      "epoch": 0.7408496732026144,
-      "grad_norm": 1.258966326713562,
-      "learning_rate": 3.3207640966883704e-06,
-      "loss": 0.9752,
+      "epoch": 0.0622614045206119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945522267644733e-05,
+      "loss": 1.1635,
       "step": 2267
     },
     {
-      "epoch": 0.7411764705882353,
-      "grad_norm": 1.2812533378601074,
-      "learning_rate": 3.312890255470609e-06,
-      "loss": 0.9528,
+      "epoch": 0.062288868748455135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945429505615894e-05,
+      "loss": 1.0783,
       "step": 2268
     },
     {
-      "epoch": 0.7415032679738562,
-      "grad_norm": 1.2921028137207031,
-      "learning_rate": 3.305023906470102e-06,
-      "loss": 0.8928,
+      "epoch": 0.06231633297629837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945336664894963e-05,
+      "loss": 1.2086,
       "step": 2269
     },
     {
-      "epoch": 0.7418300653594772,
-      "grad_norm": 1.2371723651885986,
-      "learning_rate": 3.29716505850028e-06,
-      "loss": 0.9179,
+      "epoch": 0.062343797204141606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994524374548268e-05,
+      "loss": 1.0996,
       "step": 2270
     },
     {
-      "epoch": 0.7421568627450981,
-      "grad_norm": 1.3093807697296143,
-      "learning_rate": 3.2893137203661798e-06,
-      "loss": 0.9037,
+      "epoch": 0.06237126143198484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945150747379775e-05,
+      "loss": 1.1136,
       "step": 2271
     },
     {
-      "epoch": 0.742483660130719,
-      "grad_norm": 1.3286569118499756,
-      "learning_rate": 3.2814699008644068e-06,
-      "loss": 0.9436,
+      "epoch": 0.062398725659828076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994505767058698e-05,
+      "loss": 1.2339,
       "step": 2272
     },
     {
-      "epoch": 0.7428104575163399,
-      "grad_norm": 1.2706778049468994,
-      "learning_rate": 3.2736336087831576e-06,
-      "loss": 0.9311,
+      "epoch": 0.06242618988767131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944964515105043e-05,
+      "loss": 1.0697,
       "step": 2273
     },
     {
-      "epoch": 0.7431372549019608,
-      "grad_norm": 1.2694809436798096,
-      "learning_rate": 3.265804852902188e-06,
-      "loss": 0.9278,
+      "epoch": 0.06245365411551454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944871280934693e-05,
+      "loss": 1.1164,
       "step": 2274
     },
     {
-      "epoch": 0.7434640522875817,
-      "grad_norm": 1.2477871179580688,
-      "learning_rate": 3.2579836419928135e-06,
-      "loss": 0.9107,
+      "epoch": 0.062481118343357775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944777968076666e-05,
+      "loss": 1.0825,
       "step": 2275
     },
     {
-      "epoch": 0.7437908496732026,
-      "grad_norm": 1.356742024421692,
-      "learning_rate": 3.250169984817897e-06,
-      "loss": 0.9399,
+      "epoch": 0.06250858257120101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944684576531707e-05,
+      "loss": 1.135,
       "step": 2276
     },
     {
-      "epoch": 0.7441176470588236,
-      "grad_norm": 1.263956069946289,
-      "learning_rate": 3.242363890131831e-06,
-      "loss": 0.9353,
+      "epoch": 0.06253604679904425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994459110630055e-05,
+      "loss": 1.102,
       "step": 2277
     },
     {
-      "epoch": 0.7444444444444445,
-      "grad_norm": 1.3146030902862549,
-      "learning_rate": 3.234565366680542e-06,
-      "loss": 0.92,
+      "epoch": 0.06256351102688748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944497557383933e-05,
+      "loss": 1.1274,
       "step": 2278
     },
     {
-      "epoch": 0.7447712418300654,
-      "grad_norm": 1.201258659362793,
-      "learning_rate": 3.226774423201472e-06,
-      "loss": 0.9628,
+      "epoch": 0.06259097525473072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944403929782604e-05,
+      "loss": 1.0977,
       "step": 2279
     },
     {
-      "epoch": 0.7450980392156863,
-      "grad_norm": 1.2567166090011597,
-      "learning_rate": 3.2189910684235705e-06,
-      "loss": 0.9739,
+      "epoch": 0.06261843948257395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944310223497296e-05,
+      "loss": 1.0978,
       "step": 2280
     },
     {
-      "epoch": 0.7454248366013072,
-      "grad_norm": 1.3559985160827637,
-      "learning_rate": 3.211215311067284e-06,
-      "loss": 0.906,
+      "epoch": 0.06264590371041719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944216438528755e-05,
+      "loss": 1.1446,
       "step": 2281
     },
     {
-      "epoch": 0.7457516339869281,
-      "grad_norm": 1.2795518636703491,
-      "learning_rate": 3.2034471598445505e-06,
-      "loss": 0.9283,
+      "epoch": 0.06267336793826042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994412257487772e-05,
+      "loss": 1.1321,
       "step": 2282
     },
     {
-      "epoch": 0.746078431372549,
-      "grad_norm": 1.3245009183883667,
-      "learning_rate": 3.195686623458777e-06,
-      "loss": 0.8858,
+      "epoch": 0.06270083216610366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944028632544937e-05,
+      "loss": 1.1515,
       "step": 2283
     },
     {
-      "epoch": 0.74640522875817,
-      "grad_norm": 1.3577768802642822,
-      "learning_rate": 3.187933710604847e-06,
-      "loss": 0.8989,
+      "epoch": 0.06272829639394688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943934611531147e-05,
+      "loss": 1.1316,
       "step": 2284
     },
     {
-      "epoch": 0.7467320261437909,
-      "grad_norm": 1.2622170448303223,
-      "learning_rate": 3.180188429969101e-06,
-      "loss": 0.9368,
+      "epoch": 0.06275576062179011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994384051183709e-05,
+      "loss": 1.0649,
       "step": 2285
     },
     {
-      "epoch": 0.7470588235294118,
-      "grad_norm": 1.2169404029846191,
-      "learning_rate": 3.1724507902293267e-06,
-      "loss": 0.9301,
+      "epoch": 0.06278322484963335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994374633346352e-05,
+      "loss": 1.1636,
       "step": 2286
     },
     {
-      "epoch": 0.7473856209150327,
-      "grad_norm": 1.2475335597991943,
-      "learning_rate": 3.1647208000547535e-06,
-      "loss": 0.9227,
+      "epoch": 0.06281068907747658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994365207641118e-05,
+      "loss": 1.2123,
       "step": 2287
     },
     {
-      "epoch": 0.7477124183006536,
-      "grad_norm": 1.2803884744644165,
-      "learning_rate": 3.1569984681060375e-06,
-      "loss": 0.9522,
+      "epoch": 0.06283815330531982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943557740680807e-05,
+      "loss": 1.0256,
       "step": 2288
     },
     {
-      "epoch": 0.7480392156862745,
-      "grad_norm": 1.3047538995742798,
-      "learning_rate": 3.149283803035256e-06,
-      "loss": 0.9076,
+      "epoch": 0.06286561753316305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943463326273156e-05,
+      "loss": 1.1424,
       "step": 2289
     },
     {
-      "epoch": 0.7483660130718954,
-      "grad_norm": 1.2296929359436035,
-      "learning_rate": 3.1415768134858947e-06,
-      "loss": 0.932,
+      "epoch": 0.06289308176100629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943368833188968e-05,
+      "loss": 1.0976,
       "step": 2290
     },
     {
-      "epoch": 0.7486928104575163,
-      "grad_norm": 1.272104263305664,
-      "learning_rate": 3.133877508092844e-06,
-      "loss": 0.9317,
+      "epoch": 0.06292054598884952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943274261429e-05,
+      "loss": 1.0003,
       "step": 2291
     },
     {
-      "epoch": 0.7490196078431373,
-      "grad_norm": 1.2717616558074951,
-      "learning_rate": 3.1261858954823798e-06,
-      "loss": 0.9616,
+      "epoch": 0.06294801021669276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994317961099399e-05,
+      "loss": 1.0625,
       "step": 2292
     },
     {
-      "epoch": 0.7493464052287582,
-      "grad_norm": 1.3122819662094116,
-      "learning_rate": 3.1185019842721664e-06,
-      "loss": 0.9555,
+      "epoch": 0.062975474444536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943084881884696e-05,
+      "loss": 1.1772,
       "step": 2293
     },
     {
-      "epoch": 0.7496732026143791,
-      "grad_norm": 1.2782644033432007,
-      "learning_rate": 3.1108257830712253e-06,
-      "loss": 0.9398,
+      "epoch": 0.06300293867237923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994299007410186e-05,
+      "loss": 1.1043,
       "step": 2294
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 1.2781953811645508,
-      "learning_rate": 3.103157300479954e-06,
-      "loss": 0.9511,
+      "epoch": 0.06303040290022247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942895187646232e-05,
+      "loss": 0.9762,
       "step": 2295
     },
     {
-      "epoch": 0.7503267973856209,
-      "grad_norm": 1.2587476968765259,
-      "learning_rate": 3.0954965450900963e-06,
-      "loss": 0.8865,
+      "epoch": 0.0630578671280657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994280022251857e-05,
+      "loss": 1.1748,
       "step": 2296
     },
     {
-      "epoch": 0.7506535947712418,
-      "grad_norm": 1.309575080871582,
-      "learning_rate": 3.0878435254847393e-06,
-      "loss": 0.9182,
+      "epoch": 0.06308533135590892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994270517871962e-05,
+      "loss": 1.006,
       "step": 2297
     },
     {
-      "epoch": 0.7509803921568627,
-      "grad_norm": 1.3021526336669922,
-      "learning_rate": 3.080198250238302e-06,
-      "loss": 0.9683,
+      "epoch": 0.06311279558375216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942610056250135e-05,
+      "loss": 1.0995,
       "step": 2298
     },
     {
-      "epoch": 0.7513071895424837,
-      "grad_norm": 1.2413824796676636,
-      "learning_rate": 3.0725607279165315e-06,
-      "loss": 0.9014,
+      "epoch": 0.06314025981159539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942514855110866e-05,
+      "loss": 1.0977,
       "step": 2299
     },
     {
-      "epoch": 0.7516339869281046,
-      "grad_norm": 1.2684258222579956,
-      "learning_rate": 3.064930967076477e-06,
-      "loss": 0.9194,
+      "epoch": 0.06316772403943863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942419575302568e-05,
+      "loss": 1.0933,
       "step": 2300
     },
     {
-      "epoch": 0.7519607843137255,
-      "grad_norm": 1.3159880638122559,
-      "learning_rate": 3.057308976266503e-06,
-      "loss": 0.9629,
+      "epoch": 0.06319518826728186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942324216826e-05,
+      "loss": 1.0768,
       "step": 2301
     },
     {
-      "epoch": 0.7522875816993464,
-      "grad_norm": 1.2627214193344116,
-      "learning_rate": 3.0496947640262655e-06,
-      "loss": 0.9167,
+      "epoch": 0.0632226524951251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942228779681905e-05,
+      "loss": 1.1106,
       "step": 2302
     },
     {
-      "epoch": 0.7526143790849673,
-      "grad_norm": 1.2537089586257935,
-      "learning_rate": 3.042088338886702e-06,
-      "loss": 0.9159,
+      "epoch": 0.06325011672296833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942133263871048e-05,
+      "loss": 1.1952,
       "step": 2303
     },
     {
-      "epoch": 0.7529411764705882,
-      "grad_norm": 1.2983770370483398,
-      "learning_rate": 3.0344897093700333e-06,
-      "loss": 0.9566,
+      "epoch": 0.06327758095081157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942037669394185e-05,
+      "loss": 1.0564,
       "step": 2304
     },
     {
-      "epoch": 0.7532679738562091,
-      "grad_norm": 1.3200887441635132,
-      "learning_rate": 3.026898883989734e-06,
-      "loss": 0.9056,
+      "epoch": 0.0633050451786548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994194199625206e-05,
+      "loss": 0.9582,
       "step": 2305
     },
     {
-      "epoch": 0.75359477124183,
-      "grad_norm": 1.1931440830230713,
-      "learning_rate": 3.019315871250543e-06,
-      "loss": 0.8954,
+      "epoch": 0.06333250940649804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941846244445447e-05,
+      "loss": 0.9651,
       "step": 2306
     },
     {
-      "epoch": 0.753921568627451,
-      "grad_norm": 1.2083357572555542,
-      "learning_rate": 3.011740679648447e-06,
-      "loss": 0.8975,
+      "epoch": 0.06335997363434127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994175041397509e-05,
+      "loss": 1.0361,
       "step": 2307
     },
     {
-      "epoch": 0.7542483660130719,
-      "grad_norm": 1.2307372093200684,
-      "learning_rate": 3.0041733176706668e-06,
-      "loss": 0.9523,
+      "epoch": 0.06338743786218451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941654504841757e-05,
+      "loss": 1.029,
       "step": 2308
     },
     {
-      "epoch": 0.7545751633986928,
-      "grad_norm": 1.2680959701538086,
-      "learning_rate": 2.996613793795646e-06,
-      "loss": 0.9293,
+      "epoch": 0.06341490209002774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99415585170462e-05,
+      "loss": 1.0599,
       "step": 2309
     },
     {
-      "epoch": 0.7549019607843137,
-      "grad_norm": 1.299697756767273,
-      "learning_rate": 2.9890621164930656e-06,
-      "loss": 0.9277,
+      "epoch": 0.06344236631787098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941462450589183e-05,
+      "loss": 1.1113,
       "step": 2310
     },
     {
-      "epoch": 0.7552287581699346,
-      "grad_norm": 1.2849456071853638,
-      "learning_rate": 2.9815182942237885e-06,
-      "loss": 0.9374,
+      "epoch": 0.0634698305457142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994136630547146e-05,
+      "loss": 1.1484,
       "step": 2311
     },
     {
-      "epoch": 0.7555555555555555,
-      "grad_norm": 1.275316834449768,
-      "learning_rate": 2.9739823354398958e-06,
-      "loss": 0.9155,
+      "epoch": 0.06349729477355744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99412700816938e-05,
+      "loss": 1.0274,
       "step": 2312
     },
     {
-      "epoch": 0.7558823529411764,
-      "grad_norm": 1.3317190408706665,
-      "learning_rate": 2.9664542485846505e-06,
-      "loss": 0.9534,
+      "epoch": 0.06352475900140067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994117377925696e-05,
+      "loss": 1.0959,
       "step": 2313
     },
     {
-      "epoch": 0.7562091503267974,
-      "grad_norm": 1.2938636541366577,
-      "learning_rate": 2.9589340420924983e-06,
-      "loss": 0.9322,
+      "epoch": 0.0635522232292439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99410773981617e-05,
+      "loss": 1.0257,
       "step": 2314
     },
     {
-      "epoch": 0.7565359477124183,
-      "grad_norm": 1.270707607269287,
-      "learning_rate": 2.9514217243890573e-06,
-      "loss": 0.9256,
+      "epoch": 0.06357968745708714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940980938408784e-05,
+      "loss": 1.079,
       "step": 2315
     },
     {
-      "epoch": 0.7568627450980392,
-      "grad_norm": 1.291558861732483,
-      "learning_rate": 2.943917303891107e-06,
-      "loss": 0.9598,
+      "epoch": 0.06360715168493038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994088439999898e-05,
+      "loss": 1.1307,
       "step": 2316
     },
     {
-      "epoch": 0.7571895424836601,
-      "grad_norm": 1.287424087524414,
-      "learning_rate": 2.936420789006572e-06,
-      "loss": 0.9272,
+      "epoch": 0.06363461591277361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940787782933043e-05,
+      "loss": 1.1654,
       "step": 2317
     },
     {
-      "epoch": 0.757516339869281,
-      "grad_norm": 1.2474398612976074,
-      "learning_rate": 2.9289321881345257e-06,
-      "loss": 0.9513,
+      "epoch": 0.06366208014061685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940691087211745e-05,
+      "loss": 1.126,
       "step": 2318
     },
     {
-      "epoch": 0.7578431372549019,
-      "grad_norm": 1.3043147325515747,
-      "learning_rate": 2.921451509665175e-06,
-      "loss": 0.9125,
+      "epoch": 0.06368954436846008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994059431283585e-05,
+      "loss": 1.028,
       "step": 2319
     },
     {
-      "epoch": 0.7581699346405228,
-      "grad_norm": 1.2723067998886108,
-      "learning_rate": 2.913978761979849e-06,
-      "loss": 0.9242,
+      "epoch": 0.06371700859630332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994049745980612e-05,
+      "loss": 1.1224,
       "step": 2320
     },
     {
-      "epoch": 0.7584967320261438,
-      "grad_norm": 1.3041671514511108,
-      "learning_rate": 2.9065139534509935e-06,
-      "loss": 0.9401,
+      "epoch": 0.06374447282414655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940400528123325e-05,
+      "loss": 0.9574,
       "step": 2321
     },
     {
-      "epoch": 0.7588235294117647,
-      "grad_norm": 1.3032827377319336,
-      "learning_rate": 2.8990570924421526e-06,
-      "loss": 0.9264,
+      "epoch": 0.06377193705198979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994030351778823e-05,
+      "loss": 1.0622,
       "step": 2322
     },
     {
-      "epoch": 0.7591503267973856,
-      "grad_norm": 1.3332056999206543,
-      "learning_rate": 2.891608187307973e-06,
-      "loss": 0.9581,
+      "epoch": 0.06379940127983302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940206428801602e-05,
+      "loss": 1.1244,
       "step": 2323
     },
     {
-      "epoch": 0.7594771241830065,
-      "grad_norm": 1.3028998374938965,
-      "learning_rate": 2.8841672463941827e-06,
-      "loss": 0.9096,
+      "epoch": 0.06382686550767624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940109261164212e-05,
+      "loss": 1.1912,
       "step": 2324
     },
     {
-      "epoch": 0.7598039215686274,
-      "grad_norm": 1.32345712184906,
-      "learning_rate": 2.876734278037593e-06,
-      "loss": 0.909,
+      "epoch": 0.06385432973551948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940012014876828e-05,
+      "loss": 1.002,
       "step": 2325
     },
     {
-      "epoch": 0.7601307189542483,
-      "grad_norm": 1.275283694267273,
-      "learning_rate": 2.8693092905660757e-06,
-      "loss": 0.9113,
+      "epoch": 0.06388179396336272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939914689940215e-05,
+      "loss": 0.9616,
       "step": 2326
     },
     {
-      "epoch": 0.7604575163398692,
-      "grad_norm": 1.2471257448196411,
-      "learning_rate": 2.8618922922985694e-06,
-      "loss": 0.9574,
+      "epoch": 0.06390925819120595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939817286355146e-05,
+      "loss": 1.1567,
       "step": 2327
     },
     {
-      "epoch": 0.7607843137254902,
-      "grad_norm": 1.408612847328186,
-      "learning_rate": 2.854483291545046e-06,
-      "loss": 0.9217,
+      "epoch": 0.06393672241904919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939719804122394e-05,
+      "loss": 0.9738,
       "step": 2328
     },
     {
-      "epoch": 0.7611111111111111,
-      "grad_norm": 1.3042173385620117,
-      "learning_rate": 2.8470822966065357e-06,
-      "loss": 0.924,
+      "epoch": 0.06396418664689242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939622243242727e-05,
+      "loss": 1.0188,
       "step": 2329
     },
     {
-      "epoch": 0.761437908496732,
-      "grad_norm": 1.3652455806732178,
-      "learning_rate": 2.8396893157750893e-06,
-      "loss": 0.9415,
+      "epoch": 0.06399165087473566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939524603716926e-05,
+      "loss": 1.0934,
       "step": 2330
     },
     {
-      "epoch": 0.7617647058823529,
-      "grad_norm": 1.2883495092391968,
-      "learning_rate": 2.832304357333777e-06,
-      "loss": 0.9365,
+      "epoch": 0.06401911510257889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993942688554575e-05,
+      "loss": 1.0827,
       "step": 2331
     },
     {
-      "epoch": 0.7620915032679738,
-      "grad_norm": 1.1887331008911133,
-      "learning_rate": 2.8249274295566863e-06,
-      "loss": 0.8838,
+      "epoch": 0.06404657933042213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993932908872998e-05,
+      "loss": 1.163,
       "step": 2332
     },
     {
-      "epoch": 0.7624183006535947,
-      "grad_norm": 1.2297059297561646,
-      "learning_rate": 2.8175585407089067e-06,
-      "loss": 0.9139,
+      "epoch": 0.06407404355826536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939231213270386e-05,
+      "loss": 1.0978,
       "step": 2333
     },
     {
-      "epoch": 0.7627450980392156,
-      "grad_norm": 1.3398977518081665,
-      "learning_rate": 2.810197699046513e-06,
-      "loss": 0.8927,
+      "epoch": 0.0641015077861086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939133259167745e-05,
+      "loss": 1.0834,
       "step": 2334
     },
     {
-      "epoch": 0.7630718954248366,
-      "grad_norm": 1.2631516456604004,
-      "learning_rate": 2.8028449128165715e-06,
-      "loss": 0.9211,
+      "epoch": 0.06412897201395183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939035226422833e-05,
+      "loss": 1.0725,
       "step": 2335
     },
     {
-      "epoch": 0.7633986928104575,
-      "grad_norm": 1.2660962343215942,
-      "learning_rate": 2.795500190257122e-06,
-      "loss": 0.9165,
+      "epoch": 0.06415643624179507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938937115036423e-05,
+      "loss": 1.1102,
       "step": 2336
     },
     {
-      "epoch": 0.7637254901960784,
-      "grad_norm": 1.400551676750183,
-      "learning_rate": 2.7881635395971686e-06,
-      "loss": 0.9431,
+      "epoch": 0.0641839004696383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938838925009293e-05,
+      "loss": 1.0433,
       "step": 2337
     },
     {
-      "epoch": 0.7640522875816993,
-      "grad_norm": 1.3197565078735352,
-      "learning_rate": 2.780834969056676e-06,
-      "loss": 0.8735,
+      "epoch": 0.06421136469748152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993874065634222e-05,
+      "loss": 1.1038,
       "step": 2338
     },
     {
-      "epoch": 0.7643790849673202,
-      "grad_norm": 1.3131808042526245,
-      "learning_rate": 2.773514486846546e-06,
-      "loss": 0.9256,
+      "epoch": 0.06423882892532476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993864230903598e-05,
+      "loss": 1.1034,
       "step": 2339
     },
     {
-      "epoch": 0.7647058823529411,
-      "grad_norm": 1.2214471101760864,
-      "learning_rate": 2.766202101168628e-06,
-      "loss": 0.9487,
+      "epoch": 0.064266293153168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938543883091353e-05,
+      "loss": 1.081,
       "step": 2340
     },
     {
-      "epoch": 0.765032679738562,
-      "grad_norm": 1.2740213871002197,
-      "learning_rate": 2.7588978202156945e-06,
-      "loss": 0.8943,
+      "epoch": 0.06429375738101123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993844537850912e-05,
+      "loss": 1.1403,
       "step": 2341
     },
     {
-      "epoch": 0.765359477124183,
-      "grad_norm": 1.2675347328186035,
-      "learning_rate": 2.7516016521714417e-06,
-      "loss": 0.9364,
+      "epoch": 0.06432122160885446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938346795290054e-05,
+      "loss": 1.0652,
       "step": 2342
     },
     {
-      "epoch": 0.765686274509804,
-      "grad_norm": 1.2153257131576538,
-      "learning_rate": 2.7443136052104723e-06,
-      "loss": 0.9252,
+      "epoch": 0.0643486858366977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938248133434936e-05,
+      "loss": 1.1189,
       "step": 2343
     },
     {
-      "epoch": 0.7660130718954249,
-      "grad_norm": 1.2413872480392456,
-      "learning_rate": 2.7370336874982952e-06,
-      "loss": 0.9481,
+      "epoch": 0.06437615006454094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938149392944553e-05,
+      "loss": 1.0372,
       "step": 2344
     },
     {
-      "epoch": 0.7663398692810458,
-      "grad_norm": 1.2568696737289429,
-      "learning_rate": 2.729761907191303e-06,
-      "loss": 0.9424,
+      "epoch": 0.06440361429238417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938050573819683e-05,
+      "loss": 1.021,
       "step": 2345
     },
     {
-      "epoch": 0.7666666666666667,
-      "grad_norm": 1.251447081565857,
-      "learning_rate": 2.7224982724367775e-06,
-      "loss": 0.9117,
+      "epoch": 0.0644310785202274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937951676061106e-05,
+      "loss": 1.1242,
       "step": 2346
     },
     {
-      "epoch": 0.7669934640522876,
-      "grad_norm": 1.2595696449279785,
-      "learning_rate": 2.7152427913728742e-06,
-      "loss": 0.9407,
+      "epoch": 0.06445854274807064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937852699669604e-05,
+      "loss": 1.0678,
       "step": 2347
     },
     {
-      "epoch": 0.7673202614379085,
-      "grad_norm": 1.2998888492584229,
-      "learning_rate": 2.7079954721286108e-06,
-      "loss": 0.9563,
+      "epoch": 0.06448600697591388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937753644645963e-05,
+      "loss": 0.9059,
       "step": 2348
     },
     {
-      "epoch": 0.7676470588235295,
-      "grad_norm": 1.3002018928527832,
-      "learning_rate": 2.700756322823861e-06,
-      "loss": 0.9338,
+      "epoch": 0.06451347120375711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937654510990967e-05,
+      "loss": 1.1128,
       "step": 2349
     },
     {
-      "epoch": 0.7679738562091504,
-      "grad_norm": 1.2875487804412842,
-      "learning_rate": 2.693525351569347e-06,
-      "loss": 0.9812,
+      "epoch": 0.06454093543160035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937555298705397e-05,
+      "loss": 1.1426,
       "step": 2350
     },
     {
-      "epoch": 0.7683006535947713,
-      "grad_norm": 1.2929884195327759,
-      "learning_rate": 2.6863025664666264e-06,
-      "loss": 0.9204,
+      "epoch": 0.06456839965944358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993745600779004e-05,
+      "loss": 1.0505,
       "step": 2351
     },
     {
-      "epoch": 0.7686274509803922,
-      "grad_norm": 1.343565821647644,
-      "learning_rate": 2.6790879756080845e-06,
-      "loss": 0.9622,
+      "epoch": 0.0645958638872868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993735663824568e-05,
+      "loss": 1.0004,
       "step": 2352
     },
     {
-      "epoch": 0.7689542483660131,
-      "grad_norm": 1.300872564315796,
-      "learning_rate": 2.671881587076929e-06,
-      "loss": 0.9392,
+      "epoch": 0.06462332811513004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937257190073108e-05,
+      "loss": 0.9846,
       "step": 2353
     },
     {
-      "epoch": 0.769281045751634,
-      "grad_norm": 1.3145891427993774,
-      "learning_rate": 2.664683408947173e-06,
-      "loss": 0.9091,
+      "epoch": 0.06465079234297327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993715766327311e-05,
+      "loss": 1.1843,
       "step": 2354
     },
     {
-      "epoch": 0.7696078431372549,
-      "grad_norm": 1.2799218893051147,
-      "learning_rate": 2.657493449283637e-06,
-      "loss": 0.9218,
+      "epoch": 0.06467825657081651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937058057846466e-05,
+      "loss": 1.1218,
       "step": 2355
     },
     {
-      "epoch": 0.7699346405228759,
-      "grad_norm": 1.2616987228393555,
-      "learning_rate": 2.6503117161419246e-06,
-      "loss": 0.9417,
+      "epoch": 0.06470572079865974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936958373793973e-05,
+      "loss": 1.0983,
       "step": 2356
     },
     {
-      "epoch": 0.7702614379084968,
-      "grad_norm": 1.2818011045455933,
-      "learning_rate": 2.6431382175684296e-06,
-      "loss": 0.9511,
+      "epoch": 0.06473318502650298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936858611116413e-05,
+      "loss": 1.0956,
       "step": 2357
     },
     {
-      "epoch": 0.7705882352941177,
-      "grad_norm": 1.2249089479446411,
-      "learning_rate": 2.6359729616003184e-06,
-      "loss": 0.9303,
+      "epoch": 0.06476064925434621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936758769814585e-05,
+      "loss": 1.1058,
       "step": 2358
     },
     {
-      "epoch": 0.7709150326797386,
-      "grad_norm": 1.2061792612075806,
-      "learning_rate": 2.6288159562655215e-06,
-      "loss": 0.9368,
+      "epoch": 0.06478811348218945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936658849889264e-05,
+      "loss": 1.1648,
       "step": 2359
     },
     {
-      "epoch": 0.7712418300653595,
-      "grad_norm": 1.2625164985656738,
-      "learning_rate": 2.6216672095827267e-06,
-      "loss": 0.9463,
+      "epoch": 0.06481557771003268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936558851341255e-05,
+      "loss": 1.157,
       "step": 2360
     },
     {
-      "epoch": 0.7715686274509804,
-      "grad_norm": 1.2779780626296997,
-      "learning_rate": 2.6145267295613685e-06,
-      "loss": 0.9323,
+      "epoch": 0.06484304193787592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936458774171342e-05,
+      "loss": 1.1735,
       "step": 2361
     },
     {
-      "epoch": 0.7718954248366013,
-      "grad_norm": 1.2837448120117188,
-      "learning_rate": 2.607394524201616e-06,
-      "loss": 0.9636,
+      "epoch": 0.06487050616571916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936358618380316e-05,
+      "loss": 1.1368,
       "step": 2362
     },
     {
-      "epoch": 0.7722222222222223,
-      "grad_norm": 1.267409324645996,
-      "learning_rate": 2.6002706014943702e-06,
-      "loss": 0.924,
+      "epoch": 0.06489797039356239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936258383968972e-05,
+      "loss": 1.0688,
       "step": 2363
     },
     {
-      "epoch": 0.7725490196078432,
-      "grad_norm": 1.2802129983901978,
-      "learning_rate": 2.5931549694212545e-06,
-      "loss": 0.9082,
+      "epoch": 0.06492543462140563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936158070938104e-05,
+      "loss": 1.0099,
       "step": 2364
     },
     {
-      "epoch": 0.7728758169934641,
-      "grad_norm": 1.2848308086395264,
-      "learning_rate": 2.5860476359546006e-06,
-      "loss": 0.9532,
+      "epoch": 0.06495289884924885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936057679288504e-05,
+      "loss": 1.1659,
       "step": 2365
     },
     {
-      "epoch": 0.773202614379085,
-      "grad_norm": 1.2615687847137451,
-      "learning_rate": 2.578948609057446e-06,
-      "loss": 0.938,
+      "epoch": 0.06498036307709208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935957209020968e-05,
+      "loss": 1.0762,
       "step": 2366
     },
     {
-      "epoch": 0.7735294117647059,
-      "grad_norm": 1.2685338258743286,
-      "learning_rate": 2.5718578966835116e-06,
-      "loss": 0.9415,
+      "epoch": 0.06500782730493532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993585666013629e-05,
+      "loss": 1.0412,
       "step": 2367
     },
     {
-      "epoch": 0.7738562091503268,
-      "grad_norm": 1.2217979431152344,
-      "learning_rate": 2.564775506777214e-06,
-      "loss": 0.9142,
+      "epoch": 0.06503529153277855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935756032635264e-05,
+      "loss": 1.0899,
       "step": 2368
     },
     {
-      "epoch": 0.7741830065359477,
-      "grad_norm": 1.3154078722000122,
-      "learning_rate": 2.5577014472736407e-06,
-      "loss": 0.9089,
+      "epoch": 0.06506275576062179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993565532651869e-05,
+      "loss": 1.0875,
       "step": 2369
     },
     {
-      "epoch": 0.7745098039215687,
-      "grad_norm": 1.252656102180481,
-      "learning_rate": 2.550635726098547e-06,
-      "loss": 0.9317,
+      "epoch": 0.06509021998846502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935554541787356e-05,
+      "loss": 1.0955,
       "step": 2370
     },
     {
-      "epoch": 0.7748366013071896,
-      "grad_norm": 1.2502864599227905,
-      "learning_rate": 2.5435783511683444e-06,
-      "loss": 0.921,
+      "epoch": 0.06511768421630826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935453678442068e-05,
+      "loss": 1.0588,
       "step": 2371
     },
     {
-      "epoch": 0.7751633986928105,
-      "grad_norm": 1.2517006397247314,
-      "learning_rate": 2.536529330390095e-06,
-      "loss": 0.9143,
+      "epoch": 0.0651451484441515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935352736483626e-05,
+      "loss": 1.2277,
       "step": 2372
     },
     {
-      "epoch": 0.7754901960784314,
-      "grad_norm": 1.2831865549087524,
-      "learning_rate": 2.5294886716615e-06,
-      "loss": 0.9035,
+      "epoch": 0.06517261267199473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993525171591282e-05,
+      "loss": 1.0802,
       "step": 2373
     },
     {
-      "epoch": 0.7758169934640523,
-      "grad_norm": 1.2684791088104248,
-      "learning_rate": 2.5224563828708904e-06,
-      "loss": 0.9251,
+      "epoch": 0.06520007689983796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935150616730457e-05,
+      "loss": 1.2573,
       "step": 2374
     },
     {
-      "epoch": 0.7761437908496732,
-      "grad_norm": 1.2689807415008545,
-      "learning_rate": 2.515432471897221e-06,
-      "loss": 0.9253,
+      "epoch": 0.0652275411276812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935049438937334e-05,
+      "loss": 1.0063,
       "step": 2375
     },
     {
-      "epoch": 0.7764705882352941,
-      "grad_norm": 1.2649152278900146,
-      "learning_rate": 2.5084169466100626e-06,
-      "loss": 0.9493,
+      "epoch": 0.06525500535552443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993494818253425e-05,
+      "loss": 1.213,
       "step": 2376
     },
     {
-      "epoch": 0.776797385620915,
-      "grad_norm": 1.2500135898590088,
-      "learning_rate": 2.5014098148695866e-06,
-      "loss": 0.9516,
+      "epoch": 0.06528246958336767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993484684752201e-05,
+      "loss": 1.1002,
       "step": 2377
     },
     {
-      "epoch": 0.777124183006536,
-      "grad_norm": 1.275122880935669,
-      "learning_rate": 2.4944110845265646e-06,
-      "loss": 0.935,
+      "epoch": 0.0653099338112109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993474543390141e-05,
+      "loss": 1.0641,
       "step": 2378
     },
     {
-      "epoch": 0.7774509803921569,
-      "grad_norm": 1.31528902053833,
-      "learning_rate": 2.487420763422348e-06,
-      "loss": 0.9079,
+      "epoch": 0.06533739803905413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934643941673258e-05,
+      "loss": 1.0889,
       "step": 2379
     },
     {
-      "epoch": 0.7777777777777778,
-      "grad_norm": 1.2789169549942017,
-      "learning_rate": 2.480438859388873e-06,
-      "loss": 0.8865,
+      "epoch": 0.06536486226689736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993454237083835e-05,
+      "loss": 1.0609,
       "step": 2380
     },
     {
-      "epoch": 0.7781045751633987,
-      "grad_norm": 1.2386566400527954,
-      "learning_rate": 2.473465380248643e-06,
-      "loss": 0.934,
+      "epoch": 0.0653923264947406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99344407213975e-05,
+      "loss": 1.0204,
       "step": 2381
     },
     {
-      "epoch": 0.7784313725490196,
-      "grad_norm": 1.2719922065734863,
-      "learning_rate": 2.466500333814723e-06,
-      "loss": 0.9226,
+      "epoch": 0.06541979072258383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934338993351506e-05,
+      "loss": 1.1087,
       "step": 2382
     },
     {
-      "epoch": 0.7787581699346405,
-      "grad_norm": 1.218661904335022,
-      "learning_rate": 2.459543727890733e-06,
-      "loss": 0.9544,
+      "epoch": 0.06544725495042707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934237186701174e-05,
+      "loss": 1.1797,
       "step": 2383
     },
     {
-      "epoch": 0.7790849673202614,
-      "grad_norm": 1.2528438568115234,
-      "learning_rate": 2.452595570270828e-06,
-      "loss": 0.928,
+      "epoch": 0.0654747191782703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934135301447308e-05,
+      "loss": 1.1411,
       "step": 2384
     },
     {
-      "epoch": 0.7794117647058824,
-      "grad_norm": 1.2952110767364502,
-      "learning_rate": 2.4456558687397034e-06,
-      "loss": 0.9396,
+      "epoch": 0.06550218340611354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934033337590713e-05,
+      "loss": 1.0629,
       "step": 2385
     },
     {
-      "epoch": 0.7797385620915033,
-      "grad_norm": 1.245591640472412,
-      "learning_rate": 2.438724631072581e-06,
-      "loss": 0.9408,
+      "epoch": 0.06552964763395677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933931295132202e-05,
+      "loss": 1.1049,
       "step": 2386
     },
     {
-      "epoch": 0.7800653594771242,
-      "grad_norm": 1.4344747066497803,
-      "learning_rate": 2.4318018650351972e-06,
-      "loss": 0.9553,
+      "epoch": 0.06555711186180001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933829174072577e-05,
+      "loss": 1.0811,
       "step": 2387
     },
     {
-      "epoch": 0.7803921568627451,
-      "grad_norm": 1.2329436540603638,
-      "learning_rate": 2.424887578383799e-06,
-      "loss": 0.9424,
+      "epoch": 0.06558457608964324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933726974412648e-05,
+      "loss": 1.0491,
       "step": 2388
     },
     {
-      "epoch": 0.780718954248366,
-      "grad_norm": 1.2247254848480225,
-      "learning_rate": 2.417981778865135e-06,
-      "loss": 0.9176,
+      "epoch": 0.06561204031748648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933624696153222e-05,
+      "loss": 1.0832,
       "step": 2389
     },
     {
-      "epoch": 0.7810457516339869,
-      "grad_norm": 1.2907459735870361,
-      "learning_rate": 2.411084474216434e-06,
-      "loss": 0.9165,
+      "epoch": 0.06563950454532971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933522339295112e-05,
+      "loss": 1.0814,
       "step": 2390
     },
     {
-      "epoch": 0.7813725490196078,
-      "grad_norm": 1.3504539728164673,
-      "learning_rate": 2.4041956721654194e-06,
-      "loss": 0.9559,
+      "epoch": 0.06566696877317295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993341990383912e-05,
+      "loss": 1.1531,
       "step": 2391
     },
     {
-      "epoch": 0.7816993464052288,
-      "grad_norm": 1.2711952924728394,
-      "learning_rate": 2.397315380430284e-06,
-      "loss": 0.9049,
+      "epoch": 0.06569443300101617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933317389786067e-05,
+      "loss": 1.1307,
       "step": 2392
     },
     {
-      "epoch": 0.7820261437908497,
-      "grad_norm": 1.2316068410873413,
-      "learning_rate": 2.3904436067196845e-06,
-      "loss": 0.9684,
+      "epoch": 0.0657218972288594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993321479713676e-05,
+      "loss": 1.2432,
       "step": 2393
     },
     {
-      "epoch": 0.7823529411764706,
-      "grad_norm": 1.2323167324066162,
-      "learning_rate": 2.3835803587327365e-06,
-      "loss": 0.9386,
+      "epoch": 0.06574936145670264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933112125892003e-05,
+      "loss": 1.168,
       "step": 2394
     },
     {
-      "epoch": 0.7826797385620915,
-      "grad_norm": 1.2921934127807617,
-      "learning_rate": 2.3767256441590015e-06,
-      "loss": 0.9223,
+      "epoch": 0.06577682568454588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993300937605262e-05,
+      "loss": 0.9971,
       "step": 2395
     },
     {
-      "epoch": 0.7830065359477124,
-      "grad_norm": 1.2060061693191528,
-      "learning_rate": 2.36987947067848e-06,
-      "loss": 0.9071,
+      "epoch": 0.06580428991238911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993290654761942e-05,
+      "loss": 1.0032,
       "step": 2396
     },
     {
-      "epoch": 0.7833333333333333,
-      "grad_norm": 1.20408296585083,
-      "learning_rate": 2.3630418459616045e-06,
-      "loss": 0.9314,
+      "epoch": 0.06583175414023235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932803640593215e-05,
+      "loss": 1.1154,
       "step": 2397
     },
     {
-      "epoch": 0.7836601307189542,
-      "grad_norm": 1.3334466218948364,
-      "learning_rate": 2.356212777669229e-06,
-      "loss": 0.9527,
+      "epoch": 0.06585921836807558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993270065497482e-05,
+      "loss": 0.9991,
       "step": 2398
     },
     {
-      "epoch": 0.7839869281045752,
-      "grad_norm": 1.2612345218658447,
-      "learning_rate": 2.3493922734526185e-06,
-      "loss": 0.9372,
+      "epoch": 0.06588668259591882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993259759076505e-05,
+      "loss": 0.989,
       "step": 2399
     },
     {
-      "epoch": 0.7843137254901961,
-      "grad_norm": 1.2328543663024902,
-      "learning_rate": 2.342580340953451e-06,
-      "loss": 0.9259,
+      "epoch": 0.06591414682376205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932494447964717e-05,
+      "loss": 1.1305,
       "step": 2400
     },
     {
-      "epoch": 0.784640522875817,
-      "grad_norm": 1.289398431777954,
-      "learning_rate": 2.3357769878037875e-06,
-      "loss": 0.9629,
+      "epoch": 0.06594161105160529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932391226574643e-05,
+      "loss": 1.1019,
       "step": 2401
     },
     {
-      "epoch": 0.7849673202614379,
-      "grad_norm": 1.2641597986221313,
-      "learning_rate": 2.328982221626087e-06,
-      "loss": 0.9093,
+      "epoch": 0.06596907527944852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932287926595643e-05,
+      "loss": 1.0761,
       "step": 2402
     },
     {
-      "epoch": 0.7852941176470588,
-      "grad_norm": 1.2405787706375122,
-      "learning_rate": 2.3221960500331853e-06,
-      "loss": 0.926,
+      "epoch": 0.06599653950729176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932184548028536e-05,
+      "loss": 1.0539,
       "step": 2403
     },
     {
-      "epoch": 0.7856209150326797,
-      "grad_norm": 1.2521872520446777,
-      "learning_rate": 2.3154184806282863e-06,
-      "loss": 0.9175,
+      "epoch": 0.06602400373513499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932081090874133e-05,
+      "loss": 1.138,
       "step": 2404
     },
     {
-      "epoch": 0.7859477124183006,
-      "grad_norm": 1.376996636390686,
-      "learning_rate": 2.308649521004961e-06,
-      "loss": 0.9087,
+      "epoch": 0.06605146796297823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931977555133256e-05,
+      "loss": 1.0485,
       "step": 2405
     },
     {
-      "epoch": 0.7862745098039216,
-      "grad_norm": 1.2366547584533691,
-      "learning_rate": 2.3018891787471308e-06,
-      "loss": 0.9074,
+      "epoch": 0.06607893219082145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993187394080673e-05,
+      "loss": 1.0493,
       "step": 2406
     },
     {
-      "epoch": 0.7866013071895425,
-      "grad_norm": 1.3165206909179688,
-      "learning_rate": 2.2951374614290587e-06,
-      "loss": 0.901,
+      "epoch": 0.06610639641866468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931770247895365e-05,
+      "loss": 1.0862,
       "step": 2407
     },
     {
-      "epoch": 0.7869281045751634,
-      "grad_norm": 1.2374179363250732,
-      "learning_rate": 2.2883943766153503e-06,
-      "loss": 0.9101,
+      "epoch": 0.06613386064650792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993166647639999e-05,
+      "loss": 1.1292,
       "step": 2408
     },
     {
-      "epoch": 0.7872549019607843,
-      "grad_norm": 1.3071197271347046,
-      "learning_rate": 2.2816599318609367e-06,
-      "loss": 0.938,
+      "epoch": 0.06616132487435115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931562626321423e-05,
+      "loss": 1.039,
       "step": 2409
     },
     {
-      "epoch": 0.7875816993464052,
-      "grad_norm": 1.273718237876892,
-      "learning_rate": 2.2749341347110697e-06,
-      "loss": 0.9094,
+      "epoch": 0.06618878910219439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931458697660485e-05,
+      "loss": 1.0203,
       "step": 2410
     },
     {
-      "epoch": 0.7879084967320261,
-      "grad_norm": 1.2722898721694946,
-      "learning_rate": 2.2682169927013153e-06,
-      "loss": 0.8956,
+      "epoch": 0.06621625333003762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931354690418e-05,
+      "loss": 1.0914,
       "step": 2411
     },
     {
-      "epoch": 0.788235294117647,
-      "grad_norm": 1.2811861038208008,
-      "learning_rate": 2.261508513357532e-06,
-      "loss": 0.9381,
+      "epoch": 0.06624371755788086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931250604594784e-05,
+      "loss": 1.0769,
       "step": 2412
     },
     {
-      "epoch": 0.788562091503268,
-      "grad_norm": 1.2351571321487427,
-      "learning_rate": 2.2548087041958834e-06,
-      "loss": 0.9013,
+      "epoch": 0.0662711817857241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993114644019167e-05,
+      "loss": 1.0647,
       "step": 2413
     },
     {
-      "epoch": 0.7888888888888889,
-      "grad_norm": 1.258510708808899,
-      "learning_rate": 2.2481175727228142e-06,
-      "loss": 0.905,
+      "epoch": 0.06629864601356733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993104219720948e-05,
+      "loss": 1.1429,
       "step": 2414
     },
     {
-      "epoch": 0.7892156862745098,
-      "grad_norm": 1.3395025730133057,
-      "learning_rate": 2.2414351264350475e-06,
-      "loss": 0.9526,
+      "epoch": 0.06632611024141057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930937875649034e-05,
+      "loss": 1.0879,
       "step": 2415
     },
     {
-      "epoch": 0.7895424836601307,
-      "grad_norm": 1.3129386901855469,
-      "learning_rate": 2.234761372819577e-06,
-      "loss": 0.9292,
+      "epoch": 0.0663535744692538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930833475511163e-05,
+      "loss": 1.0407,
       "step": 2416
     },
     {
-      "epoch": 0.7898692810457516,
-      "grad_norm": 1.2526644468307495,
-      "learning_rate": 2.2280963193536542e-06,
-      "loss": 0.9576,
+      "epoch": 0.06638103869709704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993072899679669e-05,
+      "loss": 1.0707,
       "step": 2417
     },
     {
-      "epoch": 0.7901960784313725,
-      "grad_norm": 1.296499490737915,
-      "learning_rate": 2.2214399735047856e-06,
-      "loss": 0.9181,
+      "epoch": 0.06640850292494027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930624439506444e-05,
+      "loss": 1.1651,
       "step": 2418
     },
     {
-      "epoch": 0.7905228758169934,
-      "grad_norm": 1.2378432750701904,
-      "learning_rate": 2.2147923427307196e-06,
-      "loss": 0.9229,
+      "epoch": 0.06643596715278349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930519803641245e-05,
+      "loss": 1.0961,
       "step": 2419
     },
     {
-      "epoch": 0.7908496732026143,
-      "grad_norm": 1.2521858215332031,
-      "learning_rate": 2.208153434479442e-06,
-      "loss": 0.903,
+      "epoch": 0.06646343138062673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930415089201933e-05,
+      "loss": 0.9883,
       "step": 2420
     },
     {
-      "epoch": 0.7911764705882353,
-      "grad_norm": 1.2317386865615845,
-      "learning_rate": 2.2015232561891654e-06,
-      "loss": 0.9274,
+      "epoch": 0.06649089560846996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930310296189328e-05,
+      "loss": 1.0604,
       "step": 2421
     },
     {
-      "epoch": 0.7915032679738562,
-      "grad_norm": 1.2375493049621582,
-      "learning_rate": 2.194901815288324e-06,
-      "loss": 0.9518,
+      "epoch": 0.0665183598363132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930205424604262e-05,
+      "loss": 1.0921,
       "step": 2422
     },
     {
-      "epoch": 0.7918300653594771,
-      "grad_norm": 1.3109562397003174,
-      "learning_rate": 2.1882891191955535e-06,
-      "loss": 0.9387,
+      "epoch": 0.06654582406415643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930100474447563e-05,
+      "loss": 1.0609,
       "step": 2423
     },
     {
-      "epoch": 0.792156862745098,
-      "grad_norm": 1.2846870422363281,
-      "learning_rate": 2.1816851753197023e-06,
-      "loss": 0.9083,
+      "epoch": 0.06657328829199967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992999544572006e-05,
+      "loss": 0.9861,
       "step": 2424
     },
     {
-      "epoch": 0.7924836601307189,
-      "grad_norm": 1.2539937496185303,
-      "learning_rate": 2.175089991059809e-06,
-      "loss": 0.9539,
+      "epoch": 0.0666007525198429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992989033842259e-05,
+      "loss": 1.0639,
       "step": 2425
     },
     {
-      "epoch": 0.7928104575163398,
-      "grad_norm": 1.2300678491592407,
-      "learning_rate": 2.168503573805096e-06,
-      "loss": 0.9064,
+      "epoch": 0.06662821674768614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929785152555983e-05,
+      "loss": 1.1142,
       "step": 2426
     },
     {
-      "epoch": 0.7931372549019607,
-      "grad_norm": 1.2440855503082275,
-      "learning_rate": 2.161925930934967e-06,
-      "loss": 0.9205,
+      "epoch": 0.06665568097552937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929679888121065e-05,
+      "loss": 1.0792,
       "step": 2427
     },
     {
-      "epoch": 0.7934640522875817,
-      "grad_norm": 1.3599144220352173,
-      "learning_rate": 2.155357069818995e-06,
-      "loss": 0.9155,
+      "epoch": 0.06668314520337261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929574545118674e-05,
+      "loss": 1.0829,
       "step": 2428
     },
     {
-      "epoch": 0.7937908496732026,
-      "grad_norm": 1.2876702547073364,
-      "learning_rate": 2.1487969978169053e-06,
-      "loss": 0.8885,
+      "epoch": 0.06671060943121584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929469123549647e-05,
+      "loss": 1.0837,
       "step": 2429
     },
     {
-      "epoch": 0.7941176470588235,
-      "grad_norm": 1.2620528936386108,
-      "learning_rate": 2.1422457222785875e-06,
-      "loss": 0.885,
+      "epoch": 0.06673807365905908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929363623414807e-05,
+      "loss": 1.1111,
       "step": 2430
     },
     {
-      "epoch": 0.7944444444444444,
-      "grad_norm": 1.2651883363723755,
-      "learning_rate": 2.1357032505440678e-06,
-      "loss": 0.9458,
+      "epoch": 0.06676553788690232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929258044715002e-05,
+      "loss": 1.1847,
       "step": 2431
     },
     {
-      "epoch": 0.7947712418300653,
-      "grad_norm": 1.3037853240966797,
-      "learning_rate": 2.129169589943514e-06,
-      "loss": 0.9428,
+      "epoch": 0.06679300211474555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929152387451062e-05,
+      "loss": 1.1001,
       "step": 2432
     },
     {
-      "epoch": 0.7950980392156862,
-      "grad_norm": 1.2704211473464966,
-      "learning_rate": 2.1226447477972177e-06,
-      "loss": 0.901,
+      "epoch": 0.06682046634258877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929046651623816e-05,
+      "loss": 1.1072,
       "step": 2433
     },
     {
-      "epoch": 0.7954248366013071,
-      "grad_norm": 1.2793852090835571,
-      "learning_rate": 2.1161287314155955e-06,
-      "loss": 0.944,
+      "epoch": 0.06684793057043201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992894083723411e-05,
+      "loss": 1.1661,
       "step": 2434
     },
     {
-      "epoch": 0.795751633986928,
-      "grad_norm": 1.2998136281967163,
-      "learning_rate": 2.1096215480991654e-06,
-      "loss": 0.9061,
+      "epoch": 0.06687539479827524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928834944282776e-05,
+      "loss": 1.019,
       "step": 2435
     },
     {
-      "epoch": 0.796078431372549,
-      "grad_norm": 1.2822964191436768,
-      "learning_rate": 2.1031232051385606e-06,
-      "loss": 0.9214,
+      "epoch": 0.06690285902611848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928728972770656e-05,
+      "loss": 0.9383,
       "step": 2436
     },
     {
-      "epoch": 0.7964052287581699,
-      "grad_norm": 1.3204345703125,
-      "learning_rate": 2.096633709814504e-06,
-      "loss": 0.9417,
+      "epoch": 0.06693032325396171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928622922698588e-05,
+      "loss": 1.0965,
       "step": 2437
     },
     {
-      "epoch": 0.7967320261437908,
-      "grad_norm": 1.279780626296997,
-      "learning_rate": 2.0901530693978055e-06,
-      "loss": 0.8731,
+      "epoch": 0.06695778748180495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992851679406741e-05,
+      "loss": 1.0015,
       "step": 2438
     },
     {
-      "epoch": 0.7970588235294118,
-      "grad_norm": 1.3213443756103516,
-      "learning_rate": 2.0836812911493553e-06,
-      "loss": 0.9547,
+      "epoch": 0.06698525170964818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928410586877956e-05,
+      "loss": 1.1793,
       "step": 2439
     },
     {
-      "epoch": 0.7973856209150327,
-      "grad_norm": 1.2980109453201294,
-      "learning_rate": 2.077218382320113e-06,
-      "loss": 0.9157,
+      "epoch": 0.06701271593749142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928304301131073e-05,
+      "loss": 1.1204,
       "step": 2440
     },
     {
-      "epoch": 0.7977124183006536,
-      "grad_norm": 1.3073225021362305,
-      "learning_rate": 2.0707643501511033e-06,
-      "loss": 0.8855,
+      "epoch": 0.06704018016533465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99281979368276e-05,
+      "loss": 1.0542,
       "step": 2441
     },
     {
-      "epoch": 0.7980392156862746,
-      "grad_norm": 1.2879061698913574,
-      "learning_rate": 2.0643192018734027e-06,
-      "loss": 0.9037,
+      "epoch": 0.06706764439317789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928091493968383e-05,
+      "loss": 1.0798,
       "step": 2442
     },
     {
-      "epoch": 0.7983660130718955,
-      "grad_norm": 1.239933729171753,
-      "learning_rate": 2.0578829447081363e-06,
-      "loss": 0.946,
+      "epoch": 0.06709510862102112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927984972554257e-05,
+      "loss": 1.065,
       "step": 2443
     },
     {
-      "epoch": 0.7986928104575164,
-      "grad_norm": 1.2766157388687134,
-      "learning_rate": 2.0514555858664663e-06,
-      "loss": 0.9084,
+      "epoch": 0.06712257284886436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992787837258607e-05,
+      "loss": 1.1034,
       "step": 2444
     },
     {
-      "epoch": 0.7990196078431373,
-      "grad_norm": 1.3080322742462158,
-      "learning_rate": 2.04503713254959e-06,
-      "loss": 0.8843,
+      "epoch": 0.0671500370767076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927771694064662e-05,
+      "loss": 1.1745,
       "step": 2445
     },
     {
-      "epoch": 0.7993464052287582,
-      "grad_norm": 1.2926350831985474,
-      "learning_rate": 2.038627591948715e-06,
-      "loss": 0.9369,
+      "epoch": 0.06717750130455082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992766493699088e-05,
+      "loss": 1.1372,
       "step": 2446
     },
     {
-      "epoch": 0.7996732026143791,
-      "grad_norm": 1.2722233533859253,
-      "learning_rate": 2.0322269712450758e-06,
-      "loss": 0.9402,
+      "epoch": 0.06720496553239405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927558101365565e-05,
+      "loss": 1.1458,
       "step": 2447
     },
     {
-      "epoch": 0.8,
-      "grad_norm": 1.3083336353302002,
-      "learning_rate": 2.025835277609908e-06,
-      "loss": 0.9234,
+      "epoch": 0.06723242976023729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992745118718957e-05,
+      "loss": 1.0118,
       "step": 2448
     },
     {
-      "epoch": 0.800326797385621,
-      "grad_norm": 1.2524272203445435,
-      "learning_rate": 2.0194525182044443e-06,
-      "loss": 0.9071,
+      "epoch": 0.06725989398808052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992734419446373e-05,
+      "loss": 1.0883,
       "step": 2449
     },
     {
-      "epoch": 0.8006535947712419,
-      "grad_norm": 1.354598045349121,
-      "learning_rate": 2.0130787001799103e-06,
-      "loss": 0.906,
+      "epoch": 0.06728735821592376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99272371231889e-05,
+      "loss": 1.1127,
       "step": 2450
     },
     {
-      "epoch": 0.8009803921568628,
-      "grad_norm": 1.3774058818817139,
-      "learning_rate": 2.0067138306775126e-06,
-      "loss": 0.9246,
+      "epoch": 0.06731482244376699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927129973365923e-05,
+      "loss": 0.9654,
       "step": 2451
     },
     {
-      "epoch": 0.8013071895424837,
-      "grad_norm": 1.274056077003479,
-      "learning_rate": 2.000357916828428e-06,
-      "loss": 0.9484,
+      "epoch": 0.06734228667161023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992702274499565e-05,
+      "loss": 1.1434,
       "step": 2452
     },
     {
-      "epoch": 0.8016339869281046,
-      "grad_norm": 1.2856240272521973,
-      "learning_rate": 1.9940109657538053e-06,
-      "loss": 0.9743,
+      "epoch": 0.06736975089945346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992691543807893e-05,
+      "loss": 1.0519,
       "step": 2453
     },
     {
-      "epoch": 0.8019607843137255,
-      "grad_norm": 1.3214199542999268,
-      "learning_rate": 1.987672984564749e-06,
-      "loss": 0.9223,
+      "epoch": 0.0673972151272967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926808052616605e-05,
+      "loss": 1.1125,
       "step": 2454
     },
     {
-      "epoch": 0.8022875816993464,
-      "grad_norm": 1.2718156576156616,
-      "learning_rate": 1.9813439803623135e-06,
-      "loss": 0.9696,
+      "epoch": 0.06742467935513993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992670058860953e-05,
+      "loss": 0.959,
       "step": 2455
     },
     {
-      "epoch": 0.8026143790849674,
-      "grad_norm": 1.3085253238677979,
-      "learning_rate": 1.975023960237499e-06,
-      "loss": 0.9266,
+      "epoch": 0.06745214358298317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926593046058556e-05,
+      "loss": 1.1812,
       "step": 2456
     },
     {
-      "epoch": 0.8029411764705883,
-      "grad_norm": 1.2514809370040894,
-      "learning_rate": 1.9687129312712296e-06,
-      "loss": 0.9594,
+      "epoch": 0.0674796078108264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926485424964533e-05,
+      "loss": 1.0667,
       "step": 2457
     },
     {
-      "epoch": 0.8032679738562092,
-      "grad_norm": 1.2652406692504883,
-      "learning_rate": 1.962410900534367e-06,
-      "loss": 0.9022,
+      "epoch": 0.06750707203866964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992637772532831e-05,
+      "loss": 1.0616,
       "step": 2458
     },
     {
-      "epoch": 0.8035947712418301,
-      "grad_norm": 1.3976317644119263,
-      "learning_rate": 1.9561178750876873e-06,
-      "loss": 0.9649,
+      "epoch": 0.06753453626651287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926269947150745e-05,
+      "loss": 1.1208,
       "step": 2459
     },
     {
-      "epoch": 0.803921568627451,
-      "grad_norm": 1.2796179056167603,
-      "learning_rate": 1.949833861981877e-06,
-      "loss": 0.9539,
+      "epoch": 0.0675620004943561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926162090432685e-05,
+      "loss": 1.149,
       "step": 2460
     },
     {
-      "epoch": 0.8042483660130719,
-      "grad_norm": 1.2063108682632446,
-      "learning_rate": 1.9435588682575236e-06,
-      "loss": 0.9198,
+      "epoch": 0.06758946472219933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926054155174986e-05,
+      "loss": 1.2006,
       "step": 2461
     },
     {
-      "epoch": 0.8045751633986928,
-      "grad_norm": 1.2902705669403076,
-      "learning_rate": 1.9372929009451115e-06,
-      "loss": 0.9339,
+      "epoch": 0.06761692895004257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99259461413785e-05,
+      "loss": 1.0485,
       "step": 2462
     },
     {
-      "epoch": 0.8049019607843138,
-      "grad_norm": 1.3115543127059937,
-      "learning_rate": 1.9310359670650127e-06,
-      "loss": 0.9222,
+      "epoch": 0.0676443931778858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925838049044084e-05,
+      "loss": 1.0391,
       "step": 2463
     },
     {
-      "epoch": 0.8052287581699347,
-      "grad_norm": 1.3245325088500977,
-      "learning_rate": 1.924788073627474e-06,
-      "loss": 0.9228,
+      "epoch": 0.06767185740572904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925729878172593e-05,
+      "loss": 1.0475,
       "step": 2464
     },
     {
-      "epoch": 0.8055555555555556,
-      "grad_norm": 1.219936728477478,
-      "learning_rate": 1.918549227632619e-06,
-      "loss": 0.9367,
+      "epoch": 0.06769932163357227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992562162876488e-05,
+      "loss": 1.0158,
       "step": 2465
     },
     {
-      "epoch": 0.8058823529411765,
-      "grad_norm": 1.2851636409759521,
-      "learning_rate": 1.9123194360704288e-06,
-      "loss": 0.8784,
+      "epoch": 0.0677267858614155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925513300821806e-05,
+      "loss": 1.018,
       "step": 2466
     },
     {
-      "epoch": 0.8062091503267974,
-      "grad_norm": 1.215273380279541,
-      "learning_rate": 1.906098705920748e-06,
-      "loss": 0.8971,
+      "epoch": 0.06775425008925874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925404894344228e-05,
+      "loss": 0.9663,
       "step": 2467
     },
     {
-      "epoch": 0.8065359477124183,
-      "grad_norm": 1.2453721761703491,
-      "learning_rate": 1.8998870441532569e-06,
-      "loss": 0.9483,
+      "epoch": 0.06778171431710198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925296409332998e-05,
+      "loss": 1.2128,
       "step": 2468
     },
     {
-      "epoch": 0.8068627450980392,
-      "grad_norm": 1.2820699214935303,
-      "learning_rate": 1.8936844577274849e-06,
-      "loss": 0.9397,
+      "epoch": 0.06780917854494521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992518784578898e-05,
+      "loss": 1.0249,
       "step": 2469
     },
     {
-      "epoch": 0.8071895424836601,
-      "grad_norm": 1.2658807039260864,
-      "learning_rate": 1.887490953592791e-06,
-      "loss": 0.9005,
+      "epoch": 0.06783664277278845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925079203713032e-05,
+      "loss": 1.0515,
       "step": 2470
     },
     {
-      "epoch": 0.8075163398692811,
-      "grad_norm": 1.2363642454147339,
-      "learning_rate": 1.8813065386883588e-06,
-      "loss": 0.9432,
+      "epoch": 0.06786410700063168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992497048310601e-05,
+      "loss": 1.1494,
       "step": 2471
     },
     {
-      "epoch": 0.807843137254902,
-      "grad_norm": 1.2711753845214844,
-      "learning_rate": 1.875131219943187e-06,
-      "loss": 0.9372,
+      "epoch": 0.06789157122847492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992486168396878e-05,
+      "loss": 1.0495,
       "step": 2472
     },
     {
-      "epoch": 0.8081699346405229,
-      "grad_norm": 1.251408576965332,
-      "learning_rate": 1.8689650042760865e-06,
-      "loss": 0.9263,
+      "epoch": 0.06791903545631815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99247528063022e-05,
+      "loss": 1.0318,
       "step": 2473
     },
     {
-      "epoch": 0.8084967320261438,
-      "grad_norm": 1.2155338525772095,
-      "learning_rate": 1.8628078985956622e-06,
-      "loss": 0.943,
+      "epoch": 0.06794649968416137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924643850107128e-05,
+      "loss": 1.0432,
       "step": 2474
     },
     {
-      "epoch": 0.8088235294117647,
-      "grad_norm": 1.4260796308517456,
-      "learning_rate": 1.856659909800318e-06,
-      "loss": 0.9415,
+      "epoch": 0.06797396391200461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992453481538443e-05,
+      "loss": 1.1065,
       "step": 2475
     },
     {
-      "epoch": 0.8091503267973856,
-      "grad_norm": 1.2938611507415771,
-      "learning_rate": 1.8505210447782418e-06,
-      "loss": 0.9319,
+      "epoch": 0.06800142813984784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924425702134972e-05,
+      "loss": 1.2001,
       "step": 2476
     },
     {
-      "epoch": 0.8094771241830065,
-      "grad_norm": 1.3026469945907593,
-      "learning_rate": 1.8443913104073984e-06,
-      "loss": 0.9413,
+      "epoch": 0.06802889236769108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992431651035961e-05,
+      "loss": 0.9718,
       "step": 2477
     },
     {
-      "epoch": 0.8098039215686275,
-      "grad_norm": 1.247814655303955,
-      "learning_rate": 1.8382707135555255e-06,
-      "loss": 0.9156,
+      "epoch": 0.06805635659553431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992420724005921e-05,
+      "loss": 1.1703,
       "step": 2478
     },
     {
-      "epoch": 0.8101307189542484,
-      "grad_norm": 1.1740690469741821,
-      "learning_rate": 1.832159261080122e-06,
-      "loss": 0.9385,
+      "epoch": 0.06808382082337755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924097891234643e-05,
+      "loss": 1.0421,
       "step": 2479
     },
     {
-      "epoch": 0.8104575163398693,
-      "grad_norm": 1.339051604270935,
-      "learning_rate": 1.8260569598284361e-06,
-      "loss": 0.926,
+      "epoch": 0.06811128505122079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923988463886767e-05,
+      "loss": 1.0094,
       "step": 2480
     },
     {
-      "epoch": 0.8107843137254902,
-      "grad_norm": 1.2642390727996826,
-      "learning_rate": 1.81996381663747e-06,
-      "loss": 0.9029,
+      "epoch": 0.06813874927906402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992387895801645e-05,
+      "loss": 1.1234,
       "step": 2481
     },
     {
-      "epoch": 0.8111111111111111,
-      "grad_norm": 1.3194509744644165,
-      "learning_rate": 1.8138798383339618e-06,
-      "loss": 0.9547,
+      "epoch": 0.06816621350690726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923769373624555e-05,
+      "loss": 1.1024,
       "step": 2482
     },
     {
-      "epoch": 0.811437908496732,
-      "grad_norm": 1.289520263671875,
-      "learning_rate": 1.807805031734382e-06,
-      "loss": 0.9597,
+      "epoch": 0.06819367773475049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923659710711954e-05,
+      "loss": 0.9959,
       "step": 2483
     },
     {
-      "epoch": 0.8117647058823529,
-      "grad_norm": 1.2713595628738403,
-      "learning_rate": 1.8017394036449276e-06,
-      "loss": 0.9251,
+      "epoch": 0.06822114196259373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923549969279518e-05,
+      "loss": 1.114,
       "step": 2484
     },
     {
-      "epoch": 0.8120915032679739,
-      "grad_norm": 1.2834360599517822,
-      "learning_rate": 1.7956829608615067e-06,
-      "loss": 0.9222,
+      "epoch": 0.06824860619043696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992344014932811e-05,
+      "loss": 1.0325,
       "step": 2485
     },
     {
-      "epoch": 0.8124183006535948,
-      "grad_norm": 1.230804204940796,
-      "learning_rate": 1.7896357101697405e-06,
-      "loss": 0.983,
+      "epoch": 0.0682760704182802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923330250858594e-05,
+      "loss": 1.1561,
       "step": 2486
     },
     {
-      "epoch": 0.8127450980392157,
-      "grad_norm": 1.2444958686828613,
-      "learning_rate": 1.783597658344951e-06,
-      "loss": 0.943,
+      "epoch": 0.06830353464612342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923220273871848e-05,
+      "loss": 1.0612,
       "step": 2487
     },
     {
-      "epoch": 0.8130718954248366,
-      "grad_norm": 1.2800990343093872,
-      "learning_rate": 1.7775688121521518e-06,
-      "loss": 0.8984,
+      "epoch": 0.06833099887396665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923110218368743e-05,
+      "loss": 1.0841,
       "step": 2488
     },
     {
-      "epoch": 0.8133986928104575,
-      "grad_norm": 1.2413051128387451,
-      "learning_rate": 1.7715491783460447e-06,
-      "loss": 0.9263,
+      "epoch": 0.06835846310180989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923000084350142e-05,
+      "loss": 1.1021,
       "step": 2489
     },
     {
-      "epoch": 0.8137254901960784,
-      "grad_norm": 1.245615005493164,
-      "learning_rate": 1.7655387636710097e-06,
-      "loss": 0.9345,
+      "epoch": 0.06838592732965312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992288987181692e-05,
+      "loss": 1.0625,
       "step": 2490
     },
     {
-      "epoch": 0.8140522875816993,
-      "grad_norm": 1.2100154161453247,
-      "learning_rate": 1.7595375748610943e-06,
-      "loss": 0.9344,
+      "epoch": 0.06841339155749636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992277958076995e-05,
+      "loss": 1.1077,
       "step": 2491
     },
     {
-      "epoch": 0.8143790849673203,
-      "grad_norm": 1.2270171642303467,
-      "learning_rate": 1.7535456186400123e-06,
-      "loss": 0.9137,
+      "epoch": 0.0684408557853396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922669211210105e-05,
+      "loss": 1.058,
       "step": 2492
     },
     {
-      "epoch": 0.8147058823529412,
-      "grad_norm": 1.2654831409454346,
-      "learning_rate": 1.7475629017211348e-06,
-      "loss": 0.8963,
+      "epoch": 0.06846832001318283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922558763138257e-05,
+      "loss": 0.9676,
       "step": 2493
     },
     {
-      "epoch": 0.8150326797385621,
-      "grad_norm": 1.214009165763855,
-      "learning_rate": 1.7415894308074777e-06,
-      "loss": 0.9242,
+      "epoch": 0.06849578424102606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922448236555284e-05,
+      "loss": 1.1082,
       "step": 2494
     },
     {
-      "epoch": 0.815359477124183,
-      "grad_norm": 1.2837978601455688,
-      "learning_rate": 1.7356252125917018e-06,
-      "loss": 0.8952,
+      "epoch": 0.0685232484688693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992233763146205e-05,
+      "loss": 0.9923,
       "step": 2495
     },
     {
-      "epoch": 0.8156862745098039,
-      "grad_norm": 1.2899352312088013,
-      "learning_rate": 1.7296702537560994e-06,
-      "loss": 0.9569,
+      "epoch": 0.06855071269671253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922226947859445e-05,
+      "loss": 1.0873,
       "step": 2496
     },
     {
-      "epoch": 0.8160130718954248,
-      "grad_norm": 1.253238320350647,
-      "learning_rate": 1.723724560972585e-06,
-      "loss": 0.9107,
+      "epoch": 0.06857817692455577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922116185748332e-05,
+      "loss": 0.9955,
       "step": 2497
     },
     {
-      "epoch": 0.8163398692810457,
-      "grad_norm": 1.230694055557251,
-      "learning_rate": 1.7177881409026954e-06,
-      "loss": 0.9162,
+      "epoch": 0.068605641152399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922005345129592e-05,
+      "loss": 1.0898,
       "step": 2498
     },
     {
-      "epoch": 0.8166666666666667,
-      "grad_norm": 1.2592300176620483,
-      "learning_rate": 1.7118610001975777e-06,
-      "loss": 0.926,
+      "epoch": 0.06863310538024224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921894426004104e-05,
+      "loss": 1.0475,
       "step": 2499
     },
     {
-      "epoch": 0.8169934640522876,
-      "grad_norm": 1.2765823602676392,
-      "learning_rate": 1.7059431454979825e-06,
-      "loss": 0.9021,
+      "epoch": 0.06866056960808548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921783428372744e-05,
+      "loss": 0.9718,
       "step": 2500
     },
     {
-      "epoch": 0.8173202614379085,
-      "grad_norm": 1.2469711303710938,
-      "learning_rate": 1.7000345834342592e-06,
-      "loss": 0.9168,
+      "epoch": 0.0686880338359287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992167235223639e-05,
+      "loss": 1.1513,
       "step": 2501
     },
     {
-      "epoch": 0.8176470588235294,
-      "grad_norm": 1.3738840818405151,
-      "learning_rate": 1.6941353206263366e-06,
-      "loss": 0.9382,
+      "epoch": 0.06871549806377193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921561197595923e-05,
+      "loss": 1.051,
       "step": 2502
     },
     {
-      "epoch": 0.8179738562091503,
-      "grad_norm": 1.3343967199325562,
-      "learning_rate": 1.6882453636837338e-06,
-      "loss": 0.9332,
+      "epoch": 0.06874296229161517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921449964452218e-05,
+      "loss": 1.0877,
       "step": 2503
     },
     {
-      "epoch": 0.8183006535947712,
-      "grad_norm": 1.32282292842865,
-      "learning_rate": 1.682364719205537e-06,
-      "loss": 0.9286,
+      "epoch": 0.0687704265194584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921338652806157e-05,
+      "loss": 1.0597,
       "step": 2504
     },
     {
-      "epoch": 0.8186274509803921,
-      "grad_norm": 1.235652208328247,
-      "learning_rate": 1.6764933937804083e-06,
-      "loss": 0.9578,
+      "epoch": 0.06879789074730164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921227262658624e-05,
+      "loss": 1.1329,
       "step": 2505
     },
     {
-      "epoch": 0.818954248366013,
-      "grad_norm": 1.2428207397460938,
-      "learning_rate": 1.6706313939865604e-06,
-      "loss": 0.9163,
+      "epoch": 0.06882535497514487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921115794010496e-05,
+      "loss": 1.0934,
       "step": 2506
     },
     {
-      "epoch": 0.819281045751634,
-      "grad_norm": 1.273455262184143,
-      "learning_rate": 1.6647787263917615e-06,
-      "loss": 0.9319,
+      "epoch": 0.06885281920298811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921004246862656e-05,
+      "loss": 1.1306,
       "step": 2507
     },
     {
-      "epoch": 0.8196078431372549,
-      "grad_norm": 1.267368197441101,
-      "learning_rate": 1.6589353975533174e-06,
-      "loss": 0.9448,
+      "epoch": 0.06888028343083134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992089262121599e-05,
+      "loss": 1.0812,
       "step": 2508
     },
     {
-      "epoch": 0.8199346405228758,
-      "grad_norm": 1.290126085281372,
-      "learning_rate": 1.653101414018079e-06,
-      "loss": 0.887,
+      "epoch": 0.06890774765867458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920780917071377e-05,
+      "loss": 1.1402,
       "step": 2509
     },
     {
-      "epoch": 0.8202614379084967,
-      "grad_norm": 1.4156934022903442,
-      "learning_rate": 1.6472767823224235e-06,
-      "loss": 0.9329,
+      "epoch": 0.06893521188651781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99206691344297e-05,
+      "loss": 1.1331,
       "step": 2510
     },
     {
-      "epoch": 0.8205882352941176,
-      "grad_norm": 1.3654197454452515,
-      "learning_rate": 1.6414615089922504e-06,
-      "loss": 0.9365,
+      "epoch": 0.06896267611436105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992055727329185e-05,
+      "loss": 1.0605,
       "step": 2511
     },
     {
-      "epoch": 0.8209150326797385,
-      "grad_norm": 1.2668730020523071,
-      "learning_rate": 1.6356556005429758e-06,
-      "loss": 0.9496,
+      "epoch": 0.06899014034220428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920445333658705e-05,
+      "loss": 1.2794,
       "step": 2512
     },
     {
-      "epoch": 0.8212418300653594,
-      "grad_norm": 1.3182365894317627,
-      "learning_rate": 1.6298590634795186e-06,
-      "loss": 0.9557,
+      "epoch": 0.06901760457004752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920333315531154e-05,
+      "loss": 1.059,
       "step": 2513
     },
     {
-      "epoch": 0.8215686274509804,
-      "grad_norm": 1.234191656112671,
-      "learning_rate": 1.6240719042963004e-06,
-      "loss": 0.9302,
+      "epoch": 0.06904506879789074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920221218910085e-05,
+      "loss": 1.1665,
       "step": 2514
     },
     {
-      "epoch": 0.8218954248366013,
-      "grad_norm": 1.236884355545044,
-      "learning_rate": 1.6182941294772402e-06,
-      "loss": 0.9316,
+      "epoch": 0.06907253302573398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920109043796385e-05,
+      "loss": 1.2139,
       "step": 2515
     },
     {
-      "epoch": 0.8222222222222222,
-      "grad_norm": 1.2898272275924683,
-      "learning_rate": 1.6125257454957365e-06,
-      "loss": 0.9393,
+      "epoch": 0.06909999725357721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919996790190933e-05,
+      "loss": 1.098,
       "step": 2516
     },
     {
-      "epoch": 0.8225490196078431,
-      "grad_norm": 1.40749990940094,
-      "learning_rate": 1.6067667588146707e-06,
-      "loss": 0.9511,
+      "epoch": 0.06912746148142045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991988445809463e-05,
+      "loss": 1.1144,
       "step": 2517
     },
     {
-      "epoch": 0.822875816993464,
-      "grad_norm": 1.3073242902755737,
-      "learning_rate": 1.6010171758863958e-06,
-      "loss": 0.9598,
+      "epoch": 0.06915492570926368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919772047508355e-05,
+      "loss": 1.1187,
       "step": 2518
     },
     {
-      "epoch": 0.8232026143790849,
-      "grad_norm": 1.307183027267456,
-      "learning_rate": 1.595277003152722e-06,
-      "loss": 0.9554,
+      "epoch": 0.06918238993710692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919659558433002e-05,
+      "loss": 1.0134,
       "step": 2519
     },
     {
-      "epoch": 0.8235294117647058,
-      "grad_norm": 1.2258145809173584,
-      "learning_rate": 1.589546247044923e-06,
-      "loss": 0.915,
+      "epoch": 0.06920985416495015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991954699086946e-05,
+      "loss": 1.1634,
       "step": 2520
     },
     {
-      "epoch": 0.8238562091503268,
-      "grad_norm": 1.3120123147964478,
-      "learning_rate": 1.5838249139837204e-06,
-      "loss": 0.9066,
+      "epoch": 0.06923731839279339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991943434481862e-05,
+      "loss": 1.1604,
       "step": 2521
     },
     {
-      "epoch": 0.8241830065359477,
-      "grad_norm": 1.2677026987075806,
-      "learning_rate": 1.5781130103792797e-06,
-      "loss": 0.8949,
+      "epoch": 0.06926478262063662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919321620281372e-05,
+      "loss": 1.0041,
       "step": 2522
     },
     {
-      "epoch": 0.8245098039215686,
-      "grad_norm": 1.2530571222305298,
-      "learning_rate": 1.5724105426311987e-06,
-      "loss": 0.9222,
+      "epoch": 0.06929224684847986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991920881725861e-05,
+      "loss": 1.0792,
       "step": 2523
     },
     {
-      "epoch": 0.8248366013071895,
-      "grad_norm": 1.2881101369857788,
-      "learning_rate": 1.5667175171285054e-06,
-      "loss": 0.9041,
+      "epoch": 0.06931971107632309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919095935751226e-05,
+      "loss": 1.044,
       "step": 2524
     },
     {
-      "epoch": 0.8251633986928104,
-      "grad_norm": 1.2486236095428467,
-      "learning_rate": 1.5610339402496478e-06,
-      "loss": 0.9279,
+      "epoch": 0.06934717530416633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991898297576011e-05,
+      "loss": 1.0608,
       "step": 2525
     },
     {
-      "epoch": 0.8254901960784313,
-      "grad_norm": 1.268133521080017,
-      "learning_rate": 1.5553598183624885e-06,
-      "loss": 0.9349,
+      "epoch": 0.06937463953200956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918869937286163e-05,
+      "loss": 1.1415,
       "step": 2526
     },
     {
-      "epoch": 0.8258169934640522,
-      "grad_norm": 1.2701220512390137,
-      "learning_rate": 1.549695157824296e-06,
-      "loss": 0.9367,
+      "epoch": 0.0694021037598528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991875682033027e-05,
+      "loss": 1.0312,
       "step": 2527
     },
     {
-      "epoch": 0.8261437908496732,
-      "grad_norm": 1.229984998703003,
-      "learning_rate": 1.5440399649817384e-06,
-      "loss": 0.9253,
+      "epoch": 0.06942956798769602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918643624893336e-05,
+      "loss": 1.1516,
       "step": 2528
     },
     {
-      "epoch": 0.8264705882352941,
-      "grad_norm": 1.3237847089767456,
-      "learning_rate": 1.538394246170879e-06,
-      "loss": 0.9353,
+      "epoch": 0.06945703221553925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918530350976248e-05,
+      "loss": 0.9898,
       "step": 2529
     },
     {
-      "epoch": 0.826797385620915,
-      "grad_norm": 1.238158106803894,
-      "learning_rate": 1.5327580077171589e-06,
-      "loss": 0.9004,
+      "epoch": 0.06948449644338249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918416998579907e-05,
+      "loss": 1.0712,
       "step": 2530
     },
     {
-      "epoch": 0.8271241830065359,
-      "grad_norm": 1.244997262954712,
-      "learning_rate": 1.5271312559354057e-06,
-      "loss": 0.9473,
+      "epoch": 0.06951196067122573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918303567705205e-05,
+      "loss": 0.8809,
       "step": 2531
     },
     {
-      "epoch": 0.8274509803921568,
-      "grad_norm": 1.232391595840454,
-      "learning_rate": 1.5215139971298131e-06,
-      "loss": 0.9407,
+      "epoch": 0.06953942489906896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991819005835305e-05,
+      "loss": 1.1584,
       "step": 2532
     },
     {
-      "epoch": 0.8277777777777777,
-      "grad_norm": 1.287225365638733,
-      "learning_rate": 1.5159062375939414e-06,
-      "loss": 0.9037,
+      "epoch": 0.0695668891269122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918076470524328e-05,
+      "loss": 1.0686,
       "step": 2533
     },
     {
-      "epoch": 0.8281045751633986,
-      "grad_norm": 1.285146951675415,
-      "learning_rate": 1.5103079836107082e-06,
-      "loss": 0.8895,
+      "epoch": 0.06959435335475543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917962804219946e-05,
+      "loss": 1.2073,
       "step": 2534
     },
     {
-      "epoch": 0.8284313725490197,
-      "grad_norm": 1.334654688835144,
-      "learning_rate": 1.5047192414523814e-06,
-      "loss": 0.9771,
+      "epoch": 0.06962181758259867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99178490594408e-05,
+      "loss": 0.9807,
       "step": 2535
     },
     {
-      "epoch": 0.8287581699346406,
-      "grad_norm": 1.235201120376587,
-      "learning_rate": 1.499140017380566e-06,
-      "loss": 0.9013,
+      "epoch": 0.0696492818104419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917735236187788e-05,
+      "loss": 1.0618,
       "step": 2536
     },
     {
-      "epoch": 0.8290849673202615,
-      "grad_norm": 1.3004368543624878,
-      "learning_rate": 1.49357031764621e-06,
-      "loss": 0.9153,
+      "epoch": 0.06967674603828514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917621334461813e-05,
+      "loss": 1.1707,
       "step": 2537
     },
     {
-      "epoch": 0.8294117647058824,
-      "grad_norm": 1.2786165475845337,
-      "learning_rate": 1.4880101484895882e-06,
-      "loss": 0.9541,
+      "epoch": 0.06970421026612837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917507354263778e-05,
+      "loss": 1.0225,
       "step": 2538
     },
     {
-      "epoch": 0.8297385620915033,
-      "grad_norm": 1.2525187730789185,
-      "learning_rate": 1.4824595161402965e-06,
-      "loss": 0.9212,
+      "epoch": 0.06973167449397161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917393295594584e-05,
+      "loss": 1.127,
       "step": 2539
     },
     {
-      "epoch": 0.8300653594771242,
-      "grad_norm": 1.4153162240982056,
-      "learning_rate": 1.4769184268172465e-06,
-      "loss": 0.9422,
+      "epoch": 0.06975913872181484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991727915845513e-05,
+      "loss": 1.0257,
       "step": 2540
     },
     {
-      "epoch": 0.8303921568627451,
-      "grad_norm": 1.3215506076812744,
-      "learning_rate": 1.4713868867286606e-06,
-      "loss": 0.9217,
+      "epoch": 0.06978660294965806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917164942846324e-05,
+      "loss": 1.107,
       "step": 2541
     },
     {
-      "epoch": 0.8307189542483661,
-      "grad_norm": 1.2581850290298462,
-      "learning_rate": 1.4658649020720539e-06,
-      "loss": 0.9556,
+      "epoch": 0.0698140671775013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917050648769066e-05,
+      "loss": 1.0578,
       "step": 2542
     },
     {
-      "epoch": 0.831045751633987,
-      "grad_norm": 1.2921464443206787,
-      "learning_rate": 1.460352479034245e-06,
-      "loss": 0.9115,
+      "epoch": 0.06984153140534453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916936276224262e-05,
+      "loss": 0.9871,
       "step": 2543
     },
     {
-      "epoch": 0.8313725490196079,
-      "grad_norm": 1.2434353828430176,
-      "learning_rate": 1.4548496237913335e-06,
-      "loss": 0.9248,
+      "epoch": 0.06986899563318777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991682182521282e-05,
+      "loss": 1.034,
       "step": 2544
     },
     {
-      "epoch": 0.8316993464052288,
-      "grad_norm": 1.2290167808532715,
-      "learning_rate": 1.449356342508702e-06,
-      "loss": 0.9402,
+      "epoch": 0.069896459861031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991670729573564e-05,
+      "loss": 1.0264,
       "step": 2545
     },
     {
-      "epoch": 0.8320261437908497,
-      "grad_norm": 1.27121901512146,
-      "learning_rate": 1.4438726413410077e-06,
-      "loss": 0.9281,
+      "epoch": 0.06992392408887424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991659268779363e-05,
+      "loss": 1.1647,
       "step": 2546
     },
     {
-      "epoch": 0.8323529411764706,
-      "grad_norm": 1.2455936670303345,
-      "learning_rate": 1.4383985264321708e-06,
-      "loss": 0.9253,
+      "epoch": 0.06995138831671747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99164780013877e-05,
+      "loss": 1.166,
       "step": 2547
     },
     {
-      "epoch": 0.8326797385620915,
-      "grad_norm": 1.238486886024475,
-      "learning_rate": 1.4329340039153738e-06,
-      "loss": 0.9351,
+      "epoch": 0.06997885254456071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991636323651875e-05,
+      "loss": 1.0738,
       "step": 2548
     },
     {
-      "epoch": 0.8330065359477125,
-      "grad_norm": 1.3052226305007935,
-      "learning_rate": 1.42747907991305e-06,
-      "loss": 0.9029,
+      "epoch": 0.07000631677240395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916248393187695e-05,
+      "loss": 1.099,
       "step": 2549
     },
     {
-      "epoch": 0.8333333333333334,
-      "grad_norm": 1.2347023487091064,
-      "learning_rate": 1.4220337605368816e-06,
-      "loss": 0.9024,
+      "epoch": 0.07003378100024718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991613347139544e-05,
+      "loss": 1.1089,
       "step": 2550
     },
     {
-      "epoch": 0.8336601307189543,
-      "grad_norm": 1.2405554056167603,
-      "learning_rate": 1.4165980518877853e-06,
-      "loss": 0.9431,
+      "epoch": 0.07006124522809042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916018471142896e-05,
+      "loss": 1.053,
       "step": 2551
     },
     {
-      "epoch": 0.8339869281045752,
-      "grad_norm": 1.3104034662246704,
-      "learning_rate": 1.4111719600559159e-06,
-      "loss": 0.9001,
+      "epoch": 0.07008870945593365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915903392430976e-05,
+      "loss": 1.0657,
       "step": 2552
     },
     {
-      "epoch": 0.8343137254901961,
-      "grad_norm": 1.237195372581482,
-      "learning_rate": 1.405755491120645e-06,
-      "loss": 0.9206,
+      "epoch": 0.07011617368377689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915788235260586e-05,
+      "loss": 1.1153,
       "step": 2553
     },
     {
-      "epoch": 0.834640522875817,
-      "grad_norm": 1.2994108200073242,
-      "learning_rate": 1.4003486511505715e-06,
-      "loss": 0.9148,
+      "epoch": 0.07014363791162012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915672999632638e-05,
+      "loss": 1.0791,
       "step": 2554
     },
     {
-      "epoch": 0.8349673202614379,
-      "grad_norm": 1.2617149353027344,
-      "learning_rate": 1.3949514462035008e-06,
-      "loss": 0.9763,
+      "epoch": 0.07017110213946334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915557685548042e-05,
+      "loss": 1.1215,
       "step": 2555
     },
     {
-      "epoch": 0.8352941176470589,
-      "grad_norm": 1.279016137123108,
-      "learning_rate": 1.3895638823264447e-06,
-      "loss": 0.915,
+      "epoch": 0.07019856636730658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915442293007715e-05,
+      "loss": 1.0444,
       "step": 2556
     },
     {
-      "epoch": 0.8356209150326798,
-      "grad_norm": 1.3476613759994507,
-      "learning_rate": 1.3841859655556166e-06,
-      "loss": 0.9377,
+      "epoch": 0.07022603059514981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991532682201257e-05,
+      "loss": 1.0671,
       "step": 2557
     },
     {
-      "epoch": 0.8359477124183007,
-      "grad_norm": 1.2920169830322266,
-      "learning_rate": 1.3788177019164117e-06,
-      "loss": 0.9094,
+      "epoch": 0.07025349482299305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915211272563516e-05,
+      "loss": 1.0356,
       "step": 2558
     },
     {
-      "epoch": 0.8362745098039216,
-      "grad_norm": 1.257989764213562,
-      "learning_rate": 1.3734590974234208e-06,
-      "loss": 0.9184,
+      "epoch": 0.07028095905083628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991509564466147e-05,
+      "loss": 1.04,
       "step": 2559
     },
     {
-      "epoch": 0.8366013071895425,
-      "grad_norm": 1.2586430311203003,
-      "learning_rate": 1.3681101580804057e-06,
-      "loss": 0.9068,
+      "epoch": 0.07030842327867952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914979938307345e-05,
+      "loss": 1.1252,
       "step": 2560
     },
     {
-      "epoch": 0.8369281045751634,
-      "grad_norm": 1.3634520769119263,
-      "learning_rate": 1.3627708898803038e-06,
-      "loss": 0.9235,
+      "epoch": 0.07033588750652275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991486415350206e-05,
+      "loss": 1.1205,
       "step": 2561
     },
     {
-      "epoch": 0.8372549019607843,
-      "grad_norm": 1.270514726638794,
-      "learning_rate": 1.3574412988052132e-06,
-      "loss": 0.9266,
+      "epoch": 0.07036335173436599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914748290246528e-05,
+      "loss": 1.0973,
       "step": 2562
     },
     {
-      "epoch": 0.8375816993464053,
-      "grad_norm": 1.2610677480697632,
-      "learning_rate": 1.3521213908263931e-06,
-      "loss": 0.9545,
+      "epoch": 0.07039081596220922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991463234854167e-05,
+      "loss": 1.2068,
       "step": 2563
     },
     {
-      "epoch": 0.8379084967320262,
-      "grad_norm": 1.1846345663070679,
-      "learning_rate": 1.3468111719042497e-06,
-      "loss": 0.953,
+      "epoch": 0.07041828019005246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914516328388396e-05,
+      "loss": 1.1735,
       "step": 2564
     },
     {
-      "epoch": 0.8382352941176471,
-      "grad_norm": 1.2262256145477295,
-      "learning_rate": 1.3415106479883366e-06,
-      "loss": 0.9063,
+      "epoch": 0.0704457444178957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914400229787628e-05,
+      "loss": 1.0961,
       "step": 2565
     },
     {
-      "epoch": 0.838562091503268,
-      "grad_norm": 1.245896577835083,
-      "learning_rate": 1.3362198250173454e-06,
-      "loss": 0.9119,
+      "epoch": 0.07047320864573893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914284052740286e-05,
+      "loss": 1.1719,
       "step": 2566
     },
     {
-      "epoch": 0.8388888888888889,
-      "grad_norm": 1.2760426998138428,
-      "learning_rate": 1.3309387089190973e-06,
-      "loss": 0.9496,
+      "epoch": 0.07050067287358217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991416779724729e-05,
+      "loss": 1.1396,
       "step": 2567
     },
     {
-      "epoch": 0.8392156862745098,
-      "grad_norm": 1.242173671722412,
-      "learning_rate": 1.3256673056105385e-06,
-      "loss": 0.8577,
+      "epoch": 0.07052813710142539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991405146330956e-05,
+      "loss": 1.0206,
       "step": 2568
     },
     {
-      "epoch": 0.8395424836601307,
-      "grad_norm": 1.245143175125122,
-      "learning_rate": 1.3204056209977345e-06,
-      "loss": 0.954,
+      "epoch": 0.07055560132926862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991393505092801e-05,
+      "loss": 1.1346,
       "step": 2569
     },
     {
-      "epoch": 0.8398692810457516,
-      "grad_norm": 1.2576987743377686,
-      "learning_rate": 1.3151536609758586e-06,
-      "loss": 0.9323,
+      "epoch": 0.07058306555711186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913818560103567e-05,
+      "loss": 1.117,
       "step": 2570
     },
     {
-      "epoch": 0.8401960784313726,
-      "grad_norm": 1.271610140800476,
-      "learning_rate": 1.3099114314291927e-06,
-      "loss": 0.9626,
+      "epoch": 0.07061052978495509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913701990837147e-05,
+      "loss": 1.0635,
       "step": 2571
     },
     {
-      "epoch": 0.8405228758169935,
-      "grad_norm": 1.2318177223205566,
-      "learning_rate": 1.3046789382311132e-06,
-      "loss": 0.9205,
+      "epoch": 0.07063799401279833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913585343129683e-05,
+      "loss": 1.0403,
       "step": 2572
     },
     {
-      "epoch": 0.8408496732026144,
-      "grad_norm": 1.2818197011947632,
-      "learning_rate": 1.2994561872440893e-06,
-      "loss": 0.9434,
+      "epoch": 0.07066545824064156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991346861698209e-05,
+      "loss": 1.0539,
       "step": 2573
     },
     {
-      "epoch": 0.8411764705882353,
-      "grad_norm": 1.3024132251739502,
-      "learning_rate": 1.294243184319679e-06,
-      "loss": 0.9427,
+      "epoch": 0.0706929224684848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913351812395294e-05,
+      "loss": 0.9869,
       "step": 2574
     },
     {
-      "epoch": 0.8415032679738562,
-      "grad_norm": 1.2664527893066406,
-      "learning_rate": 1.2890399352985094e-06,
-      "loss": 0.9161,
+      "epoch": 0.07072038669632803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913234929370215e-05,
+      "loss": 1.0511,
       "step": 2575
     },
     {
-      "epoch": 0.8418300653594771,
-      "grad_norm": 1.222354531288147,
-      "learning_rate": 1.2838464460102862e-06,
-      "loss": 0.9502,
+      "epoch": 0.07074785092417127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913117967907785e-05,
+      "loss": 1.0831,
       "step": 2576
     },
     {
-      "epoch": 0.842156862745098,
-      "grad_norm": 1.279577374458313,
-      "learning_rate": 1.2786627222737801e-06,
-      "loss": 0.9526,
+      "epoch": 0.0707753151520145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991300092800892e-05,
+      "loss": 1.0792,
       "step": 2577
     },
     {
-      "epoch": 0.842483660130719,
-      "grad_norm": 1.241087555885315,
-      "learning_rate": 1.273488769896819e-06,
-      "loss": 0.9266,
+      "epoch": 0.07080277937985774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991288380967456e-05,
+      "loss": 0.9586,
       "step": 2578
     },
     {
-      "epoch": 0.8428104575163399,
-      "grad_norm": 1.2717089653015137,
-      "learning_rate": 1.2683245946762824e-06,
-      "loss": 0.9291,
+      "epoch": 0.07083024360770097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912766612905618e-05,
+      "loss": 1.0596,
       "step": 2579
     },
     {
-      "epoch": 0.8431372549019608,
-      "grad_norm": 1.27631676197052,
-      "learning_rate": 1.2631702023980997e-06,
-      "loss": 0.9277,
+      "epoch": 0.07085770783554421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912649337703027e-05,
+      "loss": 1.0592,
       "step": 2580
     },
     {
-      "epoch": 0.8434640522875817,
-      "grad_norm": 1.3116049766540527,
-      "learning_rate": 1.2580255988372315e-06,
-      "loss": 0.9103,
+      "epoch": 0.07088517206338744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912531984067717e-05,
+      "loss": 1.1275,
       "step": 2581
     },
     {
-      "epoch": 0.8437908496732026,
-      "grad_norm": 1.258224606513977,
-      "learning_rate": 1.252890789757678e-06,
-      "loss": 0.9075,
+      "epoch": 0.07091263629123067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912414552000615e-05,
+      "loss": 1.1045,
       "step": 2582
     },
     {
-      "epoch": 0.8441176470588235,
-      "grad_norm": 1.2783156633377075,
-      "learning_rate": 1.2477657809124632e-06,
-      "loss": 0.9324,
+      "epoch": 0.0709401005190739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912297041502647e-05,
+      "loss": 1.0473,
       "step": 2583
     },
     {
-      "epoch": 0.8444444444444444,
-      "grad_norm": 1.329314947128296,
-      "learning_rate": 1.2426505780436326e-06,
-      "loss": 0.9391,
+      "epoch": 0.07096756474691714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912179452574743e-05,
+      "loss": 1.0285,
       "step": 2584
     },
     {
-      "epoch": 0.8447712418300654,
-      "grad_norm": 1.2848305702209473,
-      "learning_rate": 1.2375451868822418e-06,
-      "loss": 0.9248,
+      "epoch": 0.07099502897476037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912061785217837e-05,
+      "loss": 1.0869,
       "step": 2585
     },
     {
-      "epoch": 0.8450980392156863,
-      "grad_norm": 1.3426408767700195,
-      "learning_rate": 1.2324496131483588e-06,
-      "loss": 0.9372,
+      "epoch": 0.0710224932026036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991194403943286e-05,
+      "loss": 1.1362,
       "step": 2586
     },
     {
-      "epoch": 0.8454248366013072,
-      "grad_norm": 1.308241844177246,
-      "learning_rate": 1.2273638625510441e-06,
-      "loss": 0.9373,
+      "epoch": 0.07104995743044684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991182621522074e-05,
+      "loss": 1.0498,
       "step": 2587
     },
     {
-      "epoch": 0.8457516339869281,
-      "grad_norm": 1.269370198249817,
-      "learning_rate": 1.2222879407883592e-06,
-      "loss": 0.9006,
+      "epoch": 0.07107742165829008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911708312582415e-05,
+      "loss": 1.0917,
       "step": 2588
     },
     {
-      "epoch": 0.846078431372549,
-      "grad_norm": 1.25048828125,
-      "learning_rate": 1.2172218535473524e-06,
-      "loss": 0.9982,
+      "epoch": 0.07110488588613331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911590331518813e-05,
+      "loss": 1.1228,
       "step": 2589
     },
     {
-      "epoch": 0.8464052287581699,
-      "grad_norm": 1.3376715183258057,
-      "learning_rate": 1.2121656065040498e-06,
-      "loss": 0.9462,
+      "epoch": 0.07113235011397655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991147227203087e-05,
+      "loss": 1.0336,
       "step": 2590
     },
     {
-      "epoch": 0.8467320261437908,
-      "grad_norm": 1.333617091178894,
-      "learning_rate": 1.2071192053234581e-06,
-      "loss": 0.9243,
+      "epoch": 0.07115981434181978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911354134119517e-05,
+      "loss": 1.0073,
       "step": 2591
     },
     {
-      "epoch": 0.8470588235294118,
-      "grad_norm": 1.2359426021575928,
-      "learning_rate": 1.2020826556595477e-06,
-      "loss": 0.9353,
+      "epoch": 0.07118727856966302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991123591778569e-05,
+      "loss": 1.0904,
       "step": 2592
     },
     {
-      "epoch": 0.8473856209150327,
-      "grad_norm": 1.2946254014968872,
-      "learning_rate": 1.1970559631552537e-06,
-      "loss": 0.9258,
+      "epoch": 0.07121474279750625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911117623030324e-05,
+      "loss": 1.0218,
       "step": 2593
     },
     {
-      "epoch": 0.8477124183006536,
-      "grad_norm": 1.3344491720199585,
-      "learning_rate": 1.1920391334424675e-06,
-      "loss": 0.9631,
+      "epoch": 0.07124220702534949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991099924985436e-05,
+      "loss": 1.0292,
       "step": 2594
     },
     {
-      "epoch": 0.8480392156862745,
-      "grad_norm": 1.2986805438995361,
-      "learning_rate": 1.1870321721420297e-06,
-      "loss": 0.9218,
+      "epoch": 0.07126967125319271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991088079825873e-05,
+      "loss": 1.0377,
       "step": 2595
     },
     {
-      "epoch": 0.8483660130718954,
-      "grad_norm": 1.2520842552185059,
-      "learning_rate": 1.182035084863724e-06,
-      "loss": 0.9021,
+      "epoch": 0.07129713548103594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910762268244372e-05,
+      "loss": 1.1017,
       "step": 2596
     },
     {
-      "epoch": 0.8486928104575163,
-      "grad_norm": 1.263347864151001,
-      "learning_rate": 1.1770478772062733e-06,
-      "loss": 0.8954,
+      "epoch": 0.07132459970887918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910643659812225e-05,
+      "loss": 1.0414,
       "step": 2597
     },
     {
-      "epoch": 0.8490196078431372,
-      "grad_norm": 1.2558622360229492,
-      "learning_rate": 1.1720705547573263e-06,
-      "loss": 0.9372,
+      "epoch": 0.07135206393672242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910524972963224e-05,
+      "loss": 1.0808,
       "step": 2598
     },
     {
-      "epoch": 0.8493464052287582,
-      "grad_norm": 1.2917793989181519,
-      "learning_rate": 1.1671031230934627e-06,
-      "loss": 0.9609,
+      "epoch": 0.07137952816456565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991040620769831e-05,
+      "loss": 1.0962,
       "step": 2599
     },
     {
-      "epoch": 0.8496732026143791,
-      "grad_norm": 1.260246992111206,
-      "learning_rate": 1.1621455877801757e-06,
-      "loss": 0.9628,
+      "epoch": 0.07140699239240889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910287364018422e-05,
+      "loss": 1.1166,
       "step": 2600
     },
     {
-      "epoch": 0.85,
-      "grad_norm": 1.3364595174789429,
-      "learning_rate": 1.1571979543718747e-06,
-      "loss": 0.9413,
+      "epoch": 0.07143445662025212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910168441924506e-05,
+      "loss": 1.0919,
       "step": 2601
     },
     {
-      "epoch": 0.8503267973856209,
-      "grad_norm": 1.2791861295700073,
-      "learning_rate": 1.1522602284118733e-06,
-      "loss": 0.9109,
+      "epoch": 0.07146192084809536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910049441417497e-05,
+      "loss": 1.0475,
       "step": 2602
     },
     {
-      "epoch": 0.8506535947712418,
-      "grad_norm": 1.2146074771881104,
-      "learning_rate": 1.1473324154323817e-06,
-      "loss": 0.9173,
+      "epoch": 0.07148938507593859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909930362498337e-05,
+      "loss": 0.9493,
       "step": 2603
     },
     {
-      "epoch": 0.8509803921568627,
-      "grad_norm": 1.2449605464935303,
-      "learning_rate": 1.1424145209545079e-06,
-      "loss": 0.937,
+      "epoch": 0.07151684930378183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990981120516797e-05,
+      "loss": 1.0276,
       "step": 2604
     },
     {
-      "epoch": 0.8513071895424836,
-      "grad_norm": 1.2609751224517822,
-      "learning_rate": 1.1375065504882465e-06,
-      "loss": 0.9575,
+      "epoch": 0.07154431353162506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909691969427337e-05,
+      "loss": 1.1204,
       "step": 2605
     },
     {
-      "epoch": 0.8516339869281045,
-      "grad_norm": 1.2887301445007324,
-      "learning_rate": 1.1326085095324712e-06,
-      "loss": 0.9145,
+      "epoch": 0.0715717777594683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909572655277383e-05,
+      "loss": 1.0877,
       "step": 2606
     },
     {
-      "epoch": 0.8519607843137255,
-      "grad_norm": 1.3085263967514038,
-      "learning_rate": 1.1277204035749334e-06,
-      "loss": 0.9669,
+      "epoch": 0.07159924198731153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990945326271905e-05,
+      "loss": 1.0777,
       "step": 2607
     },
     {
-      "epoch": 0.8522875816993464,
-      "grad_norm": 1.2644747495651245,
-      "learning_rate": 1.1228422380922533e-06,
-      "loss": 0.9112,
+      "epoch": 0.07162670621515477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909333791753286e-05,
+      "loss": 0.9225,
       "step": 2608
     },
     {
-      "epoch": 0.8526143790849673,
-      "grad_norm": 1.2316032648086548,
-      "learning_rate": 1.1179740185499077e-06,
-      "loss": 0.9324,
+      "epoch": 0.07165417044299799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909214242381037e-05,
+      "loss": 1.0415,
       "step": 2609
     },
     {
-      "epoch": 0.8529411764705882,
-      "grad_norm": 1.2026655673980713,
-      "learning_rate": 1.1131157504022394e-06,
-      "loss": 0.9228,
+      "epoch": 0.07168163467084122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909094614603243e-05,
+      "loss": 1.0499,
       "step": 2610
     },
     {
-      "epoch": 0.8532679738562091,
-      "grad_norm": 1.255262017250061,
-      "learning_rate": 1.1082674390924352e-06,
-      "loss": 0.937,
+      "epoch": 0.07170909889868446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990897490842085e-05,
+      "loss": 1.1094,
       "step": 2611
     },
     {
-      "epoch": 0.85359477124183,
-      "grad_norm": 1.2536380290985107,
-      "learning_rate": 1.1034290900525279e-06,
-      "loss": 0.9517,
+      "epoch": 0.0717365631265277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908855123834816e-05,
+      "loss": 1.0692,
       "step": 2612
     },
     {
-      "epoch": 0.8539215686274509,
-      "grad_norm": 1.2326328754425049,
-      "learning_rate": 1.0986007087033901e-06,
-      "loss": 0.9085,
+      "epoch": 0.07176402735437093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990873526084608e-05,
+      "loss": 1.0664,
       "step": 2613
     },
     {
-      "epoch": 0.8542483660130719,
-      "grad_norm": 1.2794170379638672,
-      "learning_rate": 1.0937823004547254e-06,
-      "loss": 0.9266,
+      "epoch": 0.07179149158221416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990861531945559e-05,
+      "loss": 1.0757,
       "step": 2614
     },
     {
-      "epoch": 0.8545751633986928,
-      "grad_norm": 1.2699018716812134,
-      "learning_rate": 1.0889738707050634e-06,
-      "loss": 0.9635,
+      "epoch": 0.0718189558100574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908495299664303e-05,
+      "loss": 1.0802,
       "step": 2615
     },
     {
-      "epoch": 0.8549019607843137,
-      "grad_norm": 1.2440378665924072,
-      "learning_rate": 1.0841754248417535e-06,
-      "loss": 0.9314,
+      "epoch": 0.07184642003790064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990837520147316e-05,
+      "loss": 1.1152,
       "step": 2616
     },
     {
-      "epoch": 0.8552287581699346,
-      "grad_norm": 1.2506433725357056,
-      "learning_rate": 1.0793869682409596e-06,
-      "loss": 0.9165,
+      "epoch": 0.07187388426574387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990825502488311e-05,
+      "loss": 1.1052,
       "step": 2617
     },
     {
-      "epoch": 0.8555555555555555,
-      "grad_norm": 1.2906996011734009,
-      "learning_rate": 1.0746085062676548e-06,
-      "loss": 0.9115,
+      "epoch": 0.0719013484935871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908134769895117e-05,
+      "loss": 1.0096,
       "step": 2618
     },
     {
-      "epoch": 0.8558823529411764,
-      "grad_norm": 1.2778472900390625,
-      "learning_rate": 1.069840044275615e-06,
-      "loss": 0.9031,
+      "epoch": 0.07192881272143034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908014436510116e-05,
+      "loss": 1.0651,
       "step": 2619
     },
     {
-      "epoch": 0.8562091503267973,
-      "grad_norm": 1.2150437831878662,
-      "learning_rate": 1.065081587607406e-06,
-      "loss": 0.9382,
+      "epoch": 0.07195627694927358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907894024729073e-05,
+      "loss": 1.0825,
       "step": 2620
     },
     {
-      "epoch": 0.8565359477124183,
-      "grad_norm": 1.2650495767593384,
-      "learning_rate": 1.0603331415943906e-06,
-      "loss": 0.9178,
+      "epoch": 0.07198374117711681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907773534552927e-05,
+      "loss": 1.1292,
       "step": 2621
     },
     {
-      "epoch": 0.8568627450980392,
-      "grad_norm": 1.2262901067733765,
-      "learning_rate": 1.0555947115567123e-06,
-      "loss": 0.9402,
+      "epoch": 0.07201120540496005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907652965982646e-05,
+      "loss": 1.1716,
       "step": 2622
     },
     {
-      "epoch": 0.8571895424836601,
-      "grad_norm": 1.2599016427993774,
-      "learning_rate": 1.050866302803294e-06,
-      "loss": 0.925,
+      "epoch": 0.07203866963280327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990753231901917e-05,
+      "loss": 1.1015,
       "step": 2623
     },
     {
-      "epoch": 0.857516339869281,
-      "grad_norm": 1.270525336265564,
-      "learning_rate": 1.0461479206318303e-06,
-      "loss": 0.9025,
+      "epoch": 0.0720661338606465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907411593663467e-05,
+      "loss": 1.1017,
       "step": 2624
     },
     {
-      "epoch": 0.8578431372549019,
-      "grad_norm": 1.2549470663070679,
-      "learning_rate": 1.041439570328785e-06,
-      "loss": 0.9082,
+      "epoch": 0.07209359808848974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907290789916483e-05,
+      "loss": 1.0772,
       "step": 2625
     },
     {
-      "epoch": 0.8581699346405228,
-      "grad_norm": 1.2663320302963257,
-      "learning_rate": 1.0367412571693746e-06,
-      "loss": 0.9007,
+      "epoch": 0.07212106231633297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907169907779175e-05,
+      "loss": 1.0257,
       "step": 2626
     },
     {
-      "epoch": 0.8584967320261437,
-      "grad_norm": 1.2762349843978882,
-      "learning_rate": 1.0320529864175766e-06,
-      "loss": 0.8877,
+      "epoch": 0.07214852654417621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907048947252502e-05,
+      "loss": 1.0663,
       "step": 2627
     },
     {
-      "epoch": 0.8588235294117647,
-      "grad_norm": 1.27090322971344,
-      "learning_rate": 1.0273747633261144e-06,
-      "loss": 0.9217,
+      "epoch": 0.07217599077201944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906927908337422e-05,
+      "loss": 1.1592,
       "step": 2628
     },
     {
-      "epoch": 0.8591503267973856,
-      "grad_norm": 1.2649651765823364,
-      "learning_rate": 1.0227065931364566e-06,
-      "loss": 0.9397,
+      "epoch": 0.07220345499986268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906806791034888e-05,
+      "loss": 1.0522,
       "step": 2629
     },
     {
-      "epoch": 0.8594771241830066,
-      "grad_norm": 1.2254953384399414,
-      "learning_rate": 1.0180484810788082e-06,
-      "loss": 0.9298,
+      "epoch": 0.07223091922770591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990668559534586e-05,
+      "loss": 1.1626,
       "step": 2630
     },
     {
-      "epoch": 0.8598039215686275,
-      "grad_norm": 1.2577524185180664,
-      "learning_rate": 1.013400432372098e-06,
-      "loss": 0.9291,
+      "epoch": 0.07225838345554915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906564321271304e-05,
+      "loss": 1.1046,
       "step": 2631
     },
     {
-      "epoch": 0.8601307189542484,
-      "grad_norm": 1.3017067909240723,
-      "learning_rate": 1.0087624522239902e-06,
-      "loss": 0.9075,
+      "epoch": 0.07228584768339238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906442968812173e-05,
+      "loss": 1.0996,
       "step": 2632
     },
     {
-      "epoch": 0.8604575163398693,
-      "grad_norm": 1.2761616706848145,
-      "learning_rate": 1.0041345458308615e-06,
-      "loss": 0.8988,
+      "epoch": 0.07231331191123562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906321537969422e-05,
+      "loss": 1.0445,
       "step": 2633
     },
     {
-      "epoch": 0.8607843137254902,
-      "grad_norm": 1.2676093578338623,
-      "learning_rate": 9.995167183778055e-07,
-      "loss": 0.9404,
+      "epoch": 0.07234077613907886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990620002874402e-05,
+      "loss": 1.1243,
       "step": 2634
     },
     {
-      "epoch": 0.8611111111111112,
-      "grad_norm": 1.2793949842453003,
-      "learning_rate": 9.94908975038621e-07,
-      "loss": 0.9115,
+      "epoch": 0.07236824036692209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906078441136926e-05,
+      "loss": 1.1724,
       "step": 2635
     },
     {
-      "epoch": 0.8614379084967321,
-      "grad_norm": 1.2842150926589966,
-      "learning_rate": 9.903113209758098e-07,
-      "loss": 0.9059,
+      "epoch": 0.07239570459476531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905956775149105e-05,
+      "loss": 1.1942,
       "step": 2636
     },
     {
-      "epoch": 0.861764705882353,
-      "grad_norm": 1.2319809198379517,
-      "learning_rate": 9.857237613405679e-07,
-      "loss": 0.9139,
+      "epoch": 0.07242316882260855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905835030781514e-05,
+      "loss": 1.03,
       "step": 2637
     },
     {
-      "epoch": 0.8620915032679739,
-      "grad_norm": 1.2802468538284302,
-      "learning_rate": 9.811463012727863e-07,
-      "loss": 0.9628,
+      "epoch": 0.07245063305045178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990571320803512e-05,
+      "loss": 1.0693,
       "step": 2638
     },
     {
-      "epoch": 0.8624183006535948,
-      "grad_norm": 1.224134087562561,
-      "learning_rate": 9.765789459010343e-07,
-      "loss": 0.9444,
+      "epoch": 0.07247809727829502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905591306910886e-05,
+      "loss": 1.1079,
       "step": 2639
     },
     {
-      "epoch": 0.8627450980392157,
-      "grad_norm": 1.2821130752563477,
-      "learning_rate": 9.720217003425648e-07,
-      "loss": 0.9281,
+      "epoch": 0.07250556150613825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905469327409777e-05,
+      "loss": 1.139,
       "step": 2640
     },
     {
-      "epoch": 0.8630718954248366,
-      "grad_norm": 1.2534217834472656,
-      "learning_rate": 9.674745697032995e-07,
-      "loss": 0.8738,
+      "epoch": 0.07253302573398149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905347269532758e-05,
+      "loss": 1.1578,
       "step": 2641
     },
     {
-      "epoch": 0.8633986928104576,
-      "grad_norm": 1.2449150085449219,
-      "learning_rate": 9.629375590778344e-07,
-      "loss": 0.9371,
+      "epoch": 0.07256048996182472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905225133280792e-05,
+      "loss": 1.0741,
       "step": 2642
     },
     {
-      "epoch": 0.8637254901960785,
-      "grad_norm": 1.2354578971862793,
-      "learning_rate": 9.58410673549416e-07,
-      "loss": 0.9611,
+      "epoch": 0.07258795418966796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905102918654847e-05,
+      "loss": 1.1162,
       "step": 2643
     },
     {
-      "epoch": 0.8640522875816994,
-      "grad_norm": 1.2398710250854492,
-      "learning_rate": 9.538939181899565e-07,
-      "loss": 0.9628,
+      "epoch": 0.0726154184175112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904980625655894e-05,
+      "loss": 1.0639,
       "step": 2644
     },
     {
-      "epoch": 0.8643790849673203,
-      "grad_norm": 1.2976208925247192,
-      "learning_rate": 9.493872980600138e-07,
-      "loss": 0.9152,
+      "epoch": 0.07264288264535443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904858254284896e-05,
+      "loss": 1.0674,
       "step": 2645
     },
     {
-      "epoch": 0.8647058823529412,
-      "grad_norm": 1.2427151203155518,
-      "learning_rate": 9.448908182087913e-07,
-      "loss": 0.9418,
+      "epoch": 0.07267034687319766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904735804542825e-05,
+      "loss": 1.0701,
       "step": 2646
     },
     {
-      "epoch": 0.8650326797385621,
-      "grad_norm": 1.344389796257019,
-      "learning_rate": 9.404044836741343e-07,
-      "loss": 0.9182,
+      "epoch": 0.0726978111010409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904613276430647e-05,
+      "loss": 1.0188,
       "step": 2647
     },
     {
-      "epoch": 0.865359477124183,
-      "grad_norm": 1.231285810470581,
-      "learning_rate": 9.359282994825125e-07,
-      "loss": 0.9338,
+      "epoch": 0.07272527532888413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990449066994933e-05,
+      "loss": 0.9471,
       "step": 2648
     },
     {
-      "epoch": 0.865686274509804,
-      "grad_norm": 1.2618991136550903,
-      "learning_rate": 9.314622706490328e-07,
-      "loss": 0.9313,
+      "epoch": 0.07275273955672737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904367985099848e-05,
+      "loss": 1.0373,
       "step": 2649
     },
     {
-      "epoch": 0.8660130718954249,
-      "grad_norm": 1.2322951555252075,
-      "learning_rate": 9.270064021774194e-07,
-      "loss": 0.9198,
+      "epoch": 0.07278020378457059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990424522188317e-05,
+      "loss": 0.9811,
       "step": 2650
     },
     {
-      "epoch": 0.8663398692810458,
-      "grad_norm": 1.3478643894195557,
-      "learning_rate": 9.225606990600134e-07,
-      "loss": 0.9384,
+      "epoch": 0.07280766801241383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904122380300265e-05,
+      "loss": 1.2004,
       "step": 2651
     },
     {
-      "epoch": 0.8666666666666667,
-      "grad_norm": 1.275430679321289,
-      "learning_rate": 9.181251662777668e-07,
-      "loss": 0.9537,
+      "epoch": 0.07283513224025706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990399946035211e-05,
+      "loss": 1.0863,
       "step": 2652
     },
     {
-      "epoch": 0.8669934640522876,
-      "grad_norm": 1.3156771659851074,
-      "learning_rate": 9.136998088002391e-07,
-      "loss": 0.9537,
+      "epoch": 0.0728625964681003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990387646203967e-05,
+      "loss": 1.0552,
       "step": 2653
     },
     {
-      "epoch": 0.8673202614379085,
-      "grad_norm": 1.3228421211242676,
-      "learning_rate": 9.092846315855841e-07,
-      "loss": 0.9617,
+      "epoch": 0.07289006069594353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990375338536393e-05,
+      "loss": 1.1146,
       "step": 2654
     },
     {
-      "epoch": 0.8676470588235294,
-      "grad_norm": 1.2504810094833374,
-      "learning_rate": 9.048796395805537e-07,
-      "loss": 0.945,
+      "epoch": 0.07291752492378677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903630230325852e-05,
+      "loss": 1.0873,
       "step": 2655
     },
     {
-      "epoch": 0.8679738562091504,
-      "grad_norm": 1.22694993019104,
-      "learning_rate": 9.004848377204878e-07,
-      "loss": 0.9061,
+      "epoch": 0.07294498915163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903506996926415e-05,
+      "loss": 1.1005,
       "step": 2656
     },
     {
-      "epoch": 0.8683006535947713,
-      "grad_norm": 1.2320703268051147,
-      "learning_rate": 8.961002309293099e-07,
-      "loss": 0.9066,
+      "epoch": 0.07297245337947324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903383685166598e-05,
+      "loss": 1.0798,
       "step": 2657
     },
     {
-      "epoch": 0.8686274509803922,
-      "grad_norm": 1.253768801689148,
-      "learning_rate": 8.917258241195181e-07,
-      "loss": 0.9074,
+      "epoch": 0.07299991760731647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990326029504737e-05,
+      "loss": 1.0857,
       "step": 2658
     },
     {
-      "epoch": 0.8689542483660131,
-      "grad_norm": 1.3209984302520752,
-      "learning_rate": 8.873616221921866e-07,
-      "loss": 0.9811,
+      "epoch": 0.07302738183515971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990313682656971e-05,
+      "loss": 1.017,
       "step": 2659
     },
     {
-      "epoch": 0.869281045751634,
-      "grad_norm": 1.2350627183914185,
-      "learning_rate": 8.830076300369517e-07,
-      "loss": 0.9089,
+      "epoch": 0.07305484606300294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903013279734597e-05,
+      "loss": 1.0985,
       "step": 2660
     },
     {
-      "epoch": 0.8696078431372549,
-      "grad_norm": 1.2529828548431396,
-      "learning_rate": 8.786638525320146e-07,
-      "loss": 0.9685,
+      "epoch": 0.07308231029084618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902889654543008e-05,
+      "loss": 1.0767,
       "step": 2661
     },
     {
-      "epoch": 0.8699346405228758,
-      "grad_norm": 1.2313716411590576,
-      "learning_rate": 8.74330294544129e-07,
-      "loss": 0.9385,
+      "epoch": 0.07310977451868941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990276595099592e-05,
+      "loss": 1.0363,
       "step": 2662
     },
     {
-      "epoch": 0.8702614379084967,
-      "grad_norm": 1.254014015197754,
-      "learning_rate": 8.700069609286e-07,
-      "loss": 0.9281,
+      "epoch": 0.07313723874653263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990264216909431e-05,
+      "loss": 1.025,
       "step": 2663
     },
     {
-      "epoch": 0.8705882352941177,
-      "grad_norm": 1.2683019638061523,
-      "learning_rate": 8.656938565292805e-07,
-      "loss": 0.9431,
+      "epoch": 0.07316470297437587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902518308839157e-05,
+      "loss": 1.2172,
       "step": 2664
     },
     {
-      "epoch": 0.8709150326797386,
-      "grad_norm": 1.2454928159713745,
-      "learning_rate": 8.613909861785541e-07,
-      "loss": 0.9286,
+      "epoch": 0.0731921672022191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902394370231448e-05,
+      "loss": 1.0619,
       "step": 2665
     },
     {
-      "epoch": 0.8712418300653595,
-      "grad_norm": 1.2659786939620972,
-      "learning_rate": 8.570983546973466e-07,
-      "loss": 0.9081,
+      "epoch": 0.07321963143006234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990227035327216e-05,
+      "loss": 1.0343,
       "step": 2666
     },
     {
-      "epoch": 0.8715686274509804,
-      "grad_norm": 1.2812650203704834,
-      "learning_rate": 8.528159668951097e-07,
-      "loss": 0.9369,
+      "epoch": 0.07324709565790558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990214625796227e-05,
+      "loss": 1.1074,
       "step": 2667
     },
     {
-      "epoch": 0.8718954248366013,
-      "grad_norm": 1.2951239347457886,
-      "learning_rate": 8.485438275698154e-07,
-      "loss": 0.9263,
+      "epoch": 0.07327455988574881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990202208430276e-05,
+      "loss": 1.1417,
       "step": 2668
     },
     {
-      "epoch": 0.8722222222222222,
-      "grad_norm": 1.275812029838562,
-      "learning_rate": 8.442819415079573e-07,
-      "loss": 0.8884,
+      "epoch": 0.07330202411359205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990189783229462e-05,
+      "loss": 1.0985,
       "step": 2669
     },
     {
-      "epoch": 0.8725490196078431,
-      "grad_norm": 1.267166018486023,
-      "learning_rate": 8.4003031348454e-07,
-      "loss": 0.9208,
+      "epoch": 0.07332948834143528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901773501938825e-05,
+      "loss": 1.0984,
       "step": 2670
     },
     {
-      "epoch": 0.8728758169934641,
-      "grad_norm": 1.2838608026504517,
-      "learning_rate": 8.357889482630721e-07,
-      "loss": 0.9307,
+      "epoch": 0.07335695256927852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901649093236366e-05,
+      "loss": 1.1187,
       "step": 2671
     },
     {
-      "epoch": 0.873202614379085,
-      "grad_norm": 1.2377196550369263,
-      "learning_rate": 8.315578505955679e-07,
-      "loss": 0.9023,
+      "epoch": 0.07338441679712175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901524606188224e-05,
+      "loss": 1.0956,
       "step": 2672
     },
     {
-      "epoch": 0.8735294117647059,
-      "grad_norm": 1.254073977470398,
-      "learning_rate": 8.273370252225354e-07,
-      "loss": 0.9847,
+      "epoch": 0.07341188102496499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901400040795386e-05,
+      "loss": 1.1503,
       "step": 2673
     },
     {
-      "epoch": 0.8738562091503268,
-      "grad_norm": 1.2674115896224976,
-      "learning_rate": 8.231264768729753e-07,
-      "loss": 0.9118,
+      "epoch": 0.07343934525280822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990127539705883e-05,
+      "loss": 1.0558,
       "step": 2674
     },
     {
-      "epoch": 0.8741830065359477,
-      "grad_norm": 1.237385630607605,
-      "learning_rate": 8.189262102643747e-07,
-      "loss": 0.9178,
+      "epoch": 0.07346680948065146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901150674979548e-05,
+      "loss": 1.0673,
       "step": 2675
     },
     {
-      "epoch": 0.8745098039215686,
-      "grad_norm": 1.2555899620056152,
-      "learning_rate": 8.14736230102694e-07,
-      "loss": 0.9383,
+      "epoch": 0.07349427370849469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990102587455853e-05,
+      "loss": 1.0192,
       "step": 2676
     },
     {
-      "epoch": 0.8748366013071895,
-      "grad_norm": 1.243852138519287,
-      "learning_rate": 8.105565410823791e-07,
-      "loss": 0.9601,
+      "epoch": 0.07352173793633791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900900995796755e-05,
+      "loss": 1.023,
       "step": 2677
     },
     {
-      "epoch": 0.8751633986928105,
-      "grad_norm": 1.2326081991195679,
-      "learning_rate": 8.063871478863383e-07,
-      "loss": 0.9171,
+      "epoch": 0.07354920216418115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990077603869522e-05,
+      "loss": 1.1438,
       "step": 2678
     },
     {
-      "epoch": 0.8754901960784314,
-      "grad_norm": 1.307738184928894,
-      "learning_rate": 8.022280551859484e-07,
-      "loss": 0.9204,
+      "epoch": 0.07357666639202438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990065100325491e-05,
+      "loss": 1.0384,
       "step": 2679
     },
     {
-      "epoch": 0.8758169934640523,
-      "grad_norm": 1.2536729574203491,
-      "learning_rate": 7.980792676410409e-07,
-      "loss": 0.9354,
+      "epoch": 0.07360413061986762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900525889476815e-05,
+      "loss": 1.0694,
       "step": 2680
     },
     {
-      "epoch": 0.8761437908496732,
-      "grad_norm": 1.246086597442627,
-      "learning_rate": 7.939407898999108e-07,
-      "loss": 0.9245,
+      "epoch": 0.07363159484771085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990040069736192e-05,
+      "loss": 1.0342,
       "step": 2681
     },
     {
-      "epoch": 0.8764705882352941,
-      "grad_norm": 1.2353016138076782,
-      "learning_rate": 7.898126265992912e-07,
-      "loss": 0.9005,
+      "epoch": 0.07365905907555409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900275426911224e-05,
+      "loss": 1.2064,
       "step": 2682
     },
     {
-      "epoch": 0.876797385620915,
-      "grad_norm": 1.2722978591918945,
-      "learning_rate": 7.856947823643645e-07,
-      "loss": 0.911,
+      "epoch": 0.07368652330339732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900150078125714e-05,
+      "loss": 1.0797,
       "step": 2683
     },
     {
-      "epoch": 0.8771241830065359,
-      "grad_norm": 1.242780089378357,
-      "learning_rate": 7.815872618087506e-07,
-      "loss": 0.9313,
+      "epoch": 0.07371398753124056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900024651006377e-05,
+      "loss": 1.0807,
       "step": 2684
     },
     {
-      "epoch": 0.8774509803921569,
-      "grad_norm": 1.2464956045150757,
-      "learning_rate": 7.774900695345044e-07,
-      "loss": 0.9345,
+      "epoch": 0.0737414517590838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899899145554212e-05,
+      "loss": 1.1365,
       "step": 2685
     },
     {
-      "epoch": 0.8777777777777778,
-      "grad_norm": 1.2380897998809814,
-      "learning_rate": 7.734032101321054e-07,
-      "loss": 0.9203,
+      "epoch": 0.07376891598692703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989977356177021e-05,
+      "loss": 1.1346,
       "step": 2686
     },
     {
-      "epoch": 0.8781045751633987,
-      "grad_norm": 1.218306303024292,
-      "learning_rate": 7.693266881804618e-07,
-      "loss": 0.9259,
+      "epoch": 0.07379638021477027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899647899655367e-05,
+      "loss": 1.1082,
       "step": 2687
     },
     {
-      "epoch": 0.8784313725490196,
-      "grad_norm": 1.2372385263442993,
-      "learning_rate": 7.652605082468912e-07,
-      "loss": 0.9299,
+      "epoch": 0.0738238444426135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899522159210673e-05,
+      "loss": 1.0576,
       "step": 2688
     },
     {
-      "epoch": 0.8787581699346405,
-      "grad_norm": 1.2648085355758667,
-      "learning_rate": 7.612046748871327e-07,
-      "loss": 0.8891,
+      "epoch": 0.07385130867045674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899396340437124e-05,
+      "loss": 1.1025,
       "step": 2689
     },
     {
-      "epoch": 0.8790849673202614,
-      "grad_norm": 1.2616935968399048,
-      "learning_rate": 7.571591926453292e-07,
-      "loss": 0.9612,
+      "epoch": 0.07387877289829996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989927044333572e-05,
+      "loss": 1.1094,
       "step": 2690
     },
     {
-      "epoch": 0.8794117647058823,
-      "grad_norm": 1.2924431562423706,
-      "learning_rate": 7.531240660540251e-07,
-      "loss": 0.9421,
+      "epoch": 0.07390623712614319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899144467907452e-05,
+      "loss": 1.067,
       "step": 2691
     },
     {
-      "epoch": 0.8797385620915033,
-      "grad_norm": 1.2935622930526733,
-      "learning_rate": 7.490992996341662e-07,
-      "loss": 0.9226,
+      "epoch": 0.07393370135398643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899018414153317e-05,
+      "loss": 1.0154,
       "step": 2692
     },
     {
-      "epoch": 0.8800653594771242,
-      "grad_norm": 1.2652051448822021,
-      "learning_rate": 7.450848978950864e-07,
-      "loss": 0.9444,
+      "epoch": 0.07396116558182966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898892282074316e-05,
+      "loss": 1.0011,
       "step": 2693
     },
     {
-      "epoch": 0.8803921568627451,
-      "grad_norm": 1.2621649503707886,
-      "learning_rate": 7.410808653345092e-07,
-      "loss": 0.9498,
+      "epoch": 0.0739886298096729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898766071671442e-05,
+      "loss": 1.0199,
       "step": 2694
     },
     {
-      "epoch": 0.880718954248366,
-      "grad_norm": 1.3639044761657715,
-      "learning_rate": 7.370872064385414e-07,
-      "loss": 0.9275,
+      "epoch": 0.07401609403751613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98986397829457e-05,
+      "loss": 1.101,
       "step": 2695
     },
     {
-      "epoch": 0.8810457516339869,
-      "grad_norm": 1.272204041481018,
-      "learning_rate": 7.331039256816664e-07,
-      "loss": 0.9361,
+      "epoch": 0.07404355826535937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989851341589808e-05,
+      "loss": 1.1004,
       "step": 2696
     },
     {
-      "epoch": 0.8813725490196078,
-      "grad_norm": 1.2545020580291748,
-      "learning_rate": 7.291310275267383e-07,
-      "loss": 0.9226,
+      "epoch": 0.0740710224932026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898386970529593e-05,
+      "loss": 1.0959,
       "step": 2697
     },
     {
-      "epoch": 0.8816993464052287,
-      "grad_norm": 1.254132628440857,
-      "learning_rate": 7.251685164249823e-07,
-      "loss": 0.9396,
+      "epoch": 0.07409848672104584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898260446841233e-05,
+      "loss": 1.1217,
       "step": 2698
     },
     {
-      "epoch": 0.8820261437908496,
-      "grad_norm": 1.2690503597259521,
-      "learning_rate": 7.212163968159802e-07,
-      "loss": 0.9358,
+      "epoch": 0.07412595094888907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898133844834e-05,
+      "loss": 1.0607,
       "step": 2699
     },
     {
-      "epoch": 0.8823529411764706,
-      "grad_norm": 1.2229745388031006,
-      "learning_rate": 7.17274673127677e-07,
-      "loss": 0.9131,
+      "epoch": 0.07415341517673231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98980071645089e-05,
+      "loss": 1.0928,
       "step": 2700
     },
     {
-      "epoch": 0.8826797385620915,
-      "grad_norm": 1.262070894241333,
-      "learning_rate": 7.133433497763686e-07,
-      "loss": 0.9636,
+      "epoch": 0.07418087940457554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897880405866933e-05,
+      "loss": 1.0657,
       "step": 2701
     },
     {
-      "epoch": 0.8830065359477124,
-      "grad_norm": 1.2345844507217407,
-      "learning_rate": 7.094224311666953e-07,
-      "loss": 0.8843,
+      "epoch": 0.07420834363241878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897753568909102e-05,
+      "loss": 1.1337,
       "step": 2702
     },
     {
-      "epoch": 0.8833333333333333,
-      "grad_norm": 1.3319556713104248,
-      "learning_rate": 7.055119216916451e-07,
-      "loss": 0.9533,
+      "epoch": 0.07423580786026202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989762665363641e-05,
+      "loss": 1.0989,
       "step": 2703
     },
     {
-      "epoch": 0.8836601307189542,
-      "grad_norm": 1.2641284465789795,
-      "learning_rate": 7.016118257325399e-07,
-      "loss": 0.9552,
+      "epoch": 0.07426327208810524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897499660049863e-05,
+      "loss": 0.9547,
       "step": 2704
     },
     {
-      "epoch": 0.8839869281045751,
-      "grad_norm": 1.310497522354126,
-      "learning_rate": 6.977221476590334e-07,
-      "loss": 0.9387,
+      "epoch": 0.07429073631594847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897372588150463e-05,
+      "loss": 1.1437,
       "step": 2705
     },
     {
-      "epoch": 0.884313725490196,
-      "grad_norm": 1.2581020593643188,
-      "learning_rate": 6.938428918291107e-07,
-      "loss": 0.9282,
+      "epoch": 0.07431820054379171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989724543793922e-05,
+      "loss": 1.0847,
       "step": 2706
     },
     {
-      "epoch": 0.884640522875817,
-      "grad_norm": 1.249563455581665,
-      "learning_rate": 6.899740625890772e-07,
-      "loss": 0.9213,
+      "epoch": 0.07434566477163494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989711820941713e-05,
+      "loss": 1.0598,
       "step": 2707
     },
     {
-      "epoch": 0.8849673202614379,
-      "grad_norm": 1.231027603149414,
-      "learning_rate": 6.861156642735578e-07,
-      "loss": 0.8998,
+      "epoch": 0.07437312899947818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896990902585216e-05,
+      "loss": 1.1427,
       "step": 2708
     },
     {
-      "epoch": 0.8852941176470588,
-      "grad_norm": 1.2011014223098755,
-      "learning_rate": 6.822677012054924e-07,
-      "loss": 0.9425,
+      "epoch": 0.07440059322732141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896863517444473e-05,
+      "loss": 1.0441,
       "step": 2709
     },
     {
-      "epoch": 0.8856209150326797,
-      "grad_norm": 1.2784180641174316,
-      "learning_rate": 6.784301776961222e-07,
-      "loss": 0.9031,
+      "epoch": 0.07442805745516465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989673605399591e-05,
+      "loss": 1.0492,
       "step": 2710
     },
     {
-      "epoch": 0.8859477124183006,
-      "grad_norm": 1.262056589126587,
-      "learning_rate": 6.746030980450002e-07,
-      "loss": 0.9373,
+      "epoch": 0.07445552168300788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896608512240543e-05,
+      "loss": 1.0224,
       "step": 2711
     },
     {
-      "epoch": 0.8862745098039215,
-      "grad_norm": 1.2727148532867432,
-      "learning_rate": 6.707864665399733e-07,
-      "loss": 0.9222,
+      "epoch": 0.07448298591085112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989648089217937e-05,
+      "loss": 1.0473,
       "step": 2712
     },
     {
-      "epoch": 0.8866013071895424,
-      "grad_norm": 1.3255398273468018,
-      "learning_rate": 6.66980287457184e-07,
-      "loss": 0.9276,
+      "epoch": 0.07451045013869435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989635319381341e-05,
+      "loss": 1.1303,
       "step": 2713
     },
     {
-      "epoch": 0.8869281045751634,
-      "grad_norm": 1.271846055984497,
-      "learning_rate": 6.631845650610647e-07,
-      "loss": 0.9332,
+      "epoch": 0.07453791436653759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896225417143668e-05,
+      "loss": 1.0895,
       "step": 2714
     },
     {
-      "epoch": 0.8872549019607843,
-      "grad_norm": 1.2473992109298706,
-      "learning_rate": 6.593993036043311e-07,
-      "loss": 0.9098,
+      "epoch": 0.07456537859438082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989609756217116e-05,
+      "loss": 1.1898,
       "step": 2715
     },
     {
-      "epoch": 0.8875816993464052,
-      "grad_norm": 1.2631304264068604,
-      "learning_rate": 6.556245073279777e-07,
-      "loss": 0.9332,
+      "epoch": 0.07459284282222406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989596962889689e-05,
+      "loss": 1.1652,
       "step": 2716
     },
     {
-      "epoch": 0.8879084967320261,
-      "grad_norm": 1.3208658695220947,
-      "learning_rate": 6.518601804612734e-07,
-      "loss": 0.9469,
+      "epoch": 0.07462030705006728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989584161732188e-05,
+      "loss": 1.12,
       "step": 2717
     },
     {
-      "epoch": 0.888235294117647,
-      "grad_norm": 1.259142279624939,
-      "learning_rate": 6.481063272217603e-07,
-      "loss": 0.9209,
+      "epoch": 0.07464777127791052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895713527447137e-05,
+      "loss": 1.1631,
       "step": 2718
     },
     {
-      "epoch": 0.8885620915032679,
-      "grad_norm": 1.2566546201705933,
-      "learning_rate": 6.443629518152439e-07,
-      "loss": 0.9198,
+      "epoch": 0.07467523550575375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895585359273672e-05,
+      "loss": 1.1077,
       "step": 2719
     },
     {
-      "epoch": 0.8888888888888888,
-      "grad_norm": 1.2736214399337769,
-      "learning_rate": 6.406300584357905e-07,
-      "loss": 0.9442,
+      "epoch": 0.07470269973359699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895457112802507e-05,
+      "loss": 1.1702,
       "step": 2720
     },
     {
-      "epoch": 0.8892156862745098,
-      "grad_norm": 1.3069831132888794,
-      "learning_rate": 6.369076512657224e-07,
-      "loss": 0.9591,
+      "epoch": 0.07473016396144022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989532878803465e-05,
+      "loss": 1.2433,
       "step": 2721
     },
     {
-      "epoch": 0.8895424836601307,
-      "grad_norm": 1.2684086561203003,
-      "learning_rate": 6.331957344756123e-07,
-      "loss": 0.9261,
+      "epoch": 0.07475762818928346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895200384971117e-05,
+      "loss": 1.1076,
       "step": 2722
     },
     {
-      "epoch": 0.8898692810457516,
-      "grad_norm": 1.2447372674942017,
-      "learning_rate": 6.294943122242802e-07,
-      "loss": 0.9307,
+      "epoch": 0.07478509241712669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895071903612925e-05,
+      "loss": 1.1314,
       "step": 2723
     },
     {
-      "epoch": 0.8901960784313725,
-      "grad_norm": 1.2508488893508911,
-      "learning_rate": 6.258033886587911e-07,
-      "loss": 0.9084,
+      "epoch": 0.07481255664496993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894943343961092e-05,
+      "loss": 1.0743,
       "step": 2724
     },
     {
-      "epoch": 0.8905228758169934,
-      "grad_norm": 1.2724894285202026,
-      "learning_rate": 6.221229679144413e-07,
-      "loss": 0.9273,
+      "epoch": 0.07484002087281316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894814706016635e-05,
+      "loss": 1.0608,
       "step": 2725
     },
     {
-      "epoch": 0.8908496732026144,
-      "grad_norm": 1.2188353538513184,
-      "learning_rate": 6.184530541147649e-07,
-      "loss": 0.8986,
+      "epoch": 0.0748674851006564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894685989780574e-05,
+      "loss": 1.0151,
       "step": 2726
     },
     {
-      "epoch": 0.8911764705882353,
-      "grad_norm": 1.2957991361618042,
-      "learning_rate": 6.147936513715203e-07,
-      "loss": 0.9328,
+      "epoch": 0.07489494932849963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894557195253922e-05,
+      "loss": 1.0978,
       "step": 2727
     },
     {
-      "epoch": 0.8915032679738563,
-      "grad_norm": 1.2794489860534668,
-      "learning_rate": 6.111447637846923e-07,
-      "loss": 0.9493,
+      "epoch": 0.07492241355634287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98944283224377e-05,
+      "loss": 1.0151,
       "step": 2728
     },
     {
-      "epoch": 0.8918300653594772,
-      "grad_norm": 1.2501661777496338,
-      "learning_rate": 6.075063954424831e-07,
-      "loss": 0.8982,
+      "epoch": 0.0749498777841861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989429937133293e-05,
+      "loss": 1.0331,
       "step": 2729
     },
     {
-      "epoch": 0.8921568627450981,
-      "grad_norm": 1.2566659450531006,
-      "learning_rate": 6.038785504213096e-07,
-      "loss": 0.9301,
+      "epoch": 0.07497734201202934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989417034194063e-05,
+      "loss": 0.9747,
       "step": 2730
     },
     {
-      "epoch": 0.892483660130719,
-      "grad_norm": 1.195156455039978,
-      "learning_rate": 6.002612327857971e-07,
-      "loss": 0.9379,
+      "epoch": 0.07500480623987256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989404123426182e-05,
+      "loss": 1.0255,
       "step": 2731
     },
     {
-      "epoch": 0.8928104575163399,
-      "grad_norm": 1.2679314613342285,
-      "learning_rate": 5.966544465887803e-07,
-      "loss": 0.9389,
+      "epoch": 0.0750322704677158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893912048297523e-05,
+      "loss": 1.0334,
       "step": 2732
     },
     {
-      "epoch": 0.8931372549019608,
-      "grad_norm": 1.258272409439087,
-      "learning_rate": 5.930581958712878e-07,
-      "loss": 0.9191,
+      "epoch": 0.07505973469555903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893782784048766e-05,
+      "loss": 1.0983,
       "step": 2733
     },
     {
-      "epoch": 0.8934640522875817,
-      "grad_norm": 1.330064058303833,
-      "learning_rate": 5.894724846625499e-07,
-      "loss": 0.91,
+      "epoch": 0.07508719892340227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989365344151656e-05,
+      "loss": 1.0459,
       "step": 2734
     },
     {
-      "epoch": 0.8937908496732027,
-      "grad_norm": 1.2527068853378296,
-      "learning_rate": 5.858973169799853e-07,
-      "loss": 0.901,
+      "epoch": 0.0751146631512455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893524020701943e-05,
+      "loss": 0.9366,
       "step": 2735
     },
     {
-      "epoch": 0.8941176470588236,
-      "grad_norm": 1.28890860080719,
-      "learning_rate": 5.823326968292009e-07,
-      "loss": 0.9288,
+      "epoch": 0.07514212737908874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989339452160593e-05,
+      "loss": 1.1158,
       "step": 2736
     },
     {
-      "epoch": 0.8944444444444445,
-      "grad_norm": 1.2555824518203735,
-      "learning_rate": 5.787786282039887e-07,
-      "loss": 0.9236,
+      "epoch": 0.07516959160693197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893264944229544e-05,
+      "loss": 1.0409,
       "step": 2737
     },
     {
-      "epoch": 0.8947712418300654,
-      "grad_norm": 1.2087335586547852,
-      "learning_rate": 5.752351150863133e-07,
-      "loss": 0.935,
+      "epoch": 0.0751970558347752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893135288573812e-05,
+      "loss": 1.1351,
       "step": 2738
     },
     {
-      "epoch": 0.8950980392156863,
-      "grad_norm": 1.2583866119384766,
-      "learning_rate": 5.71702161446317e-07,
-      "loss": 0.9234,
+      "epoch": 0.07522452006261844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893005554639764e-05,
+      "loss": 1.085,
       "step": 2739
     },
     {
-      "epoch": 0.8954248366013072,
-      "grad_norm": 1.2879211902618408,
-      "learning_rate": 5.681797712423099e-07,
-      "loss": 0.944,
+      "epoch": 0.07525198429046168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892875742428426e-05,
+      "loss": 1.0823,
       "step": 2740
     },
     {
-      "epoch": 0.8957516339869281,
-      "grad_norm": 1.2292206287384033,
-      "learning_rate": 5.646679484207674e-07,
-      "loss": 0.9178,
+      "epoch": 0.07527944851830491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989274585194082e-05,
+      "loss": 1.1304,
       "step": 2741
     },
     {
-      "epoch": 0.8960784313725491,
-      "grad_norm": 1.252079725265503,
-      "learning_rate": 5.611666969163243e-07,
-      "loss": 0.9114,
+      "epoch": 0.07530691274614815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989261588317798e-05,
+      "loss": 1.0138,
       "step": 2742
     },
     {
-      "epoch": 0.89640522875817,
-      "grad_norm": 1.2552968263626099,
-      "learning_rate": 5.57676020651774e-07,
-      "loss": 0.9468,
+      "epoch": 0.07533437697399138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892485836140926e-05,
+      "loss": 1.1705,
       "step": 2743
     },
     {
-      "epoch": 0.8967320261437909,
-      "grad_norm": 1.2461203336715698,
-      "learning_rate": 5.541959235380578e-07,
-      "loss": 0.8998,
+      "epoch": 0.07536184120183462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892355710830696e-05,
+      "loss": 1.1455,
       "step": 2744
     },
     {
-      "epoch": 0.8970588235294118,
-      "grad_norm": 1.3250987529754639,
-      "learning_rate": 5.50726409474267e-07,
-      "loss": 0.9153,
+      "epoch": 0.07538930542967784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892225507248314e-05,
+      "loss": 1.0606,
       "step": 2745
     },
     {
-      "epoch": 0.8973856209150327,
-      "grad_norm": 1.2983301877975464,
-      "learning_rate": 5.472674823476343e-07,
-      "loss": 0.9558,
+      "epoch": 0.07541676965752107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892095225394815e-05,
+      "loss": 1.0998,
       "step": 2746
     },
     {
-      "epoch": 0.8977124183006536,
-      "grad_norm": 1.2494629621505737,
-      "learning_rate": 5.438191460335296e-07,
-      "loss": 0.9023,
+      "epoch": 0.07544423388536431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891964865271224e-05,
+      "loss": 1.0906,
       "step": 2747
     },
     {
-      "epoch": 0.8980392156862745,
-      "grad_norm": 1.2517595291137695,
-      "learning_rate": 5.403814043954592e-07,
-      "loss": 0.9239,
+      "epoch": 0.07547169811320754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891834426878572e-05,
+      "loss": 1.1346,
       "step": 2748
     },
     {
-      "epoch": 0.8983660130718955,
-      "grad_norm": 1.2146055698394775,
-      "learning_rate": 5.369542612850609e-07,
-      "loss": 0.9305,
+      "epoch": 0.07549916234105078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891703910217898e-05,
+      "loss": 1.148,
       "step": 2749
     },
     {
-      "epoch": 0.8986928104575164,
-      "grad_norm": 1.2924662828445435,
-      "learning_rate": 5.335377205420911e-07,
-      "loss": 0.9037,
+      "epoch": 0.07552662656889401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989157331529023e-05,
+      "loss": 1.0588,
       "step": 2750
     },
     {
-      "epoch": 0.8990196078431373,
-      "grad_norm": 1.2957134246826172,
-      "learning_rate": 5.301317859944333e-07,
-      "loss": 0.9393,
+      "epoch": 0.07555409079673725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891442642096604e-05,
+      "loss": 1.0656,
       "step": 2751
     },
     {
-      "epoch": 0.8993464052287582,
-      "grad_norm": 1.2331953048706055,
-      "learning_rate": 5.267364614580861e-07,
-      "loss": 0.9403,
+      "epoch": 0.07558155502458049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989131189063805e-05,
+      "loss": 1.049,
       "step": 2752
     },
     {
-      "epoch": 0.8996732026143791,
-      "grad_norm": 1.29071044921875,
-      "learning_rate": 5.233517507371611e-07,
-      "loss": 0.9313,
+      "epoch": 0.07560901925242372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989118106091561e-05,
+      "loss": 1.0712,
       "step": 2753
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 1.2715486288070679,
-      "learning_rate": 5.199776576238791e-07,
-      "loss": 0.9129,
+      "epoch": 0.07563648348026696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891050152930308e-05,
+      "loss": 1.1428,
       "step": 2754
     },
     {
-      "epoch": 0.9003267973856209,
-      "grad_norm": 1.289405107498169,
-      "learning_rate": 5.166141858985596e-07,
-      "loss": 0.9357,
+      "epoch": 0.07566394770811019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989091916668319e-05,
+      "loss": 1.0663,
       "step": 2755
     },
     {
-      "epoch": 0.9006535947712418,
-      "grad_norm": 1.2185696363449097,
-      "learning_rate": 5.132613393296293e-07,
-      "loss": 0.9466,
+      "epoch": 0.07569141193595343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890788102175284e-05,
+      "loss": 1.1256,
       "step": 2756
     },
     {
-      "epoch": 0.9009803921568628,
-      "grad_norm": 1.1906970739364624,
-      "learning_rate": 5.099191216736065e-07,
-      "loss": 0.925,
+      "epoch": 0.07571887616379666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890656959407635e-05,
+      "loss": 1.1087,
       "step": 2757
     },
     {
-      "epoch": 0.9013071895424837,
-      "grad_norm": 1.3140509128570557,
-      "learning_rate": 5.065875366751016e-07,
-      "loss": 0.9457,
+      "epoch": 0.07574634039163988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890525738381274e-05,
+      "loss": 1.1396,
       "step": 2758
     },
     {
-      "epoch": 0.9016339869281046,
-      "grad_norm": 1.2899158000946045,
-      "learning_rate": 5.03266588066812e-07,
-      "loss": 0.9644,
+      "epoch": 0.07577380461948312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890394439097244e-05,
+      "loss": 1.0466,
       "step": 2759
     },
     {
-      "epoch": 0.9019607843137255,
-      "grad_norm": 1.3413043022155762,
-      "learning_rate": 4.999562795695201e-07,
-      "loss": 0.9286,
+      "epoch": 0.07580126884732635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890263061556578e-05,
+      "loss": 1.1848,
       "step": 2760
     },
     {
-      "epoch": 0.9022875816993464,
-      "grad_norm": 1.243686556816101,
-      "learning_rate": 4.966566148920815e-07,
-      "loss": 0.9406,
+      "epoch": 0.07582873307516959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890131605760324e-05,
+      "loss": 1.1213,
       "step": 2761
     },
     {
-      "epoch": 0.9026143790849673,
-      "grad_norm": 1.2461599111557007,
-      "learning_rate": 4.933675977314334e-07,
-      "loss": 0.9162,
+      "epoch": 0.07585619730301282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989000007170952e-05,
+      "loss": 1.1282,
       "step": 2762
     },
     {
-      "epoch": 0.9029411764705882,
-      "grad_norm": 1.2679201364517212,
-      "learning_rate": 4.900892317725803e-07,
-      "loss": 0.9382,
+      "epoch": 0.07588366153085606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98898684594052e-05,
+      "loss": 1.1395,
       "step": 2763
     },
     {
-      "epoch": 0.9032679738562092,
-      "grad_norm": 1.1814968585968018,
-      "learning_rate": 4.868215206885918e-07,
-      "loss": 0.9208,
+      "epoch": 0.0759111257586993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988973676884841e-05,
+      "loss": 1.174,
       "step": 2764
     },
     {
-      "epoch": 0.9035947712418301,
-      "grad_norm": 1.2741985321044922,
-      "learning_rate": 4.835644681406027e-07,
-      "loss": 0.9289,
+      "epoch": 0.07593858998654253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988960500004019e-05,
+      "loss": 1.0916,
       "step": 2765
     },
     {
-      "epoch": 0.903921568627451,
-      "grad_norm": 1.246484637260437,
-      "learning_rate": 4.80318077777805e-07,
-      "loss": 0.9249,
+      "epoch": 0.07596605421438576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988947315298159e-05,
+      "loss": 1.0152,
       "step": 2766
     },
     {
-      "epoch": 0.9042483660130719,
-      "grad_norm": 1.2597062587738037,
-      "learning_rate": 4.770823532374436e-07,
-      "loss": 0.9157,
+      "epoch": 0.075993518442229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889341227673642e-05,
+      "loss": 1.0829,
       "step": 2767
     },
     {
-      "epoch": 0.9045751633986928,
-      "grad_norm": 1.2482421398162842,
-      "learning_rate": 4.738572981448153e-07,
-      "loss": 0.8996,
+      "epoch": 0.07602098267007223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889209224117397e-05,
+      "loss": 1.0278,
       "step": 2768
     },
     {
-      "epoch": 0.9049019607843137,
-      "grad_norm": 1.2951656579971313,
-      "learning_rate": 4.706429161132608e-07,
-      "loss": 0.9251,
+      "epoch": 0.07604844689791547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889077142313898e-05,
+      "loss": 1.0559,
       "step": 2769
     },
     {
-      "epoch": 0.9052287581699346,
-      "grad_norm": 1.2994251251220703,
-      "learning_rate": 4.6743921074416387e-07,
-      "loss": 0.9386,
+      "epoch": 0.0760759111257587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888944982264192e-05,
+      "loss": 1.084,
       "step": 2770
     },
     {
-      "epoch": 0.9055555555555556,
-      "grad_norm": 1.32150137424469,
-      "learning_rate": 4.6424618562694866e-07,
-      "loss": 0.9399,
+      "epoch": 0.07610337535360194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988881274396932e-05,
+      "loss": 1.0812,
       "step": 2771
     },
     {
-      "epoch": 0.9058823529411765,
-      "grad_norm": 1.3015503883361816,
-      "learning_rate": 4.61063844339068e-07,
-      "loss": 0.9345,
+      "epoch": 0.07613083958144516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888680427430335e-05,
+      "loss": 1.0657,
       "step": 2772
     },
     {
-      "epoch": 0.9062091503267974,
-      "grad_norm": 1.2445698976516724,
-      "learning_rate": 4.578921904460076e-07,
-      "loss": 0.9306,
+      "epoch": 0.0761583038092884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888548032648278e-05,
+      "loss": 1.0336,
       "step": 2773
     },
     {
-      "epoch": 0.9065359477124183,
-      "grad_norm": 1.2169032096862793,
-      "learning_rate": 4.547312275012794e-07,
-      "loss": 0.9287,
+      "epoch": 0.07618576803713163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888415559624197e-05,
+      "loss": 1.124,
       "step": 2774
     },
     {
-      "epoch": 0.9068627450980392,
-      "grad_norm": 1.2824161052703857,
-      "learning_rate": 4.51580959046416e-07,
-      "loss": 0.9378,
+      "epoch": 0.07621323226497487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888283008359142e-05,
+      "loss": 1.0223,
       "step": 2775
     },
     {
-      "epoch": 0.9071895424836601,
-      "grad_norm": 1.302354335784912,
-      "learning_rate": 4.4844138861096954e-07,
-      "loss": 0.9042,
+      "epoch": 0.0762406964928181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988815037885416e-05,
+      "loss": 1.038,
       "step": 2776
     },
     {
-      "epoch": 0.907516339869281,
-      "grad_norm": 1.3240875005722046,
-      "learning_rate": 4.453125197125052e-07,
-      "loss": 0.9467,
+      "epoch": 0.07626816072066134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888017671110306e-05,
+      "loss": 1.2093,
       "step": 2777
     },
     {
-      "epoch": 0.907843137254902,
-      "grad_norm": 1.3048969507217407,
-      "learning_rate": 4.421943558565966e-07,
-      "loss": 0.9436,
+      "epoch": 0.07629562494850457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887884885128623e-05,
+      "loss": 1.1145,
       "step": 2778
     },
     {
-      "epoch": 0.9081699346405229,
-      "grad_norm": 1.2552735805511475,
-      "learning_rate": 4.390869005368259e-07,
-      "loss": 0.8779,
+      "epoch": 0.07632308917634781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887752020910165e-05,
+      "loss": 1.0729,
       "step": 2779
     },
     {
-      "epoch": 0.9084967320261438,
-      "grad_norm": 1.2898192405700684,
-      "learning_rate": 4.359901572347758e-07,
-      "loss": 0.9075,
+      "epoch": 0.07635055340419104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887619078455984e-05,
+      "loss": 1.1282,
       "step": 2780
     },
     {
-      "epoch": 0.9088235294117647,
-      "grad_norm": 1.2205445766448975,
-      "learning_rate": 4.329041294200298e-07,
-      "loss": 0.9227,
+      "epoch": 0.07637801763203428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887486057767133e-05,
+      "loss": 1.0693,
       "step": 2781
     },
     {
-      "epoch": 0.9091503267973856,
-      "grad_norm": 1.2342127561569214,
-      "learning_rate": 4.298288205501633e-07,
-      "loss": 0.9127,
+      "epoch": 0.07640548185987751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887352958844658e-05,
+      "loss": 1.0999,
       "step": 2782
     },
     {
-      "epoch": 0.9094771241830065,
-      "grad_norm": 1.3402910232543945,
-      "learning_rate": 4.2676423407074117e-07,
-      "loss": 0.9275,
+      "epoch": 0.07643294608772075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887219781689614e-05,
+      "loss": 1.001,
       "step": 2783
     },
     {
-      "epoch": 0.9098039215686274,
-      "grad_norm": 1.2397226095199585,
-      "learning_rate": 4.2371037341531805e-07,
-      "loss": 0.9057,
+      "epoch": 0.07646041031556398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887086526303064e-05,
+      "loss": 1.2277,
       "step": 2784
     },
     {
-      "epoch": 0.9101307189542484,
-      "grad_norm": 1.2681400775909424,
-      "learning_rate": 4.206672420054292e-07,
-      "loss": 0.9312,
+      "epoch": 0.0764878745434072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886953192686053e-05,
+      "loss": 1.0227,
       "step": 2785
     },
     {
-      "epoch": 0.9104575163398693,
-      "grad_norm": 1.2425963878631592,
-      "learning_rate": 4.176348432505917e-07,
-      "loss": 0.8786,
+      "epoch": 0.07651533877125044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886819780839636e-05,
+      "loss": 1.1455,
       "step": 2786
     },
     {
-      "epoch": 0.9107843137254902,
-      "grad_norm": 1.2621220350265503,
-      "learning_rate": 4.1461318054829446e-07,
-      "loss": 0.9207,
+      "epoch": 0.07654280299909368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988668629076487e-05,
+      "loss": 1.078,
       "step": 2787
     },
     {
-      "epoch": 0.9111111111111111,
-      "grad_norm": 1.2874048948287964,
-      "learning_rate": 4.116022572839984e-07,
-      "loss": 0.9034,
+      "epoch": 0.07657026722693691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886552722462816e-05,
+      "loss": 1.037,
       "step": 2788
     },
     {
-      "epoch": 0.911437908496732,
-      "grad_norm": 1.3084028959274292,
-      "learning_rate": 4.0860207683113383e-07,
-      "loss": 0.8928,
+      "epoch": 0.07659773145478015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988641907593453e-05,
+      "loss": 1.1508,
       "step": 2789
     },
     {
-      "epoch": 0.9117647058823529,
-      "grad_norm": 1.285574197769165,
-      "learning_rate": 4.0561264255109424e-07,
-      "loss": 0.9051,
+      "epoch": 0.07662519568262338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988628535118106e-05,
+      "loss": 1.168,
       "step": 2790
     },
     {
-      "epoch": 0.9120915032679738,
-      "grad_norm": 1.271628975868225,
-      "learning_rate": 4.0263395779323143e-07,
-      "loss": 0.9223,
+      "epoch": 0.07665265991046662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886151548203476e-05,
+      "loss": 1.2169,
       "step": 2791
     },
     {
-      "epoch": 0.9124183006535947,
-      "grad_norm": 1.2965457439422607,
-      "learning_rate": 3.996660258948548e-07,
-      "loss": 0.9167,
+      "epoch": 0.07668012413830985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988601766700283e-05,
+      "loss": 1.1651,
       "step": 2792
     },
     {
-      "epoch": 0.9127450980392157,
-      "grad_norm": 1.2786386013031006,
-      "learning_rate": 3.9670885018122886e-07,
-      "loss": 0.9181,
+      "epoch": 0.07670758836615309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885883707580183e-05,
+      "loss": 1.0696,
       "step": 2793
     },
     {
-      "epoch": 0.9130718954248366,
-      "grad_norm": 1.26300847530365,
-      "learning_rate": 3.937624339655599e-07,
-      "loss": 0.9081,
+      "epoch": 0.07673505259399632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885749669936592e-05,
+      "loss": 1.0862,
       "step": 2794
     },
     {
-      "epoch": 0.9133986928104575,
-      "grad_norm": 1.232082724571228,
-      "learning_rate": 3.908267805490051e-07,
-      "loss": 0.9349,
+      "epoch": 0.07676251682183956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988561555407312e-05,
+      "loss": 1.0852,
       "step": 2795
     },
     {
-      "epoch": 0.9137254901960784,
-      "grad_norm": 1.2317700386047363,
-      "learning_rate": 3.879018932206624e-07,
-      "loss": 0.9424,
+      "epoch": 0.07678998104968279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988548135999083e-05,
+      "loss": 1.0726,
       "step": 2796
     },
     {
-      "epoch": 0.9140522875816993,
-      "grad_norm": 1.2482713460922241,
-      "learning_rate": 3.849877752575659e-07,
-      "loss": 0.9299,
+      "epoch": 0.07681744527752603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885347087690783e-05,
+      "loss": 1.0764,
       "step": 2797
     },
     {
-      "epoch": 0.9143790849673202,
-      "grad_norm": 1.1988390684127808,
-      "learning_rate": 3.820844299246851e-07,
-      "loss": 0.9268,
+      "epoch": 0.07684490950536926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988521273717404e-05,
+      "loss": 1.1021,
       "step": 2798
     },
     {
-      "epoch": 0.9147058823529411,
-      "grad_norm": 1.239059329032898,
-      "learning_rate": 3.7919186047491895e-07,
-      "loss": 0.9306,
+      "epoch": 0.07687237373321248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885078308441668e-05,
+      "loss": 1.1298,
       "step": 2799
     },
     {
-      "epoch": 0.9150326797385621,
-      "grad_norm": 1.2839527130126953,
-      "learning_rate": 3.7631007014909294e-07,
-      "loss": 0.9214,
+      "epoch": 0.07689983796105572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884943801494722e-05,
+      "loss": 1.1115,
       "step": 2800
     },
     {
-      "epoch": 0.915359477124183,
-      "grad_norm": 1.3207298517227173,
-      "learning_rate": 3.734390621759565e-07,
-      "loss": 0.9124,
+      "epoch": 0.07692730218889895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884809216334276e-05,
+      "loss": 1.0315,
       "step": 2801
     },
     {
-      "epoch": 0.9156862745098039,
-      "grad_norm": 1.2520211935043335,
-      "learning_rate": 3.705788397721777e-07,
-      "loss": 0.9292,
+      "epoch": 0.07695476641674219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988467455296139e-05,
+      "loss": 1.0107,
       "step": 2802
     },
     {
-      "epoch": 0.9160130718954248,
-      "grad_norm": 1.1882442235946655,
-      "learning_rate": 3.6772940614234197e-07,
-      "loss": 0.9041,
+      "epoch": 0.07698223064458543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884539811377125e-05,
+      "loss": 1.0117,
       "step": 2803
     },
     {
-      "epoch": 0.9163398692810457,
-      "grad_norm": 1.3762990236282349,
-      "learning_rate": 3.6489076447894456e-07,
-      "loss": 0.9089,
+      "epoch": 0.07700969487242866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884404991582555e-05,
+      "loss": 1.0614,
       "step": 2804
     },
     {
-      "epoch": 0.9166666666666666,
-      "grad_norm": 1.2468937635421753,
-      "learning_rate": 3.620629179623936e-07,
-      "loss": 0.9331,
+      "epoch": 0.0770371591002719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884270093578743e-05,
+      "loss": 1.1203,
       "step": 2805
     },
     {
-      "epoch": 0.9169934640522875,
-      "grad_norm": 1.3120876550674438,
-      "learning_rate": 3.5924586976099683e-07,
-      "loss": 0.9074,
+      "epoch": 0.07706462332811513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988413511736676e-05,
+      "loss": 1.1566,
       "step": 2806
     },
     {
-      "epoch": 0.9173202614379085,
-      "grad_norm": 1.270674705505371,
-      "learning_rate": 3.5643962303096635e-07,
-      "loss": 0.95,
+      "epoch": 0.07709208755595837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988400006294767e-05,
+      "loss": 1.1501,
       "step": 2807
     },
     {
-      "epoch": 0.9176470588235294,
-      "grad_norm": 1.3151239156723022,
-      "learning_rate": 3.5364418091641374e-07,
-      "loss": 0.9339,
+      "epoch": 0.0771195517838016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988386493032254e-05,
+      "loss": 1.0793,
       "step": 2808
     },
     {
-      "epoch": 0.9179738562091503,
-      "grad_norm": 1.3107588291168213,
-      "learning_rate": 3.5085954654934364e-07,
-      "loss": 0.9302,
+      "epoch": 0.07714701601164484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883729719492445e-05,
+      "loss": 1.0306,
       "step": 2809
     },
     {
-      "epoch": 0.9183006535947712,
-      "grad_norm": 1.2183864116668701,
-      "learning_rate": 3.4808572304965147e-07,
-      "loss": 0.9321,
+      "epoch": 0.07717448023948807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988359443045845e-05,
+      "loss": 0.9814,
       "step": 2810
     },
     {
-      "epoch": 0.9186274509803921,
-      "grad_norm": 1.2645806074142456,
-      "learning_rate": 3.453227135251214e-07,
-      "loss": 0.9372,
+      "epoch": 0.07720194446733131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988345906322163e-05,
+      "loss": 1.1086,
       "step": 2811
     },
     {
-      "epoch": 0.918954248366013,
-      "grad_norm": 1.2436304092407227,
-      "learning_rate": 3.425705210714192e-07,
-      "loss": 0.9043,
+      "epoch": 0.07722940869517453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883323617783053e-05,
+      "loss": 1.0674,
       "step": 2812
     },
     {
-      "epoch": 0.9192810457516339,
-      "grad_norm": 1.2682304382324219,
-      "learning_rate": 3.3982914877209396e-07,
-      "loss": 0.9167,
+      "epoch": 0.07725687292301776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988318809414379e-05,
+      "loss": 1.1568,
       "step": 2813
     },
     {
-      "epoch": 0.9196078431372549,
-      "grad_norm": 1.2047115564346313,
-      "learning_rate": 3.3709859969857096e-07,
-      "loss": 0.9083,
+      "epoch": 0.077284337150861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883052492304915e-05,
+      "loss": 1.0637,
       "step": 2814
     },
     {
-      "epoch": 0.9199346405228758,
-      "grad_norm": 1.2426236867904663,
-      "learning_rate": 3.343788769101486e-07,
-      "loss": 0.9445,
+      "epoch": 0.07731180137870423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98829168122675e-05,
+      "loss": 1.1014,
       "step": 2815
     },
     {
-      "epoch": 0.9202614379084967,
-      "grad_norm": 1.2164344787597656,
-      "learning_rate": 3.316699834539983e-07,
-      "loss": 0.9057,
+      "epoch": 0.07733926560654747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882781054032615e-05,
+      "loss": 1.0292,
       "step": 2816
     },
     {
-      "epoch": 0.9205882352941176,
-      "grad_norm": 1.3658124208450317,
-      "learning_rate": 3.289719223651511e-07,
-      "loss": 0.9423,
+      "epoch": 0.0773667298343907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882645217601343e-05,
+      "loss": 1.1179,
       "step": 2817
     },
     {
-      "epoch": 0.9209150326797385,
-      "grad_norm": 1.2375199794769287,
-      "learning_rate": 3.2628469666650897e-07,
-      "loss": 0.9264,
+      "epoch": 0.07739419406223394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882509302974754e-05,
+      "loss": 1.1915,
       "step": 2818
     },
     {
-      "epoch": 0.9212418300653594,
-      "grad_norm": 1.2850091457366943,
-      "learning_rate": 3.236083093688325e-07,
-      "loss": 0.9198,
+      "epoch": 0.07742165829007717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988237331015392e-05,
+      "loss": 1.019,
       "step": 2819
     },
     {
-      "epoch": 0.9215686274509803,
-      "grad_norm": 1.2433944940567017,
-      "learning_rate": 3.2094276347073626e-07,
-      "loss": 0.9427,
+      "epoch": 0.07744912251792041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882237239139924e-05,
+      "loss": 1.1179,
       "step": 2820
     },
     {
-      "epoch": 0.9218954248366014,
-      "grad_norm": 1.2292578220367432,
-      "learning_rate": 3.1828806195869035e-07,
-      "loss": 0.8901,
+      "epoch": 0.07747658674576365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988210108993384e-05,
+      "loss": 1.0666,
       "step": 2821
     },
     {
-      "epoch": 0.9222222222222223,
-      "grad_norm": 1.2640177011489868,
-      "learning_rate": 3.1564420780701435e-07,
-      "loss": 0.9229,
+      "epoch": 0.07750405097360688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988196486253674e-05,
+      "loss": 1.1533,
       "step": 2822
     },
     {
-      "epoch": 0.9225490196078432,
-      "grad_norm": 1.2807273864746094,
-      "learning_rate": 3.130112039778732e-07,
-      "loss": 0.9097,
+      "epoch": 0.07753151520145012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881828556949706e-05,
+      "loss": 1.1993,
       "step": 2823
     },
     {
-      "epoch": 0.9228758169934641,
-      "grad_norm": 1.3499293327331543,
-      "learning_rate": 3.103890534212783e-07,
-      "loss": 0.9712,
+      "epoch": 0.07755897942929335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988169217317382e-05,
+      "loss": 1.0724,
       "step": 2824
     },
     {
-      "epoch": 0.923202614379085,
-      "grad_norm": 1.3087131977081299,
-      "learning_rate": 3.0777775907507633e-07,
-      "loss": 0.9348,
+      "epoch": 0.07758644365713659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881555711210153e-05,
+      "loss": 1.0516,
       "step": 2825
     },
     {
-      "epoch": 0.9235294117647059,
-      "grad_norm": 1.2676881551742554,
-      "learning_rate": 3.051773238649558e-07,
-      "loss": 0.9468,
+      "epoch": 0.07761390788497981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881419171059796e-05,
+      "loss": 1.0782,
       "step": 2826
     },
     {
-      "epoch": 0.9238562091503268,
-      "grad_norm": 1.1934623718261719,
-      "learning_rate": 3.025877507044361e-07,
-      "loss": 0.9203,
+      "epoch": 0.07764137211282304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881282552723818e-05,
+      "loss": 1.1937,
       "step": 2827
     },
     {
-      "epoch": 0.9241830065359478,
-      "grad_norm": 1.2779300212860107,
-      "learning_rate": 3.000090424948665e-07,
-      "loss": 0.9223,
+      "epoch": 0.07766883634066628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881145856203307e-05,
+      "loss": 1.1119,
       "step": 2828
     },
     {
-      "epoch": 0.9245098039215687,
-      "grad_norm": 1.2955601215362549,
-      "learning_rate": 2.974412021254236e-07,
-      "loss": 0.9039,
+      "epoch": 0.07769630056850951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881009081499338e-05,
+      "loss": 1.067,
       "step": 2829
     },
     {
-      "epoch": 0.9248366013071896,
-      "grad_norm": 1.2459690570831299,
-      "learning_rate": 2.9488423247310713e-07,
-      "loss": 0.902,
+      "epoch": 0.07772376479635275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880872228613005e-05,
+      "loss": 1.1,
       "step": 2830
     },
     {
-      "epoch": 0.9251633986928105,
-      "grad_norm": 1.2361972332000732,
-      "learning_rate": 2.9233813640274e-07,
-      "loss": 0.8989,
+      "epoch": 0.07775122902419598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880735297545378e-05,
+      "loss": 1.199,
       "step": 2831
     },
     {
-      "epoch": 0.9254901960784314,
-      "grad_norm": 1.2566313743591309,
-      "learning_rate": 2.8980291676695917e-07,
-      "loss": 0.9218,
+      "epoch": 0.07777869325203922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880598288297545e-05,
+      "loss": 1.0566,
       "step": 2832
     },
     {
-      "epoch": 0.9258169934640523,
-      "grad_norm": 1.2794790267944336,
-      "learning_rate": 2.872785764062169e-07,
-      "loss": 0.9109,
+      "epoch": 0.07780615747988245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880461200870594e-05,
+      "loss": 1.0803,
       "step": 2833
     },
     {
-      "epoch": 0.9261437908496732,
-      "grad_norm": 1.3012183904647827,
-      "learning_rate": 2.8476511814877763e-07,
-      "loss": 0.9055,
+      "epoch": 0.07783362170772569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988032403526561e-05,
+      "loss": 1.0494,
       "step": 2834
     },
     {
-      "epoch": 0.9264705882352942,
-      "grad_norm": 1.3295671939849854,
-      "learning_rate": 2.822625448107108e-07,
-      "loss": 0.9515,
+      "epoch": 0.07786108593556892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880186791483668e-05,
+      "loss": 1.0537,
       "step": 2835
     },
     {
-      "epoch": 0.9267973856209151,
-      "grad_norm": 1.2670961618423462,
-      "learning_rate": 2.7977085919589253e-07,
-      "loss": 0.9395,
+      "epoch": 0.07788855016341216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880049469525863e-05,
+      "loss": 1.0607,
       "step": 2836
     },
     {
-      "epoch": 0.927124183006536,
-      "grad_norm": 1.1792900562286377,
-      "learning_rate": 2.772900640959997e-07,
-      "loss": 0.9337,
+      "epoch": 0.0779160143912554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987991206939328e-05,
+      "loss": 1.0925,
       "step": 2837
     },
     {
-      "epoch": 0.9274509803921569,
-      "grad_norm": 1.2128862142562866,
-      "learning_rate": 2.7482016229050776e-07,
-      "loss": 0.9253,
+      "epoch": 0.07794347861909863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879774591087005e-05,
+      "loss": 1.2143,
       "step": 2838
     },
     {
-      "epoch": 0.9277777777777778,
-      "grad_norm": 1.2431585788726807,
-      "learning_rate": 2.723611565466866e-07,
-      "loss": 0.9037,
+      "epoch": 0.07797094284694185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987963703460813e-05,
+      "loss": 1.1472,
       "step": 2839
     },
     {
-      "epoch": 0.9281045751633987,
-      "grad_norm": 1.2438546419143677,
-      "learning_rate": 2.6991304961959895e-07,
-      "loss": 0.9453,
+      "epoch": 0.07799840707478509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879499399957735e-05,
+      "loss": 1.0752,
       "step": 2840
     },
     {
-      "epoch": 0.9284313725490196,
-      "grad_norm": 1.3066754341125488,
-      "learning_rate": 2.674758442520964e-07,
-      "loss": 0.8894,
+      "epoch": 0.07802587130262832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879361687136916e-05,
+      "loss": 1.109,
       "step": 2841
     },
     {
-      "epoch": 0.9287581699346406,
-      "grad_norm": 1.285156488418579,
-      "learning_rate": 2.650495431748168e-07,
-      "loss": 0.9052,
+      "epoch": 0.07805333553047156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987922389614676e-05,
+      "loss": 0.9807,
       "step": 2842
     },
     {
-      "epoch": 0.9290849673202615,
-      "grad_norm": 1.3472915887832642,
-      "learning_rate": 2.626341491061801e-07,
-      "loss": 0.9205,
+      "epoch": 0.07808079975831479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987908602698836e-05,
+      "loss": 1.0494,
       "step": 2843
     },
     {
-      "epoch": 0.9294117647058824,
-      "grad_norm": 1.2183195352554321,
-      "learning_rate": 2.602296647523861e-07,
-      "loss": 0.9259,
+      "epoch": 0.07810826398615803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98789480796628e-05,
+      "loss": 1.098,
       "step": 2844
     },
     {
-      "epoch": 0.9297385620915033,
-      "grad_norm": 1.270002007484436,
-      "learning_rate": 2.5783609280741197e-07,
-      "loss": 0.9262,
+      "epoch": 0.07813572821400126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987881005417118e-05,
+      "loss": 1.0665,
       "step": 2845
     },
     {
-      "epoch": 0.9300653594771242,
-      "grad_norm": 1.244882345199585,
-      "learning_rate": 2.554534359530081e-07,
-      "loss": 0.9047,
+      "epoch": 0.0781631924418445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878671950514587e-05,
+      "loss": 1.0921,
       "step": 2846
     },
     {
-      "epoch": 0.9303921568627451,
-      "grad_norm": 1.2720301151275635,
-      "learning_rate": 2.530816968586958e-07,
-      "loss": 0.9437,
+      "epoch": 0.07819065666968773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878533768694112e-05,
+      "loss": 1.0904,
       "step": 2847
     },
     {
-      "epoch": 0.930718954248366,
-      "grad_norm": 1.2309529781341553,
-      "learning_rate": 2.507208781817638e-07,
-      "loss": 0.9429,
+      "epoch": 0.07821812089753097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878395508710853e-05,
+      "loss": 1.0881,
       "step": 2848
     },
     {
-      "epoch": 0.931045751633987,
-      "grad_norm": 1.2508732080459595,
-      "learning_rate": 2.4837098256726867e-07,
-      "loss": 0.9317,
+      "epoch": 0.0782455851253742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878257170565908e-05,
+      "loss": 1.0399,
       "step": 2849
     },
     {
-      "epoch": 0.9313725490196079,
-      "grad_norm": 1.2283730506896973,
-      "learning_rate": 2.4603201264802423e-07,
-      "loss": 0.9236,
+      "epoch": 0.07827304935321744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878118754260357e-05,
+      "loss": 1.0783,
       "step": 2850
     },
     {
-      "epoch": 0.9316993464052288,
-      "grad_norm": 1.274714469909668,
-      "learning_rate": 2.4370397104460433e-07,
-      "loss": 0.9456,
+      "epoch": 0.07830051358106067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987798025979531e-05,
+      "loss": 1.118,
       "step": 2851
     },
     {
-      "epoch": 0.9320261437908497,
-      "grad_norm": 1.2774232625961304,
-      "learning_rate": 2.413868603653413e-07,
-      "loss": 0.9506,
+      "epoch": 0.07832797780890391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877841687171853e-05,
+      "loss": 1.002,
       "step": 2852
     },
     {
-      "epoch": 0.9323529411764706,
-      "grad_norm": 1.2688169479370117,
-      "learning_rate": 2.3908068320631863e-07,
-      "loss": 0.913,
+      "epoch": 0.07835544203674713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987770303639109e-05,
+      "loss": 1.101,
       "step": 2853
     },
     {
-      "epoch": 0.9326797385620915,
-      "grad_norm": 1.25229811668396,
-      "learning_rate": 2.3678544215136934e-07,
-      "loss": 0.9177,
+      "epoch": 0.07838290626459037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987756430745411e-05,
+      "loss": 1.0948,
       "step": 2854
     },
     {
-      "epoch": 0.9330065359477124,
-      "grad_norm": 1.2211521863937378,
-      "learning_rate": 2.3450113977207534e-07,
-      "loss": 0.9171,
+      "epoch": 0.0784103704924336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987742550036202e-05,
+      "loss": 1.0807,
       "step": 2855
     },
     {
-      "epoch": 0.9333333333333333,
-      "grad_norm": 1.2482794523239136,
-      "learning_rate": 2.3222777862776046e-07,
-      "loss": 0.9532,
+      "epoch": 0.07843783472027684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877286615115908e-05,
+      "loss": 1.1031,
       "step": 2856
     },
     {
-      "epoch": 0.9336601307189543,
-      "grad_norm": 1.2442493438720703,
-      "learning_rate": 2.2996536126549395e-07,
-      "loss": 0.9283,
+      "epoch": 0.07846529894812007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987714765171688e-05,
+      "loss": 0.9706,
       "step": 2857
     },
     {
-      "epoch": 0.9339869281045752,
-      "grad_norm": 1.3515138626098633,
-      "learning_rate": 2.2771389022008038e-07,
-      "loss": 0.9283,
+      "epoch": 0.0784927631759633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877008610166036e-05,
+      "loss": 1.0499,
       "step": 2858
     },
     {
-      "epoch": 0.9343137254901961,
-      "grad_norm": 1.225803256034851,
-      "learning_rate": 2.2547336801405972e-07,
-      "loss": 0.9384,
+      "epoch": 0.07852022740380654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876869490464468e-05,
+      "loss": 1.0579,
       "step": 2859
     },
     {
-      "epoch": 0.934640522875817,
-      "grad_norm": 1.226920485496521,
-      "learning_rate": 2.2324379715770728e-07,
-      "loss": 0.9453,
+      "epoch": 0.07854769163164978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876730292613292e-05,
+      "loss": 1.086,
       "step": 2860
     },
     {
-      "epoch": 0.9349673202614379,
-      "grad_norm": 1.2547824382781982,
-      "learning_rate": 2.210251801490282e-07,
-      "loss": 0.9368,
+      "epoch": 0.07857515585949301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876591016613593e-05,
+      "loss": 1.0965,
       "step": 2861
     },
     {
-      "epoch": 0.9352941176470588,
-      "grad_norm": 1.24323570728302,
-      "learning_rate": 2.1881751947375296e-07,
-      "loss": 0.9483,
+      "epoch": 0.07860262008733625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876451662466482e-05,
+      "loss": 1.1389,
       "step": 2862
     },
     {
-      "epoch": 0.9356209150326797,
-      "grad_norm": 1.2736896276474,
-      "learning_rate": 2.1662081760533748e-07,
-      "loss": 0.9397,
+      "epoch": 0.07863008431517948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987631223017306e-05,
+      "loss": 0.9418,
       "step": 2863
     },
     {
-      "epoch": 0.9359477124183007,
-      "grad_norm": 1.247786521911621,
-      "learning_rate": 2.1443507700495968e-07,
-      "loss": 0.8997,
+      "epoch": 0.07865754854302272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987617271973443e-05,
+      "loss": 0.996,
       "step": 2864
     },
     {
-      "epoch": 0.9362745098039216,
-      "grad_norm": 1.2643455266952515,
-      "learning_rate": 2.1226030012151844e-07,
-      "loss": 0.9291,
+      "epoch": 0.07868501277086595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876033131151696e-05,
+      "loss": 1.0103,
       "step": 2865
     },
     {
-      "epoch": 0.9366013071895425,
-      "grad_norm": 1.2388579845428467,
-      "learning_rate": 2.1009648939162574e-07,
-      "loss": 0.9055,
+      "epoch": 0.07871247699870917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987589346442596e-05,
+      "loss": 1.0775,
       "step": 2866
     },
     {
-      "epoch": 0.9369281045751634,
-      "grad_norm": 1.2739453315734863,
-      "learning_rate": 2.0794364723961013e-07,
-      "loss": 0.9212,
+      "epoch": 0.07873994122655241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875753719558333e-05,
+      "loss": 1.0923,
       "step": 2867
     },
     {
-      "epoch": 0.9372549019607843,
-      "grad_norm": 1.205257773399353,
-      "learning_rate": 2.0580177607750663e-07,
-      "loss": 0.9568,
+      "epoch": 0.07876740545439564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875613896549916e-05,
+      "loss": 1.062,
       "step": 2868
     },
     {
-      "epoch": 0.9375816993464052,
-      "grad_norm": 1.2424256801605225,
-      "learning_rate": 2.0367087830506226e-07,
-      "loss": 0.9258,
+      "epoch": 0.07879486968223888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875473995401812e-05,
+      "loss": 1.1072,
       "step": 2869
     },
     {
-      "epoch": 0.9379084967320261,
-      "grad_norm": 1.2727583646774292,
-      "learning_rate": 2.0155095630972954e-07,
-      "loss": 0.9236,
+      "epoch": 0.07882233391008212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875334016115136e-05,
+      "loss": 1.1907,
       "step": 2870
     },
     {
-      "epoch": 0.9382352941176471,
-      "grad_norm": 1.2687897682189941,
-      "learning_rate": 1.9944201246666072e-07,
-      "loss": 0.9257,
+      "epoch": 0.07884979813792535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875193958690992e-05,
+      "loss": 1.1942,
       "step": 2871
     },
     {
-      "epoch": 0.938562091503268,
-      "grad_norm": 1.3258572816848755,
-      "learning_rate": 1.973440491387124e-07,
-      "loss": 0.9231,
+      "epoch": 0.07887726236576859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987505382313049e-05,
+      "loss": 1.0111,
       "step": 2872
     },
     {
-      "epoch": 0.9388888888888889,
-      "grad_norm": 1.2342180013656616,
-      "learning_rate": 1.9525706867643323e-07,
-      "loss": 0.9169,
+      "epoch": 0.07890472659361182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874913609434734e-05,
+      "loss": 1.0992,
       "step": 2873
     },
     {
-      "epoch": 0.9392156862745098,
-      "grad_norm": 1.2743037939071655,
-      "learning_rate": 1.9318107341807168e-07,
-      "loss": 0.9059,
+      "epoch": 0.07893219082145506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874773317604837e-05,
+      "loss": 1.077,
       "step": 2874
     },
     {
-      "epoch": 0.9395424836601307,
-      "grad_norm": 1.2049615383148193,
-      "learning_rate": 1.9111606568956497e-07,
-      "loss": 0.9289,
+      "epoch": 0.07895965504929829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987463294764191e-05,
+      "loss": 1.1237,
       "step": 2875
     },
     {
-      "epoch": 0.9398692810457516,
-      "grad_norm": 1.3156250715255737,
-      "learning_rate": 1.890620478045435e-07,
-      "loss": 0.9165,
+      "epoch": 0.07898711927714153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987449249954706e-05,
+      "loss": 1.0627,
       "step": 2876
     },
     {
-      "epoch": 0.9401960784313725,
-      "grad_norm": 1.2537641525268555,
-      "learning_rate": 1.8701902206432198e-07,
-      "loss": 0.9564,
+      "epoch": 0.07901458350498476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874351973321402e-05,
+      "loss": 1.0641,
       "step": 2877
     },
     {
-      "epoch": 0.9405228758169935,
-      "grad_norm": 1.2616552114486694,
-      "learning_rate": 1.8498699075789938e-07,
-      "loss": 0.9368,
+      "epoch": 0.079042047732828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874211368966044e-05,
+      "loss": 1.0668,
       "step": 2878
     },
     {
-      "epoch": 0.9408496732026144,
-      "grad_norm": 1.2401145696640015,
-      "learning_rate": 1.8296595616195789e-07,
-      "loss": 0.9163,
+      "epoch": 0.07906951196067123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874070686482102e-05,
+      "loss": 1.0562,
       "step": 2879
     },
     {
-      "epoch": 0.9411764705882353,
-      "grad_norm": 1.2224998474121094,
-      "learning_rate": 1.8095592054085843e-07,
-      "loss": 0.9459,
+      "epoch": 0.07909697618851445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873929925870687e-05,
+      "loss": 1.0829,
       "step": 2880
     },
     {
-      "epoch": 0.9415032679738562,
-      "grad_norm": 1.221230149269104,
-      "learning_rate": 1.7895688614663952e-07,
-      "loss": 0.9386,
+      "epoch": 0.07912444041635769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873789087132917e-05,
+      "loss": 0.9825,
       "step": 2881
     },
     {
-      "epoch": 0.9418300653594771,
-      "grad_norm": 1.1999045610427856,
-      "learning_rate": 1.7696885521901075e-07,
-      "loss": 0.9661,
+      "epoch": 0.07915190464420092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98736481702699e-05,
+      "loss": 1.0021,
       "step": 2882
     },
     {
-      "epoch": 0.942156862745098,
-      "grad_norm": 1.2679529190063477,
-      "learning_rate": 1.7499182998535925e-07,
-      "loss": 0.913,
+      "epoch": 0.07917936887204416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873507175282753e-05,
+      "loss": 0.9984,
       "step": 2883
     },
     {
-      "epoch": 0.9424836601307189,
-      "grad_norm": 1.2561336755752563,
-      "learning_rate": 1.7302581266073537e-07,
-      "loss": 0.9336,
+      "epoch": 0.0792068330998874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873366102172596e-05,
+      "loss": 1.109,
       "step": 2884
     },
     {
-      "epoch": 0.9428104575163399,
-      "grad_norm": 1.236271619796753,
-      "learning_rate": 1.7107080544785937e-07,
-      "loss": 0.9614,
+      "epoch": 0.07923429732773063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873224950940536e-05,
+      "loss": 1.0519,
       "step": 2885
     },
     {
-      "epoch": 0.9431372549019608,
-      "grad_norm": 1.2756356000900269,
-      "learning_rate": 1.6912681053711577e-07,
-      "loss": 0.9199,
+      "epoch": 0.07926176155557386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98730837215877e-05,
+      "loss": 1.1549,
       "step": 2886
     },
     {
-      "epoch": 0.9434640522875817,
-      "grad_norm": 1.2603354454040527,
-      "learning_rate": 1.671938301065512e-07,
-      "loss": 0.9131,
+      "epoch": 0.0792892257834171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872942414115203e-05,
+      "loss": 1.0523,
       "step": 2887
     },
     {
-      "epoch": 0.9437908496732026,
-      "grad_norm": 1.2756175994873047,
-      "learning_rate": 1.6527186632186997e-07,
-      "loss": 0.9144,
+      "epoch": 0.07931669001126034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872801028524156e-05,
+      "loss": 1.0262,
       "step": 2888
     },
     {
-      "epoch": 0.9441176470588235,
-      "grad_norm": 1.3160924911499023,
-      "learning_rate": 1.6336092133643511e-07,
-      "loss": 0.9316,
+      "epoch": 0.07934415423910357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987265956481569e-05,
+      "loss": 1.0022,
       "step": 2889
     },
     {
-      "epoch": 0.9444444444444444,
-      "grad_norm": 1.2826639413833618,
-      "learning_rate": 1.6146099729126285e-07,
-      "loss": 0.9093,
+      "epoch": 0.0793716184669468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987251802299091e-05,
+      "loss": 1.081,
       "step": 2890
     },
     {
-      "epoch": 0.9447712418300653,
-      "grad_norm": 1.2711713314056396,
-      "learning_rate": 1.5957209631502158e-07,
-      "loss": 0.9412,
+      "epoch": 0.07939908269479004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987237640305095e-05,
+      "loss": 1.1575,
       "step": 2891
     },
     {
-      "epoch": 0.9450980392156862,
-      "grad_norm": 1.2459214925765991,
-      "learning_rate": 1.5769422052403172e-07,
-      "loss": 0.917,
+      "epoch": 0.07942654692263328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987223470499692e-05,
+      "loss": 0.9735,
       "step": 2892
     },
     {
-      "epoch": 0.9454248366013072,
-      "grad_norm": 1.258628487586975,
-      "learning_rate": 1.5582737202225916e-07,
-      "loss": 0.896,
+      "epoch": 0.07945401115047651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872092928829944e-05,
+      "loss": 1.0069,
       "step": 2893
     },
     {
-      "epoch": 0.9457516339869281,
-      "grad_norm": 1.2900310754776,
-      "learning_rate": 1.5397155290131528e-07,
-      "loss": 0.9343,
+      "epoch": 0.07948147537831973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871951074551142e-05,
+      "loss": 1.0136,
       "step": 2894
     },
     {
-      "epoch": 0.946078431372549,
-      "grad_norm": 1.2127065658569336,
-      "learning_rate": 1.5212676524045345e-07,
-      "loss": 0.9347,
+      "epoch": 0.07950893960616297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871809142161646e-05,
+      "loss": 1.1481,
       "step": 2895
     },
     {
-      "epoch": 0.9464052287581699,
-      "grad_norm": 1.2298002243041992,
-      "learning_rate": 1.5029301110656923e-07,
-      "loss": 0.9208,
+      "epoch": 0.0795364038340062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871667131662565e-05,
+      "loss": 1.0634,
       "step": 2896
     },
     {
-      "epoch": 0.9467320261437908,
-      "grad_norm": 1.2317129373550415,
-      "learning_rate": 1.4847029255419365e-07,
-      "loss": 0.9358,
+      "epoch": 0.07956386806184944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871525043055033e-05,
+      "loss": 1.0391,
       "step": 2897
     },
     {
-      "epoch": 0.9470588235294117,
-      "grad_norm": 1.2704126834869385,
-      "learning_rate": 1.4665861162549644e-07,
-      "loss": 0.9301,
+      "epoch": 0.07959133228969267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871382876340174e-05,
+      "loss": 1.0367,
       "step": 2898
     },
     {
-      "epoch": 0.9473856209150326,
-      "grad_norm": 1.237702488899231,
-      "learning_rate": 1.4485797035027728e-07,
-      "loss": 0.9105,
+      "epoch": 0.07961879651753591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871240631519105e-05,
+      "loss": 1.126,
       "step": 2899
     },
     {
-      "epoch": 0.9477124183006536,
-      "grad_norm": 1.2143815755844116,
-      "learning_rate": 1.4306837074597235e-07,
-      "loss": 0.9157,
+      "epoch": 0.07964626074537914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987109830859296e-05,
+      "loss": 1.0681,
       "step": 2900
     },
     {
-      "epoch": 0.9480392156862745,
-      "grad_norm": 1.2666219472885132,
-      "learning_rate": 1.4128981481764115e-07,
-      "loss": 0.947,
+      "epoch": 0.07967372497322238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987095590756286e-05,
+      "loss": 1.1557,
       "step": 2901
     },
     {
-      "epoch": 0.9483660130718954,
-      "grad_norm": 1.2933430671691895,
-      "learning_rate": 1.39522304557973e-07,
-      "loss": 0.9236,
+      "epoch": 0.07970118920106561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987081342842993e-05,
+      "loss": 1.1095,
       "step": 2902
     },
     {
-      "epoch": 0.9486928104575163,
-      "grad_norm": 1.34320068359375,
-      "learning_rate": 1.3776584194728048e-07,
-      "loss": 0.9259,
+      "epoch": 0.07972865342890885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870670871195305e-05,
+      "loss": 1.1149,
       "step": 2903
     },
     {
-      "epoch": 0.9490196078431372,
-      "grad_norm": 1.2914093732833862,
-      "learning_rate": 1.3602042895349942e-07,
-      "loss": 0.9263,
+      "epoch": 0.07975611765675208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987052823586011e-05,
+      "loss": 1.0254,
       "step": 2904
     },
     {
-      "epoch": 0.9493464052287581,
-      "grad_norm": 1.2369648218154907,
-      "learning_rate": 1.3428606753218553e-07,
-      "loss": 0.9436,
+      "epoch": 0.07978358188459532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870385522425467e-05,
+      "loss": 0.9525,
       "step": 2905
     },
     {
-      "epoch": 0.949673202614379,
-      "grad_norm": 1.3813564777374268,
-      "learning_rate": 1.3256275962651223e-07,
-      "loss": 0.9414,
+      "epoch": 0.07981104611243856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870242730892516e-05,
+      "loss": 0.9579,
       "step": 2906
     },
     {
-      "epoch": 0.95,
-      "grad_norm": 1.2426443099975586,
-      "learning_rate": 1.308505071672661e-07,
-      "loss": 0.9314,
+      "epoch": 0.07983851034028178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870099861262378e-05,
+      "loss": 0.9949,
       "step": 2907
     },
     {
-      "epoch": 0.9503267973856209,
-      "grad_norm": 1.236682415008545,
-      "learning_rate": 1.2914931207285154e-07,
-      "loss": 0.9203,
+      "epoch": 0.07986597456812501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869956913536185e-05,
+      "loss": 1.1228,
       "step": 2908
     },
     {
-      "epoch": 0.9506535947712418,
-      "grad_norm": 1.293799638748169,
-      "learning_rate": 1.2745917624928273e-07,
-      "loss": 0.9537,
+      "epoch": 0.07989343879596825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869813887715072e-05,
+      "loss": 1.1432,
       "step": 2909
     },
     {
-      "epoch": 0.9509803921568627,
-      "grad_norm": 1.24419367313385,
-      "learning_rate": 1.2578010159018162e-07,
-      "loss": 0.9087,
+      "epoch": 0.07992090302381148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869670783800168e-05,
+      "loss": 1.0328,
       "step": 2910
     },
     {
-      "epoch": 0.9513071895424836,
-      "grad_norm": 1.2715831995010376,
-      "learning_rate": 1.2411208997677783e-07,
-      "loss": 0.9225,
+      "epoch": 0.07994836725165472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869527601792603e-05,
+      "loss": 1.1714,
       "step": 2911
     },
     {
-      "epoch": 0.9516339869281045,
-      "grad_norm": 1.2506669759750366,
-      "learning_rate": 1.2245514327790863e-07,
-      "loss": 0.9087,
+      "epoch": 0.07997583147949795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869384341693517e-05,
+      "loss": 1.0643,
       "step": 2912
     },
     {
-      "epoch": 0.9519607843137254,
-      "grad_norm": 1.3213900327682495,
-      "learning_rate": 1.208092633500102e-07,
-      "loss": 0.9509,
+      "epoch": 0.08000329570734119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869241003504035e-05,
+      "loss": 1.0543,
       "step": 2913
     },
     {
-      "epoch": 0.9522875816993464,
-      "grad_norm": 1.2950198650360107,
-      "learning_rate": 1.1917445203712186e-07,
-      "loss": 0.9339,
+      "epoch": 0.08003075993518442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869097587225297e-05,
+      "loss": 1.0172,
       "step": 2914
     },
     {
-      "epoch": 0.9526143790849673,
-      "grad_norm": 1.247329592704773,
-      "learning_rate": 1.1755071117088068e-07,
-      "loss": 0.9268,
+      "epoch": 0.08005822416302766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868954092858436e-05,
+      "loss": 1.0849,
       "step": 2915
     },
     {
-      "epoch": 0.9529411764705882,
-      "grad_norm": 1.2774913311004639,
-      "learning_rate": 1.1593804257052143e-07,
-      "loss": 0.8965,
+      "epoch": 0.0800856883908709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868810520404587e-05,
+      "loss": 0.9665,
       "step": 2916
     },
     {
-      "epoch": 0.9532679738562092,
-      "grad_norm": 1.23866868019104,
-      "learning_rate": 1.1433644804287324e-07,
-      "loss": 0.9457,
+      "epoch": 0.08011315261871413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986866686986489e-05,
+      "loss": 1.1229,
       "step": 2917
     },
     {
-      "epoch": 0.9535947712418301,
-      "grad_norm": 1.2896881103515625,
-      "learning_rate": 1.127459293823574e-07,
-      "loss": 0.934,
+      "epoch": 0.08014061684655736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986852314124047e-05,
+      "loss": 1.135,
       "step": 2918
     },
     {
-      "epoch": 0.953921568627451,
-      "grad_norm": 1.2288728952407837,
-      "learning_rate": 1.111664883709862e-07,
-      "loss": 0.9302,
+      "epoch": 0.0801680810744006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868379334532478e-05,
+      "loss": 1.1136,
       "step": 2919
     },
     {
-      "epoch": 0.954248366013072,
-      "grad_norm": 1.238110065460205,
-      "learning_rate": 1.0959812677835968e-07,
-      "loss": 0.8954,
+      "epoch": 0.08019554530224383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868235449742043e-05,
+      "loss": 1.2059,
       "step": 2920
     },
     {
-      "epoch": 0.9545751633986929,
-      "grad_norm": 1.2876003980636597,
-      "learning_rate": 1.0804084636166779e-07,
-      "loss": 0.9223,
+      "epoch": 0.08022300953008706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986809148687031e-05,
+      "loss": 1.0847,
       "step": 2921
     },
     {
-      "epoch": 0.9549019607843138,
-      "grad_norm": 1.3085627555847168,
-      "learning_rate": 1.064946488656804e-07,
-      "loss": 0.9513,
+      "epoch": 0.08025047375793029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867947445918413e-05,
+      "loss": 1.1019,
       "step": 2922
     },
     {
-      "epoch": 0.9552287581699347,
-      "grad_norm": 1.2046713829040527,
-      "learning_rate": 1.04959536022754e-07,
-      "loss": 0.9582,
+      "epoch": 0.08027793798577353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867803326887495e-05,
+      "loss": 1.2917,
       "step": 2923
     },
     {
-      "epoch": 0.9555555555555556,
-      "grad_norm": 1.2059134244918823,
-      "learning_rate": 1.0343550955282278e-07,
-      "loss": 0.9497,
+      "epoch": 0.08030540221361676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867659129778692e-05,
+      "loss": 1.0764,
       "step": 2924
     },
     {
-      "epoch": 0.9558823529411765,
-      "grad_norm": 1.252514123916626,
-      "learning_rate": 1.0192257116340198e-07,
-      "loss": 0.9384,
+      "epoch": 0.08033286644146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867514854593147e-05,
+      "loss": 1.0951,
       "step": 2925
     },
     {
-      "epoch": 0.9562091503267974,
-      "grad_norm": 1.2704881429672241,
-      "learning_rate": 1.0042072254958346e-07,
-      "loss": 0.9299,
+      "epoch": 0.08036033066930323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867370501332005e-05,
+      "loss": 1.0839,
       "step": 2926
     },
     {
-      "epoch": 0.9565359477124183,
-      "grad_norm": 1.2529617547988892,
-      "learning_rate": 9.892996539403232e-08,
-      "loss": 0.9438,
+      "epoch": 0.08038779489714647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867226069996404e-05,
+      "loss": 0.9939,
       "step": 2927
     },
     {
-      "epoch": 0.9568627450980393,
-      "grad_norm": 1.2381125688552856,
-      "learning_rate": 9.74503013669914e-08,
-      "loss": 0.917,
+      "epoch": 0.0804152591249897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986708156058749e-05,
+      "loss": 1.1259,
       "step": 2928
     },
     {
-      "epoch": 0.9571895424836602,
-      "grad_norm": 1.291129469871521,
-      "learning_rate": 9.598173212626904e-08,
-      "loss": 0.9169,
+      "epoch": 0.08044272335283294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866936973106398e-05,
+      "loss": 0.9421,
       "step": 2929
     },
     {
-      "epoch": 0.9575163398692811,
-      "grad_norm": 1.1997593641281128,
-      "learning_rate": 9.452425931724685e-08,
-      "loss": 0.9071,
+      "epoch": 0.08047018758067617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866792307554284e-05,
+      "loss": 1.0851,
       "step": 2930
     },
     {
-      "epoch": 0.957843137254902,
-      "grad_norm": 1.2585723400115967,
-      "learning_rate": 9.307788457287304e-08,
-      "loss": 0.9362,
+      "epoch": 0.08049765180851941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986664756393229e-05,
+      "loss": 1.0847,
       "step": 2931
     },
     {
-      "epoch": 0.9581699346405229,
-      "grad_norm": 1.2452504634857178,
-      "learning_rate": 9.164260951366021e-08,
-      "loss": 0.9319,
+      "epoch": 0.08052511603636264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866502742241554e-05,
+      "loss": 1.0847,
       "step": 2932
     },
     {
-      "epoch": 0.9584967320261438,
-      "grad_norm": 1.2922323942184448,
-      "learning_rate": 9.021843574768651e-08,
-      "loss": 0.9001,
+      "epoch": 0.08055258026420588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866357842483227e-05,
+      "loss": 1.0827,
       "step": 2933
     },
     {
-      "epoch": 0.9588235294117647,
-      "grad_norm": 1.2332851886749268,
-      "learning_rate": 8.880536487059333e-08,
-      "loss": 0.9425,
+      "epoch": 0.0805800444920491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866212864658455e-05,
+      "loss": 1.1302,
       "step": 2934
     },
     {
-      "epoch": 0.9591503267973857,
-      "grad_norm": 1.2237526178359985,
-      "learning_rate": 8.740339846557977e-08,
-      "loss": 0.9169,
+      "epoch": 0.08060750871989233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866067808768385e-05,
+      "loss": 0.9087,
       "step": 2935
     },
     {
-      "epoch": 0.9594771241830066,
-      "grad_norm": 1.2299292087554932,
-      "learning_rate": 8.601253810340493e-08,
-      "loss": 0.9257,
+      "epoch": 0.08063497294773557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865922674814167e-05,
+      "loss": 1.1342,
       "step": 2936
     },
     {
-      "epoch": 0.9598039215686275,
-      "grad_norm": 1.230087399482727,
-      "learning_rate": 8.46327853423834e-08,
-      "loss": 0.9215,
+      "epoch": 0.0806624371755788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986577746279694e-05,
+      "loss": 1.1509,
       "step": 2937
     },
     {
-      "epoch": 0.9601307189542484,
-      "grad_norm": 1.227476954460144,
-      "learning_rate": 8.326414172838749e-08,
-      "loss": 0.9289,
+      "epoch": 0.08068990140342204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865632172717867e-05,
+      "loss": 1.1204,
       "step": 2938
     },
     {
-      "epoch": 0.9604575163398693,
-      "grad_norm": 1.3103216886520386,
-      "learning_rate": 8.190660879484058e-08,
-      "loss": 0.882,
+      "epoch": 0.08071736563126528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986548680457809e-05,
+      "loss": 1.0613,
       "step": 2939
     },
     {
-      "epoch": 0.9607843137254902,
-      "grad_norm": 1.2492274045944214,
-      "learning_rate": 8.056018806271937e-08,
-      "loss": 0.9408,
+      "epoch": 0.08074482985910851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986534135837876e-05,
+      "loss": 1.0925,
       "step": 2940
     },
     {
-      "epoch": 0.9611111111111111,
-      "grad_norm": 1.2148364782333374,
-      "learning_rate": 7.922488104054826e-08,
-      "loss": 0.9479,
+      "epoch": 0.08077229408695175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865195834121028e-05,
+      "loss": 1.1146,
       "step": 2941
     },
     {
-      "epoch": 0.961437908496732,
-      "grad_norm": 1.2693687677383423,
-      "learning_rate": 7.790068922440163e-08,
-      "loss": 0.9235,
+      "epoch": 0.08079975831479498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986505023180604e-05,
+      "loss": 1.1906,
       "step": 2942
     },
     {
-      "epoch": 0.961764705882353,
-      "grad_norm": 1.280561923980713,
-      "learning_rate": 7.658761409789939e-08,
-      "loss": 0.9223,
+      "epoch": 0.08082722254263822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864904551434957e-05,
+      "loss": 1.0392,
       "step": 2943
     },
     {
-      "epoch": 0.9620915032679739,
-      "grad_norm": 1.2846943140029907,
-      "learning_rate": 7.528565713220803e-08,
-      "loss": 0.906,
+      "epoch": 0.08085468677048145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986475879300893e-05,
+      "loss": 1.1277,
       "step": 2944
     },
     {
-      "epoch": 0.9624183006535948,
-      "grad_norm": 1.2633920907974243,
-      "learning_rate": 7.399481978603518e-08,
-      "loss": 0.883,
+      "epoch": 0.08088215099832469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864612956529105e-05,
+      "loss": 1.0253,
       "step": 2945
     },
     {
-      "epoch": 0.9627450980392157,
-      "grad_norm": 1.2043060064315796,
-      "learning_rate": 7.271510350563171e-08,
-      "loss": 0.949,
+      "epoch": 0.08090961522616792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864467041996645e-05,
+      "loss": 1.1689,
       "step": 2946
     },
     {
-      "epoch": 0.9630718954248366,
-      "grad_norm": 1.3267790079116821,
-      "learning_rate": 7.144650972478739e-08,
-      "loss": 0.9438,
+      "epoch": 0.08093707945401116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98643210494127e-05,
+      "loss": 1.0785,
       "step": 2947
     },
     {
-      "epoch": 0.9633986928104575,
-      "grad_norm": 1.3110997676849365,
-      "learning_rate": 7.018903986483083e-08,
-      "loss": 0.8944,
+      "epoch": 0.08096454368185438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864174978778427e-05,
+      "loss": 1.1572,
       "step": 2948
     },
     {
-      "epoch": 0.9637254901960784,
-      "grad_norm": 1.201145052909851,
-      "learning_rate": 6.894269533462728e-08,
-      "loss": 0.928,
+      "epoch": 0.08099200790969761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864028830094977e-05,
+      "loss": 1.1059,
       "step": 2949
     },
     {
-      "epoch": 0.9640522875816994,
-      "grad_norm": 1.2305066585540771,
-      "learning_rate": 6.770747753057749e-08,
-      "loss": 0.8958,
+      "epoch": 0.08101947213754085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863882603363512e-05,
+      "loss": 1.0686,
       "step": 2950
     },
     {
-      "epoch": 0.9643790849673203,
-      "grad_norm": 1.1992011070251465,
-      "learning_rate": 6.648338783661668e-08,
-      "loss": 0.9642,
+      "epoch": 0.08104693636538408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863736298585188e-05,
+      "loss": 1.0165,
       "step": 2951
     },
     {
-      "epoch": 0.9647058823529412,
-      "grad_norm": 1.2637802362442017,
-      "learning_rate": 6.527042762420888e-08,
-      "loss": 0.946,
+      "epoch": 0.08107440059322732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986358991576116e-05,
+      "loss": 1.0974,
       "step": 2952
     },
     {
-      "epoch": 0.9650326797385621,
-      "grad_norm": 1.2906533479690552,
-      "learning_rate": 6.406859825235257e-08,
-      "loss": 0.9559,
+      "epoch": 0.08110186482107055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986344345489259e-05,
+      "loss": 1.1392,
       "step": 2953
     },
     {
-      "epoch": 0.965359477124183,
-      "grad_norm": 1.2911957502365112,
-      "learning_rate": 6.287790106757396e-08,
-      "loss": 0.9528,
+      "epoch": 0.08112932904891379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863296915980632e-05,
+      "loss": 1.0935,
       "step": 2954
     },
     {
-      "epoch": 0.9656862745098039,
-      "grad_norm": 1.2863143682479858,
-      "learning_rate": 6.169833740392594e-08,
-      "loss": 0.9396,
+      "epoch": 0.08115679327675702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986315029902645e-05,
+      "loss": 1.1117,
       "step": 2955
     },
     {
-      "epoch": 0.9660130718954248,
-      "grad_norm": 1.2749435901641846,
-      "learning_rate": 6.052990858298801e-08,
-      "loss": 0.9111,
+      "epoch": 0.08118425750460026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98630036040312e-05,
+      "loss": 1.0717,
       "step": 2956
     },
     {
-      "epoch": 0.9663398692810458,
-      "grad_norm": 1.2576189041137695,
-      "learning_rate": 5.937261591386523e-08,
-      "loss": 0.8966,
+      "epoch": 0.0812117217324435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862856830996048e-05,
+      "loss": 1.0632,
       "step": 2957
     },
     {
-      "epoch": 0.9666666666666667,
-      "grad_norm": 1.2506078481674194,
-      "learning_rate": 5.8226460693183715e-08,
-      "loss": 0.9214,
+      "epoch": 0.08123918596028673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986270997992215e-05,
+      "loss": 1.089,
       "step": 2958
     },
     {
-      "epoch": 0.9669934640522876,
-      "grad_norm": 1.2651448249816895,
-      "learning_rate": 5.709144420509183e-08,
-      "loss": 0.9318,
+      "epoch": 0.08126665018812997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862563050810668e-05,
+      "loss": 1.0253,
       "step": 2959
     },
     {
-      "epoch": 0.9673202614379085,
-      "grad_norm": 1.23804771900177,
-      "learning_rate": 5.5967567721258996e-08,
-      "loss": 0.9121,
+      "epoch": 0.0812941144159732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986241604366277e-05,
+      "loss": 1.1848,
       "step": 2960
     },
     {
-      "epoch": 0.9676470588235294,
-      "grad_norm": 1.2750968933105469,
-      "learning_rate": 5.4854832500873534e-08,
-      "loss": 0.9362,
+      "epoch": 0.08132157864381642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862268958479617e-05,
+      "loss": 1.0305,
       "step": 2961
     },
     {
-      "epoch": 0.9679738562091503,
-      "grad_norm": 1.2885894775390625,
-      "learning_rate": 5.375323979063929e-08,
-      "loss": 0.9409,
+      "epoch": 0.08134904287165966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862121795262366e-05,
+      "loss": 1.1145,
       "step": 2962
     },
     {
-      "epoch": 0.9683006535947712,
-      "grad_norm": 1.341187596321106,
-      "learning_rate": 5.266279082477899e-08,
-      "loss": 0.9554,
+      "epoch": 0.08137650709950289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861974554012192e-05,
+      "loss": 1.0562,
       "step": 2963
     },
     {
-      "epoch": 0.9686274509803922,
-      "grad_norm": 1.2524917125701904,
-      "learning_rate": 5.158348682502756e-08,
-      "loss": 0.9585,
+      "epoch": 0.08140397132734613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861827234730256e-05,
+      "loss": 1.1326,
       "step": 2964
     },
     {
-      "epoch": 0.9689542483660131,
-      "grad_norm": 1.2829262018203735,
-      "learning_rate": 5.051532900063438e-08,
-      "loss": 0.9425,
+      "epoch": 0.08143143555518936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861679837417723e-05,
+      "loss": 1.1201,
       "step": 2965
     },
     {
-      "epoch": 0.969281045751634,
-      "grad_norm": 1.3411271572113037,
-      "learning_rate": 4.945831854835881e-08,
-      "loss": 0.9146,
+      "epoch": 0.0814588997830326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861532362075757e-05,
+      "loss": 1.056,
       "step": 2966
     },
     {
-      "epoch": 0.9696078431372549,
-      "grad_norm": 1.2276219129562378,
-      "learning_rate": 4.841245665247352e-08,
-      "loss": 0.9124,
+      "epoch": 0.08148636401087583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861384808705525e-05,
+      "loss": 1.1111,
       "step": 2967
     },
     {
-      "epoch": 0.9699346405228758,
-      "grad_norm": 1.233949899673462,
-      "learning_rate": 4.737774448475896e-08,
-      "loss": 0.9479,
+      "epoch": 0.08151382823871907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98612371773082e-05,
+      "loss": 1.0401,
       "step": 2968
     },
     {
-      "epoch": 0.9702614379084967,
-      "grad_norm": 1.247559666633606,
-      "learning_rate": 4.6354183204503355e-08,
-      "loss": 0.9193,
+      "epoch": 0.0815412924665623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861089467884946e-05,
+      "loss": 1.1599,
       "step": 2969
     },
     {
-      "epoch": 0.9705882352941176,
-      "grad_norm": 1.2837650775909424,
-      "learning_rate": 4.534177395850048e-08,
-      "loss": 0.931,
+      "epoch": 0.08156875669440554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986094168043693e-05,
+      "loss": 1.0276,
       "step": 2970
     },
     {
-      "epoch": 0.9709150326797386,
-      "grad_norm": 1.2611311674118042,
-      "learning_rate": 4.434051788105076e-08,
-      "loss": 0.9236,
+      "epoch": 0.08159622092224877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860793814965326e-05,
+      "loss": 1.0246,
       "step": 2971
     },
     {
-      "epoch": 0.9712418300653595,
-      "grad_norm": 1.2409662008285522,
-      "learning_rate": 4.335041609396018e-08,
-      "loss": 0.9295,
+      "epoch": 0.08162368515009201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98606458714713e-05,
+      "loss": 1.1413,
       "step": 2972
     },
     {
-      "epoch": 0.9715686274509804,
-      "grad_norm": 1.225548505783081,
-      "learning_rate": 4.237146970653361e-08,
-      "loss": 0.9281,
+      "epoch": 0.08165114937793524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860497849956023e-05,
+      "loss": 1.0073,
       "step": 2973
     },
     {
-      "epoch": 0.9718954248366013,
-      "grad_norm": 1.2250770330429077,
-      "learning_rate": 4.140367981558147e-08,
-      "loss": 0.9438,
+      "epoch": 0.08167861360577848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860349750420672e-05,
+      "loss": 1.1096,
       "step": 2974
     },
     {
-      "epoch": 0.9722222222222222,
-      "grad_norm": 1.228743076324463,
-      "learning_rate": 4.044704750541084e-08,
-      "loss": 0.9146,
+      "epoch": 0.0817060778336217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986020157286641e-05,
+      "loss": 1.1659,
       "step": 2975
     },
     {
-      "epoch": 0.9725490196078431,
-      "grad_norm": 1.2991398572921753,
-      "learning_rate": 3.950157384783104e-08,
-      "loss": 0.9074,
+      "epoch": 0.08173354206146494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860053317294417e-05,
+      "loss": 1.0757,
       "step": 2976
     },
     {
-      "epoch": 0.972875816993464,
-      "grad_norm": 1.2687703371047974,
-      "learning_rate": 3.8567259902146936e-08,
-      "loss": 0.9361,
+      "epoch": 0.08176100628930817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985990498370586e-05,
+      "loss": 1.0578,
       "step": 2977
     },
     {
-      "epoch": 0.973202614379085,
-      "grad_norm": 1.2624174356460571,
-      "learning_rate": 3.7644106715161164e-08,
-      "loss": 0.899,
+      "epoch": 0.08178847051715141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859756572101915e-05,
+      "loss": 1.0745,
       "step": 2978
     },
     {
-      "epoch": 0.9735294117647059,
-      "grad_norm": 1.2400588989257812,
-      "learning_rate": 3.6732115321173044e-08,
-      "loss": 0.9061,
+      "epoch": 0.08181593474499464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859608082483757e-05,
+      "loss": 1.1703,
       "step": 2979
     },
     {
-      "epoch": 0.9738562091503268,
-      "grad_norm": 1.2320042848587036,
-      "learning_rate": 3.5831286741973006e-08,
-      "loss": 0.9405,
+      "epoch": 0.08184339897283788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859459514852565e-05,
+      "loss": 1.0878,
       "step": 2980
     },
     {
-      "epoch": 0.9741830065359477,
-      "grad_norm": 1.2547532320022583,
-      "learning_rate": 3.494162198684703e-08,
-      "loss": 0.9223,
+      "epoch": 0.08187086320068111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985931086920951e-05,
+      "loss": 1.0508,
       "step": 2981
     },
     {
-      "epoch": 0.9745098039215686,
-      "grad_norm": 1.238123893737793,
-      "learning_rate": 3.406312205257334e-08,
-      "loss": 0.9165,
+      "epoch": 0.08189832742852435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859162145555762e-05,
+      "loss": 1.0629,
       "step": 2982
     },
     {
-      "epoch": 0.9748366013071895,
-      "grad_norm": 1.2480792999267578,
-      "learning_rate": 3.319578792342126e-08,
-      "loss": 0.9268,
+      "epoch": 0.08192579165636758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859013343892506e-05,
+      "loss": 1.0294,
       "step": 2983
     },
     {
-      "epoch": 0.9751633986928104,
-      "grad_norm": 1.2685017585754395,
-      "learning_rate": 3.2339620571146814e-08,
-      "loss": 0.9373,
+      "epoch": 0.08195325588421082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985886446422092e-05,
+      "loss": 1.1021,
       "step": 2984
     },
     {
-      "epoch": 0.9754901960784313,
-      "grad_norm": 1.2430567741394043,
-      "learning_rate": 3.149462095500044e-08,
-      "loss": 0.9429,
+      "epoch": 0.08198072011205405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985871550654218e-05,
+      "loss": 1.2103,
       "step": 2985
     },
     {
-      "epoch": 0.9758169934640523,
-      "grad_norm": 1.2629387378692627,
-      "learning_rate": 3.0660790021714846e-08,
-      "loss": 0.9216,
+      "epoch": 0.08200818433989729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985856647085746e-05,
+      "loss": 1.004,
       "step": 2986
     },
     {
-      "epoch": 0.9761437908496732,
-      "grad_norm": 1.3042864799499512,
-      "learning_rate": 2.983812870551273e-08,
-      "loss": 0.926,
+      "epoch": 0.08203564856774052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9858417357167945e-05,
+      "loss": 1.031,
       "step": 2987
     },
     {
-      "epoch": 0.9764705882352941,
-      "grad_norm": 1.2855888605117798,
-      "learning_rate": 2.902663792810012e-08,
-      "loss": 0.9144,
+      "epoch": 0.08206311279558375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9858268165474813e-05,
+      "loss": 1.192,
       "step": 2988
     },
     {
-      "epoch": 0.976797385620915,
-      "grad_norm": 1.2476388216018677,
-      "learning_rate": 2.8226318598670866e-08,
-      "loss": 0.9466,
+      "epoch": 0.08209057702342698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9858118895779248e-05,
+      "loss": 0.9609,
       "step": 2989
     },
     {
-      "epoch": 0.9771241830065359,
-      "grad_norm": 1.2977923154830933,
-      "learning_rate": 2.7437171613898806e-08,
-      "loss": 0.9076,
+      "epoch": 0.08211804125127022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985796954808242e-05,
+      "loss": 1.1872,
       "step": 2990
     },
     {
-      "epoch": 0.9774509803921568,
-      "grad_norm": 1.2652438879013062,
-      "learning_rate": 2.6659197857941134e-08,
-      "loss": 0.9023,
+      "epoch": 0.08214550547911345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857820122385523e-05,
+      "loss": 1.0444,
       "step": 2991
     },
     {
-      "epoch": 0.9777777777777777,
-      "grad_norm": 1.2535902261734009,
-      "learning_rate": 2.5892398202439494e-08,
-      "loss": 0.9273,
+      "epoch": 0.08217296970695669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857670618689737e-05,
+      "loss": 1.1118,
       "step": 2992
     },
     {
-      "epoch": 0.9781045751633987,
-      "grad_norm": 1.312806248664856,
-      "learning_rate": 2.5136773506511113e-08,
-      "loss": 0.9014,
+      "epoch": 0.08220043393479992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985752103699624e-05,
+      "loss": 1.2046,
       "step": 2993
     },
     {
-      "epoch": 0.9784313725490196,
-      "grad_norm": 1.2147274017333984,
-      "learning_rate": 2.4392324616758777e-08,
-      "loss": 0.8927,
+      "epoch": 0.08222789816264316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857371377306216e-05,
+      "loss": 1.1524,
       "step": 2994
     },
     {
-      "epoch": 0.9787581699346405,
-      "grad_norm": 1.2400981187820435,
-      "learning_rate": 2.365905236725752e-08,
-      "loss": 0.9106,
+      "epoch": 0.08225536239048639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857221639620854e-05,
+      "loss": 1.1603,
       "step": 2995
     },
     {
-      "epoch": 0.9790849673202614,
-      "grad_norm": 1.282895803451538,
-      "learning_rate": 2.293695757956571e-08,
-      "loss": 0.938,
+      "epoch": 0.08228282661832963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857071823941336e-05,
+      "loss": 1.1069,
       "step": 2996
     },
     {
-      "epoch": 0.9794117647058823,
-      "grad_norm": 1.2666411399841309,
-      "learning_rate": 2.2226041062715087e-08,
-      "loss": 0.9198,
+      "epoch": 0.08231029084617286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856921930268846e-05,
+      "loss": 0.9835,
       "step": 2997
     },
     {
-      "epoch": 0.9797385620915032,
-      "grad_norm": 1.2039028406143188,
-      "learning_rate": 2.1526303613214063e-08,
-      "loss": 0.9489,
+      "epoch": 0.0823377550740161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985677195860457e-05,
+      "loss": 1.114,
       "step": 2998
     },
     {
-      "epoch": 0.9800653594771241,
-      "grad_norm": 1.2359743118286133,
-      "learning_rate": 2.0837746015046623e-08,
-      "loss": 0.9147,
+      "epoch": 0.08236521930185933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856621908949704e-05,
+      "loss": 1.1204,
       "step": 2999
     },
     {
-      "epoch": 0.9803921568627451,
-      "grad_norm": 1.2589095830917358,
-      "learning_rate": 2.0160369039670115e-08,
-      "loss": 0.9297,
+      "epoch": 0.08239268352970257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856471781305423e-05,
+      "loss": 1.0548,
       "step": 3000
     },
     {
-      "epoch": 0.980718954248366,
-      "grad_norm": 1.224442481994629,
-      "learning_rate": 1.9494173446017452e-08,
-      "loss": 0.9284,
+      "epoch": 0.0824201477575458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856321575672918e-05,
+      "loss": 1.1765,
       "step": 3001
     },
     {
-      "epoch": 0.9810457516339869,
-      "grad_norm": 1.2154706716537476,
-      "learning_rate": 1.883915998049046e-08,
-      "loss": 0.9165,
+      "epoch": 0.08244761198538902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856171292053378e-05,
+      "loss": 1.1649,
       "step": 3002
     },
     {
-      "epoch": 0.9813725490196078,
-      "grad_norm": 1.3387328386306763,
-      "learning_rate": 1.819532937696433e-08,
-      "loss": 0.901,
+      "epoch": 0.08247507621323226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856020930447992e-05,
+      "loss": 1.0675,
       "step": 3003
     },
     {
-      "epoch": 0.9816993464052287,
-      "grad_norm": 1.3553398847579956,
-      "learning_rate": 1.7562682356786488e-08,
-      "loss": 0.9065,
+      "epoch": 0.0825025404410755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855870490857953e-05,
+      "loss": 1.1388,
       "step": 3004
     },
     {
-      "epoch": 0.9820261437908496,
-      "grad_norm": 1.360330581665039,
-      "learning_rate": 1.6941219628769934e-08,
-      "loss": 0.9476,
+      "epoch": 0.08253000466891873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855719973284447e-05,
+      "loss": 1.0844,
       "step": 3005
     },
     {
-      "epoch": 0.9823529411764705,
-      "grad_norm": 1.3150663375854492,
-      "learning_rate": 1.6330941889202145e-08,
-      "loss": 0.945,
+      "epoch": 0.08255746889676197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985556937772867e-05,
+      "loss": 1.0867,
       "step": 3006
     },
     {
-      "epoch": 0.9826797385620915,
-      "grad_norm": 1.2501200437545776,
-      "learning_rate": 1.5731849821833955e-08,
-      "loss": 0.8808,
+      "epoch": 0.0825849331246052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855418704191807e-05,
+      "loss": 1.0892,
       "step": 3007
     },
     {
-      "epoch": 0.9830065359477124,
-      "grad_norm": 1.2569761276245117,
-      "learning_rate": 1.514394409788955e-08,
-      "loss": 0.906,
+      "epoch": 0.08261239735244844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855267952675055e-05,
+      "loss": 1.0957,
       "step": 3008
     },
     {
-      "epoch": 0.9833333333333333,
-      "grad_norm": 1.2934633493423462,
-      "learning_rate": 1.456722537605315e-08,
-      "loss": 0.9304,
+      "epoch": 0.08263986158029167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855117123179606e-05,
+      "loss": 1.0897,
       "step": 3009
     },
     {
-      "epoch": 0.9836601307189542,
-      "grad_norm": 1.246471643447876,
-      "learning_rate": 1.4001694302481216e-08,
-      "loss": 0.8976,
+      "epoch": 0.0826673258081349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854966215706653e-05,
+      "loss": 1.0484,
       "step": 3010
     },
     {
-      "epoch": 0.9839869281045751,
-      "grad_norm": 1.3419761657714844,
-      "learning_rate": 1.3447351510792461e-08,
-      "loss": 0.9285,
+      "epoch": 0.08269479003597814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985481523025739e-05,
+      "loss": 1.0574,
       "step": 3011
     },
     {
-      "epoch": 0.984313725490196,
-      "grad_norm": 1.2551084756851196,
-      "learning_rate": 1.290419762207007e-08,
-      "loss": 0.9443,
+      "epoch": 0.08272225426382138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854664166833005e-05,
+      "loss": 0.9767,
       "step": 3012
     },
     {
-      "epoch": 0.984640522875817,
-      "grad_norm": 1.2327992916107178,
-      "learning_rate": 1.2372233244862809e-08,
-      "loss": 0.9227,
+      "epoch": 0.08274971849166461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985451302543471e-05,
+      "loss": 1.0728,
       "step": 3013
     },
     {
-      "epoch": 0.984967320261438,
-      "grad_norm": 1.2566817998886108,
-      "learning_rate": 1.1851458975181695e-08,
-      "loss": 0.9344,
+      "epoch": 0.08277718271950785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854361806063686e-05,
+      "loss": 1.014,
       "step": 3014
     },
     {
-      "epoch": 0.9852941176470589,
-      "grad_norm": 1.2114917039871216,
-      "learning_rate": 1.1341875396500002e-08,
-      "loss": 0.9584,
+      "epoch": 0.08280464694735108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985421050872113e-05,
+      "loss": 1.2236,
       "step": 3015
     },
     {
-      "epoch": 0.9856209150326798,
-      "grad_norm": 1.2800182104110718,
-      "learning_rate": 1.0843483079755468e-08,
-      "loss": 0.9773,
+      "epoch": 0.0828321111751943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985405913340825e-05,
+      "loss": 1.1014,
       "step": 3016
     },
     {
-      "epoch": 0.9859477124183007,
-      "grad_norm": 1.2389096021652222,
-      "learning_rate": 1.0356282583342536e-08,
-      "loss": 0.9062,
+      "epoch": 0.08285957540303754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853907680126236e-05,
+      "loss": 1.0457,
       "step": 3017
     },
     {
-      "epoch": 0.9862745098039216,
-      "grad_norm": 1.2918764352798462,
-      "learning_rate": 9.88027445312123e-09,
-      "loss": 0.924,
+      "epoch": 0.08288703963088077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853756148876285e-05,
+      "loss": 1.1655,
       "step": 3018
     },
     {
-      "epoch": 0.9866013071895425,
-      "grad_norm": 1.2447881698608398,
-      "learning_rate": 9.415459222409384e-09,
-      "loss": 0.9521,
+      "epoch": 0.08291450385872401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98536045396596e-05,
+      "loss": 1.1362,
       "step": 3019
     },
     {
-      "epoch": 0.9869281045751634,
-      "grad_norm": 1.2139034271240234,
-      "learning_rate": 8.961837411982643e-09,
-      "loss": 0.9296,
+      "epoch": 0.08294196808656724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853452852477378e-05,
+      "loss": 1.0234,
       "step": 3020
     },
     {
-      "epoch": 0.9872549019607844,
-      "grad_norm": 1.2161638736724854,
-      "learning_rate": 8.519409530080014e-09,
-      "loss": 0.9309,
+      "epoch": 0.08296943231441048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985330108733082e-05,
+      "loss": 1.0969,
       "step": 3021
     },
     {
-      "epoch": 0.9875816993464053,
-      "grad_norm": 1.2810288667678833,
-      "learning_rate": 8.088176072396092e-09,
-      "loss": 0.9115,
+      "epoch": 0.08299689654225371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853149244221126e-05,
+      "loss": 1.0607,
       "step": 3022
     },
     {
-      "epoch": 0.9879084967320262,
-      "grad_norm": 1.244753122329712,
-      "learning_rate": 7.668137522082176e-09,
-      "loss": 0.9263,
+      "epoch": 0.08302436077009695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98529973231495e-05,
+      "loss": 1.1458,
       "step": 3023
     },
     {
-      "epoch": 0.9882352941176471,
-      "grad_norm": 1.2982666492462158,
-      "learning_rate": 7.25929434974848e-09,
-      "loss": 0.9264,
+      "epoch": 0.08305182499794019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852845324117144e-05,
+      "loss": 1.0889,
       "step": 3024
     },
     {
-      "epoch": 0.988562091503268,
-      "grad_norm": 1.2737797498703003,
-      "learning_rate": 6.861647013461925e-09,
-      "loss": 0.9368,
+      "epoch": 0.08307928922578342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852693247125255e-05,
+      "loss": 1.1539,
       "step": 3025
     },
     {
-      "epoch": 0.9888888888888889,
-      "grad_norm": 1.2095872163772583,
-      "learning_rate": 6.475195958745018e-09,
-      "loss": 0.9297,
+      "epoch": 0.08310675345362666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852541092175042e-05,
+      "loss": 1.1089,
       "step": 3026
     },
     {
-      "epoch": 0.9892156862745098,
-      "grad_norm": 1.2623018026351929,
-      "learning_rate": 6.099941618576966e-09,
-      "loss": 0.924,
+      "epoch": 0.08313421768146989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852388859267712e-05,
+      "loss": 0.9694,
       "step": 3027
     },
     {
-      "epoch": 0.9895424836601308,
-      "grad_norm": 1.2859172821044922,
-      "learning_rate": 5.735884413391457e-09,
-      "loss": 0.9258,
+      "epoch": 0.08316168190931313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985223654840446e-05,
+      "loss": 1.1222,
       "step": 3028
     },
     {
-      "epoch": 0.9898692810457517,
-      "grad_norm": 1.3006476163864136,
-      "learning_rate": 5.38302475107666e-09,
-      "loss": 0.8924,
+      "epoch": 0.08318914613715635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852084159586495e-05,
+      "loss": 1.0349,
       "step": 3029
     },
     {
-      "epoch": 0.9901960784313726,
-      "grad_norm": 1.2645727396011353,
-      "learning_rate": 5.041363026976331e-09,
-      "loss": 0.9502,
+      "epoch": 0.08321661036499958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851931692815028e-05,
+      "loss": 1.1009,
       "step": 3030
     },
     {
-      "epoch": 0.9905228758169935,
-      "grad_norm": 1.3012253046035767,
-      "learning_rate": 4.710899623886489e-09,
-      "loss": 0.9765,
+      "epoch": 0.08324407459284282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985177914809126e-05,
+      "loss": 1.1598,
       "step": 3031
     },
     {
-      "epoch": 0.9908496732026144,
-      "grad_norm": 1.2175376415252686,
-      "learning_rate": 4.39163491205652e-09,
-      "loss": 0.9037,
+      "epoch": 0.08327153882068605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98516265254164e-05,
+      "loss": 1.031,
       "step": 3032
     },
     {
-      "epoch": 0.9911764705882353,
-      "grad_norm": 1.2316147089004517,
-      "learning_rate": 4.083569249191399e-09,
-      "loss": 0.9495,
+      "epoch": 0.08329900304852929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985147382479165e-05,
+      "loss": 1.0952,
       "step": 3033
     },
     {
-      "epoch": 0.9915032679738562,
-      "grad_norm": 1.3054842948913574,
-      "learning_rate": 3.786702980445034e-09,
-      "loss": 0.9208,
+      "epoch": 0.08332646727637252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851321046218226e-05,
+      "loss": 1.0835,
       "step": 3034
     },
     {
-      "epoch": 0.9918300653594772,
-      "grad_norm": 1.2581192255020142,
-      "learning_rate": 3.5010364384280293e-09,
-      "loss": 0.8923,
+      "epoch": 0.08335393150421576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851168189697335e-05,
+      "loss": 1.1344,
       "step": 3035
     },
     {
-      "epoch": 0.9921568627450981,
-      "grad_norm": 1.2894705533981323,
-      "learning_rate": 3.226569943197699e-09,
-      "loss": 0.9121,
+      "epoch": 0.083381395732059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851015255230185e-05,
+      "loss": 1.0311,
       "step": 3036
     },
     {
-      "epoch": 0.992483660130719,
-      "grad_norm": 1.1880016326904297,
-      "learning_rate": 2.963303802266948e-09,
-      "loss": 0.9294,
+      "epoch": 0.08340885995990223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850862242817986e-05,
+      "loss": 1.0995,
       "step": 3037
     },
     {
-      "epoch": 0.9928104575163399,
-      "grad_norm": 1.2762552499771118,
-      "learning_rate": 2.711238310597608e-09,
-      "loss": 0.9389,
+      "epoch": 0.08343632418774546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985070915246195e-05,
+      "loss": 1.0792,
       "step": 3038
     },
     {
-      "epoch": 0.9931372549019608,
-      "grad_norm": 1.2670321464538574,
-      "learning_rate": 2.4703737506037715e-09,
-      "loss": 0.9665,
+      "epoch": 0.0834637884155887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985055598416328e-05,
+      "loss": 1.1079,
       "step": 3039
     },
     {
-      "epoch": 0.9934640522875817,
-      "grad_norm": 1.1829358339309692,
-      "learning_rate": 2.2407103921484595e-09,
-      "loss": 0.9046,
+      "epoch": 0.08349125264343193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850402737923206e-05,
+      "loss": 1.0456,
       "step": 3040
     },
     {
-      "epoch": 0.9937908496732026,
-      "grad_norm": 1.1963225603103638,
-      "learning_rate": 2.022248492546952e-09,
-      "loss": 0.9295,
+      "epoch": 0.08351871687127517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850249413742926e-05,
+      "loss": 1.0442,
       "step": 3041
     },
     {
-      "epoch": 0.9941176470588236,
-      "grad_norm": 1.2797281742095947,
-      "learning_rate": 1.8149882965623478e-09,
-      "loss": 0.9532,
+      "epoch": 0.0835461810991184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850096011623658e-05,
+      "loss": 1.1168,
       "step": 3042
     },
     {
-      "epoch": 0.9944444444444445,
-      "grad_norm": 1.252914547920227,
-      "learning_rate": 1.6189300364088945e-09,
-      "loss": 0.9242,
+      "epoch": 0.08357364532696163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984994253156661e-05,
+      "loss": 1.1582,
       "step": 3043
     },
     {
-      "epoch": 0.9947712418300654,
-      "grad_norm": 1.296935796737671,
-      "learning_rate": 1.4340739317497688e-09,
-      "loss": 0.9612,
+      "epoch": 0.08360110955480486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849788973573007e-05,
+      "loss": 1.0541,
       "step": 3044
     },
     {
-      "epoch": 0.9950980392156863,
-      "grad_norm": 1.237506628036499,
-      "learning_rate": 1.2604201896970758e-09,
-      "loss": 0.9434,
+      "epoch": 0.0836285737826481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849635337644053e-05,
+      "loss": 1.2164,
       "step": 3045
     },
     {
-      "epoch": 0.9954248366013072,
-      "grad_norm": 1.2566847801208496,
-      "learning_rate": 1.0979690048107395e-09,
-      "loss": 0.942,
+      "epoch": 0.08365603801049133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984948162378097e-05,
+      "loss": 0.9941,
       "step": 3046
     },
     {
-      "epoch": 0.9957516339869281,
-      "grad_norm": 1.2504136562347412,
-      "learning_rate": 9.467205591018324e-10,
-      "loss": 0.9436,
+      "epoch": 0.08368350223833457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849327831984974e-05,
+      "loss": 1.1453,
       "step": 3047
     },
     {
-      "epoch": 0.996078431372549,
-      "grad_norm": 1.2101879119873047,
-      "learning_rate": 8.066750220292463e-10,
-      "loss": 0.9139,
+      "epoch": 0.0837109664661778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984917396225728e-05,
+      "loss": 1.0413,
       "step": 3048
     },
     {
-      "epoch": 0.99640522875817,
-      "grad_norm": 1.2871382236480713,
-      "learning_rate": 6.778325504974704e-10,
-      "loss": 0.9404,
+      "epoch": 0.08373843069402104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849020014599108e-05,
+      "loss": 0.9892,
       "step": 3049
     },
     {
-      "epoch": 0.9967320261437909,
-      "grad_norm": 1.2015436887741089,
-      "learning_rate": 5.601932888632533e-10,
-      "loss": 0.9201,
+      "epoch": 0.08376589492186427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848865989011678e-05,
+      "loss": 1.1133,
       "step": 3050
     },
     {
-      "epoch": 0.9970588235294118,
-      "grad_norm": 1.2414661645889282,
-      "learning_rate": 4.537573689278318e-10,
-      "loss": 0.9304,
+      "epoch": 0.08379335914970751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98487118854962e-05,
+      "loss": 1.1216,
       "step": 3051
     },
     {
-      "epoch": 0.9973856209150327,
-      "grad_norm": 1.2801110744476318,
-      "learning_rate": 3.585249099435917e-10,
-      "loss": 0.9407,
+      "epoch": 0.08382082337755074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848557704053898e-05,
+      "loss": 1.0205,
       "step": 3052
     },
     {
-      "epoch": 0.9977124183006536,
-      "grad_norm": 1.2112396955490112,
-      "learning_rate": 2.7449601860629614e-10,
-      "loss": 0.9271,
+      "epoch": 0.08384828760539398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848403444685998e-05,
+      "loss": 1.1614,
       "step": 3053
     },
     {
-      "epoch": 0.9980392156862745,
-      "grad_norm": 1.2668055295944214,
-      "learning_rate": 2.0167078906285776e-10,
-      "loss": 0.9346,
+      "epoch": 0.08387575183323721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984824910739371e-05,
+      "loss": 1.0644,
       "step": 3054
     },
     {
-      "epoch": 0.9983660130718954,
-      "grad_norm": 1.2903761863708496,
-      "learning_rate": 1.4004930290578699e-10,
-      "loss": 0.9534,
+      "epoch": 0.08390321606108045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848094692178263e-05,
+      "loss": 1.0045,
       "step": 3055
     },
     {
-      "epoch": 0.9986928104575163,
-      "grad_norm": 1.3056076765060425,
-      "learning_rate": 8.963162917763335e-11,
-      "loss": 0.9544,
+      "epoch": 0.08393068028892367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847940199040874e-05,
+      "loss": 1.1191,
       "step": 3056
     },
     {
-      "epoch": 0.9990196078431373,
-      "grad_norm": 1.266438364982605,
-      "learning_rate": 5.041782436321363e-11,
-      "loss": 0.9439,
+      "epoch": 0.0839581445167669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984778562798277e-05,
+      "loss": 1.0703,
       "step": 3057
     },
     {
-      "epoch": 0.9993464052287582,
-      "grad_norm": 1.2873599529266357,
-      "learning_rate": 2.2407932399604038e-11,
-      "loss": 0.9434,
+      "epoch": 0.08398560874461014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847630979005168e-05,
+      "loss": 1.0857,
       "step": 3058
     },
     {
-      "epoch": 0.9996732026143791,
-      "grad_norm": 1.1944462060928345,
-      "learning_rate": 5.601984669478811e-12,
-      "loss": 0.9185,
+      "epoch": 0.08401307297245338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847476252109294e-05,
+      "loss": 1.1287,
       "step": 3059
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 1.2318390607833862,
-      "learning_rate": 0.0,
-      "loss": 0.9196,
+      "epoch": 0.08404053720029661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847321447296378e-05,
+      "loss": 1.1118,
       "step": 3060
     },
+    {
+      "epoch": 0.08406800142813985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847166564567638e-05,
+      "loss": 1.0519,
+      "step": 3061
+    },
+    {
+      "epoch": 0.08409546565598308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98470116039243e-05,
+      "loss": 0.9716,
+      "step": 3062
+    },
+    {
+      "epoch": 0.08412292988382632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846856565367595e-05,
+      "loss": 1.0764,
+      "step": 3063
+    },
+    {
+      "epoch": 0.08415039411166955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846701448898748e-05,
+      "loss": 0.9832,
+      "step": 3064
+    },
+    {
+      "epoch": 0.08417785833951279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984654625451898e-05,
+      "loss": 1.0386,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08420532256735602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846390982229526e-05,
+      "loss": 1.0816,
+      "step": 3066
+    },
+    {
+      "epoch": 0.08423278679519926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846235632031613e-05,
+      "loss": 1.0908,
+      "step": 3067
+    },
+    {
+      "epoch": 0.08426025102304249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984608020392647e-05,
+      "loss": 1.0471,
+      "step": 3068
+    },
+    {
+      "epoch": 0.08428771525088573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845924697915323e-05,
+      "loss": 0.9647,
+      "step": 3069
+    },
+    {
+      "epoch": 0.08431517947872895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984576911399941e-05,
+      "loss": 1.0162,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08434264370657218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845613452179954e-05,
+      "loss": 1.041,
+      "step": 3071
+    },
+    {
+      "epoch": 0.08437010793441542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845457712458196e-05,
+      "loss": 1.0031,
+      "step": 3072
+    },
+    {
+      "epoch": 0.08439757216225865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845301894835357e-05,
+      "loss": 1.1264,
+      "step": 3073
+    },
+    {
+      "epoch": 0.08442503639010189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845145999312677e-05,
+      "loss": 1.0098,
+      "step": 3074
+    },
+    {
+      "epoch": 0.08445250061794513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984499002589139e-05,
+      "loss": 1.182,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08447996484578836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984483397457273e-05,
+      "loss": 1.1343,
+      "step": 3076
+    },
+    {
+      "epoch": 0.0845074290736316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984467784535793e-05,
+      "loss": 1.1034,
+      "step": 3077
+    },
+    {
+      "epoch": 0.08453489330147483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984452163824822e-05,
+      "loss": 1.1573,
+      "step": 3078
+    },
+    {
+      "epoch": 0.08456235752931807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9844365353244845e-05,
+      "loss": 1.1104,
+      "step": 3079
+    },
+    {
+      "epoch": 0.0845898217571613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984420899034904e-05,
+      "loss": 1.1182,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08461728598500454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9844052549562038e-05,
+      "loss": 1.0133,
+      "step": 3081
+    },
+    {
+      "epoch": 0.08464475021284777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984389603088508e-05,
+      "loss": 1.0242,
+      "step": 3082
+    },
+    {
+      "epoch": 0.084672214440691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98437394343194e-05,
+      "loss": 1.0349,
+      "step": 3083
+    },
+    {
+      "epoch": 0.08469967866853423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843582759866242e-05,
+      "loss": 1.1285,
+      "step": 3084
+    },
+    {
+      "epoch": 0.08472714289637746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843426007526845e-05,
+      "loss": 1.0561,
+      "step": 3085
+    },
+    {
+      "epoch": 0.0847546071242207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984326917730245e-05,
+      "loss": 1.1252,
+      "step": 3086
+    },
+    {
+      "epoch": 0.08478207135206393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843112269194297e-05,
+      "loss": 1.1526,
+      "step": 3087
+    },
+    {
+      "epoch": 0.08480953557990717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842955283203623e-05,
+      "loss": 1.0888,
+      "step": 3088
+    },
+    {
+      "epoch": 0.0848369998077504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842798219331674e-05,
+      "loss": 1.1057,
+      "step": 3089
+    },
+    {
+      "epoch": 0.08486446403559364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842641077579694e-05,
+      "loss": 1.1611,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08489192826343687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842483857948927e-05,
+      "loss": 1.1477,
+      "step": 3091
+    },
+    {
+      "epoch": 0.08491939249128011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842326560440616e-05,
+      "loss": 1.1476,
+      "step": 3092
+    },
+    {
+      "epoch": 0.08494685671912335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842169185056e-05,
+      "loss": 1.1813,
+      "step": 3093
+    },
+    {
+      "epoch": 0.08497432094696658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984201173179633e-05,
+      "loss": 1.0475,
+      "step": 3094
+    },
+    {
+      "epoch": 0.08500178517480982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841854200662852e-05,
+      "loss": 1.1733,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08502924940265305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841696591656812e-05,
+      "loss": 1.0333,
+      "step": 3096
+    },
+    {
+      "epoch": 0.08505671363049627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841538904779453e-05,
+      "loss": 1.0259,
+      "step": 3097
+    },
+    {
+      "epoch": 0.08508417785833951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841381140032033e-05,
+      "loss": 1.0765,
+      "step": 3098
+    },
+    {
+      "epoch": 0.08511164208618274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841223297415787e-05,
+      "loss": 1.0956,
+      "step": 3099
+    },
+    {
+      "epoch": 0.08513910631402598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841065376931972e-05,
+      "loss": 1.1147,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08516657054186921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840907378581834e-05,
+      "loss": 1.028,
+      "step": 3101
+    },
+    {
+      "epoch": 0.08519403476971245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840749302366627e-05,
+      "loss": 1.0308,
+      "step": 3102
+    },
+    {
+      "epoch": 0.08522149899755568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840591148287598e-05,
+      "loss": 1.1617,
+      "step": 3103
+    },
+    {
+      "epoch": 0.08524896322539892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840432916346e-05,
+      "loss": 1.0201,
+      "step": 3104
+    },
+    {
+      "epoch": 0.08527642745324215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840274606543088e-05,
+      "loss": 1.1258,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08530389168108539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840116218880107e-05,
+      "loss": 1.1186,
+      "step": 3106
+    },
+    {
+      "epoch": 0.08533135590892862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983995775335832e-05,
+      "loss": 1.0807,
+      "step": 3107
+    },
+    {
+      "epoch": 0.08535882013677186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839799209978973e-05,
+      "loss": 0.9334,
+      "step": 3108
+    },
+    {
+      "epoch": 0.0853862843646151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983964058874332e-05,
+      "loss": 1.1203,
+      "step": 3109
+    },
+    {
+      "epoch": 0.08541374859245832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839481889652623e-05,
+      "loss": 1.0788,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08544121282030155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839323112708133e-05,
+      "loss": 1.0566,
+      "step": 3111
+    },
+    {
+      "epoch": 0.08546867704814479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839164257911107e-05,
+      "loss": 0.9679,
+      "step": 3112
+    },
+    {
+      "epoch": 0.08549614127598802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839005325262805e-05,
+      "loss": 0.9211,
+      "step": 3113
+    },
+    {
+      "epoch": 0.08552360550383126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838846314764477e-05,
+      "loss": 1.1197,
+      "step": 3114
+    },
+    {
+      "epoch": 0.08555106973167449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838687226417387e-05,
+      "loss": 0.9723,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08557853395951773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983852806022279e-05,
+      "loss": 1.0734,
+      "step": 3116
+    },
+    {
+      "epoch": 0.08560599818736096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838368816181953e-05,
+      "loss": 1.1573,
+      "step": 3117
+    },
+    {
+      "epoch": 0.0856334624152042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983820949429613e-05,
+      "loss": 1.1481,
+      "step": 3118
+    },
+    {
+      "epoch": 0.08566092664304743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983805009456658e-05,
+      "loss": 1.0747,
+      "step": 3119
+    },
+    {
+      "epoch": 0.08568839087089067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837890616994565e-05,
+      "loss": 1.069,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0857158550987339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837731061581355e-05,
+      "loss": 1.0486,
+      "step": 3121
+    },
+    {
+      "epoch": 0.08574331932657714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837571428328202e-05,
+      "loss": 1.1029,
+      "step": 3122
+    },
+    {
+      "epoch": 0.08577078355442037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837411717236374e-05,
+      "loss": 1.0631,
+      "step": 3123
+    },
+    {
+      "epoch": 0.0857982477822636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837251928307134e-05,
+      "loss": 1.1212,
+      "step": 3124
+    },
+    {
+      "epoch": 0.08582571201010683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837092061541746e-05,
+      "loss": 0.9597,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08585317623795007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836932116941474e-05,
+      "loss": 1.077,
+      "step": 3126
+    },
+    {
+      "epoch": 0.0858806404657933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836772094507582e-05,
+      "loss": 1.0762,
+      "step": 3127
+    },
+    {
+      "epoch": 0.08590810469363654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836611994241343e-05,
+      "loss": 1.1129,
+      "step": 3128
+    },
+    {
+      "epoch": 0.08593556892147977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983645181614402e-05,
+      "loss": 1.1175,
+      "step": 3129
+    },
+    {
+      "epoch": 0.085963033149323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983629156021688e-05,
+      "loss": 1.1403,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08599049737716624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836131226461187e-05,
+      "loss": 1.1586,
+      "step": 3131
+    },
+    {
+      "epoch": 0.08601796160500948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983597081487822e-05,
+      "loss": 0.9891,
+      "step": 3132
+    },
+    {
+      "epoch": 0.08604542583285271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835810325469237e-05,
+      "loss": 1.1964,
+      "step": 3133
+    },
+    {
+      "epoch": 0.08607289006069595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835649758235513e-05,
+      "loss": 1.2438,
+      "step": 3134
+    },
+    {
+      "epoch": 0.08610035428853918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983548911317832e-05,
+      "loss": 1.0974,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08612781851638242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835328390298928e-05,
+      "loss": 0.9794,
+      "step": 3136
+    },
+    {
+      "epoch": 0.08615528274422565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835167589598604e-05,
+      "loss": 1.1102,
+      "step": 3137
+    },
+    {
+      "epoch": 0.08618274697206887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835006711078625e-05,
+      "loss": 1.0916,
+      "step": 3138
+    },
+    {
+      "epoch": 0.08621021119991211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834845754740266e-05,
+      "loss": 0.995,
+      "step": 3139
+    },
+    {
+      "epoch": 0.08623767542775534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834684720584797e-05,
+      "loss": 1.0616,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08626513965559858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834523608613492e-05,
+      "loss": 1.072,
+      "step": 3141
+    },
+    {
+      "epoch": 0.08629260388344182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834362418827625e-05,
+      "loss": 1.1121,
+      "step": 3142
+    },
+    {
+      "epoch": 0.08632006811128505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834201151228476e-05,
+      "loss": 1.1312,
+      "step": 3143
+    },
+    {
+      "epoch": 0.08634753233912829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834039805817317e-05,
+      "loss": 1.0499,
+      "step": 3144
+    },
+    {
+      "epoch": 0.08637499656697152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833878382595427e-05,
+      "loss": 1.1015,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08640246079481476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983371688156408e-05,
+      "loss": 1.081,
+      "step": 3146
+    },
+    {
+      "epoch": 0.08642992502265799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833555302724557e-05,
+      "loss": 1.0929,
+      "step": 3147
+    },
+    {
+      "epoch": 0.08645738925050123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833393646078134e-05,
+      "loss": 1.1589,
+      "step": 3148
+    },
+    {
+      "epoch": 0.08648485347834446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833231911626094e-05,
+      "loss": 1.1369,
+      "step": 3149
+    },
+    {
+      "epoch": 0.0865123177061877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983307009936971e-05,
+      "loss": 1.0232,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08653978193403092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983290820931027e-05,
+      "loss": 1.111,
+      "step": 3151
+    },
+    {
+      "epoch": 0.08656724616187415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983274624144905e-05,
+      "loss": 1.1809,
+      "step": 3152
+    },
+    {
+      "epoch": 0.08659471038971739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832584195787332e-05,
+      "loss": 1.1145,
+      "step": 3153
+    },
+    {
+      "epoch": 0.08662217461756062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832422072326402e-05,
+      "loss": 1.1465,
+      "step": 3154
+    },
+    {
+      "epoch": 0.08664963884540386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832259871067535e-05,
+      "loss": 1.0928,
+      "step": 3155
+    },
+    {
+      "epoch": 0.0866771030732471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832097592012023e-05,
+      "loss": 1.1333,
+      "step": 3156
+    },
+    {
+      "epoch": 0.08670456730109033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831935235161145e-05,
+      "loss": 1.1211,
+      "step": 3157
+    },
+    {
+      "epoch": 0.08673203152893356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983177280051619e-05,
+      "loss": 1.0683,
+      "step": 3158
+    },
+    {
+      "epoch": 0.0867594957567768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831610288078435e-05,
+      "loss": 1.0387,
+      "step": 3159
+    },
+    {
+      "epoch": 0.08678695998462004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831447697849174e-05,
+      "loss": 1.0443,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08681442421246327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983128502982969e-05,
+      "loss": 1.0669,
+      "step": 3161
+    },
+    {
+      "epoch": 0.0868418884403065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831122284021268e-05,
+      "loss": 1.1565,
+      "step": 3162
+    },
+    {
+      "epoch": 0.08686935266814974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830959460425204e-05,
+      "loss": 1.0659,
+      "step": 3163
+    },
+    {
+      "epoch": 0.08689681689599298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830796559042778e-05,
+      "loss": 1.0622,
+      "step": 3164
+    },
+    {
+      "epoch": 0.0869242811238362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983063357987528e-05,
+      "loss": 0.9948,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08695174535167943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830470522924e-05,
+      "loss": 1.1194,
+      "step": 3166
+    },
+    {
+      "epoch": 0.08697920957952267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830307388190234e-05,
+      "loss": 1.0877,
+      "step": 3167
+    },
+    {
+      "epoch": 0.0870066738073659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830144175675262e-05,
+      "loss": 1.0881,
+      "step": 3168
+    },
+    {
+      "epoch": 0.08703413803520914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829980885380385e-05,
+      "loss": 1.0192,
+      "step": 3169
+    },
+    {
+      "epoch": 0.08706160226305237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982981751730689e-05,
+      "loss": 1.0663,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08708906649089561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829654071456074e-05,
+      "loss": 1.0256,
+      "step": 3171
+    },
+    {
+      "epoch": 0.08711653071873884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829490547829224e-05,
+      "loss": 1.1127,
+      "step": 3172
+    },
+    {
+      "epoch": 0.08714399494658208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982932694642764e-05,
+      "loss": 0.9688,
+      "step": 3173
+    },
+    {
+      "epoch": 0.08717145917442531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829163267252612e-05,
+      "loss": 1.0392,
+      "step": 3174
+    },
+    {
+      "epoch": 0.08719892340226855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982899951030544e-05,
+      "loss": 1.1716,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08722638763011178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828835675587412e-05,
+      "loss": 1.1307,
+      "step": 3176
+    },
+    {
+      "epoch": 0.08725385185795502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982867176309983e-05,
+      "loss": 1.0859,
+      "step": 3177
+    },
+    {
+      "epoch": 0.08728131608579824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828507772843992e-05,
+      "loss": 1.0571,
+      "step": 3178
+    },
+    {
+      "epoch": 0.08730878031364148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982834370482119e-05,
+      "loss": 1.1443,
+      "step": 3179
+    },
+    {
+      "epoch": 0.08733624454148471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828179559032726e-05,
+      "loss": 1.1377,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08736370876932795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828015335479903e-05,
+      "loss": 1.1432,
+      "step": 3181
+    },
+    {
+      "epoch": 0.08739117299717118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982785103416401e-05,
+      "loss": 1.101,
+      "step": 3182
+    },
+    {
+      "epoch": 0.08741863722501442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827686655086355e-05,
+      "loss": 1.0756,
+      "step": 3183
+    },
+    {
+      "epoch": 0.08744610145285765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827522198248234e-05,
+      "loss": 1.1096,
+      "step": 3184
+    },
+    {
+      "epoch": 0.08747356568070089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827357663650956e-05,
+      "loss": 1.0496,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08750102990854412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827193051295814e-05,
+      "loss": 1.0539,
+      "step": 3186
+    },
+    {
+      "epoch": 0.08752849413638736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827028361184114e-05,
+      "loss": 1.0084,
+      "step": 3187
+    },
+    {
+      "epoch": 0.0875559583642306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826863593317157e-05,
+      "loss": 0.9778,
+      "step": 3188
+    },
+    {
+      "epoch": 0.08758342259207383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982669874769625e-05,
+      "loss": 1.0827,
+      "step": 3189
+    },
+    {
+      "epoch": 0.08761088681991706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826533824322697e-05,
+      "loss": 1.1164,
+      "step": 3190
+    },
+    {
+      "epoch": 0.0876383510477603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826368823197802e-05,
+      "loss": 1.0861,
+      "step": 3191
+    },
+    {
+      "epoch": 0.08766581527560352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982620374432287e-05,
+      "loss": 1.0892,
+      "step": 3192
+    },
+    {
+      "epoch": 0.08769327950344676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826038587699208e-05,
+      "loss": 1.1062,
+      "step": 3193
+    },
+    {
+      "epoch": 0.08772074373128999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982587335332812e-05,
+      "loss": 1.02,
+      "step": 3194
+    },
+    {
+      "epoch": 0.08774820795913323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982570804121092e-05,
+      "loss": 1.0449,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08777567218697646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982554265134891e-05,
+      "loss": 1.1821,
+      "step": 3196
+    },
+    {
+      "epoch": 0.0878031364148197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98253771837434e-05,
+      "loss": 1.1758,
+      "step": 3197
+    },
+    {
+      "epoch": 0.08783060064266293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98252116383957e-05,
+      "loss": 1.0787,
+      "step": 3198
+    },
+    {
+      "epoch": 0.08785806487050617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9825046015307115e-05,
+      "loss": 0.9708,
+      "step": 3199
+    },
+    {
+      "epoch": 0.0878855290983494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982488031447897e-05,
+      "loss": 1.0147,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08791299332619264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824714535912557e-05,
+      "loss": 1.0186,
+      "step": 3201
+    },
+    {
+      "epoch": 0.08794045755403587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824548679609202e-05,
+      "loss": 1.0095,
+      "step": 3202
+    },
+    {
+      "epoch": 0.08796792178187911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982438274557021e-05,
+      "loss": 1.0613,
+      "step": 3203
+    },
+    {
+      "epoch": 0.08799538600972234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98242167337969e-05,
+      "loss": 1.0906,
+      "step": 3204
+    },
+    {
+      "epoch": 0.08802285023756556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982405064429058e-05,
+      "loss": 1.0612,
+      "step": 3205
+    },
+    {
+      "epoch": 0.0880503144654088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982388447705256e-05,
+      "loss": 1.0008,
+      "step": 3206
+    },
+    {
+      "epoch": 0.08807777869325203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9823718232084165e-05,
+      "loss": 1.0094,
+      "step": 3207
+    },
+    {
+      "epoch": 0.08810524292109527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9823551909386705e-05,
+      "loss": 1.0446,
+      "step": 3208
+    },
+    {
+      "epoch": 0.0881327071489385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98233855089615e-05,
+      "loss": 1.1655,
+      "step": 3209
+    },
+    {
+      "epoch": 0.08816017137678174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982321903080986e-05,
+      "loss": 0.9816,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08818763560462498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982305247493311e-05,
+      "loss": 1.0936,
+      "step": 3211
+    },
+    {
+      "epoch": 0.08821509983246821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822885841332563e-05,
+      "loss": 1.0221,
+      "step": 3212
+    },
+    {
+      "epoch": 0.08824256406031145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822719130009536e-05,
+      "loss": 1.0689,
+      "step": 3213
+    },
+    {
+      "epoch": 0.08827002828815468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822552340965348e-05,
+      "loss": 1.0247,
+      "step": 3214
+    },
+    {
+      "epoch": 0.08829749251599792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822385474201327e-05,
+      "loss": 1.1177,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08832495674384115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822218529718783e-05,
+      "loss": 1.1652,
+      "step": 3216
+    },
+    {
+      "epoch": 0.08835242097168439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982205150751904e-05,
+      "loss": 0.9358,
+      "step": 3217
+    },
+    {
+      "epoch": 0.08837988519952762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821884407603425e-05,
+      "loss": 1.1482,
+      "step": 3218
+    },
+    {
+      "epoch": 0.08840734942737084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821717229973254e-05,
+      "loss": 1.1301,
+      "step": 3219
+    },
+    {
+      "epoch": 0.08843481365521408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821549974629852e-05,
+      "loss": 1.0662,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08846227788305731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821382641574542e-05,
+      "loss": 1.1271,
+      "step": 3221
+    },
+    {
+      "epoch": 0.08848974211090055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982121523080865e-05,
+      "loss": 1.0931,
+      "step": 3222
+    },
+    {
+      "epoch": 0.08851720633874378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821047742333492e-05,
+      "loss": 1.1083,
+      "step": 3223
+    },
+    {
+      "epoch": 0.08854467056658702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982088017615041e-05,
+      "loss": 1.1457,
+      "step": 3224
+    },
+    {
+      "epoch": 0.08857213479443025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9820712532260712e-05,
+      "loss": 1.0065,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08859959902227349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9820544810665735e-05,
+      "loss": 1.1475,
+      "step": 3226
+    },
+    {
+      "epoch": 0.08862706325011672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98203770113668e-05,
+      "loss": 1.111,
+      "step": 3227
+    },
+    {
+      "epoch": 0.08865452747795996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982020913436524e-05,
+      "loss": 1.1215,
+      "step": 3228
+    },
+    {
+      "epoch": 0.0886819917058032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9820041179662385e-05,
+      "loss": 1.0313,
+      "step": 3229
+    },
+    {
+      "epoch": 0.08870945593364643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9819873147259557e-05,
+      "loss": 1.1664,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08873692016148967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981970503715809e-05,
+      "loss": 1.1617,
+      "step": 3231
+    },
+    {
+      "epoch": 0.08876438438933289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981953684935931e-05,
+      "loss": 0.9781,
+      "step": 3232
+    },
+    {
+      "epoch": 0.08879184861717612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9819368583864555e-05,
+      "loss": 1.0762,
+      "step": 3233
+    },
+    {
+      "epoch": 0.08881931284501936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981920024067515e-05,
+      "loss": 1.1039,
+      "step": 3234
+    },
+    {
+      "epoch": 0.08884677707286259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9819031819792428e-05,
+      "loss": 0.9794,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08887424130070583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818863321217726e-05,
+      "loss": 1.074,
+      "step": 3236
+    },
+    {
+      "epoch": 0.08890170552854906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818694744952373e-05,
+      "loss": 1.2264,
+      "step": 3237
+    },
+    {
+      "epoch": 0.0889291697563923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818526090997703e-05,
+      "loss": 1.0728,
+      "step": 3238
+    },
+    {
+      "epoch": 0.08895663398423553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818357359355048e-05,
+      "loss": 1.0795,
+      "step": 3239
+    },
+    {
+      "epoch": 0.08898409821207877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818188550025754e-05,
+      "loss": 1.0157,
+      "step": 3240
+    },
+    {
+      "epoch": 0.089011562439922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818019663011144e-05,
+      "loss": 1.1211,
+      "step": 3241
+    },
+    {
+      "epoch": 0.08903902666776524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817850698312558e-05,
+      "loss": 1.072,
+      "step": 3242
+    },
+    {
+      "epoch": 0.08906649089560847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817681655931333e-05,
+      "loss": 1.1442,
+      "step": 3243
+    },
+    {
+      "epoch": 0.08909395512345171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981751253586881e-05,
+      "loss": 1.0787,
+      "step": 3244
+    },
+    {
+      "epoch": 0.08912141935129494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981734333812633e-05,
+      "loss": 1.1537,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08914888357913817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981717406270522e-05,
+      "loss": 1.108,
+      "step": 3246
+    },
+    {
+      "epoch": 0.0891763478069814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981700470960683e-05,
+      "loss": 1.1076,
+      "step": 3247
+    },
+    {
+      "epoch": 0.08920381203482464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981683527883249e-05,
+      "loss": 1.0981,
+      "step": 3248
+    },
+    {
+      "epoch": 0.08923127626266787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981666577038355e-05,
+      "loss": 0.9941,
+      "step": 3249
+    },
+    {
+      "epoch": 0.08925874049051111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981649618426135e-05,
+      "loss": 1.0168,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08928620471835434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9816326520467228e-05,
+      "loss": 1.1688,
+      "step": 3251
+    },
+    {
+      "epoch": 0.08931366894619758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9816156779002532e-05,
+      "loss": 0.9781,
+      "step": 3252
+    },
+    {
+      "epoch": 0.08934113317404081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815986959868594e-05,
+      "loss": 1.0062,
+      "step": 3253
+    },
+    {
+      "epoch": 0.08936859740188405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981581706306677e-05,
+      "loss": 1.0706,
+      "step": 3254
+    },
+    {
+      "epoch": 0.08939606162972728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815647088598395e-05,
+      "loss": 1.0695,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08942352585757052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981547703646482e-05,
+      "loss": 1.0452,
+      "step": 3256
+    },
+    {
+      "epoch": 0.08945099008541375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981530690666739e-05,
+      "loss": 1.0946,
+      "step": 3257
+    },
+    {
+      "epoch": 0.08947845431325699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815136699207447e-05,
+      "loss": 1.0399,
+      "step": 3258
+    },
+    {
+      "epoch": 0.08950591854110021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814966414086344e-05,
+      "loss": 0.9695,
+      "step": 3259
+    },
+    {
+      "epoch": 0.08953338276894345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814796051305424e-05,
+      "loss": 1.0505,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08956084699678668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814625610866033e-05,
+      "loss": 1.0876,
+      "step": 3261
+    },
+    {
+      "epoch": 0.08958831122462992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814455092769523e-05,
+      "loss": 1.0865,
+      "step": 3262
+    },
+    {
+      "epoch": 0.08961577545247315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814284497017243e-05,
+      "loss": 1.0394,
+      "step": 3263
+    },
+    {
+      "epoch": 0.08964323968031639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814113823610544e-05,
+      "loss": 1.0671,
+      "step": 3264
+    },
+    {
+      "epoch": 0.08967070390815962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813943072550775e-05,
+      "loss": 1.0453,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08969816813600286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981377224383928e-05,
+      "loss": 1.0834,
+      "step": 3266
+    },
+    {
+      "epoch": 0.08972563236384609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813601337477427e-05,
+      "loss": 1.0667,
+      "step": 3267
+    },
+    {
+      "epoch": 0.08975309659168933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813430353466555e-05,
+      "loss": 1.153,
+      "step": 3268
+    },
+    {
+      "epoch": 0.08978056081953256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813259291808017e-05,
+      "loss": 1.1395,
+      "step": 3269
+    },
+    {
+      "epoch": 0.0898080250473758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813088152503175e-05,
+      "loss": 1.0964,
+      "step": 3270
+    },
+    {
+      "epoch": 0.08983548927521903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812916935553377e-05,
+      "loss": 1.0375,
+      "step": 3271
+    },
+    {
+      "epoch": 0.08986295350306227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981274564095998e-05,
+      "loss": 0.9809,
+      "step": 3272
+    },
+    {
+      "epoch": 0.08989041773090549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981257426872434e-05,
+      "loss": 1.0844,
+      "step": 3273
+    },
+    {
+      "epoch": 0.08991788195874872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812402818847806e-05,
+      "loss": 1.0211,
+      "step": 3274
+    },
+    {
+      "epoch": 0.08994534618659196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812231291331742e-05,
+      "loss": 1.1312,
+      "step": 3275
+    },
+    {
+      "epoch": 0.0899728104144352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812059686177502e-05,
+      "loss": 1.023,
+      "step": 3276
+    },
+    {
+      "epoch": 0.09000027464227843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811888003386448e-05,
+      "loss": 1.0824,
+      "step": 3277
+    },
+    {
+      "epoch": 0.09002773887012167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811716242959935e-05,
+      "loss": 1.0285,
+      "step": 3278
+    },
+    {
+      "epoch": 0.0900552030979649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981154440489932e-05,
+      "loss": 0.9988,
+      "step": 3279
+    },
+    {
+      "epoch": 0.09008266732580814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811372489205967e-05,
+      "loss": 1.1681,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09011013155365137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811200495881234e-05,
+      "loss": 1.0523,
+      "step": 3281
+    },
+    {
+      "epoch": 0.0901375957814946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811028424926482e-05,
+      "loss": 1.1281,
+      "step": 3282
+    },
+    {
+      "epoch": 0.09016506000933784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810856276343073e-05,
+      "loss": 1.0986,
+      "step": 3283
+    },
+    {
+      "epoch": 0.09019252423718108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981068405013237e-05,
+      "loss": 1.0645,
+      "step": 3284
+    },
+    {
+      "epoch": 0.09021998846502431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810511746295732e-05,
+      "loss": 1.1156,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09024745269286755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810339364834525e-05,
+      "loss": 1.0271,
+      "step": 3286
+    },
+    {
+      "epoch": 0.09027491692071077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810166905750114e-05,
+      "loss": 1.0817,
+      "step": 3287
+    },
+    {
+      "epoch": 0.090302381148554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980999436904386e-05,
+      "loss": 1.0042,
+      "step": 3288
+    },
+    {
+      "epoch": 0.09032984537639724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809821754717132e-05,
+      "loss": 1.0173,
+      "step": 3289
+    },
+    {
+      "epoch": 0.09035730960424047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809649062771296e-05,
+      "loss": 1.0134,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09038477383208371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809476293207717e-05,
+      "loss": 1.1054,
+      "step": 3291
+    },
+    {
+      "epoch": 0.09041223805992694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980930344602776e-05,
+      "loss": 1.1785,
+      "step": 3292
+    },
+    {
+      "epoch": 0.09043970228777018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809130521232795e-05,
+      "loss": 1.1816,
+      "step": 3293
+    },
+    {
+      "epoch": 0.09046716651561341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808957518824186e-05,
+      "loss": 1.1009,
+      "step": 3294
+    },
+    {
+      "epoch": 0.09049463074345665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980878443880331e-05,
+      "loss": 1.1542,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09052209497129989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808611281171532e-05,
+      "loss": 1.1254,
+      "step": 3296
+    },
+    {
+      "epoch": 0.09054955919914312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808438045930218e-05,
+      "loss": 0.9977,
+      "step": 3297
+    },
+    {
+      "epoch": 0.09057702342698636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808264733080747e-05,
+      "loss": 0.897,
+      "step": 3298
+    },
+    {
+      "epoch": 0.09060448765482959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980809134262448e-05,
+      "loss": 1.0322,
+      "step": 3299
+    },
+    {
+      "epoch": 0.09063195188267281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807917874562803e-05,
+      "loss": 1.1071,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09065941611051605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807744328897076e-05,
+      "loss": 0.9636,
+      "step": 3301
+    },
+    {
+      "epoch": 0.09068688033835928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980757070562868e-05,
+      "loss": 1.1699,
+      "step": 3302
+    },
+    {
+      "epoch": 0.09071434456620252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980739700475898e-05,
+      "loss": 1.0562,
+      "step": 3303
+    },
+    {
+      "epoch": 0.09074180879404575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807223226289363e-05,
+      "loss": 1.0686,
+      "step": 3304
+    },
+    {
+      "epoch": 0.09076927302188899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980704937022119e-05,
+      "loss": 1.0981,
+      "step": 3305
+    },
+    {
+      "epoch": 0.09079673724973222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806875436555848e-05,
+      "loss": 1.0477,
+      "step": 3306
+    },
+    {
+      "epoch": 0.09082420147757546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806701425294705e-05,
+      "loss": 1.0579,
+      "step": 3307
+    },
+    {
+      "epoch": 0.0908516657054187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806527336439147e-05,
+      "loss": 1.0942,
+      "step": 3308
+    },
+    {
+      "epoch": 0.09087912993326193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806353169990543e-05,
+      "loss": 1.1126,
+      "step": 3309
+    },
+    {
+      "epoch": 0.09090659416110516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806178925950272e-05,
+      "loss": 1.1514,
+      "step": 3310
+    },
+    {
+      "epoch": 0.0909340583889484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806004604319716e-05,
+      "loss": 1.0757,
+      "step": 3311
+    },
+    {
+      "epoch": 0.09096152261679163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805830205100254e-05,
+      "loss": 1.0242,
+      "step": 3312
+    },
+    {
+      "epoch": 0.09098898684463487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805655728293266e-05,
+      "loss": 1.1402,
+      "step": 3313
+    },
+    {
+      "epoch": 0.09101645107247809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980548117390013e-05,
+      "loss": 1.0378,
+      "step": 3314
+    },
+    {
+      "epoch": 0.09104391530032133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805306541922228e-05,
+      "loss": 0.9981,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09107137952816456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805131832360943e-05,
+      "loss": 1.0685,
+      "step": 3316
+    },
+    {
+      "epoch": 0.0910988437560078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804957045217658e-05,
+      "loss": 1.1909,
+      "step": 3317
+    },
+    {
+      "epoch": 0.09112630798385103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804782180493754e-05,
+      "loss": 1.0108,
+      "step": 3318
+    },
+    {
+      "epoch": 0.09115377221169427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804607238190617e-05,
+      "loss": 1.0674,
+      "step": 3319
+    },
+    {
+      "epoch": 0.0911812364395375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804432218309627e-05,
+      "loss": 1.0914,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09120870066738074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980425712085217e-05,
+      "loss": 1.1237,
+      "step": 3321
+    },
+    {
+      "epoch": 0.09123616489522397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980408194581964e-05,
+      "loss": 1.0782,
+      "step": 3322
+    },
+    {
+      "epoch": 0.09126362912306721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980390669321341e-05,
+      "loss": 0.9896,
+      "step": 3323
+    },
+    {
+      "epoch": 0.09129109335091044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980373136303487e-05,
+      "loss": 1.1343,
+      "step": 3324
+    },
+    {
+      "epoch": 0.09131855757875368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9803555955285416e-05,
+      "loss": 1.0918,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09134602180659691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980338046996643e-05,
+      "loss": 0.9997,
+      "step": 3326
+    },
+    {
+      "epoch": 0.09137348603444014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9803204907079294e-05,
+      "loss": 1.0628,
+      "step": 3327
+    },
+    {
+      "epoch": 0.09140095026228337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9803029266625405e-05,
+      "loss": 1.0557,
+      "step": 3328
+    },
+    {
+      "epoch": 0.0914284144901266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802853548606155e-05,
+      "loss": 1.1197,
+      "step": 3329
+    },
+    {
+      "epoch": 0.09145587871796984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980267775302293e-05,
+      "loss": 1.0793,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09148334294581308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980250187987712e-05,
+      "loss": 1.1887,
+      "step": 3331
+    },
+    {
+      "epoch": 0.09151080717365631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802325929170115e-05,
+      "loss": 1.0457,
+      "step": 3332
+    },
+    {
+      "epoch": 0.09153827140149955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802149900903313e-05,
+      "loss": 1.0202,
+      "step": 3333
+    },
+    {
+      "epoch": 0.09156573562934278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801973795078105e-05,
+      "loss": 1.1938,
+      "step": 3334
+    },
+    {
+      "epoch": 0.09159319985718602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801797611695884e-05,
+      "loss": 1.1318,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09162066408502925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980162135075804e-05,
+      "loss": 1.0275,
+      "step": 3336
+    },
+    {
+      "epoch": 0.09164812831287249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801445012265975e-05,
+      "loss": 1.2257,
+      "step": 3337
+    },
+    {
+      "epoch": 0.09167559254071572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801268596221076e-05,
+      "loss": 1.0591,
+      "step": 3338
+    },
+    {
+      "epoch": 0.09170305676855896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801092102624747e-05,
+      "loss": 1.121,
+      "step": 3339
+    },
+    {
+      "epoch": 0.09173052099640219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800915531478378e-05,
+      "loss": 1.1469,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09175798522424541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980073888278337e-05,
+      "loss": 1.147,
+      "step": 3341
+    },
+    {
+      "epoch": 0.09178544945208865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800562156541122e-05,
+      "loss": 1.0312,
+      "step": 3342
+    },
+    {
+      "epoch": 0.09181291367993188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800385352753027e-05,
+      "loss": 1.0145,
+      "step": 3343
+    },
+    {
+      "epoch": 0.09184037790777512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980020847142049e-05,
+      "loss": 1.0332,
+      "step": 3344
+    },
+    {
+      "epoch": 0.09186784213561835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800031512544903e-05,
+      "loss": 1.1068,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09189530636346159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799854476127673e-05,
+      "loss": 1.0142,
+      "step": 3346
+    },
+    {
+      "epoch": 0.09192277059130483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799677362170194e-05,
+      "loss": 1.0659,
+      "step": 3347
+    },
+    {
+      "epoch": 0.09195023481914806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799500170673875e-05,
+      "loss": 1.0113,
+      "step": 3348
+    },
+    {
+      "epoch": 0.0919776990469913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799322901640113e-05,
+      "loss": 1.0227,
+      "step": 3349
+    },
+    {
+      "epoch": 0.09200516327483453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799145555070317e-05,
+      "loss": 1.0764,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09203262750267777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979896813096588e-05,
+      "loss": 1.0417,
+      "step": 3351
+    },
+    {
+      "epoch": 0.092060091730521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798790629328217e-05,
+      "loss": 1.0768,
+      "step": 3352
+    },
+    {
+      "epoch": 0.09208755595836424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798613050158722e-05,
+      "loss": 1.0233,
+      "step": 3353
+    },
+    {
+      "epoch": 0.09211502018620746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798435393458807e-05,
+      "loss": 1.0369,
+      "step": 3354
+    },
+    {
+      "epoch": 0.0921424844140507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798257659229875e-05,
+      "loss": 1.1478,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09216994864189393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798079847473335e-05,
+      "loss": 1.0353,
+      "step": 3356
+    },
+    {
+      "epoch": 0.09219741286973716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797901958190587e-05,
+      "loss": 1.1338,
+      "step": 3357
+    },
+    {
+      "epoch": 0.0922248770975804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979772399138305e-05,
+      "loss": 0.952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.09225234132542363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979754594705212e-05,
+      "loss": 1.0323,
+      "step": 3359
+    },
+    {
+      "epoch": 0.09227980555326687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797367825199214e-05,
+      "loss": 1.1021,
+      "step": 3360
+    },
+    {
+      "epoch": 0.0923072697811101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979718962582574e-05,
+      "loss": 1.1758,
+      "step": 3361
+    },
+    {
+      "epoch": 0.09233473400895334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797011348933105e-05,
+      "loss": 1.0937,
+      "step": 3362
+    },
+    {
+      "epoch": 0.09236219823679657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979683299452272e-05,
+      "loss": 0.9698,
+      "step": 3363
+    },
+    {
+      "epoch": 0.09238966246463981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796654562595998e-05,
+      "loss": 1.0852,
+      "step": 3364
+    },
+    {
+      "epoch": 0.09241712669248305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796476053154347e-05,
+      "loss": 1.0385,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09244459092032628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979629746619919e-05,
+      "loss": 0.9878,
+      "step": 3366
+    },
+    {
+      "epoch": 0.09247205514816952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979611880173193e-05,
+      "loss": 1.0606,
+      "step": 3367
+    },
+    {
+      "epoch": 0.09249951937601274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795940059753977e-05,
+      "loss": 0.9639,
+      "step": 3368
+    },
+    {
+      "epoch": 0.09252698360385597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979576124026676e-05,
+      "loss": 1.1482,
+      "step": 3369
+    },
+    {
+      "epoch": 0.09255444783169921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979558234327168e-05,
+      "loss": 1.0771,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09258191205954244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795403368770164e-05,
+      "loss": 1.1435,
+      "step": 3371
+    },
+    {
+      "epoch": 0.09260937628738568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795224316763616e-05,
+      "loss": 1.1344,
+      "step": 3372
+    },
+    {
+      "epoch": 0.09263684051522891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795045187253464e-05,
+      "loss": 1.0538,
+      "step": 3373
+    },
+    {
+      "epoch": 0.09266430474307215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794865980241116e-05,
+      "loss": 1.0134,
+      "step": 3374
+    },
+    {
+      "epoch": 0.09269176897091538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794686695727994e-05,
+      "loss": 1.1521,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09271923319875862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979450733371552e-05,
+      "loss": 1.0775,
+      "step": 3376
+    },
+    {
+      "epoch": 0.09274669742660185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794327894205108e-05,
+      "loss": 0.9966,
+      "step": 3377
+    },
+    {
+      "epoch": 0.09277416165444509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979414837719818e-05,
+      "loss": 0.8972,
+      "step": 3378
+    },
+    {
+      "epoch": 0.09280162588228832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979396878269616e-05,
+      "loss": 1.1727,
+      "step": 3379
+    },
+    {
+      "epoch": 0.09282909011013156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793789110700457e-05,
+      "loss": 1.1138,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09285655433797478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793609361212505e-05,
+      "loss": 1.1433,
+      "step": 3381
+    },
+    {
+      "epoch": 0.09288401856581802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793429534233724e-05,
+      "loss": 1.0528,
+      "step": 3382
+    },
+    {
+      "epoch": 0.09291148279366125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793249629765533e-05,
+      "loss": 1.0294,
+      "step": 3383
+    },
+    {
+      "epoch": 0.09293894702150449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979306964780936e-05,
+      "loss": 1.1976,
+      "step": 3384
+    },
+    {
+      "epoch": 0.09296641124934772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792889588366624e-05,
+      "loss": 1.1393,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09299387547719096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792709451438755e-05,
+      "loss": 1.0616,
+      "step": 3386
+    },
+    {
+      "epoch": 0.09302133970503419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979252923702717e-05,
+      "loss": 1.1343,
+      "step": 3387
+    },
+    {
+      "epoch": 0.09304880393287743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792348945133303e-05,
+      "loss": 1.0099,
+      "step": 3388
+    },
+    {
+      "epoch": 0.09307626816072066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979216857575858e-05,
+      "loss": 1.0237,
+      "step": 3389
+    },
+    {
+      "epoch": 0.0931037323885639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791988128904425e-05,
+      "loss": 1.042,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09313119661640713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791807604572265e-05,
+      "loss": 1.1206,
+      "step": 3391
+    },
+    {
+      "epoch": 0.09315866084425037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791627002763532e-05,
+      "loss": 1.0603,
+      "step": 3392
+    },
+    {
+      "epoch": 0.0931861250720936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791446323479654e-05,
+      "loss": 1.1697,
+      "step": 3393
+    },
+    {
+      "epoch": 0.09321358929993684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791265566722056e-05,
+      "loss": 1.1342,
+      "step": 3394
+    },
+    {
+      "epoch": 0.09324105352778006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791084732492175e-05,
+      "loss": 0.971,
+      "step": 3395
+    },
+    {
+      "epoch": 0.0932685177556233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979090382079144e-05,
+      "loss": 1.0862,
+      "step": 3396
+    },
+    {
+      "epoch": 0.09329598198346653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979072283162128e-05,
+      "loss": 0.9369,
+      "step": 3397
+    },
+    {
+      "epoch": 0.09332344621130977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9790541764983127e-05,
+      "loss": 0.9895,
+      "step": 3398
+    },
+    {
+      "epoch": 0.093350910439153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9790360620878417e-05,
+      "loss": 1.1346,
+      "step": 3399
+    },
+    {
+      "epoch": 0.09337837466699624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979017939930858e-05,
+      "loss": 1.0616,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09340583889483947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789998100275053e-05,
+      "loss": 1.1,
+      "step": 3401
+    },
+    {
+      "epoch": 0.0934333031226827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978981672377927e-05,
+      "loss": 1.1458,
+      "step": 3402
+    },
+    {
+      "epoch": 0.09346076735052594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978963526982266e-05,
+      "loss": 0.9918,
+      "step": 3403
+    },
+    {
+      "epoch": 0.09348823157836918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789453738406666e-05,
+      "loss": 0.9984,
+      "step": 3404
+    },
+    {
+      "epoch": 0.09351569580621241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978927212953272e-05,
+      "loss": 1.0521,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09354316003405565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789090443202267e-05,
+      "loss": 1.1069,
+      "step": 3406
+    },
+    {
+      "epoch": 0.09357062426189888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788908679416733e-05,
+      "loss": 1.0876,
+      "step": 3407
+    },
+    {
+      "epoch": 0.09359808848974212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788726838177564e-05,
+      "loss": 1.1299,
+      "step": 3408
+    },
+    {
+      "epoch": 0.09362555271758534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788544919486193e-05,
+      "loss": 1.0005,
+      "step": 3409
+    },
+    {
+      "epoch": 0.09365301694542857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788362923344068e-05,
+      "loss": 1.1495,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09368048117327181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978818084975262e-05,
+      "loss": 1.0161,
+      "step": 3411
+    },
+    {
+      "epoch": 0.09370794540111504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787998698713296e-05,
+      "loss": 1.0385,
+      "step": 3412
+    },
+    {
+      "epoch": 0.09373540962895828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787816470227533e-05,
+      "loss": 1.0598,
+      "step": 3413
+    },
+    {
+      "epoch": 0.09376287385680152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787634164296775e-05,
+      "loss": 1.0557,
+      "step": 3414
+    },
+    {
+      "epoch": 0.09379033808464475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787451780922465e-05,
+      "loss": 1.1175,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09381780231248799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787269320106048e-05,
+      "loss": 1.0326,
+      "step": 3416
+    },
+    {
+      "epoch": 0.09384526654033122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978708678184896e-05,
+      "loss": 1.0943,
+      "step": 3417
+    },
+    {
+      "epoch": 0.09387273076817446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978690416615265e-05,
+      "loss": 1.0814,
+      "step": 3418
+    },
+    {
+      "epoch": 0.09390019499601769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9786721473018565e-05,
+      "loss": 1.1119,
+      "step": 3419
+    },
+    {
+      "epoch": 0.09392765922386093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9786538702448152e-05,
+      "loss": 1.1682,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09395512345170416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978635585444285e-05,
+      "loss": 1.1348,
+      "step": 3421
+    },
+    {
+      "epoch": 0.09398258767954738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978617292900411e-05,
+      "loss": 1.1401,
+      "step": 3422
+    },
+    {
+      "epoch": 0.09401005190739062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785989926133376e-05,
+      "loss": 1.0832,
+      "step": 3423
+    },
+    {
+      "epoch": 0.09403751613523385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785806845832105e-05,
+      "loss": 1.1465,
+      "step": 3424
+    },
+    {
+      "epoch": 0.09406498036307709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785623688101735e-05,
+      "loss": 1.127,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09409244459092032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785440452943724e-05,
+      "loss": 1.1149,
+      "step": 3426
+    },
+    {
+      "epoch": 0.09411990881876356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785257140359513e-05,
+      "loss": 0.9699,
+      "step": 3427
+    },
+    {
+      "epoch": 0.0941473730466068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978507375035056e-05,
+      "loss": 1.05,
+      "step": 3428
+    },
+    {
+      "epoch": 0.09417483727445003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784890282918315e-05,
+      "loss": 1.062,
+      "step": 3429
+    },
+    {
+      "epoch": 0.09420230150229326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784706738064222e-05,
+      "loss": 1.1112,
+      "step": 3430
+    },
+    {
+      "epoch": 0.0942297657301365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784523115789742e-05,
+      "loss": 1.0717,
+      "step": 3431
+    },
+    {
+      "epoch": 0.09425722995797974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784339416096326e-05,
+      "loss": 1.1103,
+      "step": 3432
+    },
+    {
+      "epoch": 0.09428469418582297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784155638985426e-05,
+      "loss": 1.07,
+      "step": 3433
+    },
+    {
+      "epoch": 0.0943121584136662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783971784458496e-05,
+      "loss": 1.1101,
+      "step": 3434
+    },
+    {
+      "epoch": 0.09433962264150944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978378785251699e-05,
+      "loss": 1.0393,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09436708686935266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783603843162365e-05,
+      "loss": 1.0723,
+      "step": 3436
+    },
+    {
+      "epoch": 0.0943945510971959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978341975639608e-05,
+      "loss": 1.1,
+      "step": 3437
+    },
+    {
+      "epoch": 0.09442201532503913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978323559221959e-05,
+      "loss": 1.0771,
+      "step": 3438
+    },
+    {
+      "epoch": 0.09444947955288237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978305135063434e-05,
+      "loss": 1.0306,
+      "step": 3439
+    },
+    {
+      "epoch": 0.0944769437807256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782867031641806e-05,
+      "loss": 1.0379,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09450440800856884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782682635243438e-05,
+      "loss": 1.03,
+      "step": 3441
+    },
+    {
+      "epoch": 0.09453187223641207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782498161440692e-05,
+      "loss": 1.1011,
+      "step": 3442
+    },
+    {
+      "epoch": 0.09455933646425531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782313610235035e-05,
+      "loss": 1.0284,
+      "step": 3443
+    },
+    {
+      "epoch": 0.09458680069209854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978212898162792e-05,
+      "loss": 1.1037,
+      "step": 3444
+    },
+    {
+      "epoch": 0.09461426491994178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781944275620816e-05,
+      "loss": 1.0781,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09464172914778501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781759492215174e-05,
+      "loss": 1.0054,
+      "step": 3446
+    },
+    {
+      "epoch": 0.09466919337562825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781574631412467e-05,
+      "loss": 1.015,
+      "step": 3447
+    },
+    {
+      "epoch": 0.09469665760347148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978138969321415e-05,
+      "loss": 1.1207,
+      "step": 3448
+    },
+    {
+      "epoch": 0.0947241218313147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978120467762169e-05,
+      "loss": 1.0786,
+      "step": 3449
+    },
+    {
+      "epoch": 0.09475158605915794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781019584636548e-05,
+      "loss": 1.0468,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09477905028700118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978083441426019e-05,
+      "loss": 1.1702,
+      "step": 3451
+    },
+    {
+      "epoch": 0.09480651451484441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9780649166494083e-05,
+      "loss": 1.1559,
+      "step": 3452
+    },
+    {
+      "epoch": 0.09483397874268765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978046384133969e-05,
+      "loss": 1.0565,
+      "step": 3453
+    },
+    {
+      "epoch": 0.09486144297053088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978027843879848e-05,
+      "loss": 1.0412,
+      "step": 3454
+    },
+    {
+      "epoch": 0.09488890719837412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9780092958871917e-05,
+      "loss": 1.1513,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09491637142621735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977990740156147e-05,
+      "loss": 0.979,
+      "step": 3456
+    },
+    {
+      "epoch": 0.09494383565406059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9779721766868608e-05,
+      "loss": 1.1011,
+      "step": 3457
+    },
+    {
+      "epoch": 0.09497129988190382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97795360547948e-05,
+      "loss": 1.0834,
+      "step": 3458
+    },
+    {
+      "epoch": 0.09499876410974706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9779350265341513e-05,
+      "loss": 1.1188,
+      "step": 3459
+    },
+    {
+      "epoch": 0.0950262283375903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9779164398510218e-05,
+      "loss": 1.0754,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09505369256543353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977897845430239e-05,
+      "loss": 1.0701,
+      "step": 3461
+    },
+    {
+      "epoch": 0.09508115679327676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9778792432719493e-05,
+      "loss": 1.0628,
+      "step": 3462
+    },
+    {
+      "epoch": 0.09510862102111999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9778606333763003e-05,
+      "loss": 1.0832,
+      "step": 3463
+    },
+    {
+      "epoch": 0.09513608524896322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977842015743439e-05,
+      "loss": 1.0607,
+      "step": 3464
+    },
+    {
+      "epoch": 0.09516354947680646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9778233903735132e-05,
+      "loss": 1.0467,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09519101370464969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9778047572666695e-05,
+      "loss": 1.0815,
+      "step": 3466
+    },
+    {
+      "epoch": 0.09521847793249293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777861164230563e-05,
+      "loss": 1.0734,
+      "step": 3467
+    },
+    {
+      "epoch": 0.09524594216033616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777674678428204e-05,
+      "loss": 1.0893,
+      "step": 3468
+    },
+    {
+      "epoch": 0.0952734063881794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777488115261098e-05,
+      "loss": 1.0823,
+      "step": 3469
+    },
+    {
+      "epoch": 0.09530087061602263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777301474730715e-05,
+      "loss": 0.9962,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09532833484386587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777114756838537e-05,
+      "loss": 1.1869,
+      "step": 3471
+    },
+    {
+      "epoch": 0.0953557990717091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776927961586036e-05,
+      "loss": 1.0937,
+      "step": 3472
+    },
+    {
+      "epoch": 0.09538326329955234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776741088974698e-05,
+      "loss": 0.9869,
+      "step": 3473
+    },
+    {
+      "epoch": 0.09541072752739557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776554139005998e-05,
+      "loss": 1.1164,
+      "step": 3474
+    },
+    {
+      "epoch": 0.09543819175523881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977636711168141e-05,
+      "loss": 1.143,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09546565598308203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776180007002422e-05,
+      "loss": 1.0762,
+      "step": 3476
+    },
+    {
+      "epoch": 0.09549312021092526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977599282497051e-05,
+      "loss": 1.0924,
+      "step": 3477
+    },
+    {
+      "epoch": 0.0955205844387685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9775805565587157e-05,
+      "loss": 1.0441,
+      "step": 3478
+    },
+    {
+      "epoch": 0.09554804866661173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9775618228853842e-05,
+      "loss": 1.0081,
+      "step": 3479
+    },
+    {
+      "epoch": 0.09557551289445497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977543081477205e-05,
+      "loss": 1.0338,
+      "step": 3480
+    },
+    {
+      "epoch": 0.0956029771222982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977524332334326e-05,
+      "loss": 1.0116,
+      "step": 3481
+    },
+    {
+      "epoch": 0.09563044135014144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9775055754568964e-05,
+      "loss": 1.073,
+      "step": 3482
+    },
+    {
+      "epoch": 0.09565790557798468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774868108450634e-05,
+      "loss": 1.1019,
+      "step": 3483
+    },
+    {
+      "epoch": 0.09568536980582791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774680384989767e-05,
+      "loss": 1.0487,
+      "step": 3484
+    },
+    {
+      "epoch": 0.09571283403367115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977449258418784e-05,
+      "loss": 0.8994,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09574029826151438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774304706046338e-05,
+      "loss": 1.0656,
+      "step": 3486
+    },
+    {
+      "epoch": 0.09576776248935762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774116750566755e-05,
+      "loss": 0.9382,
+      "step": 3487
+    },
+    {
+      "epoch": 0.09579522671720085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9773928717750573e-05,
+      "loss": 0.9743,
+      "step": 3488
+    },
+    {
+      "epoch": 0.09582269094504409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977374060759928e-05,
+      "loss": 1.1026,
+      "step": 3489
+    },
+    {
+      "epoch": 0.09585015517288731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9773552420114364e-05,
+      "loss": 1.1807,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09587761940073054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977336415529732e-05,
+      "loss": 1.0843,
+      "step": 3491
+    },
+    {
+      "epoch": 0.09590508362857378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977317581314963e-05,
+      "loss": 1.0249,
+      "step": 3492
+    },
+    {
+      "epoch": 0.09593254785641701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977298739367279e-05,
+      "loss": 1.1597,
+      "step": 3493
+    },
+    {
+      "epoch": 0.09596001208426025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772798896868283e-05,
+      "loss": 0.9745,
+      "step": 3494
+    },
+    {
+      "epoch": 0.09598747631210348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977261032273761e-05,
+      "loss": 1.0095,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09601494053994672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772421671282255e-05,
+      "loss": 1.0358,
+      "step": 3496
+    },
+    {
+      "epoch": 0.09604240476778995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772232942503716e-05,
+      "loss": 0.9941,
+      "step": 3497
+    },
+    {
+      "epoch": 0.09606986899563319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977204413640349e-05,
+      "loss": 1.0139,
+      "step": 3498
+    },
+    {
+      "epoch": 0.09609733322347642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771855252983056e-05,
+      "loss": 1.0757,
+      "step": 3499
+    },
+    {
+      "epoch": 0.09612479745131966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771666292243924e-05,
+      "loss": 1.1071,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0961522616791629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977147725418758e-05,
+      "loss": 0.9867,
+      "step": 3501
+    },
+    {
+      "epoch": 0.09617972590700613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771288138815526e-05,
+      "loss": 1.0638,
+      "step": 3502
+    },
+    {
+      "epoch": 0.09620719013484935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771098946129252e-05,
+      "loss": 1.1243,
+      "step": 3503
+    },
+    {
+      "epoch": 0.09623465436269259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977090967613026e-05,
+      "loss": 0.931,
+      "step": 3504
+    },
+    {
+      "epoch": 0.09626211859053582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9770720328820042e-05,
+      "loss": 1.1152,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09628958281837906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9770530904200104e-05,
+      "loss": 0.9695,
+      "step": 3506
+    },
+    {
+      "epoch": 0.09631704704622229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9770341402271944e-05,
+      "loss": 1.0092,
+      "step": 3507
+    },
+    {
+      "epoch": 0.09634451127406553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977015182303705e-05,
+      "loss": 1.0258,
+      "step": 3508
+    },
+    {
+      "epoch": 0.09637197550190876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769962166496938e-05,
+      "loss": 1.0326,
+      "step": 3509
+    },
+    {
+      "epoch": 0.096399439729752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769772432653095e-05,
+      "loss": 1.0213,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09642690395759523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976958262150703e-05,
+      "loss": 1.0707,
+      "step": 3511
+    },
+    {
+      "epoch": 0.09645436818543847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769392733060245e-05,
+      "loss": 0.9705,
+      "step": 3512
+    },
+    {
+      "epoch": 0.0964818324132817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769202767314234e-05,
+      "loss": 1.1101,
+      "step": 3513
+    },
+    {
+      "epoch": 0.09650929664112494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769012724270513e-05,
+      "loss": 1.0981,
+      "step": 3514
+    },
+    {
+      "epoch": 0.09653676086896817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9768822603930578e-05,
+      "loss": 1.1548,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09656422509681141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976863240629593e-05,
+      "loss": 1.0096,
+      "step": 3516
+    },
+    {
+      "epoch": 0.09659168932465463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9768442131368085e-05,
+      "loss": 1.1659,
+      "step": 3517
+    },
+    {
+      "epoch": 0.09661915355249787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976825177914854e-05,
+      "loss": 1.1169,
+      "step": 3518
+    },
+    {
+      "epoch": 0.0966466177803411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97680613496388e-05,
+      "loss": 1.0306,
+      "step": 3519
+    },
+    {
+      "epoch": 0.09667408200818434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767870842840377e-05,
+      "loss": 1.1332,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09670154623602757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767680258754778e-05,
+      "loss": 1.0439,
+      "step": 3521
+    },
+    {
+      "epoch": 0.09672901046387081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767489597383505e-05,
+      "loss": 1.0674,
+      "step": 3522
+    },
+    {
+      "epoch": 0.09675647469171404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767298858728074e-05,
+      "loss": 1.0415,
+      "step": 3523
+    },
+    {
+      "epoch": 0.09678393891955728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767108042789988e-05,
+      "loss": 1.0349,
+      "step": 3524
+    },
+    {
+      "epoch": 0.09681140314740051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976691714957076e-05,
+      "loss": 1.015,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09683886737524375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766726179071906e-05,
+      "loss": 1.1306,
+      "step": 3526
+    },
+    {
+      "epoch": 0.09686633160308698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766535131294925e-05,
+      "loss": 1.0398,
+      "step": 3527
+    },
+    {
+      "epoch": 0.09689379583093022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976634400624134e-05,
+      "loss": 0.9379,
+      "step": 3528
+    },
+    {
+      "epoch": 0.09692126005877345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766152803912654e-05,
+      "loss": 1.0633,
+      "step": 3529
+    },
+    {
+      "epoch": 0.09694872428661667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765961524310386e-05,
+      "loss": 1.1737,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09697618851445991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765770167436043e-05,
+      "loss": 1.1099,
+      "step": 3531
+    },
+    {
+      "epoch": 0.09700365274230315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765578733291146e-05,
+      "loss": 1.1089,
+      "step": 3532
+    },
+    {
+      "epoch": 0.09703111697014638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976538722187721e-05,
+      "loss": 1.076,
+      "step": 3533
+    },
+    {
+      "epoch": 0.09705858119798962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765195633195744e-05,
+      "loss": 1.0808,
+      "step": 3534
+    },
+    {
+      "epoch": 0.09708604542583285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976500396724827e-05,
+      "loss": 1.0853,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09711350965367609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97648122240363e-05,
+      "loss": 1.148,
+      "step": 3536
+    },
+    {
+      "epoch": 0.09714097388151932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9764620403561355e-05,
+      "loss": 0.9788,
+      "step": 3537
+    },
+    {
+      "epoch": 0.09716843810936256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9764428505824952e-05,
+      "loss": 1.021,
+      "step": 3538
+    },
+    {
+      "epoch": 0.09719590233720579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9764236530828604e-05,
+      "loss": 1.0881,
+      "step": 3539
+    },
+    {
+      "epoch": 0.09722336656504903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976404447857384e-05,
+      "loss": 1.1755,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09725083079289226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976385234906217e-05,
+      "loss": 1.0439,
+      "step": 3541
+    },
+    {
+      "epoch": 0.0972782950207355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763660142295117e-05,
+      "loss": 1.0944,
+      "step": 3542
+    },
+    {
+      "epoch": 0.09730575924857873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763467858274207e-05,
+      "loss": 0.9975,
+      "step": 3543
+    },
+    {
+      "epoch": 0.09733322347642195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763275497000952e-05,
+      "loss": 0.993,
+      "step": 3544
+    },
+    {
+      "epoch": 0.09736068770426519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763083058476882e-05,
+      "loss": 1.038,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09738815193210842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976289054270352e-05,
+      "loss": 1.1549,
+      "step": 3546
+    },
+    {
+      "epoch": 0.09741561615995166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976269794968238e-05,
+      "loss": 1.0487,
+      "step": 3547
+    },
+    {
+      "epoch": 0.0974430803877949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762505279414996e-05,
+      "loss": 1.13,
+      "step": 3548
+    },
+    {
+      "epoch": 0.09747054461563813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762312531902886e-05,
+      "loss": 1.0715,
+      "step": 3549
+    },
+    {
+      "epoch": 0.09749800884348137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976211970714758e-05,
+      "loss": 1.0776,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0975254730713246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9761926805150603e-05,
+      "loss": 1.085,
+      "step": 3551
+    },
+    {
+      "epoch": 0.09755293729916784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976173382591348e-05,
+      "loss": 0.9973,
+      "step": 3552
+    },
+    {
+      "epoch": 0.09758040152701107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9761540769437736e-05,
+      "loss": 1.0293,
+      "step": 3553
+    },
+    {
+      "epoch": 0.0976078657548543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97613476357249e-05,
+      "loss": 1.0967,
+      "step": 3554
+    },
+    {
+      "epoch": 0.09763532998269754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97611544247765e-05,
+      "loss": 1.0978,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09766279421054078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976096113659406e-05,
+      "loss": 1.1451,
+      "step": 3556
+    },
+    {
+      "epoch": 0.09769025843838401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760767771179123e-05,
+      "loss": 1.105,
+      "step": 3557
+    },
+    {
+      "epoch": 0.09771772266622723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760574328533205e-05,
+      "loss": 1.1168,
+      "step": 3558
+    },
+    {
+      "epoch": 0.09774518689407047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760380808657845e-05,
+      "loss": 1.1434,
+      "step": 3559
+    },
+    {
+      "epoch": 0.0977726511219137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976018721155457e-05,
+      "loss": 1.0515,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09780011534975694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759993537224912e-05,
+      "loss": 1.0949,
+      "step": 3561
+    },
+    {
+      "epoch": 0.09782757957760017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759799785670405e-05,
+      "loss": 1.0982,
+      "step": 3562
+    },
+    {
+      "epoch": 0.09785504380544341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759605956892585e-05,
+      "loss": 1.1389,
+      "step": 3563
+    },
+    {
+      "epoch": 0.09788250803328664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759412050892977e-05,
+      "loss": 1.1405,
+      "step": 3564
+    },
+    {
+      "epoch": 0.09790997226112988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759218067673126e-05,
+      "loss": 1.1156,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09793743648897311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759024007234557e-05,
+      "loss": 1.1799,
+      "step": 3566
+    },
+    {
+      "epoch": 0.09796490071681635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975882986957881e-05,
+      "loss": 1.1184,
+      "step": 3567
+    },
+    {
+      "epoch": 0.09799236494465959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975863565470742e-05,
+      "loss": 1.0102,
+      "step": 3568
+    },
+    {
+      "epoch": 0.09801982917250282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975844136262193e-05,
+      "loss": 1.199,
+      "step": 3569
+    },
+    {
+      "epoch": 0.09804729340034606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758246993323868e-05,
+      "loss": 1.0816,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09807475762818928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758052546814776e-05,
+      "loss": 1.0776,
+      "step": 3571
+    },
+    {
+      "epoch": 0.09810222185603251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975785802309619e-05,
+      "loss": 1.1487,
+      "step": 3572
+    },
+    {
+      "epoch": 0.09812968608387575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975766342216966e-05,
+      "loss": 1.139,
+      "step": 3573
+    },
+    {
+      "epoch": 0.09815715031171898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757468744036706e-05,
+      "loss": 1.0294,
+      "step": 3574
+    },
+    {
+      "epoch": 0.09818461453956222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757273988698883e-05,
+      "loss": 1.1145,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09821207876740545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757079156157734e-05,
+      "loss": 1.0585,
+      "step": 3576
+    },
+    {
+      "epoch": 0.09823954299524869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975688424641479e-05,
+      "loss": 1.1417,
+      "step": 3577
+    },
+    {
+      "epoch": 0.09826700722309192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97566892594716e-05,
+      "loss": 1.1053,
+      "step": 3578
+    },
+    {
+      "epoch": 0.09829447145093516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9756494195329704e-05,
+      "loss": 0.9548,
+      "step": 3579
+    },
+    {
+      "epoch": 0.0983219356787784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975629905399065e-05,
+      "loss": 1.075,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09834939990662163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9756103835455976e-05,
+      "loss": 1.0613,
+      "step": 3581
+    },
+    {
+      "epoch": 0.09837686413446486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975590853972723e-05,
+      "loss": 1.1255,
+      "step": 3582
+    },
+    {
+      "epoch": 0.0984043283623081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9755713166805958e-05,
+      "loss": 0.9997,
+      "step": 3583
+    },
+    {
+      "epoch": 0.09843179259015133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97555177166937e-05,
+      "loss": 1.1789,
+      "step": 3584
+    },
+    {
+      "epoch": 0.09845925681799456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975532218939201e-05,
+      "loss": 1.0987,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09848672104583779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975512658490243e-05,
+      "loss": 1.0244,
+      "step": 3586
+    },
+    {
+      "epoch": 0.09851418527368103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754930903226514e-05,
+      "loss": 1.0335,
+      "step": 3587
+    },
+    {
+      "epoch": 0.09854164950152426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754735144365803e-05,
+      "loss": 1.1091,
+      "step": 3588
+    },
+    {
+      "epoch": 0.0985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754539308321848e-05,
+      "loss": 1.1248,
+      "step": 3589
+    },
+    {
+      "epoch": 0.09859657795721073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97543433950962e-05,
+      "loss": 1.0714,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09862404218505397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975414740469041e-05,
+      "loss": 1.0956,
+      "step": 3591
+    },
+    {
+      "epoch": 0.0986515064128972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753951337106025e-05,
+      "loss": 1.001,
+      "step": 3592
+    },
+    {
+      "epoch": 0.09867897064074044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97537551923446e-05,
+      "loss": 1.1239,
+      "step": 3593
+    },
+    {
+      "epoch": 0.09870643486858367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753558970407686e-05,
+      "loss": 1.0687,
+      "step": 3594
+    },
+    {
+      "epoch": 0.09873389909642691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753362671296835e-05,
+      "loss": 1.0349,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09876136332427014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753166295013603e-05,
+      "loss": 1.0147,
+      "step": 3596
+    },
+    {
+      "epoch": 0.09878882755211338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9752969841559536e-05,
+      "loss": 1.1724,
+      "step": 3597
+    },
+    {
+      "epoch": 0.0988162917799566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9752773310936198e-05,
+      "loss": 1.0145,
+      "step": 3598
+    },
+    {
+      "epoch": 0.09884375600779984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975257670314514e-05,
+      "loss": 1.0691,
+      "step": 3599
+    },
+    {
+      "epoch": 0.09887122023564307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9752380018187914e-05,
+      "loss": 1.1315,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0988986844634863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975218325606608e-05,
+      "loss": 1.0463,
+      "step": 3601
+    },
+    {
+      "epoch": 0.09892614869132954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751986416781198e-05,
+      "loss": 1.1548,
+      "step": 3602
+    },
+    {
+      "epoch": 0.09895361291917278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975178950033482e-05,
+      "loss": 1.1205,
+      "step": 3603
+    },
+    {
+      "epoch": 0.09898107714701601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751592506728503e-05,
+      "loss": 1.0879,
+      "step": 3604
+    },
+    {
+      "epoch": 0.09900854137485925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975139543596381e-05,
+      "loss": 1.1406,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09903600560270248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97511982880423e-05,
+      "loss": 1.0512,
+      "step": 3606
+    },
+    {
+      "epoch": 0.09906346983054572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751001062965533e-05,
+      "loss": 1.0773,
+      "step": 3607
+    },
+    {
+      "epoch": 0.09909093405838895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750803760735068e-05,
+      "loss": 1.0763,
+      "step": 3608
+    },
+    {
+      "epoch": 0.09911839828623219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750606381352468e-05,
+      "loss": 1.032,
+      "step": 3609
+    },
+    {
+      "epoch": 0.09914586251407542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750408924819288e-05,
+      "loss": 1.0585,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09917332674191866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97502113911371e-05,
+      "loss": 1.0458,
+      "step": 3611
+    },
+    {
+      "epoch": 0.09920079096976188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975001378030746e-05,
+      "loss": 1.0957,
+      "step": 3612
+    },
+    {
+      "epoch": 0.09922825519760511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974981609233194e-05,
+      "loss": 1.1617,
+      "step": 3613
+    },
+    {
+      "epoch": 0.09925571942544835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974961832721209e-05,
+      "loss": 1.0202,
+      "step": 3614
+    },
+    {
+      "epoch": 0.09928318365329158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974942048494949e-05,
+      "loss": 1.0929,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09931064788113482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749222565545695e-05,
+      "loss": 1.1172,
+      "step": 3616
+    },
+    {
+      "epoch": 0.09933811210897805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749024569002276e-05,
+      "loss": 1.0703,
+      "step": 3617
+    },
+    {
+      "epoch": 0.09936557633682129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748826495320796e-05,
+      "loss": 1.1501,
+      "step": 3618
+    },
+    {
+      "epoch": 0.09939304056466453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748628344502827e-05,
+      "loss": 1.0259,
+      "step": 3619
+    },
+    {
+      "epoch": 0.09942050479250776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748430116549932e-05,
+      "loss": 1.0491,
+      "step": 3620
+    },
+    {
+      "epoch": 0.099447969020351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748231811463686e-05,
+      "loss": 1.1882,
+      "step": 3621
+    },
+    {
+      "epoch": 0.09947543324819423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974803342924565e-05,
+      "loss": 1.0059,
+      "step": 3622
+    },
+    {
+      "epoch": 0.09950289747603747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9747834969897398e-05,
+      "loss": 1.0864,
+      "step": 3623
+    },
+    {
+      "epoch": 0.0995303617038807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9747636433420497e-05,
+      "loss": 1.2733,
+      "step": 3624
+    },
+    {
+      "epoch": 0.09955782593172392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9747437819816526e-05,
+      "loss": 0.93,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09958529015956716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9747239129087048e-05,
+      "loss": 1.1223,
+      "step": 3626
+    },
+    {
+      "epoch": 0.0996127543874104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974704036123364e-05,
+      "loss": 1.0468,
+      "step": 3627
+    },
+    {
+      "epoch": 0.09964021861525363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974684151625787e-05,
+      "loss": 1.0195,
+      "step": 3628
+    },
+    {
+      "epoch": 0.09966768284309686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746642594161317e-05,
+      "loss": 1.0121,
+      "step": 3629
+    },
+    {
+      "epoch": 0.0996951470709401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746443594945553e-05,
+      "loss": 0.9934,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09972261129878333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746244518612152e-05,
+      "loss": 1.1086,
+      "step": 3631
+    },
+    {
+      "epoch": 0.09975007552662657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746045365162687e-05,
+      "loss": 1.1094,
+      "step": 3632
+    },
+    {
+      "epoch": 0.0997775397544698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9745846134598735e-05,
+      "loss": 1.246,
+      "step": 3633
+    },
+    {
+      "epoch": 0.09980500398231304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9745646826921876e-05,
+      "loss": 1.0504,
+      "step": 3634
+    },
+    {
+      "epoch": 0.09983246821015627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9745447442133683e-05,
+      "loss": 1.0524,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09985993243799951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9745247980235736e-05,
+      "loss": 1.0716,
+      "step": 3636
+    },
+    {
+      "epoch": 0.09988739666584275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974504844122961e-05,
+      "loss": 1.0739,
+      "step": 3637
+    },
+    {
+      "epoch": 0.09991486089368598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744848825116887e-05,
+      "loss": 0.9706,
+      "step": 3638
+    },
+    {
+      "epoch": 0.0999423251215292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744649131899145e-05,
+      "loss": 1.0169,
+      "step": 3639
+    },
+    {
+      "epoch": 0.09996978934937244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744449361577963e-05,
+      "loss": 0.9448,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09999725357721567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744249514154925e-05,
+      "loss": 1.0425,
+      "step": 3641
+    },
+    {
+      "epoch": 0.10002471780505891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744049589631608e-05,
+      "loss": 1.1222,
+      "step": 3642
+    },
+    {
+      "epoch": 0.10005218203290214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97438495880096e-05,
+      "loss": 1.1885,
+      "step": 3643
+    },
+    {
+      "epoch": 0.10007964626074538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743649509290478e-05,
+      "loss": 0.9141,
+      "step": 3644
+    },
+    {
+      "epoch": 0.10010711048858861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743449353475826e-05,
+      "loss": 1.0815,
+      "step": 3645
+    },
+    {
+      "epoch": 0.10013457471643185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974324912056723e-05,
+      "loss": 1.1072,
+      "step": 3646
+    },
+    {
+      "epoch": 0.10016203894427508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974304881056627e-05,
+      "loss": 1.0523,
+      "step": 3647
+    },
+    {
+      "epoch": 0.10018950317211832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9742848423474537e-05,
+      "loss": 1.08,
+      "step": 3648
+    },
+    {
+      "epoch": 0.10021696739996155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974264795929361e-05,
+      "loss": 1.0835,
+      "step": 3649
+    },
+    {
+      "epoch": 0.10024443162780479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974244741802508e-05,
+      "loss": 1.1405,
+      "step": 3650
+    },
+    {
+      "epoch": 0.10027189585564802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974224679967053e-05,
+      "loss": 1.0753,
+      "step": 3651
+    },
+    {
+      "epoch": 0.10029936008349125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9742046104231552e-05,
+      "loss": 1.0856,
+      "step": 3652
+    },
+    {
+      "epoch": 0.10032682431133448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974184533170973e-05,
+      "loss": 0.9622,
+      "step": 3653
+    },
+    {
+      "epoch": 0.10035428853917772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741644482106655e-05,
+      "loss": 1.1505,
+      "step": 3654
+    },
+    {
+      "epoch": 0.10038175276702095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741443555423916e-05,
+      "loss": 1.0427,
+      "step": 3655
+    },
+    {
+      "epoch": 0.10040921699486419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741242551663104e-05,
+      "loss": 1.1001,
+      "step": 3656
+    },
+    {
+      "epoch": 0.10043668122270742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741041470825807e-05,
+      "loss": 1.0341,
+      "step": 3657
+    },
+    {
+      "epoch": 0.10046414545055066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974084031291361e-05,
+      "loss": 1.0309,
+      "step": 3658
+    },
+    {
+      "epoch": 0.10049160967839389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9740639077928117e-05,
+      "loss": 1.0323,
+      "step": 3659
+    },
+    {
+      "epoch": 0.10051907390623713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9740437765870914e-05,
+      "loss": 1.0967,
+      "step": 3660
+    },
+    {
+      "epoch": 0.10054653813408036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9740236376743592e-05,
+      "loss": 0.9521,
+      "step": 3661
+    },
+    {
+      "epoch": 0.1005740023619236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974003491054775e-05,
+      "loss": 1.0794,
+      "step": 3662
+    },
+    {
+      "epoch": 0.10060146658976683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739833367284976e-05,
+      "loss": 1.0167,
+      "step": 3663
+    },
+    {
+      "epoch": 0.10062893081761007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973963174695687e-05,
+      "loss": 1.1185,
+      "step": 3664
+    },
+    {
+      "epoch": 0.1006563950454533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739430049565022e-05,
+      "loss": 1.0935,
+      "step": 3665
+    },
+    {
+      "epoch": 0.10068385927329652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973922827511103e-05,
+      "loss": 0.9282,
+      "step": 3666
+    },
+    {
+      "epoch": 0.10071132350113976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739026423596498e-05,
+      "loss": 1.0503,
+      "step": 3667
+    },
+    {
+      "epoch": 0.100738787728983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738824495023012e-05,
+      "loss": 1.0856,
+      "step": 3668
+    },
+    {
+      "epoch": 0.10076625195682623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738622489392175e-05,
+      "loss": 1.2155,
+      "step": 3669
+    },
+    {
+      "epoch": 0.10079371618466947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738420406705582e-05,
+      "loss": 1.0446,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1008211804125127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973821824696484e-05,
+      "loss": 1.0227,
+      "step": 3671
+    },
+    {
+      "epoch": 0.10084864464035594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738016010171538e-05,
+      "loss": 1.0563,
+      "step": 3672
+    },
+    {
+      "epoch": 0.10087610886819917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737813696327278e-05,
+      "loss": 1.0999,
+      "step": 3673
+    },
+    {
+      "epoch": 0.1009035730960424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973761130543367e-05,
+      "loss": 1.0587,
+      "step": 3674
+    },
+    {
+      "epoch": 0.10093103732388564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737408837492307e-05,
+      "loss": 0.9572,
+      "step": 3675
+    },
+    {
+      "epoch": 0.10095850155172888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737206292504794e-05,
+      "loss": 1.0201,
+      "step": 3676
+    },
+    {
+      "epoch": 0.10098596577957211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737003670472737e-05,
+      "loss": 1.0204,
+      "step": 3677
+    },
+    {
+      "epoch": 0.10101343000741535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973680097139773e-05,
+      "loss": 1.1299,
+      "step": 3678
+    },
+    {
+      "epoch": 0.10104089423525858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9736598195281384e-05,
+      "loss": 1.0783,
+      "step": 3679
+    },
+    {
+      "epoch": 0.1010683584631018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9736395342125297e-05,
+      "loss": 1.0829,
+      "step": 3680
+    },
+    {
+      "epoch": 0.10109582269094504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9736192411931086e-05,
+      "loss": 1.1953,
+      "step": 3681
+    },
+    {
+      "epoch": 0.10112328691878827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735989404700343e-05,
+      "loss": 0.9933,
+      "step": 3682
+    },
+    {
+      "epoch": 0.10115075114663151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735786320434685e-05,
+      "loss": 1.123,
+      "step": 3683
+    },
+    {
+      "epoch": 0.10117821537447474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735583159135713e-05,
+      "loss": 1.0576,
+      "step": 3684
+    },
+    {
+      "epoch": 0.10120567960231798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735379920805034e-05,
+      "loss": 1.0547,
+      "step": 3685
+    },
+    {
+      "epoch": 0.10123314383016122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735176605444258e-05,
+      "loss": 1.0408,
+      "step": 3686
+    },
+    {
+      "epoch": 0.10126060805800445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9734973213054996e-05,
+      "loss": 1.0986,
+      "step": 3687
+    },
+    {
+      "epoch": 0.10128807228584769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9734769743638852e-05,
+      "loss": 1.0107,
+      "step": 3688
+    },
+    {
+      "epoch": 0.10131553651369092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973456619719744e-05,
+      "loss": 1.0505,
+      "step": 3689
+    },
+    {
+      "epoch": 0.10134300074153416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973436257373237e-05,
+      "loss": 1.0175,
+      "step": 3690
+    },
+    {
+      "epoch": 0.10137046496937739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9734158873245253e-05,
+      "loss": 1.1392,
+      "step": 3691
+    },
+    {
+      "epoch": 0.10139792919722063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97339550957377e-05,
+      "loss": 1.0361,
+      "step": 3692
+    },
+    {
+      "epoch": 0.10142539342506385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9733751241211324e-05,
+      "loss": 1.0081,
+      "step": 3693
+    },
+    {
+      "epoch": 0.10145285765290708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9733547309667736e-05,
+      "loss": 1.0101,
+      "step": 3694
+    },
+    {
+      "epoch": 0.10148032188075032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9733343301108555e-05,
+      "loss": 1.1102,
+      "step": 3695
+    },
+    {
+      "epoch": 0.10150778610859355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973313921553539e-05,
+      "loss": 1.022,
+      "step": 3696
+    },
+    {
+      "epoch": 0.10153525033643679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732935052949857e-05,
+      "loss": 1.0388,
+      "step": 3697
+    },
+    {
+      "epoch": 0.10156271456428002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973273081335357e-05,
+      "loss": 1.1127,
+      "step": 3698
+    },
+    {
+      "epoch": 0.10159017879212326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732526496748146e-05,
+      "loss": 1.1145,
+      "step": 3699
+    },
+    {
+      "epoch": 0.1016176430199665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732322103135204e-05,
+      "loss": 1.122,
+      "step": 3700
+    },
+    {
+      "epoch": 0.10164510724780973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973211763251636e-05,
+      "loss": 1.0216,
+      "step": 3701
+    },
+    {
+      "epoch": 0.10167257147565296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731913084893232e-05,
+      "loss": 1.1284,
+      "step": 3702
+    },
+    {
+      "epoch": 0.1017000357034962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731708460267436e-05,
+      "loss": 1.0416,
+      "step": 3703
+    },
+    {
+      "epoch": 0.10172749993133944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731503758640594e-05,
+      "loss": 1.1461,
+      "step": 3704
+    },
+    {
+      "epoch": 0.10175496415918267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731298980014325e-05,
+      "loss": 1.1265,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1017824283870259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973109412439025e-05,
+      "loss": 1.1096,
+      "step": 3706
+    },
+    {
+      "epoch": 0.10180989261486913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9730889191769985e-05,
+      "loss": 0.989,
+      "step": 3707
+    },
+    {
+      "epoch": 0.10183735684271236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9730684182155157e-05,
+      "loss": 1.0684,
+      "step": 3708
+    },
+    {
+      "epoch": 0.1018648210705556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973047909554739e-05,
+      "loss": 1.145,
+      "step": 3709
+    },
+    {
+      "epoch": 0.10189228529839883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9730273931948298e-05,
+      "loss": 1.1227,
+      "step": 3710
+    },
+    {
+      "epoch": 0.10191974952624207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973006869135951e-05,
+      "loss": 1.1874,
+      "step": 3711
+    },
+    {
+      "epoch": 0.1019472137540853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729863373782654e-05,
+      "loss": 1.0632,
+      "step": 3712
+    },
+    {
+      "epoch": 0.10197467798192854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729657979219348e-05,
+      "loss": 1.037,
+      "step": 3713
+    },
+    {
+      "epoch": 0.10200214220977177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729452507671216e-05,
+      "loss": 0.9328,
+      "step": 3714
+    },
+    {
+      "epoch": 0.10202960643761501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729246959139892e-05,
+      "loss": 1.2003,
+      "step": 3715
+    },
+    {
+      "epoch": 0.10205707066545824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729041333626995e-05,
+      "loss": 0.9849,
+      "step": 3716
+    },
+    {
+      "epoch": 0.10208453489330148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972883563113415e-05,
+      "loss": 1.0458,
+      "step": 3717
+    },
+    {
+      "epoch": 0.10211199912114471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728629851662994e-05,
+      "loss": 1.108,
+      "step": 3718
+    },
+    {
+      "epoch": 0.10213946334898795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728423995215153e-05,
+      "loss": 1.1487,
+      "step": 3719
+    },
+    {
+      "epoch": 0.10216692757683117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728218061792247e-05,
+      "loss": 1.0635,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1021943918046744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728012051395914e-05,
+      "loss": 1.0444,
+      "step": 3721
+    },
+    {
+      "epoch": 0.10222185603251764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972780596402778e-05,
+      "loss": 1.1222,
+      "step": 3722
+    },
+    {
+      "epoch": 0.10224932026036088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9727599799689475e-05,
+      "loss": 1.074,
+      "step": 3723
+    },
+    {
+      "epoch": 0.10227678448820411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9727393558382636e-05,
+      "loss": 1.0565,
+      "step": 3724
+    },
+    {
+      "epoch": 0.10230424871604735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972718724010889e-05,
+      "loss": 1.0431,
+      "step": 3725
+    },
+    {
+      "epoch": 0.10233171294389058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972698084486987e-05,
+      "loss": 1.0088,
+      "step": 3726
+    },
+    {
+      "epoch": 0.10235917717173382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9726774372667213e-05,
+      "loss": 0.9819,
+      "step": 3727
+    },
+    {
+      "epoch": 0.10238664139957705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9726567823502546e-05,
+      "loss": 1.1208,
+      "step": 3728
+    },
+    {
+      "epoch": 0.10241410562742029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9726361197377506e-05,
+      "loss": 1.0224,
+      "step": 3729
+    },
+    {
+      "epoch": 0.10244156985526352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972615449429373e-05,
+      "loss": 0.8806,
+      "step": 3730
+    },
+    {
+      "epoch": 0.10246903408310676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972594771425285e-05,
+      "loss": 1.1317,
+      "step": 3731
+    },
+    {
+      "epoch": 0.10249649831095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972574085725651e-05,
+      "loss": 1.0659,
+      "step": 3732
+    },
+    {
+      "epoch": 0.10252396253879323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9725533923306336e-05,
+      "loss": 1.0995,
+      "step": 3733
+    },
+    {
+      "epoch": 0.10255142676663645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972532691240397e-05,
+      "loss": 1.1254,
+      "step": 3734
+    },
+    {
+      "epoch": 0.10257889099447969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9725119824551054e-05,
+      "loss": 1.1241,
+      "step": 3735
+    },
+    {
+      "epoch": 0.10260635522232292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972491265974922e-05,
+      "loss": 1.0443,
+      "step": 3736
+    },
+    {
+      "epoch": 0.10263381945016616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972470541800011e-05,
+      "loss": 1.0515,
+      "step": 3737
+    },
+    {
+      "epoch": 0.10266128367800939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9724498099305362e-05,
+      "loss": 1.1336,
+      "step": 3738
+    },
+    {
+      "epoch": 0.10268874790585263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9724290703666623e-05,
+      "loss": 1.0255,
+      "step": 3739
+    },
+    {
+      "epoch": 0.10271621213369586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9724083231085526e-05,
+      "loss": 1.0522,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1027436763615391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723875681563717e-05,
+      "loss": 1.0185,
+      "step": 3741
+    },
+    {
+      "epoch": 0.10277114058938233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723668055102835e-05,
+      "loss": 0.9824,
+      "step": 3742
+    },
+    {
+      "epoch": 0.10279860481722557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723460351704525e-05,
+      "loss": 1.0457,
+      "step": 3743
+    },
+    {
+      "epoch": 0.1028260690450688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723252571370432e-05,
+      "loss": 1.049,
+      "step": 3744
+    },
+    {
+      "epoch": 0.10285353327291204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723044714102196e-05,
+      "loss": 1.0742,
+      "step": 3745
+    },
+    {
+      "epoch": 0.10288099750075527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722836779901465e-05,
+      "loss": 1.0009,
+      "step": 3746
+    },
+    {
+      "epoch": 0.1029084617285985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972262876876988e-05,
+      "loss": 1.1005,
+      "step": 3747
+    },
+    {
+      "epoch": 0.10293592595644173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722420680709094e-05,
+      "loss": 1.093,
+      "step": 3748
+    },
+    {
+      "epoch": 0.10296339018428496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722212515720752e-05,
+      "loss": 1.0199,
+      "step": 3749
+    },
+    {
+      "epoch": 0.1029908544121282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722004273806496e-05,
+      "loss": 1.0226,
+      "step": 3750
+    },
+    {
+      "epoch": 0.10301831863997143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721795954967975e-05,
+      "loss": 1.1388,
+      "step": 3751
+    },
+    {
+      "epoch": 0.10304578286781467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721587559206837e-05,
+      "loss": 1.0585,
+      "step": 3752
+    },
+    {
+      "epoch": 0.1030732470956579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721379086524734e-05,
+      "loss": 1.0222,
+      "step": 3753
+    },
+    {
+      "epoch": 0.10310071132350114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721170536923314e-05,
+      "loss": 1.1133,
+      "step": 3754
+    },
+    {
+      "epoch": 0.10312817555134438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9720961910404232e-05,
+      "loss": 1.1515,
+      "step": 3755
+    },
+    {
+      "epoch": 0.10315563977918761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9720753206969127e-05,
+      "loss": 0.978,
+      "step": 3756
+    },
+    {
+      "epoch": 0.10318310400703085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972054442661966e-05,
+      "loss": 1.108,
+      "step": 3757
+    },
+    {
+      "epoch": 0.10321056823487408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972033556935748e-05,
+      "loss": 1.0735,
+      "step": 3758
+    },
+    {
+      "epoch": 0.10323803246271732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972012663518424e-05,
+      "loss": 1.0743,
+      "step": 3759
+    },
+    {
+      "epoch": 0.10326549669056055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971991762410159e-05,
+      "loss": 1.1334,
+      "step": 3760
+    },
+    {
+      "epoch": 0.10329296091840377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971970853611119e-05,
+      "loss": 1.0776,
+      "step": 3761
+    },
+    {
+      "epoch": 0.10332042514624701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971949937121469e-05,
+      "loss": 0.9902,
+      "step": 3762
+    },
+    {
+      "epoch": 0.10334788937409024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9719290129413745e-05,
+      "loss": 1.1665,
+      "step": 3763
+    },
+    {
+      "epoch": 0.10337535360193348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971908081071001e-05,
+      "loss": 1.0311,
+      "step": 3764
+    },
+    {
+      "epoch": 0.10340281782977671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9718871415105148e-05,
+      "loss": 1.0311,
+      "step": 3765
+    },
+    {
+      "epoch": 0.10343028205761995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971866194260081e-05,
+      "loss": 1.046,
+      "step": 3766
+    },
+    {
+      "epoch": 0.10345774628546318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971845239319865e-05,
+      "loss": 1.0307,
+      "step": 3767
+    },
+    {
+      "epoch": 0.10348521051330642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9718242766900332e-05,
+      "loss": 1.0803,
+      "step": 3768
+    },
+    {
+      "epoch": 0.10351267474114965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971803306370751e-05,
+      "loss": 1.1491,
+      "step": 3769
+    },
+    {
+      "epoch": 0.10354013896899289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717823283621853e-05,
+      "loss": 1.0301,
+      "step": 3770
+    },
+    {
+      "epoch": 0.10356760319683612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717613426645008e-05,
+      "loss": 1.0602,
+      "step": 3771
+    },
+    {
+      "epoch": 0.10359506742467936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717403492778643e-05,
+      "loss": 1.067,
+      "step": 3772
+    },
+    {
+      "epoch": 0.1036225316525226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971719348202442e-05,
+      "loss": 1.0403,
+      "step": 3773
+    },
+    {
+      "epoch": 0.10364999588036582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9716983394383993e-05,
+      "loss": 1.0944,
+      "step": 3774
+    },
+    {
+      "epoch": 0.10367746010820905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9716773229859033e-05,
+      "loss": 1.0844,
+      "step": 3775
+    },
+    {
+      "epoch": 0.10370492433605229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97165629884512e-05,
+      "loss": 1.0975,
+      "step": 3776
+    },
+    {
+      "epoch": 0.10373238856389552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971635267016215e-05,
+      "loss": 1.0454,
+      "step": 3777
+    },
+    {
+      "epoch": 0.10375985279173876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971614227499356e-05,
+      "loss": 1.0829,
+      "step": 3778
+    },
+    {
+      "epoch": 0.10378731701958199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9715931802947088e-05,
+      "loss": 1.0657,
+      "step": 3779
+    },
+    {
+      "epoch": 0.10381478124742523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9715721254024397e-05,
+      "loss": 1.0994,
+      "step": 3780
+    },
+    {
+      "epoch": 0.10384224547526846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971551062822716e-05,
+      "loss": 1.0494,
+      "step": 3781
+    },
+    {
+      "epoch": 0.1038697097031117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971529992555704e-05,
+      "loss": 1.1814,
+      "step": 3782
+    },
+    {
+      "epoch": 0.10389717393095493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97150891460157e-05,
+      "loss": 1.0584,
+      "step": 3783
+    },
+    {
+      "epoch": 0.10392463815879817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714878289604813e-05,
+      "loss": 1.0585,
+      "step": 3784
+    },
+    {
+      "epoch": 0.1039521023866414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971466735632605e-05,
+      "loss": 1.0168,
+      "step": 3785
+    },
+    {
+      "epoch": 0.10397956661448464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714456346181067e-05,
+      "loss": 0.9902,
+      "step": 3786
+    },
+    {
+      "epoch": 0.10400703084232787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714245259171548e-05,
+      "loss": 1.085,
+      "step": 3787
+    },
+    {
+      "epoch": 0.1040344950701711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714034095299156e-05,
+      "loss": 1.0847,
+      "step": 3788
+    },
+    {
+      "epoch": 0.10406195929801433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971382285456556e-05,
+      "loss": 1.1244,
+      "step": 3789
+    },
+    {
+      "epoch": 0.10408942352585757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9713611536972445e-05,
+      "loss": 1.0774,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1041168877537008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9713400142521462e-05,
+      "loss": 1.0133,
+      "step": 3791
+    },
+    {
+      "epoch": 0.10414435198154404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97131886712143e-05,
+      "loss": 1.0432,
+      "step": 3792
+    },
+    {
+      "epoch": 0.10417181620938727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712977123052626e-05,
+      "loss": 0.9924,
+      "step": 3793
+    },
+    {
+      "epoch": 0.10419928043723051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712765498038117e-05,
+      "loss": 1.0171,
+      "step": 3794
+    },
+    {
+      "epoch": 0.10422674466507374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712553796172437e-05,
+      "loss": 1.0441,
+      "step": 3795
+    },
+    {
+      "epoch": 0.10425420889291698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712342017457277e-05,
+      "loss": 1.003,
+      "step": 3796
+    },
+    {
+      "epoch": 0.10428167312076021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97121301618943e-05,
+      "loss": 1.0199,
+      "step": 3797
+    },
+    {
+      "epoch": 0.10430913734860345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971191822948519e-05,
+      "loss": 0.9943,
+      "step": 3798
+    },
+    {
+      "epoch": 0.10433660157644668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971170622023162e-05,
+      "loss": 1.0806,
+      "step": 3799
+    },
+    {
+      "epoch": 0.10436406580428992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711494134135266e-05,
+      "loss": 1.1317,
+      "step": 3800
+    },
+    {
+      "epoch": 0.10439153003213314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711281971197813e-05,
+      "loss": 1.1126,
+      "step": 3801
+    },
+    {
+      "epoch": 0.10441899425997637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711069731420936e-05,
+      "loss": 1.1569,
+      "step": 3802
+    },
+    {
+      "epoch": 0.10444645848781961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971085741480631e-05,
+      "loss": 0.9762,
+      "step": 3803
+    },
+    {
+      "epoch": 0.10447392271566285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971064502135562e-05,
+      "loss": 1.1224,
+      "step": 3804
+    },
+    {
+      "epoch": 0.10450138694350608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710432551070543e-05,
+      "loss": 1.0718,
+      "step": 3805
+    },
+    {
+      "epoch": 0.10452885117134932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710220003952762e-05,
+      "loss": 1.0881,
+      "step": 3806
+    },
+    {
+      "epoch": 0.10455631539919255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710007380003963e-05,
+      "loss": 1.1182,
+      "step": 3807
+    },
+    {
+      "epoch": 0.10458377962703579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970979467922582e-05,
+      "loss": 1.0314,
+      "step": 3808
+    },
+    {
+      "epoch": 0.10461124385487902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9709581901620023e-05,
+      "loss": 1.0117,
+      "step": 3809
+    },
+    {
+      "epoch": 0.10463870808272226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9709369047188253e-05,
+      "loss": 1.1039,
+      "step": 3810
+    },
+    {
+      "epoch": 0.10466617231056549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970915611593219e-05,
+      "loss": 1.0897,
+      "step": 3811
+    },
+    {
+      "epoch": 0.10469363653840873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708943107853526e-05,
+      "loss": 1.005,
+      "step": 3812
+    },
+    {
+      "epoch": 0.10472110076625196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708730022953944e-05,
+      "loss": 1.0306,
+      "step": 3813
+    },
+    {
+      "epoch": 0.1047485649940952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708516861235128e-05,
+      "loss": 1.0972,
+      "step": 3814
+    },
+    {
+      "epoch": 0.10477602922193842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970830362269877e-05,
+      "loss": 1.1086,
+      "step": 3815
+    },
+    {
+      "epoch": 0.10480349344978165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708090307346547e-05,
+      "loss": 1.0629,
+      "step": 3816
+    },
+    {
+      "epoch": 0.10483095767762489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9707876915180155e-05,
+      "loss": 1.126,
+      "step": 3817
+    },
+    {
+      "epoch": 0.10485842190546812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970766344620128e-05,
+      "loss": 1.0715,
+      "step": 3818
+    },
+    {
+      "epoch": 0.10488588613331136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970744990041161e-05,
+      "loss": 1.0583,
+      "step": 3819
+    },
+    {
+      "epoch": 0.1049133503611546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970723627781284e-05,
+      "loss": 1.0355,
+      "step": 3820
+    },
+    {
+      "epoch": 0.10494081458899783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9707022578406653e-05,
+      "loss": 1.0703,
+      "step": 3821
+    },
+    {
+      "epoch": 0.10496827881684107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706808802194744e-05,
+      "loss": 1.0366,
+      "step": 3822
+    },
+    {
+      "epoch": 0.1049957430446843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706594949178804e-05,
+      "loss": 1.1425,
+      "step": 3823
+    },
+    {
+      "epoch": 0.10502320727252754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706381019360527e-05,
+      "loss": 1.0553,
+      "step": 3824
+    },
+    {
+      "epoch": 0.10505067150037077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706167012741602e-05,
+      "loss": 1.0947,
+      "step": 3825
+    },
+    {
+      "epoch": 0.105078135728214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705952929323722e-05,
+      "loss": 1.0179,
+      "step": 3826
+    },
+    {
+      "epoch": 0.10510559995605724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705738769108587e-05,
+      "loss": 1.0778,
+      "step": 3827
+    },
+    {
+      "epoch": 0.10513306418390048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705524532097883e-05,
+      "loss": 1.0369,
+      "step": 3828
+    },
+    {
+      "epoch": 0.1051605284117437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705310218293313e-05,
+      "loss": 1.0768,
+      "step": 3829
+    },
+    {
+      "epoch": 0.10518799263958693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705095827696566e-05,
+      "loss": 1.1475,
+      "step": 3830
+    },
+    {
+      "epoch": 0.10521545686743017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970488136030934e-05,
+      "loss": 1.1024,
+      "step": 3831
+    },
+    {
+      "epoch": 0.1052429210952734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704666816133336e-05,
+      "loss": 1.0527,
+      "step": 3832
+    },
+    {
+      "epoch": 0.10527038532311664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704452195170248e-05,
+      "loss": 1.0512,
+      "step": 3833
+    },
+    {
+      "epoch": 0.10529784955095987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704237497421777e-05,
+      "loss": 1.0982,
+      "step": 3834
+    },
+    {
+      "epoch": 0.10532531377880311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704022722889616e-05,
+      "loss": 1.0901,
+      "step": 3835
+    },
+    {
+      "epoch": 0.10535277800664634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703807871575475e-05,
+      "loss": 1.0474,
+      "step": 3836
+    },
+    {
+      "epoch": 0.10538024223448958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703592943481042e-05,
+      "loss": 0.947,
+      "step": 3837
+    },
+    {
+      "epoch": 0.10540770646233281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703377938608023e-05,
+      "loss": 1.0833,
+      "step": 3838
+    },
+    {
+      "epoch": 0.10543517069017605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703162856958118e-05,
+      "loss": 1.1363,
+      "step": 3839
+    },
+    {
+      "epoch": 0.10546263491801929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970294769853303e-05,
+      "loss": 0.9741,
+      "step": 3840
+    },
+    {
+      "epoch": 0.10549009914586252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9702732463334464e-05,
+      "loss": 1.1338,
+      "step": 3841
+    },
+    {
+      "epoch": 0.10551756337370574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9702517151364116e-05,
+      "loss": 0.9858,
+      "step": 3842
+    },
+    {
+      "epoch": 0.10554502760154898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97023017626237e-05,
+      "loss": 1.1122,
+      "step": 3843
+    },
+    {
+      "epoch": 0.10557249182939221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970208629711491e-05,
+      "loss": 1.1455,
+      "step": 3844
+    },
+    {
+      "epoch": 0.10559995605723545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9701870754839455e-05,
+      "loss": 1.1144,
+      "step": 3845
+    },
+    {
+      "epoch": 0.10562742028507868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970165513579904e-05,
+      "loss": 1.0022,
+      "step": 3846
+    },
+    {
+      "epoch": 0.10565488451292192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9701439439995375e-05,
+      "loss": 1.0375,
+      "step": 3847
+    },
+    {
+      "epoch": 0.10568234874076515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970122366743016e-05,
+      "loss": 1.0563,
+      "step": 3848
+    },
+    {
+      "epoch": 0.10570981296860839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970100781810511e-05,
+      "loss": 1.0435,
+      "step": 3849
+    },
+    {
+      "epoch": 0.10573727719645162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700791892021925e-05,
+      "loss": 1.0681,
+      "step": 3850
+    },
+    {
+      "epoch": 0.10576474142429486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700575889182317e-05,
+      "loss": 1.0258,
+      "step": 3851
+    },
+    {
+      "epoch": 0.1057922056521381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700359809587996e-05,
+      "loss": 1.0773,
+      "step": 3852
+    },
+    {
+      "epoch": 0.10581966987998133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700143653240668e-05,
+      "loss": 0.9858,
+      "step": 3853
+    },
+    {
+      "epoch": 0.10584713410782456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9699927420142045e-05,
+      "loss": 1.0527,
+      "step": 3854
+    },
+    {
+      "epoch": 0.1058745983356678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969971111029384e-05,
+      "loss": 0.9558,
+      "step": 3855
+    },
+    {
+      "epoch": 0.10590206256351102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9699494723697768e-05,
+      "loss": 1.1027,
+      "step": 3856
+    },
+    {
+      "epoch": 0.10592952679135426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9699278260355532e-05,
+      "loss": 0.9979,
+      "step": 3857
+    },
+    {
+      "epoch": 0.10595699101919749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969906172026885e-05,
+      "loss": 1.1101,
+      "step": 3858
+    },
+    {
+      "epoch": 0.10598445524704073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9698845103439438e-05,
+      "loss": 1.0575,
+      "step": 3859
+    },
+    {
+      "epoch": 0.10601191947488396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9698628409869002e-05,
+      "loss": 1.0371,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1060393837027272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969841163955926e-05,
+      "loss": 1.0144,
+      "step": 3861
+    },
+    {
+      "epoch": 0.10606684793057043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9698194792511934e-05,
+      "loss": 1.1444,
+      "step": 3862
+    },
+    {
+      "epoch": 0.10609431215841367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969797786872873e-05,
+      "loss": 1.1517,
+      "step": 3863
+    },
+    {
+      "epoch": 0.1061217763862569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969776086821137e-05,
+      "loss": 1.0652,
+      "step": 3864
+    },
+    {
+      "epoch": 0.10614924061410014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969754379096157e-05,
+      "loss": 1.0054,
+      "step": 3865
+    },
+    {
+      "epoch": 0.10617670484194337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9697326636981044e-05,
+      "loss": 1.0549,
+      "step": 3866
+    },
+    {
+      "epoch": 0.10620416906978661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9697109406271516e-05,
+      "loss": 1.0879,
+      "step": 3867
+    },
+    {
+      "epoch": 0.10623163329762984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96968920988347e-05,
+      "loss": 0.9609,
+      "step": 3868
+    },
+    {
+      "epoch": 0.10625909752547306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969667471467232e-05,
+      "loss": 1.0065,
+      "step": 3869
+    },
+    {
+      "epoch": 0.1062865617533163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9696457253786093e-05,
+      "loss": 1.0186,
+      "step": 3870
+    },
+    {
+      "epoch": 0.10631402598115954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9696239716177737e-05,
+      "loss": 0.9145,
+      "step": 3871
+    },
+    {
+      "epoch": 0.10634149020900277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9696022101848982e-05,
+      "loss": 1.0521,
+      "step": 3872
+    },
+    {
+      "epoch": 0.106368954436846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969580441080154e-05,
+      "loss": 1.0385,
+      "step": 3873
+    },
+    {
+      "epoch": 0.10639641866468924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969558664303714e-05,
+      "loss": 1.0578,
+      "step": 3874
+    },
+    {
+      "epoch": 0.10642388289253248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96953687985575e-05,
+      "loss": 1.0891,
+      "step": 3875
+    },
+    {
+      "epoch": 0.10645134712037571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9695150877364347e-05,
+      "loss": 1.1002,
+      "step": 3876
+    },
+    {
+      "epoch": 0.10647881134821895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969493287945941e-05,
+      "loss": 1.1014,
+      "step": 3877
+    },
+    {
+      "epoch": 0.10650627557606218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9694714804844402e-05,
+      "loss": 1.1756,
+      "step": 3878
+    },
+    {
+      "epoch": 0.10653373980390542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9694496653521055e-05,
+      "loss": 1.0885,
+      "step": 3879
+    },
+    {
+      "epoch": 0.10656120403174865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96942784254911e-05,
+      "loss": 1.1929,
+      "step": 3880
+    },
+    {
+      "epoch": 0.10658866825959189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969406012075626e-05,
+      "loss": 1.0966,
+      "step": 3881
+    },
+    {
+      "epoch": 0.10661613248743512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9693841739318254e-05,
+      "loss": 1.0607,
+      "step": 3882
+    },
+    {
+      "epoch": 0.10664359671527834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969362328117882e-05,
+      "loss": 0.9488,
+      "step": 3883
+    },
+    {
+      "epoch": 0.10667106094312158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9693404746339688e-05,
+      "loss": 1.1542,
+      "step": 3884
+    },
+    {
+      "epoch": 0.10669852517096481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9693186134802578e-05,
+      "loss": 1.1101,
+      "step": 3885
+    },
+    {
+      "epoch": 0.10672598939880805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9692967446569223e-05,
+      "loss": 1.1702,
+      "step": 3886
+    },
+    {
+      "epoch": 0.10675345362665128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969274868164136e-05,
+      "loss": 0.991,
+      "step": 3887
+    },
+    {
+      "epoch": 0.10678091785449452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969252984002071e-05,
+      "loss": 1.0263,
+      "step": 3888
+    },
+    {
+      "epoch": 0.10680838208233775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9692310921709014e-05,
+      "loss": 1.0641,
+      "step": 3889
+    },
+    {
+      "epoch": 0.10683584631018099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9692091926707996e-05,
+      "loss": 1.0657,
+      "step": 3890
+    },
+    {
+      "epoch": 0.10686331053802423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9691872855019394e-05,
+      "loss": 1.108,
+      "step": 3891
+    },
+    {
+      "epoch": 0.10689077476586746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9691653706644942e-05,
+      "loss": 1.1018,
+      "step": 3892
+    },
+    {
+      "epoch": 0.1069182389937107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9691434481586368e-05,
+      "loss": 1.1287,
+      "step": 3893
+    },
+    {
+      "epoch": 0.10694570322155393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969121517984541e-05,
+      "loss": 1.0053,
+      "step": 3894
+    },
+    {
+      "epoch": 0.10697316744939717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690995801423806e-05,
+      "loss": 0.9733,
+      "step": 3895
+    },
+    {
+      "epoch": 0.10700063167724039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690776346323287e-05,
+      "loss": 1.0186,
+      "step": 3896
+    },
+    {
+      "epoch": 0.10702809590508362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690556814545595e-05,
+      "loss": 1.0995,
+      "step": 3897
+    },
+    {
+      "epoch": 0.10705556013292686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969033720609246e-05,
+      "loss": 0.9712,
+      "step": 3898
+    },
+    {
+      "epoch": 0.1070830243607701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690117520965625e-05,
+      "loss": 1.1516,
+      "step": 3899
+    },
+    {
+      "epoch": 0.10711048858861333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689897759166826e-05,
+      "loss": 1.1407,
+      "step": 3900
+    },
+    {
+      "epoch": 0.10713795281645656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689677920697803e-05,
+      "loss": 1.1025,
+      "step": 3901
+    },
+    {
+      "epoch": 0.1071654170442998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689458005560297e-05,
+      "loss": 0.9449,
+      "step": 3902
+    },
+    {
+      "epoch": 0.10719288127214303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689238013756044e-05,
+      "loss": 1.1002,
+      "step": 3903
+    },
+    {
+      "epoch": 0.10722034549998627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689017945286784e-05,
+      "loss": 1.0659,
+      "step": 3904
+    },
+    {
+      "epoch": 0.1072478097278295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688797800154263e-05,
+      "loss": 1.048,
+      "step": 3905
+    },
+    {
+      "epoch": 0.10727527395567274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688577578360222e-05,
+      "loss": 1.07,
+      "step": 3906
+    },
+    {
+      "epoch": 0.10730273818351597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688357279906403e-05,
+      "loss": 1.1135,
+      "step": 3907
+    },
+    {
+      "epoch": 0.10733020241135921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688136904794547e-05,
+      "loss": 0.9792,
+      "step": 3908
+    },
+    {
+      "epoch": 0.10735766663920245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9687916453026396e-05,
+      "loss": 1.0704,
+      "step": 3909
+    },
+    {
+      "epoch": 0.10738513086704567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96876959246037e-05,
+      "loss": 1.1256,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1074125950948889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9687475319528203e-05,
+      "loss": 0.993,
+      "step": 3911
+    },
+    {
+      "epoch": 0.10744005932273214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968725463780164e-05,
+      "loss": 1.1848,
+      "step": 3912
+    },
+    {
+      "epoch": 0.10746752355057537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9687033879425778e-05,
+      "loss": 1.1044,
+      "step": 3913
+    },
+    {
+      "epoch": 0.10749498777841861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686813044402343e-05,
+      "loss": 1.0898,
+      "step": 3914
+    },
+    {
+      "epoch": 0.10752245200626184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686592132733096e-05,
+      "loss": 1.0481,
+      "step": 3915
+    },
+    {
+      "epoch": 0.10754991623410508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686371144419777e-05,
+      "loss": 1.0491,
+      "step": 3916
+    },
+    {
+      "epoch": 0.10757738046194831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686150079464136e-05,
+      "loss": 1.0419,
+      "step": 3917
+    },
+    {
+      "epoch": 0.10760484468979155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685928937867928e-05,
+      "loss": 1.0874,
+      "step": 3918
+    },
+    {
+      "epoch": 0.10763230891763478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685707719632896e-05,
+      "loss": 1.0833,
+      "step": 3919
+    },
+    {
+      "epoch": 0.10765977314547802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685486424760794e-05,
+      "loss": 1.0818,
+      "step": 3920
+    },
+    {
+      "epoch": 0.10768723737332125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685265053253368e-05,
+      "loss": 1.0555,
+      "step": 3921
+    },
+    {
+      "epoch": 0.10771470160116449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685043605112375e-05,
+      "loss": 0.9918,
+      "step": 3922
+    },
+    {
+      "epoch": 0.10774216582900771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684822080339566e-05,
+      "loss": 1.0025,
+      "step": 3923
+    },
+    {
+      "epoch": 0.10776963005685095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684600478936692e-05,
+      "loss": 1.0666,
+      "step": 3924
+    },
+    {
+      "epoch": 0.10779709428469418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968437880090551e-05,
+      "loss": 1.0209,
+      "step": 3925
+    },
+    {
+      "epoch": 0.10782455851253742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684157046247768e-05,
+      "loss": 1.0966,
+      "step": 3926
+    },
+    {
+      "epoch": 0.10785202274038065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9683935214965228e-05,
+      "loss": 1.0208,
+      "step": 3927
+    },
+    {
+      "epoch": 0.10787948696822389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968371330705964e-05,
+      "loss": 1.0357,
+      "step": 3928
+    },
+    {
+      "epoch": 0.10790695119606712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968349132253276e-05,
+      "loss": 1.0294,
+      "step": 3929
+    },
+    {
+      "epoch": 0.10793441542391036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9683269261386345e-05,
+      "loss": 1.1566,
+      "step": 3930
+    },
+    {
+      "epoch": 0.10796187965175359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9683047123622154e-05,
+      "loss": 1.0424,
+      "step": 3931
+    },
+    {
+      "epoch": 0.10798934387959683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9682824909241943e-05,
+      "loss": 1.0966,
+      "step": 3932
+    },
+    {
+      "epoch": 0.10801680810744006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968260261824747e-05,
+      "loss": 1.072,
+      "step": 3933
+    },
+    {
+      "epoch": 0.1080442723352833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9682380250640497e-05,
+      "loss": 1.1039,
+      "step": 3934
+    },
+    {
+      "epoch": 0.10807173656312653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9682157806422778e-05,
+      "loss": 0.9084,
+      "step": 3935
+    },
+    {
+      "epoch": 0.10809920079096977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681935285596075e-05,
+      "loss": 1.1388,
+      "step": 3936
+    },
+    {
+      "epoch": 0.10812666501881299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681712688162155e-05,
+      "loss": 1.0451,
+      "step": 3937
+    },
+    {
+      "epoch": 0.10815412924665622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968149001412277e-05,
+      "loss": 1.1379,
+      "step": 3938
+    },
+    {
+      "epoch": 0.10818159347449946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681267263479685e-05,
+      "loss": 0.9896,
+      "step": 3939
+    },
+    {
+      "epoch": 0.1082090577023427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681044436234664e-05,
+      "loss": 0.9443,
+      "step": 3940
+    },
+    {
+      "epoch": 0.10823652193018593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9680821532389468e-05,
+      "loss": 1.1284,
+      "step": 3941
+    },
+    {
+      "epoch": 0.10826398615802917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9680598551945866e-05,
+      "loss": 1.1155,
+      "step": 3942
+    },
+    {
+      "epoch": 0.1082914503858724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9680375494905617e-05,
+      "loss": 1.0365,
+      "step": 3943
+    },
+    {
+      "epoch": 0.10831891461371564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9680152361270486e-05,
+      "loss": 1.0428,
+      "step": 3944
+    },
+    {
+      "epoch": 0.10834637884155887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679929151042238e-05,
+      "loss": 1.0781,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1083738430694021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679705864222645e-05,
+      "loss": 1.0151,
+      "step": 3946
+    },
+    {
+      "epoch": 0.10840130729724534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679482500813466e-05,
+      "loss": 1.0165,
+      "step": 3947
+    },
+    {
+      "epoch": 0.10842877152508858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679259060816475e-05,
+      "loss": 1.0115,
+      "step": 3948
+    },
+    {
+      "epoch": 0.10845623575293181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679035544233436e-05,
+      "loss": 1.0811,
+      "step": 3949
+    },
+    {
+      "epoch": 0.10848369998077505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967881195106612e-05,
+      "loss": 1.1224,
+      "step": 3950
+    },
+    {
+      "epoch": 0.10851116420861827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9678588281316293e-05,
+      "loss": 1.108,
+      "step": 3951
+    },
+    {
+      "epoch": 0.1085386284364615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9678364534985724e-05,
+      "loss": 1.0216,
+      "step": 3952
+    },
+    {
+      "epoch": 0.10856609266430474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9678140712076187e-05,
+      "loss": 0.9738,
+      "step": 3953
+    },
+    {
+      "epoch": 0.10859355689214797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677916812589454e-05,
+      "loss": 1.0047,
+      "step": 3954
+    },
+    {
+      "epoch": 0.10862102111999121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677692836527294e-05,
+      "loss": 1.137,
+      "step": 3955
+    },
+    {
+      "epoch": 0.10864848534783444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677468783891478e-05,
+      "loss": 1.0413,
+      "step": 3956
+    },
+    {
+      "epoch": 0.10867594957567768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967724465468378e-05,
+      "loss": 0.9905,
+      "step": 3957
+    },
+    {
+      "epoch": 0.10870341380352092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677020448905974e-05,
+      "loss": 1.0093,
+      "step": 3958
+    },
+    {
+      "epoch": 0.10873087803136415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9676796166559834e-05,
+      "loss": 1.1036,
+      "step": 3959
+    },
+    {
+      "epoch": 0.10875834225920739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9676571807647136e-05,
+      "loss": 0.9985,
+      "step": 3960
+    },
+    {
+      "epoch": 0.10878580648705062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9676347372169653e-05,
+      "loss": 1.0462,
+      "step": 3961
+    },
+    {
+      "epoch": 0.10881327071489386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9676122860129157e-05,
+      "loss": 1.0319,
+      "step": 3962
+    },
+    {
+      "epoch": 0.10884073494273709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675898271527433e-05,
+      "loss": 1.0496,
+      "step": 3963
+    },
+    {
+      "epoch": 0.10886819917058031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675673606366253e-05,
+      "loss": 1.112,
+      "step": 3964
+    },
+    {
+      "epoch": 0.10889566339842355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675448864647397e-05,
+      "loss": 1.0443,
+      "step": 3965
+    },
+    {
+      "epoch": 0.10892312762626678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675224046372638e-05,
+      "loss": 1.1265,
+      "step": 3966
+    },
+    {
+      "epoch": 0.10895059185411002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967499915154376e-05,
+      "loss": 1.0581,
+      "step": 3967
+    },
+    {
+      "epoch": 0.10897805608195325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674774180162545e-05,
+      "loss": 1.0253,
+      "step": 3968
+    },
+    {
+      "epoch": 0.10900552030979649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674549132230764e-05,
+      "loss": 1.0608,
+      "step": 3969
+    },
+    {
+      "epoch": 0.10903298453763972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674324007750204e-05,
+      "loss": 1.1026,
+      "step": 3970
+    },
+    {
+      "epoch": 0.10906044876548296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674098806722646e-05,
+      "loss": 1.1516,
+      "step": 3971
+    },
+    {
+      "epoch": 0.1090879129933262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967387352914987e-05,
+      "loss": 0.9653,
+      "step": 3972
+    },
+    {
+      "epoch": 0.10911537722116943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967364817503366e-05,
+      "loss": 1.0697,
+      "step": 3973
+    },
+    {
+      "epoch": 0.10914284144901266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673422744375794e-05,
+      "loss": 1.0458,
+      "step": 3974
+    },
+    {
+      "epoch": 0.1091703056768559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673197237178067e-05,
+      "loss": 1.0771,
+      "step": 3975
+    },
+    {
+      "epoch": 0.10919776990469914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9672971653442253e-05,
+      "loss": 1.0788,
+      "step": 3976
+    },
+    {
+      "epoch": 0.10922523413254237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9672745993170138e-05,
+      "loss": 0.9946,
+      "step": 3977
+    },
+    {
+      "epoch": 0.10925269836038559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9672520256363513e-05,
+      "loss": 0.9001,
+      "step": 3978
+    },
+    {
+      "epoch": 0.10928016258822883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967229444302416e-05,
+      "loss": 1.0095,
+      "step": 3979
+    },
+    {
+      "epoch": 0.10930762681607206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967206855315387e-05,
+      "loss": 1.0498,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1093350910439153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9671842586754422e-05,
+      "loss": 1.0048,
+      "step": 3981
+    },
+    {
+      "epoch": 0.10936255527175853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967161654382761e-05,
+      "loss": 1.074,
+      "step": 3982
+    },
+    {
+      "epoch": 0.10939001949960177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9671390424375225e-05,
+      "loss": 1.119,
+      "step": 3983
+    },
+    {
+      "epoch": 0.109417483727445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967116422839905e-05,
+      "loss": 1.0909,
+      "step": 3984
+    },
+    {
+      "epoch": 0.10944494795528824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967093795590088e-05,
+      "loss": 1.0814,
+      "step": 3985
+    },
+    {
+      "epoch": 0.10947241218313147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9670711606882496e-05,
+      "loss": 1.115,
+      "step": 3986
+    },
+    {
+      "epoch": 0.10949987641097471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96704851813457e-05,
+      "loss": 1.0476,
+      "step": 3987
+    },
+    {
+      "epoch": 0.10952734063881794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967025867929228e-05,
+      "loss": 1.0651,
+      "step": 3988
+    },
+    {
+      "epoch": 0.10955480486666118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967003210072403e-05,
+      "loss": 1.0824,
+      "step": 3989
+    },
+    {
+      "epoch": 0.10958226909450441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9669805445642732e-05,
+      "loss": 0.9972,
+      "step": 3990
+    },
+    {
+      "epoch": 0.10960973332234764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9669578714050196e-05,
+      "loss": 1.0278,
+      "step": 3991
+    },
+    {
+      "epoch": 0.10963719755019087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9669351905948203e-05,
+      "loss": 1.022,
+      "step": 3992
+    },
+    {
+      "epoch": 0.1096646617780341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9669125021338552e-05,
+      "loss": 1.175,
+      "step": 3993
+    },
+    {
+      "epoch": 0.10969212600587734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966889806022304e-05,
+      "loss": 1.0494,
+      "step": 3994
+    },
+    {
+      "epoch": 0.10971959023372058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966867102260346e-05,
+      "loss": 0.9735,
+      "step": 3995
+    },
+    {
+      "epoch": 0.10974705446156381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966844390848161e-05,
+      "loss": 1.0955,
+      "step": 3996
+    },
+    {
+      "epoch": 0.10977451868940705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9668216717859287e-05,
+      "loss": 1.0167,
+      "step": 3997
+    },
+    {
+      "epoch": 0.10980198291725028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966798945073829e-05,
+      "loss": 1.0163,
+      "step": 3998
+    },
+    {
+      "epoch": 0.10982944714509352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9667762107120412e-05,
+      "loss": 1.1376,
+      "step": 3999
+    },
+    {
+      "epoch": 0.10985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966753468700746e-05,
+      "loss": 1.1115,
+      "step": 4000
+    },
+    {
+      "epoch": 0.10988437560077999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9667307190401222e-05,
+      "loss": 1.0287,
+      "step": 4001
+    },
+    {
+      "epoch": 0.10991183982862322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966707961730351e-05,
+      "loss": 1.0208,
+      "step": 4002
+    },
+    {
+      "epoch": 0.10993930405646646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666851967716117e-05,
+      "loss": 1.0592,
+      "step": 4003
+    },
+    {
+      "epoch": 0.1099667682843097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966662424164085e-05,
+      "loss": 1.1468,
+      "step": 4004
+    },
+    {
+      "epoch": 0.10999423251215291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666396439079502e-05,
+      "loss": 1.0684,
+      "step": 4005
+    },
+    {
+      "epoch": 0.11002169673999615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666168560033886e-05,
+      "loss": 0.985,
+      "step": 4006
+    },
+    {
+      "epoch": 0.11004916096783939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665940604505795e-05,
+      "loss": 1.0574,
+      "step": 4007
+    },
+    {
+      "epoch": 0.11007662519568262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665712572497043e-05,
+      "loss": 0.9964,
+      "step": 4008
+    },
+    {
+      "epoch": 0.11010408942352586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665484464009425e-05,
+      "loss": 1.1363,
+      "step": 4009
+    },
+    {
+      "epoch": 0.11013155365136909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966525627904475e-05,
+      "loss": 1.0223,
+      "step": 4010
+    },
+    {
+      "epoch": 0.11015901787921233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665028017604826e-05,
+      "loss": 1.1528,
+      "step": 4011
+    },
+    {
+      "epoch": 0.11018648210705556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9664799679691454e-05,
+      "loss": 1.1782,
+      "step": 4012
+    },
+    {
+      "epoch": 0.1102139463348988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9664571265306446e-05,
+      "loss": 1.0838,
+      "step": 4013
+    },
+    {
+      "epoch": 0.11024141056274203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9664342774451604e-05,
+      "loss": 1.0995,
+      "step": 4014
+    },
+    {
+      "epoch": 0.11026887479058527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966411420712874e-05,
+      "loss": 1.1437,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1102963390184285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966388556333966e-05,
+      "loss": 1.0047,
+      "step": 4016
+    },
+    {
+      "epoch": 0.11032380324627174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663656843086176e-05,
+      "loss": 1.035,
+      "step": 4017
+    },
+    {
+      "epoch": 0.11035126747411496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663428046370092e-05,
+      "loss": 1.0139,
+      "step": 4018
+    },
+    {
+      "epoch": 0.1103787317019582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663199173193222e-05,
+      "loss": 1.0156,
+      "step": 4019
+    },
+    {
+      "epoch": 0.11040619592980143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966297022355738e-05,
+      "loss": 1.123,
+      "step": 4020
+    },
+    {
+      "epoch": 0.11043366015764466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9662741197464376e-05,
+      "loss": 1.0524,
+      "step": 4021
+    },
+    {
+      "epoch": 0.1104611243854879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9662512094916018e-05,
+      "loss": 1.0561,
+      "step": 4022
+    },
+    {
+      "epoch": 0.11048858861333113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966228291591412e-05,
+      "loss": 1.0936,
+      "step": 4023
+    },
+    {
+      "epoch": 0.11051605284117437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9662053660460497e-05,
+      "loss": 1.1189,
+      "step": 4024
+    },
+    {
+      "epoch": 0.1105435170690176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661824328556966e-05,
+      "loss": 1.0713,
+      "step": 4025
+    },
+    {
+      "epoch": 0.11057098129686084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661594920205337e-05,
+      "loss": 0.9677,
+      "step": 4026
+    },
+    {
+      "epoch": 0.11059844552470408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661365435407426e-05,
+      "loss": 1.135,
+      "step": 4027
+    },
+    {
+      "epoch": 0.11062590975254731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966113587416505e-05,
+      "loss": 0.9952,
+      "step": 4028
+    },
+    {
+      "epoch": 0.11065337398039055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966090623648002e-05,
+      "loss": 1.081,
+      "step": 4029
+    },
+    {
+      "epoch": 0.11068083820823378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966067652235416e-05,
+      "loss": 1.0923,
+      "step": 4030
+    },
+    {
+      "epoch": 0.11070830243607702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660446731789288e-05,
+      "loss": 1.059,
+      "step": 4031
+    },
+    {
+      "epoch": 0.11073576666392024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660216864787218e-05,
+      "loss": 1.049,
+      "step": 4032
+    },
+    {
+      "epoch": 0.11076323089176347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659986921349768e-05,
+      "loss": 1.0037,
+      "step": 4033
+    },
+    {
+      "epoch": 0.11079069511960671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659756901478762e-05,
+      "loss": 1.0869,
+      "step": 4034
+    },
+    {
+      "epoch": 0.11081815934744994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659526805176016e-05,
+      "loss": 1.0223,
+      "step": 4035
+    },
+    {
+      "epoch": 0.11084562357529318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659296632443353e-05,
+      "loss": 1.1505,
+      "step": 4036
+    },
+    {
+      "epoch": 0.11087308780313641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965906638328259e-05,
+      "loss": 1.0423,
+      "step": 4037
+    },
+    {
+      "epoch": 0.11090055203097965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965883605769556e-05,
+      "loss": 1.1251,
+      "step": 4038
+    },
+    {
+      "epoch": 0.11092801625882288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9658605655684073e-05,
+      "loss": 1.1527,
+      "step": 4039
+    },
+    {
+      "epoch": 0.11095548048666612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9658375177249955e-05,
+      "loss": 1.0452,
+      "step": 4040
+    },
+    {
+      "epoch": 0.11098294471450935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9658144622395033e-05,
+      "loss": 0.9957,
+      "step": 4041
+    },
+    {
+      "epoch": 0.11101040894235259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965791399112113e-05,
+      "loss": 1.101,
+      "step": 4042
+    },
+    {
+      "epoch": 0.11103787317019582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9657683283430072e-05,
+      "loss": 1.0714,
+      "step": 4043
+    },
+    {
+      "epoch": 0.11106533739803906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965745249932368e-05,
+      "loss": 1.184,
+      "step": 4044
+    },
+    {
+      "epoch": 0.11109280162588228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965722163880378e-05,
+      "loss": 1.2,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11112026585372552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965699070187221e-05,
+      "loss": 1.1837,
+      "step": 4046
+    },
+    {
+      "epoch": 0.11114773008156875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9656759688530785e-05,
+      "loss": 1.0495,
+      "step": 4047
+    },
+    {
+      "epoch": 0.11117519430941199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9656528598781333e-05,
+      "loss": 1.1243,
+      "step": 4048
+    },
+    {
+      "epoch": 0.11120265853725522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965629743262569e-05,
+      "loss": 1.2052,
+      "step": 4049
+    },
+    {
+      "epoch": 0.11123012276509846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9656066190065682e-05,
+      "loss": 0.9822,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11125758699294169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655834871103135e-05,
+      "loss": 1.0379,
+      "step": 4051
+    },
+    {
+      "epoch": 0.11128505122078493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655603475739884e-05,
+      "loss": 1.0506,
+      "step": 4052
+    },
+    {
+      "epoch": 0.11131251544862816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655372003977756e-05,
+      "loss": 0.964,
+      "step": 4053
+    },
+    {
+      "epoch": 0.1113399796764714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655140455818587e-05,
+      "loss": 1.033,
+      "step": 4054
+    },
+    {
+      "epoch": 0.11136744390431463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654908831264205e-05,
+      "loss": 1.1672,
+      "step": 4055
+    },
+    {
+      "epoch": 0.11139490813215787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654677130316443e-05,
+      "loss": 1.0787,
+      "step": 4056
+    },
+    {
+      "epoch": 0.1114223723600011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965444535297714e-05,
+      "loss": 0.9728,
+      "step": 4057
+    },
+    {
+      "epoch": 0.11144983658784434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654213499248125e-05,
+      "loss": 1.0724,
+      "step": 4058
+    },
+    {
+      "epoch": 0.11147730081568756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653981569131232e-05,
+      "loss": 1.0618,
+      "step": 4059
+    },
+    {
+      "epoch": 0.1115047650435308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653749562628295e-05,
+      "loss": 1.1007,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11153222927137403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653517479741152e-05,
+      "loss": 1.0776,
+      "step": 4061
+    },
+    {
+      "epoch": 0.11155969349921727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653285320471642e-05,
+      "loss": 0.9823,
+      "step": 4062
+    },
+    {
+      "epoch": 0.1115871577270605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96530530848216e-05,
+      "loss": 1.0695,
+      "step": 4063
+    },
+    {
+      "epoch": 0.11161462195490374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965282077279286e-05,
+      "loss": 1.131,
+      "step": 4064
+    },
+    {
+      "epoch": 0.11164208618274697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965258838438726e-05,
+      "loss": 0.9452,
+      "step": 4065
+    },
+    {
+      "epoch": 0.11166955041059021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965235591960665e-05,
+      "loss": 1.077,
+      "step": 4066
+    },
+    {
+      "epoch": 0.11169701463843344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9652123378452854e-05,
+      "loss": 0.9802,
+      "step": 4067
+    },
+    {
+      "epoch": 0.11172447886627668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965189076092772e-05,
+      "loss": 1.0647,
+      "step": 4068
+    },
+    {
+      "epoch": 0.11175194309411991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9651658067033092e-05,
+      "loss": 0.9992,
+      "step": 4069
+    },
+    {
+      "epoch": 0.11177940732196315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96514252967708e-05,
+      "loss": 1.0568,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11180687154980638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9651192450142697e-05,
+      "loss": 0.8107,
+      "step": 4071
+    },
+    {
+      "epoch": 0.11183433577764962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965095952715062e-05,
+      "loss": 1.13,
+      "step": 4072
+    },
+    {
+      "epoch": 0.11186180000549284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9650726527796413e-05,
+      "loss": 0.9973,
+      "step": 4073
+    },
+    {
+      "epoch": 0.11188926423333607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965049345208192e-05,
+      "loss": 0.9979,
+      "step": 4074
+    },
+    {
+      "epoch": 0.11191672846117931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9650260300008985e-05,
+      "loss": 1.0295,
+      "step": 4075
+    },
+    {
+      "epoch": 0.11194419268902255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9650027071579448e-05,
+      "loss": 1.0488,
+      "step": 4076
+    },
+    {
+      "epoch": 0.11197165691686578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9649793766795163e-05,
+      "loss": 1.2349,
+      "step": 4077
+    },
+    {
+      "epoch": 0.11199912114470902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964956038565797e-05,
+      "loss": 1.1037,
+      "step": 4078
+    },
+    {
+      "epoch": 0.11202658537255225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9649326928169718e-05,
+      "loss": 1.1285,
+      "step": 4079
+    },
+    {
+      "epoch": 0.11205404960039549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9649093394332254e-05,
+      "loss": 1.1023,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11208151382823872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648859784147424e-05,
+      "loss": 1.0034,
+      "step": 4081
+    },
+    {
+      "epoch": 0.11210897805608196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648626097617076e-05,
+      "loss": 1.0644,
+      "step": 4082
+    },
+    {
+      "epoch": 0.11213644228392519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648392334743065e-05,
+      "loss": 1.1299,
+      "step": 4083
+    },
+    {
+      "epoch": 0.11216390651176843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648158495527235e-05,
+      "loss": 1.1057,
+      "step": 4084
+    },
+    {
+      "epoch": 0.11219137073961166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9647924579971436e-05,
+      "loss": 1.0297,
+      "step": 4085
+    },
+    {
+      "epoch": 0.11221883496745488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964769058807752e-05,
+      "loss": 0.9409,
+      "step": 4086
+    },
+    {
+      "epoch": 0.11224629919529812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9647456519847343e-05,
+      "loss": 1.034,
+      "step": 4087
+    },
+    {
+      "epoch": 0.11227376342314135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964722237528275e-05,
+      "loss": 1.1548,
+      "step": 4088
+    },
+    {
+      "epoch": 0.11230122765098459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96469881543856e-05,
+      "loss": 1.0853,
+      "step": 4089
+    },
+    {
+      "epoch": 0.11232869187882782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964675385715774e-05,
+      "loss": 0.9407,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11235615610667106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9646519483601024e-05,
+      "loss": 1.0043,
+      "step": 4091
+    },
+    {
+      "epoch": 0.1123836203345143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9646285033717313e-05,
+      "loss": 0.9924,
+      "step": 4092
+    },
+    {
+      "epoch": 0.11241108456235753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9646050507508455e-05,
+      "loss": 1.0866,
+      "step": 4093
+    },
+    {
+      "epoch": 0.11243854879020077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964581590497631e-05,
+      "loss": 1.1158,
+      "step": 4094
+    },
+    {
+      "epoch": 0.112466013018044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645581226122735e-05,
+      "loss": 1.1271,
+      "step": 4095
+    },
+    {
+      "epoch": 0.11249347724588724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645346470949586e-05,
+      "loss": 1.1146,
+      "step": 4096
+    },
+    {
+      "epoch": 0.11252094147373047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645111639458717e-05,
+      "loss": 1.0491,
+      "step": 4097
+    },
+    {
+      "epoch": 0.1125484057015737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964487673165199e-05,
+      "loss": 1.0007,
+      "step": 4098
+    },
+    {
+      "epoch": 0.11257586992941694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964464174753126e-05,
+      "loss": 0.9917,
+      "step": 4099
+    },
+    {
+      "epoch": 0.11260333415726016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964440668709839e-05,
+      "loss": 1.1582,
+      "step": 4100
+    },
+    {
+      "epoch": 0.1126307983851034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644171550355237e-05,
+      "loss": 1.0338,
+      "step": 4101
+    },
+    {
+      "epoch": 0.11265826261294663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643936337303666e-05,
+      "loss": 1.0989,
+      "step": 4102
+    },
+    {
+      "epoch": 0.11268572684078987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643701047945533e-05,
+      "loss": 1.0131,
+      "step": 4103
+    },
+    {
+      "epoch": 0.1127131910686331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96434656822827e-05,
+      "loss": 0.9579,
+      "step": 4104
+    },
+    {
+      "epoch": 0.11274065529647634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643230240317034e-05,
+      "loss": 1.0655,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11276811952431957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964299472205039e-05,
+      "loss": 1.0532,
+      "step": 4106
+    },
+    {
+      "epoch": 0.11279558375216281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642759127484644e-05,
+      "loss": 1.0507,
+      "step": 4107
+    },
+    {
+      "epoch": 0.11282304798000604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964252345662165e-05,
+      "loss": 0.9985,
+      "step": 4108
+    },
+    {
+      "epoch": 0.11285051220784928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964228770946327e-05,
+      "loss": 1.1289,
+      "step": 4109
+    },
+    {
+      "epoch": 0.11287797643569251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642051886011383e-05,
+      "loss": 1.0762,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11290544066353575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964181598626784e-05,
+      "loss": 1.1257,
+      "step": 4111
+    },
+    {
+      "epoch": 0.11293290489137899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9641580010234517e-05,
+      "loss": 1.0462,
+      "step": 4112
+    },
+    {
+      "epoch": 0.1129603691192222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9641343957913276e-05,
+      "loss": 1.1195,
+      "step": 4113
+    },
+    {
+      "epoch": 0.11298783334706544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964110782930599e-05,
+      "loss": 0.9795,
+      "step": 4114
+    },
+    {
+      "epoch": 0.11301529757490868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9640871624414523e-05,
+      "loss": 1.0745,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11304276180275191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9640635343240744e-05,
+      "loss": 0.9672,
+      "step": 4116
+    },
+    {
+      "epoch": 0.11307022603059515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9640398985786523e-05,
+      "loss": 1.1125,
+      "step": 4117
+    },
+    {
+      "epoch": 0.11309769025843838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964016255205373e-05,
+      "loss": 1.0395,
+      "step": 4118
+    },
+    {
+      "epoch": 0.11312515448628162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963992604204424e-05,
+      "loss": 1.013,
+      "step": 4119
+    },
+    {
+      "epoch": 0.11315261871412485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9639689455759922e-05,
+      "loss": 1.0196,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11318008294196809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9639452793202644e-05,
+      "loss": 0.9625,
+      "step": 4121
+    },
+    {
+      "epoch": 0.11320754716981132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963921605437428e-05,
+      "loss": 1.1228,
+      "step": 4122
+    },
+    {
+      "epoch": 0.11323501139765456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638979239276706e-05,
+      "loss": 1.0823,
+      "step": 4123
+    },
+    {
+      "epoch": 0.1132624756254978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96387423479118e-05,
+      "loss": 1.0641,
+      "step": 4124
+    },
+    {
+      "epoch": 0.11328993985334103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638505380281424e-05,
+      "loss": 1.0366,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11331740408118426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638268336387463e-05,
+      "loss": 0.8133,
+      "step": 4126
+    },
+    {
+      "epoch": 0.11334486830902749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963803121623179e-05,
+      "loss": 1.0772,
+      "step": 4127
+    },
+    {
+      "epoch": 0.11337233253687072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963779401981628e-05,
+      "loss": 1.0812,
+      "step": 4128
+    },
+    {
+      "epoch": 0.11339979676471396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963755674714281e-05,
+      "loss": 1.0546,
+      "step": 4129
+    },
+    {
+      "epoch": 0.11342726099255719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9637319398213256e-05,
+      "loss": 0.9886,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11345472522040043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9637081973029498e-05,
+      "loss": 0.9216,
+      "step": 4131
+    },
+    {
+      "epoch": 0.11348218944824366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9636844471593415e-05,
+      "loss": 1.0923,
+      "step": 4132
+    },
+    {
+      "epoch": 0.1135096536760869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9636606893906887e-05,
+      "loss": 1.1017,
+      "step": 4133
+    },
+    {
+      "epoch": 0.11353711790393013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9636369239971792e-05,
+      "loss": 1.0367,
+      "step": 4134
+    },
+    {
+      "epoch": 0.11356458213177337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963613150979001e-05,
+      "loss": 1.1474,
+      "step": 4135
+    },
+    {
+      "epoch": 0.1135920463596166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9635893703363424e-05,
+      "loss": 1.037,
+      "step": 4136
+    },
+    {
+      "epoch": 0.11361951058745984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963565582069391e-05,
+      "loss": 0.9857,
+      "step": 4137
+    },
+    {
+      "epoch": 0.11364697481530307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963541786178336e-05,
+      "loss": 1.0493,
+      "step": 4138
+    },
+    {
+      "epoch": 0.11367443904314631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963517982663365e-05,
+      "loss": 1.0336,
+      "step": 4139
+    },
+    {
+      "epoch": 0.11370190327098953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9634941715246665e-05,
+      "loss": 0.9846,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11372936749883276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963470352762429e-05,
+      "loss": 1.1066,
+      "step": 4141
+    },
+    {
+      "epoch": 0.113756831726676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9634465263768405e-05,
+      "loss": 1.0255,
+      "step": 4142
+    },
+    {
+      "epoch": 0.11378429595451924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96342269236809e-05,
+      "loss": 1.0662,
+      "step": 4143
+    },
+    {
+      "epoch": 0.11381176018236247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963398850736366e-05,
+      "loss": 1.0676,
+      "step": 4144
+    },
+    {
+      "epoch": 0.1138392244102057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9633750014818572e-05,
+      "loss": 1.0578,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11386668863804894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963351144604752e-05,
+      "loss": 0.9738,
+      "step": 4146
+    },
+    {
+      "epoch": 0.11389415286589218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9633272801052398e-05,
+      "loss": 1.0755,
+      "step": 4147
+    },
+    {
+      "epoch": 0.11392161709373541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9633034079835088e-05,
+      "loss": 1.0155,
+      "step": 4148
+    },
+    {
+      "epoch": 0.11394908132157865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963279528239748e-05,
+      "loss": 0.9777,
+      "step": 4149
+    },
+    {
+      "epoch": 0.11397654554942188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9632556408741467e-05,
+      "loss": 1.0517,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11400400977726512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9632317458868933e-05,
+      "loss": 1.0729,
+      "step": 4151
+    },
+    {
+      "epoch": 0.11403147400510835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9632078432781775e-05,
+      "loss": 0.9171,
+      "step": 4152
+    },
+    {
+      "epoch": 0.11405893823295159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631839330481875e-05,
+      "loss": 1.0959,
+      "step": 4153
+    },
+    {
+      "epoch": 0.11408640246079481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631600151971138e-05,
+      "loss": 1.1388,
+      "step": 4154
+    },
+    {
+      "epoch": 0.11411386668863804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631360897251447e-05,
+      "loss": 1.0352,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11414133091648128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96311215663247e-05,
+      "loss": 1.0797,
+      "step": 4156
+    },
+    {
+      "epoch": 0.11416879514432451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9630882159192784e-05,
+      "loss": 0.9808,
+      "step": 4157
+    },
+    {
+      "epoch": 0.11419625937216775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96306426758576e-05,
+      "loss": 1.0951,
+      "step": 4158
+    },
+    {
+      "epoch": 0.11422372360001098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963040311632104e-05,
+      "loss": 1.0999,
+      "step": 4159
+    },
+    {
+      "epoch": 0.11425118782785422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9630163480585003e-05,
+      "loss": 1.1254,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11427865205569745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962992376865138e-05,
+      "loss": 1.1714,
+      "step": 4161
+    },
+    {
+      "epoch": 0.11430611628354069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962968398052207e-05,
+      "loss": 1.1273,
+      "step": 4162
+    },
+    {
+      "epoch": 0.11433358051138393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962944411619897e-05,
+      "loss": 1.1327,
+      "step": 4163
+    },
+    {
+      "epoch": 0.11436104473922716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9629204175683977e-05,
+      "loss": 1.0616,
+      "step": 4164
+    },
+    {
+      "epoch": 0.1143885089670704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9628964158978993e-05,
+      "loss": 1.0098,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11441597319491363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9628724066085913e-05,
+      "loss": 1.0756,
+      "step": 4166
+    },
+    {
+      "epoch": 0.11444343742275685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962848389700664e-05,
+      "loss": 1.1049,
+      "step": 4167
+    },
+    {
+      "epoch": 0.11447090165060009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9628243651743074e-05,
+      "loss": 0.9277,
+      "step": 4168
+    },
+    {
+      "epoch": 0.11449836587844332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962800333029711e-05,
+      "loss": 1.0904,
+      "step": 4169
+    },
+    {
+      "epoch": 0.11452583010628656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9627762932670656e-05,
+      "loss": 0.9795,
+      "step": 4170
+    },
+    {
+      "epoch": 0.1145532943341298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962752245886561e-05,
+      "loss": 1.2206,
+      "step": 4171
+    },
+    {
+      "epoch": 0.11458075856197303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962728190888388e-05,
+      "loss": 1.1446,
+      "step": 4172
+    },
+    {
+      "epoch": 0.11460822278981626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9627041282727366e-05,
+      "loss": 1.0844,
+      "step": 4173
+    },
+    {
+      "epoch": 0.1146356870176595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626800580397973e-05,
+      "loss": 0.9842,
+      "step": 4174
+    },
+    {
+      "epoch": 0.11466315124550273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626559801897605e-05,
+      "loss": 1.019,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11469061547334597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626318947228164e-05,
+      "loss": 1.0172,
+      "step": 4176
+    },
+    {
+      "epoch": 0.1147180797011892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962607801639156e-05,
+      "loss": 1.0264,
+      "step": 4177
+    },
+    {
+      "epoch": 0.11474554392903244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9625837009389698e-05,
+      "loss": 1.0828,
+      "step": 4178
+    },
+    {
+      "epoch": 0.11477300815687567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9625595926224487e-05,
+      "loss": 1.0903,
+      "step": 4179
+    },
+    {
+      "epoch": 0.11480047238471891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9625354766897828e-05,
+      "loss": 1.1179,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11482793661256213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9625113531411637e-05,
+      "loss": 1.0729,
+      "step": 4181
+    },
+    {
+      "epoch": 0.11485540084040537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962487221976782e-05,
+      "loss": 1.0388,
+      "step": 4182
+    },
+    {
+      "epoch": 0.1148828650682486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9624630831968284e-05,
+      "loss": 1.0765,
+      "step": 4183
+    },
+    {
+      "epoch": 0.11491032929609184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962438936801494e-05,
+      "loss": 1.1465,
+      "step": 4184
+    },
+    {
+      "epoch": 0.11493779352393507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9624147827909697e-05,
+      "loss": 1.0287,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11496525775177831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623906211654472e-05,
+      "loss": 1.0287,
+      "step": 4186
+    },
+    {
+      "epoch": 0.11499272197962154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623664519251174e-05,
+      "loss": 1.0475,
+      "step": 4187
+    },
+    {
+      "epoch": 0.11502018620746478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623422750701712e-05,
+      "loss": 1.0527,
+      "step": 4188
+    },
+    {
+      "epoch": 0.11504765043530801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623180906008e-05,
+      "loss": 1.0294,
+      "step": 4189
+    },
+    {
+      "epoch": 0.11507511466315125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9622938985171958e-05,
+      "loss": 1.1078,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11510257889099448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962269698819549e-05,
+      "loss": 1.0775,
+      "step": 4191
+    },
+    {
+      "epoch": 0.11513004311883772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962245491508052e-05,
+      "loss": 1.0793,
+      "step": 4192
+    },
+    {
+      "epoch": 0.11515750734668095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962221276582896e-05,
+      "loss": 1.129,
+      "step": 4193
+    },
+    {
+      "epoch": 0.11518497157452418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9621970540442722e-05,
+      "loss": 1.0001,
+      "step": 4194
+    },
+    {
+      "epoch": 0.11521243580236741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9621728238923724e-05,
+      "loss": 1.1069,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11523990003021065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962148586127389e-05,
+      "loss": 0.9381,
+      "step": 4196
+    },
+    {
+      "epoch": 0.11526736425805388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962124340749513e-05,
+      "loss": 1.1446,
+      "step": 4197
+    },
+    {
+      "epoch": 0.11529482848589712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9621000877589367e-05,
+      "loss": 1.0395,
+      "step": 4198
+    },
+    {
+      "epoch": 0.11532229271374035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9620758271558518e-05,
+      "loss": 0.9554,
+      "step": 4199
+    },
+    {
+      "epoch": 0.11534975694158359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9620515589404502e-05,
+      "loss": 1.022,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11537722116942682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962027283112924e-05,
+      "loss": 0.9914,
+      "step": 4201
+    },
+    {
+      "epoch": 0.11540468539727006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9620029996734655e-05,
+      "loss": 0.968,
+      "step": 4202
+    },
+    {
+      "epoch": 0.11543214962511329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961978708622266e-05,
+      "loss": 1.0315,
+      "step": 4203
+    },
+    {
+      "epoch": 0.11545961385295653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961954409959519e-05,
+      "loss": 0.9125,
+      "step": 4204
+    },
+    {
+      "epoch": 0.11548707808079976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9619301036854156e-05,
+      "loss": 1.1552,
+      "step": 4205
+    },
+    {
+      "epoch": 0.115514542308643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961905789800149e-05,
+      "loss": 1.0938,
+      "step": 4206
+    },
+    {
+      "epoch": 0.11554200653648623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961881468303911e-05,
+      "loss": 1.1629,
+      "step": 4207
+    },
+    {
+      "epoch": 0.11556947076432945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618571391968944e-05,
+      "loss": 1.2286,
+      "step": 4208
+    },
+    {
+      "epoch": 0.11559693499217269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618328024792913e-05,
+      "loss": 1.1514,
+      "step": 4209
+    },
+    {
+      "epoch": 0.11562439922001592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618084581512944e-05,
+      "loss": 1.1476,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11565186344785916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617841062130967e-05,
+      "loss": 1.0701,
+      "step": 4211
+    },
+    {
+      "epoch": 0.1156793276757024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617597466648904e-05,
+      "loss": 1.0494,
+      "step": 4212
+    },
+    {
+      "epoch": 0.11570679190354563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617353795068687e-05,
+      "loss": 1.0353,
+      "step": 4213
+    },
+    {
+      "epoch": 0.11573425613138887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617110047392235e-05,
+      "loss": 1.0566,
+      "step": 4214
+    },
+    {
+      "epoch": 0.1157617203592321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9616866223621488e-05,
+      "loss": 1.0376,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11578918458707534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961662232375837e-05,
+      "loss": 1.0961,
+      "step": 4216
+    },
+    {
+      "epoch": 0.11581664881491857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961637834780481e-05,
+      "loss": 1.1777,
+      "step": 4217
+    },
+    {
+      "epoch": 0.1158441130427618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961613429576274e-05,
+      "loss": 1.0361,
+      "step": 4218
+    },
+    {
+      "epoch": 0.11587157727060504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961589016763409e-05,
+      "loss": 1.1036,
+      "step": 4219
+    },
+    {
+      "epoch": 0.11589904149844828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615645963420796e-05,
+      "loss": 1.049,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11592650572629151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615401683124782e-05,
+      "loss": 1.0278,
+      "step": 4221
+    },
+    {
+      "epoch": 0.11595396995413473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615157326747988e-05,
+      "loss": 1.0243,
+      "step": 4222
+    },
+    {
+      "epoch": 0.11598143418197797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961491289429234e-05,
+      "loss": 1.0674,
+      "step": 4223
+    },
+    {
+      "epoch": 0.1160088984098212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961466838575978e-05,
+      "loss": 1.0111,
+      "step": 4224
+    },
+    {
+      "epoch": 0.11603636263766444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961442380115224e-05,
+      "loss": 1.1268,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11606382686550767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9614179140471655e-05,
+      "loss": 0.9648,
+      "step": 4226
+    },
+    {
+      "epoch": 0.11609129109335091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9613934403719957e-05,
+      "loss": 1.235,
+      "step": 4227
+    },
+    {
+      "epoch": 0.11611875532119414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9613689590899086e-05,
+      "loss": 1.0378,
+      "step": 4228
+    },
+    {
+      "epoch": 0.11614621954903738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9613444702010984e-05,
+      "loss": 1.117,
+      "step": 4229
+    },
+    {
+      "epoch": 0.11617368377688062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961319973705758e-05,
+      "loss": 1.0429,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11620114800472385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612954696040818e-05,
+      "loss": 1.1887,
+      "step": 4231
+    },
+    {
+      "epoch": 0.11622861223256709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612709578962633e-05,
+      "loss": 1.1287,
+      "step": 4232
+    },
+    {
+      "epoch": 0.11625607646041032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612464385824968e-05,
+      "loss": 0.9794,
+      "step": 4233
+    },
+    {
+      "epoch": 0.11628354068825356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612219116629756e-05,
+      "loss": 1.0187,
+      "step": 4234
+    },
+    {
+      "epoch": 0.11631100491609678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9611973771378947e-05,
+      "loss": 1.0924,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11633846914394001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961172835007448e-05,
+      "loss": 1.1243,
+      "step": 4236
+    },
+    {
+      "epoch": 0.11636593337178325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961148285271829e-05,
+      "loss": 0.9854,
+      "step": 4237
+    },
+    {
+      "epoch": 0.11639339759962648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961123727931233e-05,
+      "loss": 1.0152,
+      "step": 4238
+    },
+    {
+      "epoch": 0.11642086182746972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610991629858532e-05,
+      "loss": 1.1245,
+      "step": 4239
+    },
+    {
+      "epoch": 0.11644832605531295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961074590435885e-05,
+      "loss": 1.0774,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11647579028315619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610500102815222e-05,
+      "loss": 1.0965,
+      "step": 4241
+    },
+    {
+      "epoch": 0.11650325451099942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610254225229596e-05,
+      "loss": 0.988,
+      "step": 4242
+    },
+    {
+      "epoch": 0.11653071873884266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610008271603912e-05,
+      "loss": 1.0518,
+      "step": 4243
+    },
+    {
+      "epoch": 0.1165581829666859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9609762241940124e-05,
+      "loss": 1.0995,
+      "step": 4244
+    },
+    {
+      "epoch": 0.11658564719452913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960951613624017e-05,
+      "loss": 1.0291,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11661311142237236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960926995450601e-05,
+      "loss": 1.1143,
+      "step": 4246
+    },
+    {
+      "epoch": 0.1166405756502156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9609023696739576e-05,
+      "loss": 1.1617,
+      "step": 4247
+    },
+    {
+      "epoch": 0.11666803987805884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9608777362942828e-05,
+      "loss": 0.9982,
+      "step": 4248
+    },
+    {
+      "epoch": 0.11669550410590206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960853095311771e-05,
+      "loss": 0.9971,
+      "step": 4249
+    },
+    {
+      "epoch": 0.11672296833374529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9608284467266175e-05,
+      "loss": 1.0775,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11675043256158853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960803790539017e-05,
+      "loss": 1.1982,
+      "step": 4251
+    },
+    {
+      "epoch": 0.11677789678943176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9607791267491647e-05,
+      "loss": 1.146,
+      "step": 4252
+    },
+    {
+      "epoch": 0.116805361017275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9607544553572557e-05,
+      "loss": 1.0747,
+      "step": 4253
+    },
+    {
+      "epoch": 0.11683282524511823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9607297763634855e-05,
+      "loss": 1.0093,
+      "step": 4254
+    },
+    {
+      "epoch": 0.11686028947296147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960705089768049e-05,
+      "loss": 0.9759,
+      "step": 4255
+    },
+    {
+      "epoch": 0.1168877537008047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9606803955711416e-05,
+      "loss": 1.0175,
+      "step": 4256
+    },
+    {
+      "epoch": 0.11691521792864794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9606556937729592e-05,
+      "loss": 0.9684,
+      "step": 4257
+    },
+    {
+      "epoch": 0.11694268215649117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9606309843736966e-05,
+      "loss": 1.0522,
+      "step": 4258
+    },
+    {
+      "epoch": 0.11697014638433441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9606062673735493e-05,
+      "loss": 1.1057,
+      "step": 4259
+    },
+    {
+      "epoch": 0.11699761061217764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9605815427727134e-05,
+      "loss": 0.98,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11702507484002088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960556810571384e-05,
+      "loss": 1.117,
+      "step": 4261
+    },
+    {
+      "epoch": 0.1170525390678641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9605320707697574e-05,
+      "loss": 1.0831,
+      "step": 4262
+    },
+    {
+      "epoch": 0.11708000329570734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960507323368029e-05,
+      "loss": 1.0273,
+      "step": 4263
+    },
+    {
+      "epoch": 0.11710746752355057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9604825683663938e-05,
+      "loss": 1.1623,
+      "step": 4264
+    },
+    {
+      "epoch": 0.1171349317513938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960457805765049e-05,
+      "loss": 0.9519,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11716239597923704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96043303556419e-05,
+      "loss": 1.0636,
+      "step": 4266
+    },
+    {
+      "epoch": 0.11718986020708028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9604082577640127e-05,
+      "loss": 1.0818,
+      "step": 4267
+    },
+    {
+      "epoch": 0.11721732443492351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603834723647135e-05,
+      "loss": 1.081,
+      "step": 4268
+    },
+    {
+      "epoch": 0.11724478866276675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603586793664882e-05,
+      "loss": 1.0595,
+      "step": 4269
+    },
+    {
+      "epoch": 0.11727225289060998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603338787695328e-05,
+      "loss": 1.0964,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11729971711845322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603090705740436e-05,
+      "loss": 1.0714,
+      "step": 4271
+    },
+    {
+      "epoch": 0.11732718134629645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602842547802177e-05,
+      "loss": 1.0434,
+      "step": 4272
+    },
+    {
+      "epoch": 0.11735464557413969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602594313882503e-05,
+      "loss": 1.0803,
+      "step": 4273
+    },
+    {
+      "epoch": 0.11738210980198292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602346003983382e-05,
+      "loss": 1.0333,
+      "step": 4274
+    },
+    {
+      "epoch": 0.11740957402982616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602097618106784e-05,
+      "loss": 1.085,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11743703825766938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601849156254665e-05,
+      "loss": 1.0729,
+      "step": 4276
+    },
+    {
+      "epoch": 0.11746450248551261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601600618429003e-05,
+      "loss": 0.9861,
+      "step": 4277
+    },
+    {
+      "epoch": 0.11749196671335585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601352004631753e-05,
+      "loss": 1.0417,
+      "step": 4278
+    },
+    {
+      "epoch": 0.11751943094119909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601103314864887e-05,
+      "loss": 1.0193,
+      "step": 4279
+    },
+    {
+      "epoch": 0.11754689516904232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600854549130374e-05,
+      "loss": 0.9469,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11757435939688556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600605707430178e-05,
+      "loss": 1.1528,
+      "step": 4281
+    },
+    {
+      "epoch": 0.11760182362472879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600356789766272e-05,
+      "loss": 1.1157,
+      "step": 4282
+    },
+    {
+      "epoch": 0.11762928785257203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600107796140627e-05,
+      "loss": 1.0526,
+      "step": 4283
+    },
+    {
+      "epoch": 0.11765675208041526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9599858726555206e-05,
+      "loss": 1.0724,
+      "step": 4284
+    },
+    {
+      "epoch": 0.1176842163082585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9599609581011987e-05,
+      "loss": 1.0273,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11771168053610173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9599360359512938e-05,
+      "loss": 1.1454,
+      "step": 4286
+    },
+    {
+      "epoch": 0.11773914476394497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959911106206003e-05,
+      "loss": 1.0445,
+      "step": 4287
+    },
+    {
+      "epoch": 0.1177666089917882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598861688655238e-05,
+      "loss": 0.9707,
+      "step": 4288
+    },
+    {
+      "epoch": 0.11779407321963142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598612239300535e-05,
+      "loss": 1.0523,
+      "step": 4289
+    },
+    {
+      "epoch": 0.11782153744747466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598362713997893e-05,
+      "loss": 1.0574,
+      "step": 4290
+    },
+    {
+      "epoch": 0.1178490016753179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598113112749288e-05,
+      "loss": 1.0481,
+      "step": 4291
+    },
+    {
+      "epoch": 0.11787646590316113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959786343555669e-05,
+      "loss": 1.0739,
+      "step": 4292
+    },
+    {
+      "epoch": 0.11790393013100436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9597613682422085e-05,
+      "loss": 1.0903,
+      "step": 4293
+    },
+    {
+      "epoch": 0.1179313943588476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9597363853347437e-05,
+      "loss": 1.0068,
+      "step": 4294
+    },
+    {
+      "epoch": 0.11795885858669083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959711394833473e-05,
+      "loss": 1.1833,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11798632281453407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596863967385942e-05,
+      "loss": 1.1049,
+      "step": 4296
+    },
+    {
+      "epoch": 0.1180137870423773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596613910503047e-05,
+      "loss": 0.9939,
+      "step": 4297
+    },
+    {
+      "epoch": 0.11804125127022054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596363777688027e-05,
+      "loss": 1.0591,
+      "step": 4298
+    },
+    {
+      "epoch": 0.11806871549806378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959611356894286e-05,
+      "loss": 0.956,
+      "step": 4299
+    },
+    {
+      "epoch": 0.11809617972590701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9595863284269524e-05,
+      "loss": 1.1436,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11812364395375025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959561292367e-05,
+      "loss": 1.0817,
+      "step": 4301
+    },
+    {
+      "epoch": 0.11815110818159348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959536248714627e-05,
+      "loss": 1.1323,
+      "step": 4302
+    },
+    {
+      "epoch": 0.1181785724094367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9595111974700316e-05,
+      "loss": 1.0986,
+      "step": 4303
+    },
+    {
+      "epoch": 0.11820603663727994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9594861386334117e-05,
+      "loss": 1.0341,
+      "step": 4304
+    },
+    {
+      "epoch": 0.11823350086512317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959461072204966e-05,
+      "loss": 1.0083,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11826096509296641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9594359981848928e-05,
+      "loss": 1.0691,
+      "step": 4306
+    },
+    {
+      "epoch": 0.11828842932080964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95941091657339e-05,
+      "loss": 1.0565,
+      "step": 4307
+    },
+    {
+      "epoch": 0.11831589354865288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593858273706565e-05,
+      "loss": 1.1346,
+      "step": 4308
+    },
+    {
+      "epoch": 0.11834335777649611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593607305768906e-05,
+      "loss": 1.059,
+      "step": 4309
+    },
+    {
+      "epoch": 0.11837082200433935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593356261922913e-05,
+      "loss": 1.0947,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11839828623218258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593105142170564e-05,
+      "loss": 1.1511,
+      "step": 4311
+    },
+    {
+      "epoch": 0.11842575046002582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9592853946513855e-05,
+      "loss": 0.9329,
+      "step": 4312
+    },
+    {
+      "epoch": 0.11845321468786905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959260267495477e-05,
+      "loss": 1.0938,
+      "step": 4313
+    },
+    {
+      "epoch": 0.11848067891571229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9592351327495296e-05,
+      "loss": 1.1388,
+      "step": 4314
+    },
+    {
+      "epoch": 0.11850814314355552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9592099904137424e-05,
+      "loss": 1.0543,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11853560737139875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9591848404883137e-05,
+      "loss": 0.9926,
+      "step": 4316
+    },
+    {
+      "epoch": 0.11856307159924198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9591596829734433e-05,
+      "loss": 1.0418,
+      "step": 4317
+    },
+    {
+      "epoch": 0.11859053582708522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95913451786933e-05,
+      "loss": 1.0942,
+      "step": 4318
+    },
+    {
+      "epoch": 0.11861800005492845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959109345176173e-05,
+      "loss": 1.0948,
+      "step": 4319
+    },
+    {
+      "epoch": 0.11864546428277169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959084164894171e-05,
+      "loss": 1.0258,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11867292851061492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9590589770235236e-05,
+      "loss": 0.9283,
+      "step": 4321
+    },
+    {
+      "epoch": 0.11870039273845816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9590337815644304e-05,
+      "loss": 1.0304,
+      "step": 4322
+    },
+    {
+      "epoch": 0.11872785696630139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9590085785170902e-05,
+      "loss": 1.0214,
+      "step": 4323
+    },
+    {
+      "epoch": 0.11875532119414463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9589833678817026e-05,
+      "loss": 1.0773,
+      "step": 4324
+    },
+    {
+      "epoch": 0.11878278542198786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958958149658467e-05,
+      "loss": 0.9089,
+      "step": 4325
+    },
+    {
+      "epoch": 0.1188102496498311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9589329238475833e-05,
+      "loss": 1.0125,
+      "step": 4326
+    },
+    {
+      "epoch": 0.11883771387767433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9589076904492508e-05,
+      "loss": 1.0654,
+      "step": 4327
+    },
+    {
+      "epoch": 0.11886517810551757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958882449463669e-05,
+      "loss": 1.0698,
+      "step": 4328
+    },
+    {
+      "epoch": 0.1188926423333608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958857200891038e-05,
+      "loss": 1.1146,
+      "step": 4329
+    },
+    {
+      "epoch": 0.11892010656120403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9588319447315577e-05,
+      "loss": 0.984,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11894757078904726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9588066809854272e-05,
+      "loss": 0.9895,
+      "step": 4331
+    },
+    {
+      "epoch": 0.1189750350168905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958781409652847e-05,
+      "loss": 1.0525,
+      "step": 4332
+    },
+    {
+      "epoch": 0.11900249924473373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958756130734017e-05,
+      "loss": 1.0524,
+      "step": 4333
+    },
+    {
+      "epoch": 0.11902996347257697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9587308442291372e-05,
+      "loss": 1.0045,
+      "step": 4334
+    },
+    {
+      "epoch": 0.1190574277004202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958705550138407e-05,
+      "loss": 1.0974,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11908489192826344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9586802484620278e-05,
+      "loss": 1.0484,
+      "step": 4336
+    },
+    {
+      "epoch": 0.11911235615610667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9586549392001986e-05,
+      "loss": 1.0663,
+      "step": 4337
+    },
+    {
+      "epoch": 0.11913982038394991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9586296223531204e-05,
+      "loss": 1.0621,
+      "step": 4338
+    },
+    {
+      "epoch": 0.11916728461179314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958604297920993e-05,
+      "loss": 1.0209,
+      "step": 4339
+    },
+    {
+      "epoch": 0.11919474883963638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9585789659040176e-05,
+      "loss": 0.9982,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11922221306747961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958553626302394e-05,
+      "loss": 1.0951,
+      "step": 4341
+    },
+    {
+      "epoch": 0.11924967729532285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9585282791163224e-05,
+      "loss": 1.0441,
+      "step": 4342
+    },
+    {
+      "epoch": 0.11927714152316608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958502924346004e-05,
+      "loss": 1.0297,
+      "step": 4343
+    },
+    {
+      "epoch": 0.1193046057510093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9584775619916395e-05,
+      "loss": 0.9594,
+      "step": 4344
+    },
+    {
+      "epoch": 0.11933206997885254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9584521920534288e-05,
+      "loss": 1.0742,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11935953420669577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9584268145315733e-05,
+      "loss": 1.0186,
+      "step": 4346
+    },
+    {
+      "epoch": 0.11938699843453901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958401429426273e-05,
+      "loss": 1.158,
+      "step": 4347
+    },
+    {
+      "epoch": 0.11941446266238225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9583760367377303e-05,
+      "loss": 0.9967,
+      "step": 4348
+    },
+    {
+      "epoch": 0.11944192689022548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9583506364661446e-05,
+      "loss": 1.1097,
+      "step": 4349
+    },
+    {
+      "epoch": 0.11946939111806872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9583252286117177e-05,
+      "loss": 1.0781,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11949685534591195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95829981317465e-05,
+      "loss": 1.0879,
+      "step": 4351
+    },
+    {
+      "epoch": 0.11952431957375519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582743901551432e-05,
+      "loss": 0.9738,
+      "step": 4352
+    },
+    {
+      "epoch": 0.11955178380159842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582489595533982e-05,
+      "loss": 1.0619,
+      "step": 4353
+    },
+    {
+      "epoch": 0.11957924802944166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582235213696163e-05,
+      "loss": 1.1126,
+      "step": 4354
+    },
+    {
+      "epoch": 0.11960671225728489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581980756039988e-05,
+      "loss": 1.0497,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11963417648512813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581726222567465e-05,
+      "loss": 1.0067,
+      "step": 4356
+    },
+    {
+      "epoch": 0.11966164071297135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581471613280615e-05,
+      "loss": 1.1187,
+      "step": 4357
+    },
+    {
+      "epoch": 0.11968910494081458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581216928181453e-05,
+      "loss": 1.093,
+      "step": 4358
+    },
+    {
+      "epoch": 0.11971656916865782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958096216727199e-05,
+      "loss": 0.9702,
+      "step": 4359
+    },
+    {
+      "epoch": 0.11974403339650105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958070733055424e-05,
+      "loss": 0.9551,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11977149762434429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9580452418030226e-05,
+      "loss": 1.0674,
+      "step": 4361
+    },
+    {
+      "epoch": 0.11979896185218752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958019742970196e-05,
+      "loss": 0.9762,
+      "step": 4362
+    },
+    {
+      "epoch": 0.11982642608003076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579942365571458e-05,
+      "loss": 1.0963,
+      "step": 4363
+    },
+    {
+      "epoch": 0.119853890307874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579687225640744e-05,
+      "loss": 1.102,
+      "step": 4364
+    },
+    {
+      "epoch": 0.11988135453571723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579432009911836e-05,
+      "loss": 1.0836,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11990881876356047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579176718386747e-05,
+      "loss": 1.0549,
+      "step": 4366
+    },
+    {
+      "epoch": 0.1199362829914037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578921351067503e-05,
+      "loss": 1.068,
+      "step": 4367
+    },
+    {
+      "epoch": 0.11996374721924694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957866590795612e-05,
+      "loss": 1.0024,
+      "step": 4368
+    },
+    {
+      "epoch": 0.11999121144709017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578410389054626e-05,
+      "loss": 1.0975,
+      "step": 4369
+    },
+    {
+      "epoch": 0.1200186756749334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578154794365037e-05,
+      "loss": 1.0298,
+      "step": 4370
+    },
+    {
+      "epoch": 0.12004613990277663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957789912388938e-05,
+      "loss": 1.0987,
+      "step": 4371
+    },
+    {
+      "epoch": 0.12007360413061986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957764337762967e-05,
+      "loss": 1.0133,
+      "step": 4372
+    },
+    {
+      "epoch": 0.1201010683584631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957738755558794e-05,
+      "loss": 1.0226,
+      "step": 4373
+    },
+    {
+      "epoch": 0.12012853258630633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9577131657766207e-05,
+      "loss": 1.1395,
+      "step": 4374
+    },
+    {
+      "epoch": 0.12015599681414957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95768756841665e-05,
+      "loss": 1.0641,
+      "step": 4375
+    },
+    {
+      "epoch": 0.1201834610419928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957661963479084e-05,
+      "loss": 1.1504,
+      "step": 4376
+    },
+    {
+      "epoch": 0.12021092526983604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957636350964126e-05,
+      "loss": 1.0718,
+      "step": 4377
+    },
+    {
+      "epoch": 0.12023838949767927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9576107308719782e-05,
+      "loss": 1.1426,
+      "step": 4378
+    },
+    {
+      "epoch": 0.12026585372552251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957585103202843e-05,
+      "loss": 1.0631,
+      "step": 4379
+    },
+    {
+      "epoch": 0.12029331795336574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575594679569238e-05,
+      "loss": 1.093,
+      "step": 4380
+    },
+    {
+      "epoch": 0.12032078218120898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957533825134423e-05,
+      "loss": 1.0137,
+      "step": 4381
+    },
+    {
+      "epoch": 0.12034824640905221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575081747355438e-05,
+      "loss": 1.153,
+      "step": 4382
+    },
+    {
+      "epoch": 0.12037571063689545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957482516760489e-05,
+      "loss": 1.1398,
+      "step": 4383
+    },
+    {
+      "epoch": 0.12040317486473867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574568512094615e-05,
+      "loss": 1.0017,
+      "step": 4384
+    },
+    {
+      "epoch": 0.1204306390925819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574311780826643e-05,
+      "loss": 1.0594,
+      "step": 4385
+    },
+    {
+      "epoch": 0.12045810332042514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574054973803012e-05,
+      "loss": 1.028,
+      "step": 4386
+    },
+    {
+      "epoch": 0.12048556754826838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957379809102575e-05,
+      "loss": 1.1104,
+      "step": 4387
+    },
+    {
+      "epoch": 0.12051303177611161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9573541132496885e-05,
+      "loss": 0.9916,
+      "step": 4388
+    },
+    {
+      "epoch": 0.12054049600395485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9573284098218458e-05,
+      "loss": 1.0799,
+      "step": 4389
+    },
+    {
+      "epoch": 0.12056796023179808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9573026988192497e-05,
+      "loss": 1.0517,
+      "step": 4390
+    },
+    {
+      "epoch": 0.12059542445964132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9572769802421035e-05,
+      "loss": 0.9864,
+      "step": 4391
+    },
+    {
+      "epoch": 0.12062288868748455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9572512540906118e-05,
+      "loss": 1.1547,
+      "step": 4392
+    },
+    {
+      "epoch": 0.12065035291532779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9572255203649766e-05,
+      "loss": 1.0526,
+      "step": 4393
+    },
+    {
+      "epoch": 0.12067781714317102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9571997790654028e-05,
+      "loss": 0.9991,
+      "step": 4394
+    },
+    {
+      "epoch": 0.12070528137101426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957174030192093e-05,
+      "loss": 1.0283,
+      "step": 4395
+    },
+    {
+      "epoch": 0.1207327455988575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957148273745252e-05,
+      "loss": 1.0709,
+      "step": 4396
+    },
+    {
+      "epoch": 0.12076020982670073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957122509725083e-05,
+      "loss": 0.9862,
+      "step": 4397
+    },
+    {
+      "epoch": 0.12078767405454395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95709673813179e-05,
+      "loss": 1.0914,
+      "step": 4398
+    },
+    {
+      "epoch": 0.12081513828238719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957070958965577e-05,
+      "loss": 1.1111,
+      "step": 4399
+    },
+    {
+      "epoch": 0.12084260251023042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9570451722266472e-05,
+      "loss": 1.1145,
+      "step": 4400
+    },
+    {
+      "epoch": 0.12087006673807366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9570193779152057e-05,
+      "loss": 1.0711,
+      "step": 4401
+    },
+    {
+      "epoch": 0.12089753096591689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956993576031456e-05,
+      "loss": 1.024,
+      "step": 4402
+    },
+    {
+      "epoch": 0.12092499519376013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956967766575603e-05,
+      "loss": 1.07,
+      "step": 4403
+    },
+    {
+      "epoch": 0.12095245942160336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9569419495478497e-05,
+      "loss": 1.0521,
+      "step": 4404
+    },
+    {
+      "epoch": 0.1209799236494466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9569161249484013e-05,
+      "loss": 1.0049,
+      "step": 4405
+    },
+    {
+      "epoch": 0.12100738787728983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568902927774615e-05,
+      "loss": 0.996,
+      "step": 4406
+    },
+    {
+      "epoch": 0.12103485210513307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568644530352354e-05,
+      "loss": 1.0911,
+      "step": 4407
+    },
+    {
+      "epoch": 0.1210623163329763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956838605721927e-05,
+      "loss": 1.1286,
+      "step": 4408
+    },
+    {
+      "epoch": 0.12108978056081954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956812750837741e-05,
+      "loss": 1.1232,
+      "step": 4409
+    },
+    {
+      "epoch": 0.12111724478866277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956786888382882e-05,
+      "loss": 1.0878,
+      "step": 4410
+    },
+    {
+      "epoch": 0.121144709016506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956761018357554e-05,
+      "loss": 1.0303,
+      "step": 4411
+    },
+    {
+      "epoch": 0.12117217324434923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9567351407619625e-05,
+      "loss": 1.0312,
+      "step": 4412
+    },
+    {
+      "epoch": 0.12119963747219246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956709255596312e-05,
+      "loss": 1.0384,
+      "step": 4413
+    },
+    {
+      "epoch": 0.1212271017000357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9566833628608074e-05,
+      "loss": 1.1711,
+      "step": 4414
+    },
+    {
+      "epoch": 0.12125456592787894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956657462555653e-05,
+      "loss": 0.9869,
+      "step": 4415
+    },
+    {
+      "epoch": 0.12128203015572217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9566315546810546e-05,
+      "loss": 1.0439,
+      "step": 4416
+    },
+    {
+      "epoch": 0.1213094943835654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956605639237217e-05,
+      "loss": 1.1166,
+      "step": 4417
+    },
+    {
+      "epoch": 0.12133695861140864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9565797162243445e-05,
+      "loss": 0.9873,
+      "step": 4418
+    },
+    {
+      "epoch": 0.12136442283925188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9565537856426426e-05,
+      "loss": 1.08,
+      "step": 4419
+    },
+    {
+      "epoch": 0.12139188706709511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956527847492317e-05,
+      "loss": 1.0364,
+      "step": 4420
+    },
+    {
+      "epoch": 0.12141935129493835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9565019017735727e-05,
+      "loss": 0.9729,
+      "step": 4421
+    },
+    {
+      "epoch": 0.12144681552278158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9564759484866146e-05,
+      "loss": 1.0435,
+      "step": 4422
+    },
+    {
+      "epoch": 0.12147427975062482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956449987631648e-05,
+      "loss": 1.1853,
+      "step": 4423
+    },
+    {
+      "epoch": 0.12150174397846805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956424019208879e-05,
+      "loss": 1.0947,
+      "step": 4424
+    },
+    {
+      "epoch": 0.12152920820631127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563980432185127e-05,
+      "loss": 1.0443,
+      "step": 4425
+    },
+    {
+      "epoch": 0.12155667243415451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563720596607546e-05,
+      "loss": 1.037,
+      "step": 4426
+    },
+    {
+      "epoch": 0.12158413666199774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563460685358102e-05,
+      "loss": 1.0646,
+      "step": 4427
+    },
+    {
+      "epoch": 0.12161160088984098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563200698438854e-05,
+      "loss": 1.134,
+      "step": 4428
+    },
+    {
+      "epoch": 0.12163906511768421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562940635851856e-05,
+      "loss": 1.0549,
+      "step": 4429
+    },
+    {
+      "epoch": 0.12166652934552745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562680497599168e-05,
+      "loss": 1.0148,
+      "step": 4430
+    },
+    {
+      "epoch": 0.12169399357337068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562420283682846e-05,
+      "loss": 1.1674,
+      "step": 4431
+    },
+    {
+      "epoch": 0.12172145780121392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562159994104953e-05,
+      "loss": 1.0648,
+      "step": 4432
+    },
+    {
+      "epoch": 0.12174892202905715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561899628867544e-05,
+      "loss": 1.0504,
+      "step": 4433
+    },
+    {
+      "epoch": 0.12177638625690039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561639187972684e-05,
+      "loss": 1.0128,
+      "step": 4434
+    },
+    {
+      "epoch": 0.12180385048474363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561378671422426e-05,
+      "loss": 1.0493,
+      "step": 4435
+    },
+    {
+      "epoch": 0.12183131471258686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561118079218842e-05,
+      "loss": 1.0938,
+      "step": 4436
+    },
+    {
+      "epoch": 0.1218587789404301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560857411363984e-05,
+      "loss": 0.981,
+      "step": 4437
+    },
+    {
+      "epoch": 0.12188624316827332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956059666785992e-05,
+      "loss": 1.0588,
+      "step": 4438
+    },
+    {
+      "epoch": 0.12191370739611655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560335848708713e-05,
+      "loss": 0.9717,
+      "step": 4439
+    },
+    {
+      "epoch": 0.12194117162395979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560074953912423e-05,
+      "loss": 0.975,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12196863585180302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9559813983473117e-05,
+      "loss": 1.0269,
+      "step": 4441
+    },
+    {
+      "epoch": 0.12199610007964626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9559552937392864e-05,
+      "loss": 1.0507,
+      "step": 4442
+    },
+    {
+      "epoch": 0.1220235643074895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955929181567372e-05,
+      "loss": 1.0363,
+      "step": 4443
+    },
+    {
+      "epoch": 0.12205102853533273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9559030618317757e-05,
+      "loss": 1.0371,
+      "step": 4444
+    },
+    {
+      "epoch": 0.12207849276317596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9558769345327044e-05,
+      "loss": 1.1807,
+      "step": 4445
+    },
+    {
+      "epoch": 0.1221059569910192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955850799670364e-05,
+      "loss": 1.0396,
+      "step": 4446
+    },
+    {
+      "epoch": 0.12213342121886243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9558246572449624e-05,
+      "loss": 1.0651,
+      "step": 4447
+    },
+    {
+      "epoch": 0.12216088544670567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9557985072567055e-05,
+      "loss": 1.0423,
+      "step": 4448
+    },
+    {
+      "epoch": 0.1221883496745489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955772349705801e-05,
+      "loss": 0.9967,
+      "step": 4449
+    },
+    {
+      "epoch": 0.12221581390239214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9557461845924547e-05,
+      "loss": 1.0544,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12224327813023537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955720011916875e-05,
+      "loss": 1.0551,
+      "step": 4451
+    },
+    {
+      "epoch": 0.1222707423580786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955693831679268e-05,
+      "loss": 1.1351,
+      "step": 4452
+    },
+    {
+      "epoch": 0.12229820658592183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955667643879841e-05,
+      "loss": 1.0956,
+      "step": 4453
+    },
+    {
+      "epoch": 0.12232567081376507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955641448518802e-05,
+      "loss": 1.0471,
+      "step": 4454
+    },
+    {
+      "epoch": 0.1223531350416083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9556152455963575e-05,
+      "loss": 1.0508,
+      "step": 4455
+    },
+    {
+      "epoch": 0.12238059926945154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955589035112715e-05,
+      "loss": 1.0888,
+      "step": 4456
+    },
+    {
+      "epoch": 0.12240806349729477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9555628170680818e-05,
+      "loss": 1.037,
+      "step": 4457
+    },
+    {
+      "epoch": 0.12243552772513801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9555365914626658e-05,
+      "loss": 1.1312,
+      "step": 4458
+    },
+    {
+      "epoch": 0.12246299195298124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9555103582966733e-05,
+      "loss": 1.0404,
+      "step": 4459
+    },
+    {
+      "epoch": 0.12249045618082448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554841175703134e-05,
+      "loss": 1.0163,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12251792040866771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955457869283793e-05,
+      "loss": 0.9563,
+      "step": 4461
+    },
+    {
+      "epoch": 0.12254538463651095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554316134373196e-05,
+      "loss": 0.9836,
+      "step": 4462
+    },
+    {
+      "epoch": 0.12257284886435418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554053500311013e-05,
+      "loss": 1.0429,
+      "step": 4463
+    },
+    {
+      "epoch": 0.12260031309219742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553790790653457e-05,
+      "loss": 1.1194,
+      "step": 4464
+    },
+    {
+      "epoch": 0.12262777732004064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553528005402608e-05,
+      "loss": 1.0127,
+      "step": 4465
+    },
+    {
+      "epoch": 0.12265524154788388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553265144560544e-05,
+      "loss": 1.0447,
+      "step": 4466
+    },
+    {
+      "epoch": 0.12268270577572711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553002208129347e-05,
+      "loss": 1.0829,
+      "step": 4467
+    },
+    {
+      "epoch": 0.12271017000357035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552739196111095e-05,
+      "loss": 1.0347,
+      "step": 4468
+    },
+    {
+      "epoch": 0.12273763423141358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552476108507872e-05,
+      "loss": 1.0279,
+      "step": 4469
+    },
+    {
+      "epoch": 0.12276509845925682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552212945321754e-05,
+      "loss": 0.9626,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12279256268710005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551949706554832e-05,
+      "loss": 1.0064,
+      "step": 4471
+    },
+    {
+      "epoch": 0.12282002691494329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955168639220918e-05,
+      "loss": 1.0833,
+      "step": 4472
+    },
+    {
+      "epoch": 0.12284749114278652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551423002286887e-05,
+      "loss": 1.0598,
+      "step": 4473
+    },
+    {
+      "epoch": 0.12287495537062976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551159536790035e-05,
+      "loss": 1.1426,
+      "step": 4474
+    },
+    {
+      "epoch": 0.12290241959847299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955089599572071e-05,
+      "loss": 0.9993,
+      "step": 4475
+    },
+    {
+      "epoch": 0.12292988382631623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9550632379080993e-05,
+      "loss": 1.1335,
+      "step": 4476
+    },
+    {
+      "epoch": 0.12295734805415946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9550368686872977e-05,
+      "loss": 1.1172,
+      "step": 4477
+    },
+    {
+      "epoch": 0.1229848122820027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955010491909874e-05,
+      "loss": 1.0417,
+      "step": 4478
+    },
+    {
+      "epoch": 0.12301227650984592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549841075760378e-05,
+      "loss": 1.0202,
+      "step": 4479
+    },
+    {
+      "epoch": 0.12303974073768915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549577156859972e-05,
+      "loss": 1.0309,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12306720496553239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549313162399613e-05,
+      "loss": 0.9823,
+      "step": 4481
+    },
+    {
+      "epoch": 0.12309466919337562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549049092381387e-05,
+      "loss": 1.0883,
+      "step": 4482
+    },
+    {
+      "epoch": 0.12312213342121886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9548784946807385e-05,
+      "loss": 1.02,
+      "step": 4483
+    },
+    {
+      "epoch": 0.1231495976490621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95485207256797e-05,
+      "loss": 1.0978,
+      "step": 4484
+    },
+    {
+      "epoch": 0.12317706187690533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9548256429000423e-05,
+      "loss": 1.013,
+      "step": 4485
+    },
+    {
+      "epoch": 0.12320452610474857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954799205677164e-05,
+      "loss": 1.0301,
+      "step": 4486
+    },
+    {
+      "epoch": 0.1232319903325918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9547727608995445e-05,
+      "loss": 0.886,
+      "step": 4487
+    },
+    {
+      "epoch": 0.12325945456043504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954746308567393e-05,
+      "loss": 1.0419,
+      "step": 4488
+    },
+    {
+      "epoch": 0.12328691878827827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954719848680919e-05,
+      "loss": 0.9819,
+      "step": 4489
+    },
+    {
+      "epoch": 0.1233143830161215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9546933812403318e-05,
+      "loss": 1.0087,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12334184724396474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954666906245841e-05,
+      "loss": 1.0355,
+      "step": 4491
+    },
+    {
+      "epoch": 0.12336931147180798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9546404236976553e-05,
+      "loss": 1.0299,
+      "step": 4492
+    },
+    {
+      "epoch": 0.1233967756996512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954613933595985e-05,
+      "loss": 1.1652,
+      "step": 4493
+    },
+    {
+      "epoch": 0.12342423992749443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9545874359410392e-05,
+      "loss": 1.1082,
+      "step": 4494
+    },
+    {
+      "epoch": 0.12345170415533767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954560930733028e-05,
+      "loss": 1.0502,
+      "step": 4495
+    },
+    {
+      "epoch": 0.1234791683831809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9545344179721612e-05,
+      "loss": 1.1039,
+      "step": 4496
+    },
+    {
+      "epoch": 0.12350663261102414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954507897658648e-05,
+      "loss": 0.9389,
+      "step": 4497
+    },
+    {
+      "epoch": 0.12353409683886737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544813697926988e-05,
+      "loss": 1.1989,
+      "step": 4498
+    },
+    {
+      "epoch": 0.12356156106671061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544548343745232e-05,
+      "loss": 1.0335,
+      "step": 4499
+    },
+    {
+      "epoch": 0.12358902529455384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544282914043312e-05,
+      "loss": 1.1279,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12361648952239708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954401740882333e-05,
+      "loss": 1.0035,
+      "step": 4501
+    },
+    {
+      "epoch": 0.12364395375024032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9543751828087382e-05,
+      "loss": 1.0637,
+      "step": 4502
+    },
+    {
+      "epoch": 0.12367141797808355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9543486171837576e-05,
+      "loss": 0.9999,
+      "step": 4503
+    },
+    {
+      "epoch": 0.12369888220592679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954322044007601e-05,
+      "loss": 1.0219,
+      "step": 4504
+    },
+    {
+      "epoch": 0.12372634643377002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9542954632804786e-05,
+      "loss": 1.0746,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12375381066161324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954268875002601e-05,
+      "loss": 1.1584,
+      "step": 4506
+    },
+    {
+      "epoch": 0.12378127488945648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9542422791741782e-05,
+      "loss": 1.1458,
+      "step": 4507
+    },
+    {
+      "epoch": 0.12380873911729971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954215675795421e-05,
+      "loss": 1.1152,
+      "step": 4508
+    },
+    {
+      "epoch": 0.12383620334514295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95418906486654e-05,
+      "loss": 1.1254,
+      "step": 4509
+    },
+    {
+      "epoch": 0.12386366757298618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954162446387745e-05,
+      "loss": 1.0875,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12389113180082942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9541358203592474e-05,
+      "loss": 1.0034,
+      "step": 4511
+    },
+    {
+      "epoch": 0.12391859602867265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954109186781257e-05,
+      "loss": 1.1133,
+      "step": 4512
+    },
+    {
+      "epoch": 0.12394606025651589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954082545653986e-05,
+      "loss": 1.1044,
+      "step": 4513
+    },
+    {
+      "epoch": 0.12397352448435912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9540558969776436e-05,
+      "loss": 1.0362,
+      "step": 4514
+    },
+    {
+      "epoch": 0.12400098871220236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954029240752442e-05,
+      "loss": 1.0787,
+      "step": 4515
+    },
+    {
+      "epoch": 0.1240284529400456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9540025769785912e-05,
+      "loss": 1.0914,
+      "step": 4516
+    },
+    {
+      "epoch": 0.12405591716788883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9539759056563024e-05,
+      "loss": 1.0267,
+      "step": 4517
+    },
+    {
+      "epoch": 0.12408338139573206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953949226785787e-05,
+      "loss": 1.0707,
+      "step": 4518
+    },
+    {
+      "epoch": 0.1241108456235753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9539225403672554e-05,
+      "loss": 1.004,
+      "step": 4519
+    },
+    {
+      "epoch": 0.12413830985141852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9538958464009194e-05,
+      "loss": 1.0223,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12416577407926176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9538691448869896e-05,
+      "loss": 1.058,
+      "step": 4521
+    },
+    {
+      "epoch": 0.12419323830710499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953842435825678e-05,
+      "loss": 1.0335,
+      "step": 4522
+    },
+    {
+      "epoch": 0.12422070253494823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953815719217196e-05,
+      "loss": 1.0184,
+      "step": 4523
+    },
+    {
+      "epoch": 0.12424816676279146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953788995061754e-05,
+      "loss": 1.0784,
+      "step": 4524
+    },
+    {
+      "epoch": 0.1242756309906347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537622633595638e-05,
+      "loss": 1.0797,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12430309521847793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537355241108373e-05,
+      "loss": 0.9623,
+      "step": 4526
+    },
+    {
+      "epoch": 0.12433055944632117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537087773157858e-05,
+      "loss": 1.1466,
+      "step": 4527
+    },
+    {
+      "epoch": 0.1243580236741644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9536820229746215e-05,
+      "loss": 0.9385,
+      "step": 4528
+    },
+    {
+      "epoch": 0.12438548790200764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953655261087555e-05,
+      "loss": 1.0649,
+      "step": 4529
+    },
+    {
+      "epoch": 0.12441295212985087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9536284916547993e-05,
+      "loss": 1.106,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12444041635769411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953601714676565e-05,
+      "loss": 1.1149,
+      "step": 4531
+    },
+    {
+      "epoch": 0.12446788058553734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953574930153065e-05,
+      "loss": 1.0642,
+      "step": 4532
+    },
+    {
+      "epoch": 0.12449534481338057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9535481380845106e-05,
+      "loss": 1.0246,
+      "step": 4533
+    },
+    {
+      "epoch": 0.1245228090412238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9535213384711138e-05,
+      "loss": 1.0609,
+      "step": 4534
+    },
+    {
+      "epoch": 0.12455027326906704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953494531313087e-05,
+      "loss": 1.0219,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12457773749691027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953467716610642e-05,
+      "loss": 1.0901,
+      "step": 4536
+    },
+    {
+      "epoch": 0.1246052017247535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953440894363991e-05,
+      "loss": 1.1163,
+      "step": 4537
+    },
+    {
+      "epoch": 0.12463266595259674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9534140645733466e-05,
+      "loss": 1.046,
+      "step": 4538
+    },
+    {
+      "epoch": 0.12466013018043998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9533872272389204e-05,
+      "loss": 1.0106,
+      "step": 4539
+    },
+    {
+      "epoch": 0.12468759440828321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9533603823609258e-05,
+      "loss": 1.0729,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12471505863612645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953333529939574e-05,
+      "loss": 1.1705,
+      "step": 4541
+    },
+    {
+      "epoch": 0.12474252286396968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9533066699750782e-05,
+      "loss": 1.0297,
+      "step": 4542
+    },
+    {
+      "epoch": 0.12476998709181292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9532798024676507e-05,
+      "loss": 1.0628,
+      "step": 4543
+    },
+    {
+      "epoch": 0.12479745131965615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9532529274175043e-05,
+      "loss": 1.0898,
+      "step": 4544
+    },
+    {
+      "epoch": 0.12482491554749939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953226044824851e-05,
+      "loss": 1.022,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12485237977534262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9531991546899048e-05,
+      "loss": 1.053,
+      "step": 4546
+    },
+    {
+      "epoch": 0.12487984400318584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9531722570128768e-05,
+      "loss": 1.0083,
+      "step": 4547
+    },
+    {
+      "epoch": 0.12490730823102908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9531453517939813e-05,
+      "loss": 1.0189,
+      "step": 4548
+    },
+    {
+      "epoch": 0.12493477245887231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9531184390334303e-05,
+      "loss": 0.9731,
+      "step": 4549
+    },
+    {
+      "epoch": 0.12496223668671555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953091518731437e-05,
+      "loss": 1.0477,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12498970091455879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9530645908882143e-05,
+      "loss": 1.005,
+      "step": 4551
+    },
+    {
+      "epoch": 0.12501716514240202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9530376555039758e-05,
+      "loss": 1.0633,
+      "step": 4552
+    },
+    {
+      "epoch": 0.12504462937024527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953010712578934e-05,
+      "loss": 1.0535,
+      "step": 4553
+    },
+    {
+      "epoch": 0.1250720935980885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952983762113302e-05,
+      "loss": 1.04,
+      "step": 4554
+    },
+    {
+      "epoch": 0.1250995578259317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952956804107294e-05,
+      "loss": 1.049,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12512702205377496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9529298385611218e-05,
+      "loss": 1.0915,
+      "step": 4556
+    },
+    {
+      "epoch": 0.12515448628161818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952902865475e-05,
+      "loss": 0.9926,
+      "step": 4557
+    },
+    {
+      "epoch": 0.12518195050946143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9528758848491416e-05,
+      "loss": 1.0396,
+      "step": 4558
+    },
+    {
+      "epoch": 0.12520941473730465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95284889668376e-05,
+      "loss": 1.0292,
+      "step": 4559
+    },
+    {
+      "epoch": 0.1252368789651479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952821900979069e-05,
+      "loss": 1.11,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12526434319299112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9527948977352816e-05,
+      "loss": 1.1013,
+      "step": 4561
+    },
+    {
+      "epoch": 0.12529180742083437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9527678869526127e-05,
+      "loss": 1.1233,
+      "step": 4562
+    },
+    {
+      "epoch": 0.1253192716486776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9527408686312746e-05,
+      "loss": 1.1183,
+      "step": 4563
+    },
+    {
+      "epoch": 0.12534673587652084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9527138427714818e-05,
+      "loss": 1.088,
+      "step": 4564
+    },
+    {
+      "epoch": 0.12537420010436406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952686809373448e-05,
+      "loss": 1.1312,
+      "step": 4565
+    },
+    {
+      "epoch": 0.1254016643322073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952659768437387e-05,
+      "loss": 1.028,
+      "step": 4566
+    },
+    {
+      "epoch": 0.12542912856005053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9526327199635132e-05,
+      "loss": 1.0193,
+      "step": 4567
+    },
+    {
+      "epoch": 0.12545659278789376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95260566395204e-05,
+      "loss": 1.0809,
+      "step": 4568
+    },
+    {
+      "epoch": 0.125484057015737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952578600403182e-05,
+      "loss": 1.0762,
+      "step": 4569
+    },
+    {
+      "epoch": 0.12551152124358023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952551529317153e-05,
+      "loss": 1.0673,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12553898547142348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9525244506941675e-05,
+      "loss": 1.1351,
+      "step": 4571
+    },
+    {
+      "epoch": 0.1255664496992667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524973645344394e-05,
+      "loss": 1.0602,
+      "step": 4572
+    },
+    {
+      "epoch": 0.12559391392710995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524702708381835e-05,
+      "loss": 1.0508,
+      "step": 4573
+    },
+    {
+      "epoch": 0.12562137815495317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524431696056135e-05,
+      "loss": 1.0907,
+      "step": 4574
+    },
+    {
+      "epoch": 0.12564884238279642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524160608369443e-05,
+      "loss": 1.0235,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12567630661063964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9523889445323904e-05,
+      "loss": 1.047,
+      "step": 4576
+    },
+    {
+      "epoch": 0.1257037708384829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9523618206921658e-05,
+      "loss": 1.0649,
+      "step": 4577
+    },
+    {
+      "epoch": 0.1257312350663261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952334689316486e-05,
+      "loss": 1.0963,
+      "step": 4578
+    },
+    {
+      "epoch": 0.12575869929416936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9523075504055654e-05,
+      "loss": 1.0645,
+      "step": 4579
+    },
+    {
+      "epoch": 0.12578616352201258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952280403959618e-05,
+      "loss": 0.9007,
+      "step": 4580
+    },
+    {
+      "epoch": 0.1258136277498558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9522532499788598e-05,
+      "loss": 1.1012,
+      "step": 4581
+    },
+    {
+      "epoch": 0.12584109197769905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9522260884635045e-05,
+      "loss": 0.9932,
+      "step": 4582
+    },
+    {
+      "epoch": 0.12586855620554227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9521989194137682e-05,
+      "loss": 1.1024,
+      "step": 4583
+    },
+    {
+      "epoch": 0.12589602043338552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9521717428298642e-05,
+      "loss": 1.0104,
+      "step": 4584
+    },
+    {
+      "epoch": 0.12592348466122874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9521445587120093e-05,
+      "loss": 0.9775,
+      "step": 4585
+    },
+    {
+      "epoch": 0.125950948889072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9521173670604175e-05,
+      "loss": 1.0832,
+      "step": 4586
+    },
+    {
+      "epoch": 0.1259784131169152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9520901678753045e-05,
+      "loss": 1.1025,
+      "step": 4587
+    },
+    {
+      "epoch": 0.12600587734475846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952062961156885e-05,
+      "loss": 1.064,
+      "step": 4588
+    },
+    {
+      "epoch": 0.12603334157260168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952035746905375e-05,
+      "loss": 1.1039,
+      "step": 4589
+    },
+    {
+      "epoch": 0.12606080580044493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952008525120989e-05,
+      "loss": 0.9558,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12608827002828815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951981295803943e-05,
+      "loss": 0.9731,
+      "step": 4591
+    },
+    {
+      "epoch": 0.1261157342561314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951954058954452e-05,
+      "loss": 0.9734,
+      "step": 4592
+    },
+    {
+      "epoch": 0.12614319848397462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951926814572732e-05,
+      "loss": 1.0755,
+      "step": 4593
+    },
+    {
+      "epoch": 0.12617066271181784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951899562658998e-05,
+      "loss": 1.0629,
+      "step": 4594
+    },
+    {
+      "epoch": 0.1261981269396611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9518723032134664e-05,
+      "loss": 1.0367,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12622559116750431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951845036236352e-05,
+      "loss": 0.9442,
+      "step": 4596
+    },
+    {
+      "epoch": 0.12625305539534756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951817761727871e-05,
+      "loss": 0.8425,
+      "step": 4597
+    },
+    {
+      "epoch": 0.12628051962319078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517904796882392e-05,
+      "loss": 1.0822,
+      "step": 4598
+    },
+    {
+      "epoch": 0.12630798385103403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517631901176725e-05,
+      "loss": 1.073,
+      "step": 4599
+    },
+    {
+      "epoch": 0.12633544807887725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517358930163867e-05,
+      "loss": 1.1445,
+      "step": 4600
+    },
+    {
+      "epoch": 0.1263629123067205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517085883845977e-05,
+      "loss": 1.1178,
+      "step": 4601
+    },
+    {
+      "epoch": 0.12639037653456373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9516812762225222e-05,
+      "loss": 1.0781,
+      "step": 4602
+    },
+    {
+      "epoch": 0.12641784076240697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9516539565303753e-05,
+      "loss": 1.0219,
+      "step": 4603
+    },
+    {
+      "epoch": 0.1264453049902502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951626629308374e-05,
+      "loss": 1.0023,
+      "step": 4604
+    },
+    {
+      "epoch": 0.12647276921809344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951599294556734e-05,
+      "loss": 1.1392,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12650023344593667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9515719522756716e-05,
+      "loss": 0.9837,
+      "step": 4606
+    },
+    {
+      "epoch": 0.12652769767377992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9515446024654038e-05,
+      "loss": 1.0747,
+      "step": 4607
+    },
+    {
+      "epoch": 0.12655516190162314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951517245126146e-05,
+      "loss": 1.0427,
+      "step": 4608
+    },
+    {
+      "epoch": 0.12658262612946636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9514898802581156e-05,
+      "loss": 1.1416,
+      "step": 4609
+    },
+    {
+      "epoch": 0.1266100903573096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9514625078615282e-05,
+      "loss": 0.9945,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12663755458515283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951435127936601e-05,
+      "loss": 0.9902,
+      "step": 4611
+    },
+    {
+      "epoch": 0.12666501881299608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9514077404835508e-05,
+      "loss": 1.0284,
+      "step": 4612
+    },
+    {
+      "epoch": 0.1266924830408393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513803455025938e-05,
+      "loss": 1.0171,
+      "step": 4613
+    },
+    {
+      "epoch": 0.12671994726868255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513529429939472e-05,
+      "loss": 1.0269,
+      "step": 4614
+    },
+    {
+      "epoch": 0.12674741149652577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513255329578275e-05,
+      "loss": 0.9895,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12677487572436902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9512981153944517e-05,
+      "loss": 1.0427,
+      "step": 4616
+    },
+    {
+      "epoch": 0.12680233995221224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9512706903040365e-05,
+      "loss": 1.0574,
+      "step": 4617
+    },
+    {
+      "epoch": 0.1268298041800555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9512432576867995e-05,
+      "loss": 0.9839,
+      "step": 4618
+    },
+    {
+      "epoch": 0.1268572684078987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951215817542957e-05,
+      "loss": 0.989,
+      "step": 4619
+    },
+    {
+      "epoch": 0.12688473263574196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511883698727265e-05,
+      "loss": 1.1228,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12691219686358518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511609146763253e-05,
+      "loss": 1.106,
+      "step": 4621
+    },
+    {
+      "epoch": 0.1269396610914284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511334519539703e-05,
+      "loss": 1.1261,
+      "step": 4622
+    },
+    {
+      "epoch": 0.12696712531927165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511059817058794e-05,
+      "loss": 0.9229,
+      "step": 4623
+    },
+    {
+      "epoch": 0.12699458954711487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9510785039322693e-05,
+      "loss": 1.1104,
+      "step": 4624
+    },
+    {
+      "epoch": 0.12702205377495812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9510510186333577e-05,
+      "loss": 1.0001,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12704951800280134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9510235258093623e-05,
+      "loss": 1.0292,
+      "step": 4626
+    },
+    {
+      "epoch": 0.1270769822306446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509960254604997e-05,
+      "loss": 1.0862,
+      "step": 4627
+    },
+    {
+      "epoch": 0.1271044464584878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509685175869884e-05,
+      "loss": 0.9833,
+      "step": 4628
+    },
+    {
+      "epoch": 0.12713191068633106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509410021890463e-05,
+      "loss": 0.9963,
+      "step": 4629
+    },
+    {
+      "epoch": 0.12715937491417428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95091347926689e-05,
+      "loss": 1.0201,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12718683914201753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9508859488207385e-05,
+      "loss": 0.9422,
+      "step": 4631
+    },
+    {
+      "epoch": 0.12721430336986075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950858410850809e-05,
+      "loss": 1.1502,
+      "step": 4632
+    },
+    {
+      "epoch": 0.127241767597704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950830865357319e-05,
+      "loss": 1.0315,
+      "step": 4633
+    },
+    {
+      "epoch": 0.12726923182554722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950803312340487e-05,
+      "loss": 0.9816,
+      "step": 4634
+    },
+    {
+      "epoch": 0.12729669605339045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9507757518005312e-05,
+      "loss": 0.9885,
+      "step": 4635
+    },
+    {
+      "epoch": 0.1273241602812337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9507481837376687e-05,
+      "loss": 1.1672,
+      "step": 4636
+    },
+    {
+      "epoch": 0.12735162450907692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9507206081521187e-05,
+      "loss": 1.1251,
+      "step": 4637
+    },
+    {
+      "epoch": 0.12737908873692017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950693025044099e-05,
+      "loss": 1.0463,
+      "step": 4638
+    },
+    {
+      "epoch": 0.1274065529647634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9506654344138277e-05,
+      "loss": 1.0327,
+      "step": 4639
+    },
+    {
+      "epoch": 0.12743401719260664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9506378362615235e-05,
+      "loss": 1.0718,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12746148142044986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950610230587404e-05,
+      "loss": 1.1364,
+      "step": 4641
+    },
+    {
+      "epoch": 0.1274889456482931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505826173916887e-05,
+      "loss": 1.0983,
+      "step": 4642
+    },
+    {
+      "epoch": 0.12751640987613633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505549966745952e-05,
+      "loss": 0.9949,
+      "step": 4643
+    },
+    {
+      "epoch": 0.12754387410397958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505273684363425e-05,
+      "loss": 1.0385,
+      "step": 4644
+    },
+    {
+      "epoch": 0.1275713383318228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950499732677149e-05,
+      "loss": 1.0874,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12759880255966605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504720893972333e-05,
+      "loss": 1.0269,
+      "step": 4646
+    },
+    {
+      "epoch": 0.12762626678750927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504444385968143e-05,
+      "loss": 1.0846,
+      "step": 4647
+    },
+    {
+      "epoch": 0.1276537310153525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950416780276111e-05,
+      "loss": 1.013,
+      "step": 4648
+    },
+    {
+      "epoch": 0.12768119524319574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9503891144353415e-05,
+      "loss": 1.0541,
+      "step": 4649
+    },
+    {
+      "epoch": 0.12770865947103896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9503614410747254e-05,
+      "loss": 1.0194,
+      "step": 4650
+    },
+    {
+      "epoch": 0.1277361236988822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9503337601944812e-05,
+      "loss": 1.0722,
+      "step": 4651
+    },
+    {
+      "epoch": 0.12776358792672543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9503060717948287e-05,
+      "loss": 1.1396,
+      "step": 4652
+    },
+    {
+      "epoch": 0.12779105215456868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9502783758759862e-05,
+      "loss": 1.0341,
+      "step": 4653
+    },
+    {
+      "epoch": 0.1278185163824119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950250672438173e-05,
+      "loss": 1.1138,
+      "step": 4654
+    },
+    {
+      "epoch": 0.12784598061025515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9502229614816087e-05,
+      "loss": 1.0645,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12787344483809837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950195243006512e-05,
+      "loss": 1.0735,
+      "step": 4656
+    },
+    {
+      "epoch": 0.12790090906594162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950167517013102e-05,
+      "loss": 1.1747,
+      "step": 4657
+    },
+    {
+      "epoch": 0.12792837329378484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9501397835015994e-05,
+      "loss": 1.1625,
+      "step": 4658
+    },
+    {
+      "epoch": 0.1279558375216281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9501120424722225e-05,
+      "loss": 1.0901,
+      "step": 4659
+    },
+    {
+      "epoch": 0.1279833017494713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950084293925191e-05,
+      "loss": 1.083,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12801076597731456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9500565378607246e-05,
+      "loss": 1.0868,
+      "step": 4661
+    },
+    {
+      "epoch": 0.12803823020515778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9500287742790423e-05,
+      "loss": 1.1353,
+      "step": 4662
+    },
+    {
+      "epoch": 0.128065694433001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950001003180365e-05,
+      "loss": 1.0097,
+      "step": 4663
+    },
+    {
+      "epoch": 0.12809315866084425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499732245649116e-05,
+      "loss": 0.9331,
+      "step": 4664
+    },
+    {
+      "epoch": 0.12812062288868747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499454384329022e-05,
+      "loss": 1.1037,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12814808711653072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499176447845563e-05,
+      "loss": 1.0538,
+      "step": 4666
+    },
+    {
+      "epoch": 0.12817555134437394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498898436200943e-05,
+      "loss": 1.0616,
+      "step": 4667
+    },
+    {
+      "epoch": 0.1282030155722172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498620349397355e-05,
+      "loss": 1.0638,
+      "step": 4668
+    },
+    {
+      "epoch": 0.12823047980006042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498342187437008e-05,
+      "loss": 0.9768,
+      "step": 4669
+    },
+    {
+      "epoch": 0.12825794402790366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498063950322094e-05,
+      "loss": 0.9877,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12828540825574689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949778563805482e-05,
+      "loss": 0.9265,
+      "step": 4671
+    },
+    {
+      "epoch": 0.12831287248359013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9497507250637388e-05,
+      "loss": 1.1123,
+      "step": 4672
+    },
+    {
+      "epoch": 0.12834033671143336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9497228788071997e-05,
+      "loss": 1.1128,
+      "step": 4673
+    },
+    {
+      "epoch": 0.1283678009392766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9496950250360855e-05,
+      "loss": 0.9811,
+      "step": 4674
+    },
+    {
+      "epoch": 0.12839526516711983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9496671637506164e-05,
+      "loss": 1.0007,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12842272939496305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949639294951013e-05,
+      "loss": 0.9685,
+      "step": 4676
+    },
+    {
+      "epoch": 0.1284501936228063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9496114186374956e-05,
+      "loss": 1.167,
+      "step": 4677
+    },
+    {
+      "epoch": 0.12847765785064952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9495835348102846e-05,
+      "loss": 0.9745,
+      "step": 4678
+    },
+    {
+      "epoch": 0.12850512207849277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9495556434696007e-05,
+      "loss": 1.1242,
+      "step": 4679
+    },
+    {
+      "epoch": 0.128532586306336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949527744615665e-05,
+      "loss": 1.1421,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12856005053417924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949499838248698e-05,
+      "loss": 1.0876,
+      "step": 4681
+    },
+    {
+      "epoch": 0.12858751476202246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9494719243689204e-05,
+      "loss": 1.063,
+      "step": 4682
+    },
+    {
+      "epoch": 0.1286149789898657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949444002976553e-05,
+      "loss": 0.9571,
+      "step": 4683
+    },
+    {
+      "epoch": 0.12864244321770893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949416074071817e-05,
+      "loss": 1.0455,
+      "step": 4684
+    },
+    {
+      "epoch": 0.12866990744555218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949388137654933e-05,
+      "loss": 1.0696,
+      "step": 4685
+    },
+    {
+      "epoch": 0.1286973716733954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949360193726123e-05,
+      "loss": 1.0085,
+      "step": 4686
+    },
+    {
+      "epoch": 0.12872483590123865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9493322422856068e-05,
+      "loss": 0.9246,
+      "step": 4687
+    },
+    {
+      "epoch": 0.12875230012908187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9493042833336065e-05,
+      "loss": 1.1745,
+      "step": 4688
+    },
+    {
+      "epoch": 0.1287797643569251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9492763168703426e-05,
+      "loss": 1.1251,
+      "step": 4689
+    },
+    {
+      "epoch": 0.12880722858476834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949248342896037e-05,
+      "loss": 1.0133,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12883469281261156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949220361410911e-05,
+      "loss": 0.906,
+      "step": 4691
+    },
+    {
+      "epoch": 0.1288621570404548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9491923724151857e-05,
+      "loss": 1.0958,
+      "step": 4692
+    },
+    {
+      "epoch": 0.12888962126829803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9491643759090825e-05,
+      "loss": 1.1602,
+      "step": 4693
+    },
+    {
+      "epoch": 0.12891708549614128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949136371892823e-05,
+      "loss": 1.0165,
+      "step": 4694
+    },
+    {
+      "epoch": 0.1289445497239845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949108360366629e-05,
+      "loss": 0.9876,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12897201395182775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949080341330722e-05,
+      "loss": 1.0404,
+      "step": 4696
+    },
+    {
+      "epoch": 0.12899947817967097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949052314785324e-05,
+      "loss": 1.0361,
+      "step": 4697
+    },
+    {
+      "epoch": 0.12902694240751422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949024280730656e-05,
+      "loss": 1.0861,
+      "step": 4698
+    },
+    {
+      "epoch": 0.12905440663535744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489962391669404e-05,
+      "loss": 1.14,
+      "step": 4699
+    },
+    {
+      "epoch": 0.1290818708632007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489681900943988e-05,
+      "loss": 1.1365,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12910933509104391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489401335132536e-05,
+      "loss": 0.9545,
+      "step": 4701
+    },
+    {
+      "epoch": 0.12913679931888716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489120694237264e-05,
+      "loss": 1.0472,
+      "step": 4702
+    },
+    {
+      "epoch": 0.12916426354673038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9488839978260393e-05,
+      "loss": 1.0679,
+      "step": 4703
+    },
+    {
+      "epoch": 0.1291917277745736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9488559187204147e-05,
+      "loss": 1.0704,
+      "step": 4704
+    },
+    {
+      "epoch": 0.12921919200241685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948827832107074e-05,
+      "loss": 1.1127,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12924665623026008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487997379862404e-05,
+      "loss": 0.9626,
+      "step": 4706
+    },
+    {
+      "epoch": 0.12927412045810333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487716363581354e-05,
+      "loss": 1.0796,
+      "step": 4707
+    },
+    {
+      "epoch": 0.12930158468594655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487435272229818e-05,
+      "loss": 1.2001,
+      "step": 4708
+    },
+    {
+      "epoch": 0.1293290489137898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487154105810017e-05,
+      "loss": 1.1113,
+      "step": 4709
+    },
+    {
+      "epoch": 0.12935651314163302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948687286432418e-05,
+      "loss": 1.1005,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12938397736947627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9486591547774533e-05,
+      "loss": 1.032,
+      "step": 4711
+    },
+    {
+      "epoch": 0.1294114415973195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9486310156163294e-05,
+      "loss": 0.9832,
+      "step": 4712
+    },
+    {
+      "epoch": 0.12943890582516274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94860286894927e-05,
+      "loss": 1.1608,
+      "step": 4713
+    },
+    {
+      "epoch": 0.12946637005300596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9485747147764966e-05,
+      "loss": 1.0964,
+      "step": 4714
+    },
+    {
+      "epoch": 0.1294938342808492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948546553098233e-05,
+      "loss": 0.9822,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12952129850869243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9485183839147018e-05,
+      "loss": 1.0995,
+      "step": 4716
+    },
+    {
+      "epoch": 0.12954876273653565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9484902072261254e-05,
+      "loss": 1.0463,
+      "step": 4717
+    },
+    {
+      "epoch": 0.1295762269643789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948462023032727e-05,
+      "loss": 1.0092,
+      "step": 4718
+    },
+    {
+      "epoch": 0.12960369119222212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94843383133473e-05,
+      "loss": 0.9696,
+      "step": 4719
+    },
+    {
+      "epoch": 0.12963115542006537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948405632132357e-05,
+      "loss": 1.08,
+      "step": 4720
+    },
+    {
+      "epoch": 0.1296586196479086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9483774254258312e-05,
+      "loss": 1.1443,
+      "step": 4721
+    },
+    {
+      "epoch": 0.12968608387575184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9483492112153758e-05,
+      "loss": 0.9532,
+      "step": 4722
+    },
+    {
+      "epoch": 0.12971354810359506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9483209895012144e-05,
+      "loss": 1.1463,
+      "step": 4723
+    },
+    {
+      "epoch": 0.1297410123314383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482927602835696e-05,
+      "loss": 0.972,
+      "step": 4724
+    },
+    {
+      "epoch": 0.12976847655928153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482645235626655e-05,
+      "loss": 1.0723,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12979594078712478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482362793387248e-05,
+      "loss": 1.1895,
+      "step": 4726
+    },
+    {
+      "epoch": 0.129823405014968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482080276119713e-05,
+      "loss": 1.0493,
+      "step": 4727
+    },
+    {
+      "epoch": 0.12985086924281125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948179768382629e-05,
+      "loss": 1.125,
+      "step": 4728
+    },
+    {
+      "epoch": 0.12987833347065447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9481515016509208e-05,
+      "loss": 0.9853,
+      "step": 4729
+    },
+    {
+      "epoch": 0.1299057976984977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9481232274170708e-05,
+      "loss": 0.9965,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12993326192634094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9480949456813026e-05,
+      "loss": 1.0947,
+      "step": 4731
+    },
+    {
+      "epoch": 0.12996072615418416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9480666564438397e-05,
+      "loss": 0.9999,
+      "step": 4732
+    },
+    {
+      "epoch": 0.1299881903820274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9480383597049063e-05,
+      "loss": 1.1047,
+      "step": 4733
+    },
+    {
+      "epoch": 0.13001565460987063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948010055464726e-05,
+      "loss": 1.047,
+      "step": 4734
+    },
+    {
+      "epoch": 0.13004311883771388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9479817437235232e-05,
+      "loss": 1.0538,
+      "step": 4735
+    },
+    {
+      "epoch": 0.1300705830655571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9479534244815214e-05,
+      "loss": 1.0007,
+      "step": 4736
+    },
+    {
+      "epoch": 0.13009804729340035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947925097738945e-05,
+      "loss": 1.1172,
+      "step": 4737
+    },
+    {
+      "epoch": 0.13012551152124358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947896763496018e-05,
+      "loss": 1.0783,
+      "step": 4738
+    },
+    {
+      "epoch": 0.13015297574908682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9478684217529645e-05,
+      "loss": 1.009,
+      "step": 4739
+    },
+    {
+      "epoch": 0.13018043997693005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947840072510009e-05,
+      "loss": 1.0114,
+      "step": 4740
+    },
+    {
+      "epoch": 0.1302079042047733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947811715767376e-05,
+      "loss": 1.0453,
+      "step": 4741
+    },
+    {
+      "epoch": 0.13023536843261652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947783351525289e-05,
+      "loss": 0.9026,
+      "step": 4742
+    },
+    {
+      "epoch": 0.13026283266045974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9477549797839735e-05,
+      "loss": 1.0479,
+      "step": 4743
+    },
+    {
+      "epoch": 0.130290296888303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947726600543653e-05,
+      "loss": 1.0275,
+      "step": 4744
+    },
+    {
+      "epoch": 0.1303177611161462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9476982138045527e-05,
+      "loss": 1.0954,
+      "step": 4745
+    },
+    {
+      "epoch": 0.13034522534398946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9476698195668972e-05,
+      "loss": 1.0122,
+      "step": 4746
+    },
+    {
+      "epoch": 0.13037268957183268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947641417830911e-05,
+      "loss": 1.0526,
+      "step": 4747
+    },
+    {
+      "epoch": 0.13040015379967593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9476130085968186e-05,
+      "loss": 1.1018,
+      "step": 4748
+    },
+    {
+      "epoch": 0.13042761802751915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9475845918648452e-05,
+      "loss": 1.0382,
+      "step": 4749
+    },
+    {
+      "epoch": 0.1304550822553624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9475561676352153e-05,
+      "loss": 0.9894,
+      "step": 4750
+    },
+    {
+      "epoch": 0.13048254648320562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9475277359081542e-05,
+      "loss": 1.1068,
+      "step": 4751
+    },
+    {
+      "epoch": 0.13051001071104887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9474992966838863e-05,
+      "loss": 1.106,
+      "step": 4752
+    },
+    {
+      "epoch": 0.1305374749388921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9474708499626372e-05,
+      "loss": 1.0573,
+      "step": 4753
+    },
+    {
+      "epoch": 0.13056493916673534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9474423957446315e-05,
+      "loss": 1.1395,
+      "step": 4754
+    },
+    {
+      "epoch": 0.13059240339457856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9474139340300945e-05,
+      "loss": 1.0496,
+      "step": 4755
+    },
+    {
+      "epoch": 0.1306198676224218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947385464819252e-05,
+      "loss": 1.0411,
+      "step": 4756
+    },
+    {
+      "epoch": 0.13064733185026503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9473569881123284e-05,
+      "loss": 0.9971,
+      "step": 4757
+    },
+    {
+      "epoch": 0.13067479607810825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947328503909549e-05,
+      "loss": 1.0912,
+      "step": 4758
+    },
+    {
+      "epoch": 0.1307022603059515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9473000122111395e-05,
+      "loss": 1.0554,
+      "step": 4759
+    },
+    {
+      "epoch": 0.13072972453379472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947271513017326e-05,
+      "loss": 1.0969,
+      "step": 4760
+    },
+    {
+      "epoch": 0.13075718876163797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947243006328333e-05,
+      "loss": 1.0869,
+      "step": 4761
+    },
+    {
+      "epoch": 0.1307846529894812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9472144921443863e-05,
+      "loss": 1.0221,
+      "step": 4762
+    },
+    {
+      "epoch": 0.13081211721732444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9471859704657117e-05,
+      "loss": 1.1698,
+      "step": 4763
+    },
+    {
+      "epoch": 0.13083958144516766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947157441292535e-05,
+      "loss": 1.0937,
+      "step": 4764
+    },
+    {
+      "epoch": 0.1308670456730109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9471289046250815e-05,
+      "loss": 1.0126,
+      "step": 4765
+    },
+    {
+      "epoch": 0.13089450990085413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9471003604635772e-05,
+      "loss": 1.029,
+      "step": 4766
+    },
+    {
+      "epoch": 0.13092197412869738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470718088082477e-05,
+      "loss": 1.061,
+      "step": 4767
+    },
+    {
+      "epoch": 0.1309494383565406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470432496593198e-05,
+      "loss": 1.1245,
+      "step": 4768
+    },
+    {
+      "epoch": 0.13097690258438385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470146830170186e-05,
+      "loss": 0.9995,
+      "step": 4769
+    },
+    {
+      "epoch": 0.13100436681222707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9469861088815705e-05,
+      "loss": 0.9892,
+      "step": 4770
+    },
+    {
+      "epoch": 0.1310318310400703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9469575272532013e-05,
+      "loss": 0.9922,
+      "step": 4771
+    },
+    {
+      "epoch": 0.13105929526791354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946928938132137e-05,
+      "loss": 1.0998,
+      "step": 4772
+    },
+    {
+      "epoch": 0.13108675949575677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946900341518605e-05,
+      "loss": 1.0417,
+      "step": 4773
+    },
+    {
+      "epoch": 0.13111422372360002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468717374128302e-05,
+      "loss": 1.0421,
+      "step": 4774
+    },
+    {
+      "epoch": 0.13114168795144324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468431258150394e-05,
+      "loss": 1.1021,
+      "step": 4775
+    },
+    {
+      "epoch": 0.13116915217928649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468145067254593e-05,
+      "loss": 1.0802,
+      "step": 4776
+    },
+    {
+      "epoch": 0.1311966164071297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467858801443156e-05,
+      "loss": 1.0326,
+      "step": 4777
+    },
+    {
+      "epoch": 0.13122408063497296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467572460718357e-05,
+      "loss": 1.0756,
+      "step": 4778
+    },
+    {
+      "epoch": 0.13125154486281618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467286045082455e-05,
+      "loss": 0.9513,
+      "step": 4779
+    },
+    {
+      "epoch": 0.13127900909065943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466999554537718e-05,
+      "loss": 1.1329,
+      "step": 4780
+    },
+    {
+      "epoch": 0.13130647331850265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466712989086414e-05,
+      "loss": 1.1024,
+      "step": 4781
+    },
+    {
+      "epoch": 0.1313339375463459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946642634873081e-05,
+      "loss": 1.1138,
+      "step": 4782
+    },
+    {
+      "epoch": 0.13136140177418912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466139633473173e-05,
+      "loss": 1.0715,
+      "step": 4783
+    },
+    {
+      "epoch": 0.13138886600203234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9465852843315772e-05,
+      "loss": 0.985,
+      "step": 4784
+    },
+    {
+      "epoch": 0.1314163302298756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946556597826088e-05,
+      "loss": 1.1209,
+      "step": 4785
+    },
+    {
+      "epoch": 0.1314437944577188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946527903831076e-05,
+      "loss": 0.9703,
+      "step": 4786
+    },
+    {
+      "epoch": 0.13147125868556206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464992023467687e-05,
+      "loss": 1.0387,
+      "step": 4787
+    },
+    {
+      "epoch": 0.13149872291340528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946470493373393e-05,
+      "loss": 1.0817,
+      "step": 4788
+    },
+    {
+      "epoch": 0.13152618714124853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464417769111767e-05,
+      "loss": 1.1335,
+      "step": 4789
+    },
+    {
+      "epoch": 0.13155365136909175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464130529603458e-05,
+      "loss": 0.9751,
+      "step": 4790
+    },
+    {
+      "epoch": 0.131581115596935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946384321521128e-05,
+      "loss": 1.0089,
+      "step": 4791
+    },
+    {
+      "epoch": 0.13160857982477822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9463555825937515e-05,
+      "loss": 1.0433,
+      "step": 4792
+    },
+    {
+      "epoch": 0.13163604405262147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9463268361784427e-05,
+      "loss": 1.1577,
+      "step": 4793
+    },
+    {
+      "epoch": 0.1316635082804647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9462980822754295e-05,
+      "loss": 0.8712,
+      "step": 4794
+    },
+    {
+      "epoch": 0.13169097250830794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9462693208849394e-05,
+      "loss": 1.0572,
+      "step": 4795
+    },
+    {
+      "epoch": 0.13171843673615116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9462405520071998e-05,
+      "loss": 1.1319,
+      "step": 4796
+    },
+    {
+      "epoch": 0.13174590096399438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946211775642438e-05,
+      "loss": 1.1273,
+      "step": 4797
+    },
+    {
+      "epoch": 0.13177336519183763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946182991790883e-05,
+      "loss": 1.0062,
+      "step": 4798
+    },
+    {
+      "epoch": 0.13180082941968085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9461542004527606e-05,
+      "loss": 1.0706,
+      "step": 4799
+    },
+    {
+      "epoch": 0.1318282936475241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9461254016283006e-05,
+      "loss": 1.0288,
+      "step": 4800
+    },
+    {
+      "epoch": 0.13185575787536732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9460965953177292e-05,
+      "loss": 1.0234,
+      "step": 4801
+    },
+    {
+      "epoch": 0.13188322210321057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9460677815212754e-05,
+      "loss": 1.1448,
+      "step": 4802
+    },
+    {
+      "epoch": 0.1319106863310538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9460389602391666e-05,
+      "loss": 0.988,
+      "step": 4803
+    },
+    {
+      "epoch": 0.13193815055889704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9460101314716314e-05,
+      "loss": 1.0704,
+      "step": 4804
+    },
+    {
+      "epoch": 0.13196561478674027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9459812952188976e-05,
+      "loss": 1.0339,
+      "step": 4805
+    },
+    {
+      "epoch": 0.13199307901458351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9459524514811932e-05,
+      "loss": 1.1328,
+      "step": 4806
+    },
+    {
+      "epoch": 0.13202054324242674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9459236002587466e-05,
+      "loss": 1.0981,
+      "step": 4807
+    },
+    {
+      "epoch": 0.13204800747026998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9458947415517863e-05,
+      "loss": 1.0729,
+      "step": 4808
+    },
+    {
+      "epoch": 0.1320754716981132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9458658753605402e-05,
+      "loss": 0.9626,
+      "step": 4809
+    },
+    {
+      "epoch": 0.13210293592595646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945837001685237e-05,
+      "loss": 1.0582,
+      "step": 4810
+    },
+    {
+      "epoch": 0.13213040015379968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9458081205261048e-05,
+      "loss": 1.0913,
+      "step": 4811
+    },
+    {
+      "epoch": 0.1321578643816429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457792318833725e-05,
+      "loss": 1.0615,
+      "step": 4812
+    },
+    {
+      "epoch": 0.13218532860948615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457503357572688e-05,
+      "loss": 1.0352,
+      "step": 4813
+    },
+    {
+      "epoch": 0.13221279283732937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457214321480217e-05,
+      "loss": 1.019,
+      "step": 4814
+    },
+    {
+      "epoch": 0.13224025706517262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9456925210558603e-05,
+      "loss": 0.9869,
+      "step": 4815
+    },
+    {
+      "epoch": 0.13226772129301584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9456636024810135e-05,
+      "loss": 1.1736,
+      "step": 4816
+    },
+    {
+      "epoch": 0.1322951855208591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94563467642371e-05,
+      "loss": 1.1121,
+      "step": 4817
+    },
+    {
+      "epoch": 0.1323226497487023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9456057428841787e-05,
+      "loss": 1.1137,
+      "step": 4818
+    },
+    {
+      "epoch": 0.13235011397654556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9455768018626485e-05,
+      "loss": 1.136,
+      "step": 4819
+    },
+    {
+      "epoch": 0.13237757820438878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945547853359348e-05,
+      "loss": 1.0813,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13240504243223203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9455188973745068e-05,
+      "loss": 1.0784,
+      "step": 4821
+    },
+    {
+      "epoch": 0.13243250666007525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945489933908354e-05,
+      "loss": 1.0618,
+      "step": 4822
+    },
+    {
+      "epoch": 0.1324599708879185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454609629611187e-05,
+      "loss": 1.0312,
+      "step": 4823
+    },
+    {
+      "epoch": 0.13248743511576172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454319845330296e-05,
+      "loss": 1.0618,
+      "step": 4824
+    },
+    {
+      "epoch": 0.13251489934360494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454029986243164e-05,
+      "loss": 1.0742,
+      "step": 4825
+    },
+    {
+      "epoch": 0.1325423635714482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945374005235209e-05,
+      "loss": 1.0368,
+      "step": 4826
+    },
+    {
+      "epoch": 0.1325698277992914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945345004365936e-05,
+      "loss": 1.1001,
+      "step": 4827
+    },
+    {
+      "epoch": 0.13259729202713466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945315996016727e-05,
+      "loss": 0.984,
+      "step": 4828
+    },
+    {
+      "epoch": 0.13262475625497788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945286980187812e-05,
+      "loss": 1.137,
+      "step": 4829
+    },
+    {
+      "epoch": 0.13265222048282113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9452579568794198e-05,
+      "loss": 0.9765,
+      "step": 4830
+    },
+    {
+      "epoch": 0.13267968471066435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9452289260917806e-05,
+      "loss": 1.0086,
+      "step": 4831
+    },
+    {
+      "epoch": 0.1327071489385076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451998878251238e-05,
+      "loss": 1.1174,
+      "step": 4832
+    },
+    {
+      "epoch": 0.13273461316635082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451708420796797e-05,
+      "loss": 0.9853,
+      "step": 4833
+    },
+    {
+      "epoch": 0.13276207739419407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451417888556777e-05,
+      "loss": 0.9885,
+      "step": 4834
+    },
+    {
+      "epoch": 0.1327895416220373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945112728153348e-05,
+      "loss": 1.018,
+      "step": 4835
+    },
+    {
+      "epoch": 0.13281700584988054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94508365997292e-05,
+      "loss": 1.0734,
+      "step": 4836
+    },
+    {
+      "epoch": 0.13284447007772376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9450545843146238e-05,
+      "loss": 1.1871,
+      "step": 4837
+    },
+    {
+      "epoch": 0.13287193430556699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94502550117869e-05,
+      "loss": 1.0573,
+      "step": 4838
+    },
+    {
+      "epoch": 0.13289939853341023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449964105653482e-05,
+      "loss": 1.088,
+      "step": 4839
+    },
+    {
+      "epoch": 0.13292686276125346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449673124748286e-05,
+      "loss": 0.92,
+      "step": 4840
+    },
+    {
+      "epoch": 0.1329543269890967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944938206907362e-05,
+      "loss": 1.0689,
+      "step": 4841
+    },
+    {
+      "epoch": 0.13298179121693993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449090938631777e-05,
+      "loss": 0.9636,
+      "step": 4842
+    },
+    {
+      "epoch": 0.13300925544478318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9448799733425073e-05,
+      "loss": 1.0085,
+      "step": 4843
+    },
+    {
+      "epoch": 0.1330367196726264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9448508453455802e-05,
+      "loss": 1.0563,
+      "step": 4844
+    },
+    {
+      "epoch": 0.13306418390046965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9448217098726272e-05,
+      "loss": 1.0475,
+      "step": 4845
+    },
+    {
+      "epoch": 0.13309164812831287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447925669238786e-05,
+      "loss": 1.0862,
+      "step": 4846
+    },
+    {
+      "epoch": 0.13311911235615612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447634164995658e-05,
+      "loss": 1.0738,
+      "step": 4847
+    },
+    {
+      "epoch": 0.13314657658399934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447342585999186e-05,
+      "loss": 1.0193,
+      "step": 4848
+    },
+    {
+      "epoch": 0.1331740408118426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447050932251677e-05,
+      "loss": 1.0637,
+      "step": 4849
+    },
+    {
+      "epoch": 0.1332015050396858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9446759203755445e-05,
+      "loss": 1.0897,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13322896926752906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9446467400512794e-05,
+      "loss": 1.0602,
+      "step": 4851
+    },
+    {
+      "epoch": 0.13325643349537228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9446175522526035e-05,
+      "loss": 1.0927,
+      "step": 4852
+    },
+    {
+      "epoch": 0.1332838977232155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944588356979747e-05,
+      "loss": 1.0379,
+      "step": 4853
+    },
+    {
+      "epoch": 0.13331136195105875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9445591542329424e-05,
+      "loss": 1.0906,
+      "step": 4854
+    },
+    {
+      "epoch": 0.13333882617890197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9445299440124193e-05,
+      "loss": 1.0508,
+      "step": 4855
+    },
+    {
+      "epoch": 0.13336629040674522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9445007263184097e-05,
+      "loss": 1.0408,
+      "step": 4856
+    },
+    {
+      "epoch": 0.13339375463458844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9444715011511443e-05,
+      "loss": 1.0269,
+      "step": 4857
+    },
+    {
+      "epoch": 0.1334212188624317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944442268510855e-05,
+      "loss": 1.1974,
+      "step": 4858
+    },
+    {
+      "epoch": 0.1334486830902749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944413028397772e-05,
+      "loss": 1.008,
+      "step": 4859
+    },
+    {
+      "epoch": 0.13347614731811816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9443837808121276e-05,
+      "loss": 0.9426,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13350361154596138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944354525754153e-05,
+      "loss": 1.0399,
+      "step": 4861
+    },
+    {
+      "epoch": 0.13353107577380463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9443252632240794e-05,
+      "loss": 1.1989,
+      "step": 4862
+    },
+    {
+      "epoch": 0.13355854000164785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442959932221383e-05,
+      "loss": 1.0528,
+      "step": 4863
+    },
+    {
+      "epoch": 0.1335860042294911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442667157485616e-05,
+      "loss": 1.0195,
+      "step": 4864
+    },
+    {
+      "epoch": 0.13361346845733432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944237430803581e-05,
+      "loss": 1.1148,
+      "step": 4865
+    },
+    {
+      "epoch": 0.13364093268517754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442081383874284e-05,
+      "loss": 1.0315,
+      "step": 4866
+    },
+    {
+      "epoch": 0.1336683969130208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944178838500335e-05,
+      "loss": 1.1222,
+      "step": 4867
+    },
+    {
+      "epoch": 0.13369586114086401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944149531142533e-05,
+      "loss": 1.083,
+      "step": 4868
+    },
+    {
+      "epoch": 0.13372332536870726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9441202163142542e-05,
+      "loss": 1.0036,
+      "step": 4869
+    },
+    {
+      "epoch": 0.13375078959655048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9440908940157302e-05,
+      "loss": 1.0366,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13377825382439373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9440615642471935e-05,
+      "loss": 1.0605,
+      "step": 4871
+    },
+    {
+      "epoch": 0.13380571805223695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944032227008876e-05,
+      "loss": 1.1162,
+      "step": 4872
+    },
+    {
+      "epoch": 0.1338331822800802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94400288230101e-05,
+      "loss": 1.0425,
+      "step": 4873
+    },
+    {
+      "epoch": 0.13386064650792343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9439735301238274e-05,
+      "loss": 0.9439,
+      "step": 4874
+    },
+    {
+      "epoch": 0.13388811073576667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9439441704775605e-05,
+      "loss": 0.9648,
+      "step": 4875
+    },
+    {
+      "epoch": 0.1339155749636099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9439148033624416e-05,
+      "loss": 1.0269,
+      "step": 4876
+    },
+    {
+      "epoch": 0.13394303919145314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9438854287787032e-05,
+      "loss": 1.1021,
+      "step": 4877
+    },
+    {
+      "epoch": 0.13397050341929637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943856046726578e-05,
+      "loss": 1.0123,
+      "step": 4878
+    },
+    {
+      "epoch": 0.1339979676471396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9438266572062976e-05,
+      "loss": 1.0017,
+      "step": 4879
+    },
+    {
+      "epoch": 0.13402543187498284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9437972602180956e-05,
+      "loss": 1.0343,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13405289610282606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943767855762204e-05,
+      "loss": 1.1875,
+      "step": 4881
+    },
+    {
+      "epoch": 0.1340803603306693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943738443838855e-05,
+      "loss": 1.0916,
+      "step": 4882
+    },
+    {
+      "epoch": 0.13410782455851253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9437090244482823e-05,
+      "loss": 0.9891,
+      "step": 4883
+    },
+    {
+      "epoch": 0.13413528878635578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436795975907184e-05,
+      "loss": 1.0684,
+      "step": 4884
+    },
+    {
+      "epoch": 0.134162753014199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436501632663957e-05,
+      "loss": 1.0321,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13419021724204225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436207214755474e-05,
+      "loss": 1.1243,
+      "step": 4886
+    },
+    {
+      "epoch": 0.13421768146988547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9435912722184066e-05,
+      "loss": 1.1744,
+      "step": 4887
+    },
+    {
+      "epoch": 0.13424514569772872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943561815495206e-05,
+      "loss": 1.0019,
+      "step": 4888
+    },
+    {
+      "epoch": 0.13427260992557194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9435323513061788e-05,
+      "loss": 1.0297,
+      "step": 4889
+    },
+    {
+      "epoch": 0.1343000741534152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943502879651558e-05,
+      "loss": 1.0539,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1343275383812584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434734005315774e-05,
+      "loss": 0.9424,
+      "step": 4891
+    },
+    {
+      "epoch": 0.13435500260910163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434439139464695e-05,
+      "loss": 1.0281,
+      "step": 4892
+    },
+    {
+      "epoch": 0.13438246683694488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434144198964676e-05,
+      "loss": 1.0489,
+      "step": 4893
+    },
+    {
+      "epoch": 0.1344099310647881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943384918381806e-05,
+      "loss": 1.0573,
+      "step": 4894
+    },
+    {
+      "epoch": 0.13443739529263135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9433554094027173e-05,
+      "loss": 1.0621,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13446485952047457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9433258929594347e-05,
+      "loss": 0.9376,
+      "step": 4896
+    },
+    {
+      "epoch": 0.13449232374831782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432963690521927e-05,
+      "loss": 1.0172,
+      "step": 4897
+    },
+    {
+      "epoch": 0.13451978797616104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432668376812243e-05,
+      "loss": 0.9885,
+      "step": 4898
+    },
+    {
+      "epoch": 0.1345472522040043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432372988467633e-05,
+      "loss": 1.0818,
+      "step": 4899
+    },
+    {
+      "epoch": 0.1345747164318475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432077525490435e-05,
+      "loss": 1.0987,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13460218065969076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9431781987882984e-05,
+      "loss": 1.125,
+      "step": 4901
+    },
+    {
+      "epoch": 0.13462964488753398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943148637564762e-05,
+      "loss": 1.0095,
+      "step": 4902
+    },
+    {
+      "epoch": 0.13465710911537723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9431190688786685e-05,
+      "loss": 1.1249,
+      "step": 4903
+    },
+    {
+      "epoch": 0.13468457334322045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9430894927302513e-05,
+      "loss": 1.1654,
+      "step": 4904
+    },
+    {
+      "epoch": 0.1347120375710637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943059909119745e-05,
+      "loss": 1.0954,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13473950179890692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9430303180473835e-05,
+      "loss": 0.9806,
+      "step": 4906
+    },
+    {
+      "epoch": 0.13476696602675015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9430007195134007e-05,
+      "loss": 0.9964,
+      "step": 4907
+    },
+    {
+      "epoch": 0.1347944302545934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9429711135180307e-05,
+      "loss": 1.0435,
+      "step": 4908
+    },
+    {
+      "epoch": 0.13482189448243662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9429415000615083e-05,
+      "loss": 1.1097,
+      "step": 4909
+    },
+    {
+      "epoch": 0.13484935871027987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942911879144067e-05,
+      "loss": 0.9653,
+      "step": 4910
+    },
+    {
+      "epoch": 0.1348768229381231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942882250765942e-05,
+      "loss": 0.9642,
+      "step": 4911
+    },
+    {
+      "epoch": 0.13490428716596634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9428526149273673e-05,
+      "loss": 1.1712,
+      "step": 4912
+    },
+    {
+      "epoch": 0.13493175139380956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942822971628578e-05,
+      "loss": 1.0529,
+      "step": 4913
+    },
+    {
+      "epoch": 0.1349592156216528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9427933208698074e-05,
+      "loss": 1.0837,
+      "step": 4914
+    },
+    {
+      "epoch": 0.13498667984949603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9427636626512912e-05,
+      "loss": 1.0777,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13501414407733928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9427339969732635e-05,
+      "loss": 1.0014,
+      "step": 4916
+    },
+    {
+      "epoch": 0.1350416083051825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9427043238359594e-05,
+      "loss": 1.0436,
+      "step": 4917
+    },
+    {
+      "epoch": 0.13506907253302575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9426746432396134e-05,
+      "loss": 1.0381,
+      "step": 4918
+    },
+    {
+      "epoch": 0.13509653676086897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9426449551844606e-05,
+      "loss": 1.1585,
+      "step": 4919
+    },
+    {
+      "epoch": 0.1351240009887122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9426152596707358e-05,
+      "loss": 1.0027,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13515146521655544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425855566986738e-05,
+      "loss": 1.0452,
+      "step": 4921
+    },
+    {
+      "epoch": 0.13517892944439866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425558462685097e-05,
+      "loss": 0.9985,
+      "step": 4922
+    },
+    {
+      "epoch": 0.1352063936722419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425261283804788e-05,
+      "loss": 1.005,
+      "step": 4923
+    },
+    {
+      "epoch": 0.13523385790008513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942496403034816e-05,
+      "loss": 1.0174,
+      "step": 4924
+    },
+    {
+      "epoch": 0.13526132212792838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9424666702317566e-05,
+      "loss": 1.0381,
+      "step": 4925
+    },
+    {
+      "epoch": 0.1352887863557716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942436929971536e-05,
+      "loss": 1.0318,
+      "step": 4926
+    },
+    {
+      "epoch": 0.13531625058361485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942407182254389e-05,
+      "loss": 1.05,
+      "step": 4927
+    },
+    {
+      "epoch": 0.13534371481145807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9423774270805514e-05,
+      "loss": 1.1177,
+      "step": 4928
+    },
+    {
+      "epoch": 0.13537117903930132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9423476644502588e-05,
+      "loss": 0.948,
+      "step": 4929
+    },
+    {
+      "epoch": 0.13539864326714454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9423178943637465e-05,
+      "loss": 1.0695,
+      "step": 4930
+    },
+    {
+      "epoch": 0.1354261074949878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9422881168212498e-05,
+      "loss": 0.953,
+      "step": 4931
+    },
+    {
+      "epoch": 0.135453571722831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942258331823004e-05,
+      "loss": 0.9717,
+      "step": 4932
+    },
+    {
+      "epoch": 0.13548103595067423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9422285393692458e-05,
+      "loss": 1.0271,
+      "step": 4933
+    },
+    {
+      "epoch": 0.13550850017851748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9421987394602105e-05,
+      "loss": 0.8815,
+      "step": 4934
+    },
+    {
+      "epoch": 0.1355359644063607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942168932096134e-05,
+      "loss": 1.0738,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13556342863420395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9421391172772518e-05,
+      "loss": 0.9892,
+      "step": 4936
+    },
+    {
+      "epoch": 0.13559089286204717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9421092950037996e-05,
+      "loss": 1.1127,
+      "step": 4937
+    },
+    {
+      "epoch": 0.13561835708989042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942079465276014e-05,
+      "loss": 1.0538,
+      "step": 4938
+    },
+    {
+      "epoch": 0.13564582131773364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9420496280941307e-05,
+      "loss": 1.0395,
+      "step": 4939
+    },
+    {
+      "epoch": 0.1356732855455769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9420197834583856e-05,
+      "loss": 1.1274,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13570074977342012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419899313690152e-05,
+      "loss": 1.0719,
+      "step": 4941
+    },
+    {
+      "epoch": 0.13572821400126336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419600718262556e-05,
+      "loss": 1.0792,
+      "step": 4942
+    },
+    {
+      "epoch": 0.13575567822910659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419302048303434e-05,
+      "loss": 1.0195,
+      "step": 4943
+    },
+    {
+      "epoch": 0.13578314245694983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941900330381514e-05,
+      "loss": 0.9547,
+      "step": 4944
+    },
+    {
+      "epoch": 0.13581060668479306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9418704484800044e-05,
+      "loss": 1.1274,
+      "step": 4945
+    },
+    {
+      "epoch": 0.1358380709126363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941840559126051e-05,
+      "loss": 1.0256,
+      "step": 4946
+    },
+    {
+      "epoch": 0.13586553514047953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9418106623198903e-05,
+      "loss": 1.0114,
+      "step": 4947
+    },
+    {
+      "epoch": 0.13589299936832275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941780758061759e-05,
+      "loss": 1.1376,
+      "step": 4948
+    },
+    {
+      "epoch": 0.135920463596166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9417508463518936e-05,
+      "loss": 1.1031,
+      "step": 4949
+    },
+    {
+      "epoch": 0.13594792782400922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9417209271905303e-05,
+      "loss": 1.0792,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13597539205185247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9416910005779066e-05,
+      "loss": 1.1143,
+      "step": 4951
+    },
+    {
+      "epoch": 0.1360028562796957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941661066514259e-05,
+      "loss": 1.0767,
+      "step": 4952
+    },
+    {
+      "epoch": 0.13603032050753894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941631124999824e-05,
+      "loss": 1.0013,
+      "step": 4953
+    },
+    {
+      "epoch": 0.13605778473538216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9416011760348387e-05,
+      "loss": 1.101,
+      "step": 4954
+    },
+    {
+      "epoch": 0.1360852489632254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9415712196195403e-05,
+      "loss": 0.9135,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13611271319106863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9415412557541657e-05,
+      "loss": 1.1429,
+      "step": 4956
+    },
+    {
+      "epoch": 0.13614017741891188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941511284438952e-05,
+      "loss": 0.9999,
+      "step": 4957
+    },
+    {
+      "epoch": 0.1361676416467551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9414813056741363e-05,
+      "loss": 1.0665,
+      "step": 4958
+    },
+    {
+      "epoch": 0.13619510587459835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941451319459956e-05,
+      "loss": 0.9776,
+      "step": 4959
+    },
+    {
+      "epoch": 0.13622257010244157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9414213257966476e-05,
+      "loss": 1.0511,
+      "step": 4960
+    },
+    {
+      "epoch": 0.1362500343302848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9413913246844496e-05,
+      "loss": 1.0319,
+      "step": 4961
+    },
+    {
+      "epoch": 0.13627749855812804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9413613161235983e-05,
+      "loss": 0.993,
+      "step": 4962
+    },
+    {
+      "epoch": 0.13630496278597126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941331300114332e-05,
+      "loss": 1.1208,
+      "step": 4963
+    },
+    {
+      "epoch": 0.1363324270138145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9413012766568874e-05,
+      "loss": 0.9791,
+      "step": 4964
+    },
+    {
+      "epoch": 0.13635989124165773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941271245751503e-05,
+      "loss": 1.036,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13638735546950098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9412412073984154e-05,
+      "loss": 1.1038,
+      "step": 4966
+    },
+    {
+      "epoch": 0.1364148196973442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9412111615978633e-05,
+      "loss": 0.9858,
+      "step": 4967
+    },
+    {
+      "epoch": 0.13644228392518745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9411811083500836e-05,
+      "loss": 1.0035,
+      "step": 4968
+    },
+    {
+      "epoch": 0.13646974815303067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9411510476553143e-05,
+      "loss": 1.0406,
+      "step": 4969
+    },
+    {
+      "epoch": 0.13649721238087392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941120979513793e-05,
+      "loss": 1.0918,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13652467660871714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9410909039257587e-05,
+      "loss": 1.1082,
+      "step": 4971
+    },
+    {
+      "epoch": 0.1365521408365604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941060820891448e-05,
+      "loss": 1.0284,
+      "step": 4972
+    },
+    {
+      "epoch": 0.13657960506440361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9410307304111003e-05,
+      "loss": 1.0251,
+      "step": 4973
+    },
+    {
+      "epoch": 0.13660706929224684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9410006324849524e-05,
+      "loss": 1.0603,
+      "step": 4974
+    },
+    {
+      "epoch": 0.13663453352009008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940970527113243e-05,
+      "loss": 0.9514,
+      "step": 4975
+    },
+    {
+      "epoch": 0.1366619977479333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9409404142962104e-05,
+      "loss": 0.977,
+      "step": 4976
+    },
+    {
+      "epoch": 0.13668946197577655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940910294034093e-05,
+      "loss": 1.1447,
+      "step": 4977
+    },
+    {
+      "epoch": 0.13671692620361978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9408801663271283e-05,
+      "loss": 1.0204,
+      "step": 4978
+    },
+    {
+      "epoch": 0.13674439043146303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9408500311755557e-05,
+      "loss": 0.9765,
+      "step": 4979
+    },
+    {
+      "epoch": 0.13677185465930625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940819888579613e-05,
+      "loss": 0.8552,
+      "step": 4980
+    },
+    {
+      "epoch": 0.1367993188871495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9407897385395393e-05,
+      "loss": 0.9884,
+      "step": 4981
+    },
+    {
+      "epoch": 0.13682678311499272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940759581055572e-05,
+      "loss": 0.9841,
+      "step": 4982
+    },
+    {
+      "epoch": 0.13685424734283597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940729416127951e-05,
+      "loss": 1.1321,
+      "step": 4983
+    },
+    {
+      "epoch": 0.1368817115706792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9406992437569144e-05,
+      "loss": 1.0921,
+      "step": 4984
+    },
+    {
+      "epoch": 0.13690917579852244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940669063942701e-05,
+      "loss": 0.9974,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13693664002636566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9406388766855493e-05,
+      "loss": 1.0182,
+      "step": 4986
+    },
+    {
+      "epoch": 0.13696410425420888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9406086819856988e-05,
+      "loss": 1.0679,
+      "step": 4987
+    },
+    {
+      "epoch": 0.13699156848205213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405784798433882e-05,
+      "loss": 1.0355,
+      "step": 4988
+    },
+    {
+      "epoch": 0.13701903270989535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940548270258856e-05,
+      "loss": 1.1936,
+      "step": 4989
+    },
+    {
+      "epoch": 0.1370464969377386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405180532323416e-05,
+      "loss": 0.9853,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13707396116558182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9404878287640843e-05,
+      "loss": 1.0,
+      "step": 4991
+    },
+    {
+      "epoch": 0.13710142539342507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9404575968543224e-05,
+      "loss": 0.9695,
+      "step": 4992
+    },
+    {
+      "epoch": 0.1371288896212683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940427357503296e-05,
+      "loss": 1.0661,
+      "step": 4993
+    },
+    {
+      "epoch": 0.13715635384911154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9403971107112442e-05,
+      "loss": 1.0139,
+      "step": 4994
+    },
+    {
+      "epoch": 0.13718381807695476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940366856478406e-05,
+      "loss": 0.9589,
+      "step": 4995
+    },
+    {
+      "epoch": 0.137211282304798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9403365948050214e-05,
+      "loss": 1.1179,
+      "step": 4996
+    },
+    {
+      "epoch": 0.13723874653264123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9403063256913288e-05,
+      "loss": 0.8758,
+      "step": 4997
+    },
+    {
+      "epoch": 0.13726621076048448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940276049137569e-05,
+      "loss": 1.0808,
+      "step": 4998
+    },
+    {
+      "epoch": 0.1372936749883277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9402457651439804e-05,
+      "loss": 1.1429,
+      "step": 4999
+    },
+    {
+      "epoch": 0.13732113921617095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940215473710803e-05,
+      "loss": 1.0448,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13734860344401417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9401851748382764e-05,
+      "loss": 0.9867,
+      "step": 5001
+    },
+    {
+      "epoch": 0.1373760676718574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940154868526641e-05,
+      "loss": 1.0514,
+      "step": 5002
+    },
+    {
+      "epoch": 0.13740353189970064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940124554776136e-05,
+      "loss": 1.0576,
+      "step": 5003
+    },
+    {
+      "epoch": 0.13743099612754386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940094233587001e-05,
+      "loss": 1.0163,
+      "step": 5004
+    },
+    {
+      "epoch": 0.1374584603553871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9400639049594767e-05,
+      "loss": 0.9982,
+      "step": 5005
+    },
+    {
+      "epoch": 0.13748592458323033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9400335688938026e-05,
+      "loss": 1.0719,
+      "step": 5006
+    },
+    {
+      "epoch": 0.13751338881107358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9400032253902186e-05,
+      "loss": 1.0056,
+      "step": 5007
+    },
+    {
+      "epoch": 0.1375408530389168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939972874448965e-05,
+      "loss": 0.8883,
+      "step": 5008
+    },
+    {
+      "epoch": 0.13756831726676005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939942516070282e-05,
+      "loss": 1.0057,
+      "step": 5009
+    },
+    {
+      "epoch": 0.13759578149460328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399121502544095e-05,
+      "loss": 1.0143,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13762324572244652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939881777001588e-05,
+      "loss": 0.8642,
+      "step": 5011
+    },
+    {
+      "epoch": 0.13765070995028975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939851396312058e-05,
+      "loss": 1.1205,
+      "step": 5012
+    },
+    {
+      "epoch": 0.137678174178133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9398210081860595e-05,
+      "loss": 0.9903,
+      "step": 5013
+    },
+    {
+      "epoch": 0.13770563840597622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9397906126238334e-05,
+      "loss": 1.0315,
+      "step": 5014
+    },
+    {
+      "epoch": 0.13773310263381944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9397602096256195e-05,
+      "loss": 1.0632,
+      "step": 5015
+    },
+    {
+      "epoch": 0.1377605668616627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939729799191659e-05,
+      "loss": 1.047,
+      "step": 5016
+    },
+    {
+      "epoch": 0.1377880310895059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396993813221928e-05,
+      "loss": 1.1117,
+      "step": 5017
+    },
+    {
+      "epoch": 0.13781549531734916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396689560174604e-05,
+      "loss": 1.0124,
+      "step": 5018
+    },
+    {
+      "epoch": 0.13784295954519238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396385232777036e-05,
+      "loss": 1.1006,
+      "step": 5019
+    },
+    {
+      "epoch": 0.13787042377303563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396080831031625e-05,
+      "loss": 0.8985,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13789788800087885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9395776354940787e-05,
+      "loss": 1.0852,
+      "step": 5021
+    },
+    {
+      "epoch": 0.1379253522287221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9395471804506924e-05,
+      "loss": 0.9718,
+      "step": 5022
+    },
+    {
+      "epoch": 0.13795281645656532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939516717973245e-05,
+      "loss": 1.0988,
+      "step": 5023
+    },
+    {
+      "epoch": 0.13798028068440857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394862480619775e-05,
+      "loss": 1.0552,
+      "step": 5024
+    },
+    {
+      "epoch": 0.1380077449122518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394557707171305e-05,
+      "loss": 1.0927,
+      "step": 5025
+    },
+    {
+      "epoch": 0.13803520914009504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394252859389454e-05,
+      "loss": 0.9637,
+      "step": 5026
+    },
+    {
+      "epoch": 0.13806267336793826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393947937276643e-05,
+      "loss": 1.01,
+      "step": 5027
+    },
+    {
+      "epoch": 0.13809013759578148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939364294083527e-05,
+      "loss": 1.0595,
+      "step": 5028
+    },
+    {
+      "epoch": 0.13811760182362473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393337870067758e-05,
+      "loss": 1.0586,
+      "step": 5029
+    },
+    {
+      "epoch": 0.13814506605146795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939303272497652e-05,
+      "loss": 0.8774,
+      "step": 5030
+    },
+    {
+      "epoch": 0.1381725302793112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9392727505563966e-05,
+      "loss": 1.0296,
+      "step": 5031
+    },
+    {
+      "epoch": 0.13819999450715442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9392422211832516e-05,
+      "loss": 0.9555,
+      "step": 5032
+    },
+    {
+      "epoch": 0.13822745873499767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939211684378458e-05,
+      "loss": 1.0045,
+      "step": 5033
+    },
+    {
+      "epoch": 0.1382549229628409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939181140142258e-05,
+      "loss": 1.2201,
+      "step": 5034
+    },
+    {
+      "epoch": 0.13828238719068414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9391505884748928e-05,
+      "loss": 1.1107,
+      "step": 5035
+    },
+    {
+      "epoch": 0.13830985141852736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9391200293766047e-05,
+      "loss": 0.972,
+      "step": 5036
+    },
+    {
+      "epoch": 0.1383373156463706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9390894628476347e-05,
+      "loss": 0.9688,
+      "step": 5037
+    },
+    {
+      "epoch": 0.13836477987421383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9390588888882254e-05,
+      "loss": 0.9771,
+      "step": 5038
+    },
+    {
+      "epoch": 0.13839224410205708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9390283074986183e-05,
+      "loss": 0.9652,
+      "step": 5039
+    },
+    {
+      "epoch": 0.1384197083299003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9389977186790558e-05,
+      "loss": 1.1281,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13844717255774353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938967122429779e-05,
+      "loss": 1.1887,
+      "step": 5041
+    },
+    {
+      "epoch": 0.13847463678558677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938936518751031e-05,
+      "loss": 0.9462,
+      "step": 5042
+    },
+    {
+      "epoch": 0.13850210101343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9389059076430536e-05,
+      "loss": 1.175,
+      "step": 5043
+    },
+    {
+      "epoch": 0.13852956524127324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9388752891060888e-05,
+      "loss": 0.9623,
+      "step": 5044
+    },
+    {
+      "epoch": 0.13855702946911647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938844663140379e-05,
+      "loss": 1.1286,
+      "step": 5045
+    },
+    {
+      "epoch": 0.13858449369695972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9388140297461665e-05,
+      "loss": 1.1108,
+      "step": 5046
+    },
+    {
+      "epoch": 0.13861195792480294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938783388923694e-05,
+      "loss": 1.0387,
+      "step": 5047
+    },
+    {
+      "epoch": 0.13863942215264619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9387527406732035e-05,
+      "loss": 1.0928,
+      "step": 5048
+    },
+    {
+      "epoch": 0.1386668863804894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9387220849949378e-05,
+      "loss": 1.1241,
+      "step": 5049
+    },
+    {
+      "epoch": 0.13869435060833266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9386914218891393e-05,
+      "loss": 1.1083,
+      "step": 5050
+    },
+    {
+      "epoch": 0.13872181483617588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9386607513560505e-05,
+      "loss": 1.2162,
+      "step": 5051
+    },
+    {
+      "epoch": 0.13874927906401913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9386300733959146e-05,
+      "loss": 1.1519,
+      "step": 5052
+    },
+    {
+      "epoch": 0.13877674329186235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385993880089734e-05,
+      "loss": 0.9806,
+      "step": 5053
+    },
+    {
+      "epoch": 0.1388042075197056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385686951954706e-05,
+      "loss": 0.9878,
+      "step": 5054
+    },
+    {
+      "epoch": 0.13883167174754882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938537994955649e-05,
+      "loss": 1.0073,
+      "step": 5055
+    },
+    {
+      "epoch": 0.13885913597539204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385072872897507e-05,
+      "loss": 1.0845,
+      "step": 5056
+    },
+    {
+      "epoch": 0.1388866002032353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9384765721980193e-05,
+      "loss": 1.1386,
+      "step": 5057
+    },
+    {
+      "epoch": 0.1389140644310785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938445849680698e-05,
+      "loss": 1.1503,
+      "step": 5058
+    },
+    {
+      "epoch": 0.13894152865892176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9384151197380296e-05,
+      "loss": 1.0283,
+      "step": 5059
+    },
+    {
+      "epoch": 0.13896899288676498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938384382370257e-05,
+      "loss": 0.9779,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13899645711460823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938353637577624e-05,
+      "loss": 0.9638,
+      "step": 5061
+    },
+    {
+      "epoch": 0.13902392134245145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9383228853603732e-05,
+      "loss": 1.0153,
+      "step": 5062
+    },
+    {
+      "epoch": 0.1390513855702947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9382921257187485e-05,
+      "loss": 1.1368,
+      "step": 5063
+    },
+    {
+      "epoch": 0.13907884979813792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938261358652993e-05,
+      "loss": 1.0359,
+      "step": 5064
+    },
+    {
+      "epoch": 0.13910631402598117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93823058416335e-05,
+      "loss": 1.1027,
+      "step": 5065
+    },
+    {
+      "epoch": 0.1391337782538244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9381998022500634e-05,
+      "loss": 1.035,
+      "step": 5066
+    },
+    {
+      "epoch": 0.13916124248166764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9381690129133765e-05,
+      "loss": 1.0064,
+      "step": 5067
+    },
+    {
+      "epoch": 0.13918870670951086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938138216153533e-05,
+      "loss": 1.054,
+      "step": 5068
+    },
+    {
+      "epoch": 0.13921617093735408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9381074119707763e-05,
+      "loss": 0.9613,
+      "step": 5069
+    },
+    {
+      "epoch": 0.13924363516519733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380766003653507e-05,
+      "loss": 1.0201,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13927109939304055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938045781337499e-05,
+      "loss": 1.114,
+      "step": 5071
+    },
+    {
+      "epoch": 0.1392985636208838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380149548874665e-05,
+      "loss": 1.0806,
+      "step": 5072
+    },
+    {
+      "epoch": 0.13932602784872702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937984121015496e-05,
+      "loss": 1.0928,
+      "step": 5073
+    },
+    {
+      "epoch": 0.13935349207657027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937953279721832e-05,
+      "loss": 1.053,
+      "step": 5074
+    },
+    {
+      "epoch": 0.1393809563044135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937922431006718e-05,
+      "loss": 1.0638,
+      "step": 5075
+    },
+    {
+      "epoch": 0.13940842053225674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9378915748703985e-05,
+      "loss": 1.0899,
+      "step": 5076
+    },
+    {
+      "epoch": 0.13943588476009997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9378607113131178e-05,
+      "loss": 0.9792,
+      "step": 5077
+    },
+    {
+      "epoch": 0.13946334898794321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9378298403351192e-05,
+      "loss": 1.0585,
+      "step": 5078
+    },
+    {
+      "epoch": 0.13949081321578644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9377989619366483e-05,
+      "loss": 1.102,
+      "step": 5079
+    },
+    {
+      "epoch": 0.13951827744362968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9377680761179485e-05,
+      "loss": 1.0833,
+      "step": 5080
+    },
+    {
+      "epoch": 0.1395457416714729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9377371828792644e-05,
+      "loss": 1.0825,
+      "step": 5081
+    },
+    {
+      "epoch": 0.13957320589931613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9377062822208404e-05,
+      "loss": 1.055,
+      "step": 5082
+    },
+    {
+      "epoch": 0.13960067012715938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937675374142921e-05,
+      "loss": 1.0042,
+      "step": 5083
+    },
+    {
+      "epoch": 0.1396281343550026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9376444586457515e-05,
+      "loss": 0.9723,
+      "step": 5084
+    },
+    {
+      "epoch": 0.13965559858284585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937613535729575e-05,
+      "loss": 1.1298,
+      "step": 5085
+    },
+    {
+      "epoch": 0.13968306281068907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9375826053946376e-05,
+      "loss": 1.0972,
+      "step": 5086
+    },
+    {
+      "epoch": 0.13971052703853232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937551667641183e-05,
+      "loss": 1.0106,
+      "step": 5087
+    },
+    {
+      "epoch": 0.13973799126637554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9375207224694564e-05,
+      "loss": 0.9667,
+      "step": 5088
+    },
+    {
+      "epoch": 0.1397654554942188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937489769879703e-05,
+      "loss": 1.0355,
+      "step": 5089
+    },
+    {
+      "epoch": 0.139792919722062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9374588098721676e-05,
+      "loss": 0.893,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13982038394990526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9374278424470946e-05,
+      "loss": 1.1495,
+      "step": 5091
+    },
+    {
+      "epoch": 0.13984784817774848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373968676047292e-05,
+      "loss": 1.082,
+      "step": 5092
+    },
+    {
+      "epoch": 0.13987531240559173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937365885345317e-05,
+      "loss": 1.0378,
+      "step": 5093
+    },
+    {
+      "epoch": 0.13990277663343495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373348956691026e-05,
+      "loss": 0.9831,
+      "step": 5094
+    },
+    {
+      "epoch": 0.1399302408612782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373038985763317e-05,
+      "loss": 0.9916,
+      "step": 5095
+    },
+    {
+      "epoch": 0.13995770508912142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937272894067249e-05,
+      "loss": 1.0604,
+      "step": 5096
+    },
+    {
+      "epoch": 0.13998516931696464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9372418821421002e-05,
+      "loss": 1.0688,
+      "step": 5097
+    },
+    {
+      "epoch": 0.1400126335448079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9372108628011307e-05,
+      "loss": 0.9787,
+      "step": 5098
+    },
+    {
+      "epoch": 0.1400400977726511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9371798360445856e-05,
+      "loss": 0.9762,
+      "step": 5099
+    },
+    {
+      "epoch": 0.14006756200049436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937148801872711e-05,
+      "loss": 0.9001,
+      "step": 5100
+    },
+    {
+      "epoch": 0.14009502622833758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9371177602857515e-05,
+      "loss": 1.0708,
+      "step": 5101
+    },
+    {
+      "epoch": 0.14012249045618083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9370867112839535e-05,
+      "loss": 1.0584,
+      "step": 5102
+    },
+    {
+      "epoch": 0.14014995468402405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9370556548675624e-05,
+      "loss": 1.0347,
+      "step": 5103
+    },
+    {
+      "epoch": 0.1401774189118673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9370245910368243e-05,
+      "loss": 1.1772,
+      "step": 5104
+    },
+    {
+      "epoch": 0.14020488313971052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369935197919844e-05,
+      "loss": 1.0989,
+      "step": 5105
+    },
+    {
+      "epoch": 0.14023234736755377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369624411332888e-05,
+      "loss": 0.8866,
+      "step": 5106
+    },
+    {
+      "epoch": 0.140259811595397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369313550609834e-05,
+      "loss": 1.028,
+      "step": 5107
+    },
+    {
+      "epoch": 0.14028727582324024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369002615753145e-05,
+      "loss": 1.0157,
+      "step": 5108
+    },
+    {
+      "epoch": 0.14031474005108346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9368691606765274e-05,
+      "loss": 0.9886,
+      "step": 5109
+    },
+    {
+      "epoch": 0.14034220427892669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9368380523648688e-05,
+      "loss": 0.9978,
+      "step": 5110
+    },
+    {
+      "epoch": 0.14036966850676993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9368069366405847e-05,
+      "loss": 0.992,
+      "step": 5111
+    },
+    {
+      "epoch": 0.14039713273461316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9367758135039213e-05,
+      "loss": 1.0126,
+      "step": 5112
+    },
+    {
+      "epoch": 0.1404245969624564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9367446829551244e-05,
+      "loss": 0.9392,
+      "step": 5113
+    },
+    {
+      "epoch": 0.14045206119029963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936713544994441e-05,
+      "loss": 0.9966,
+      "step": 5114
+    },
+    {
+      "epoch": 0.14047952541814288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936682399622117e-05,
+      "loss": 1.0109,
+      "step": 5115
+    },
+    {
+      "epoch": 0.1405069896459861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936651246838399e-05,
+      "loss": 1.0362,
+      "step": 5116
+    },
+    {
+      "epoch": 0.14053445387382935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936620086643534e-05,
+      "loss": 1.0329,
+      "step": 5117
+    },
+    {
+      "epoch": 0.14056191810167257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9365889190377672e-05,
+      "loss": 1.1109,
+      "step": 5118
+    },
+    {
+      "epoch": 0.14058938232951582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936557744021347e-05,
+      "loss": 1.0779,
+      "step": 5119
+    },
+    {
+      "epoch": 0.14061684655735904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9365265615945185e-05,
+      "loss": 0.9546,
+      "step": 5120
+    },
+    {
+      "epoch": 0.1406443107852023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936495371757529e-05,
+      "loss": 1.0638,
+      "step": 5121
+    },
+    {
+      "epoch": 0.1406717750130455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936464174510626e-05,
+      "loss": 0.9637,
+      "step": 5122
+    },
+    {
+      "epoch": 0.14069923924088873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9364329698540553e-05,
+      "loss": 1.0178,
+      "step": 5123
+    },
+    {
+      "epoch": 0.14072670346873198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936401757788064e-05,
+      "loss": 0.9818,
+      "step": 5124
+    },
+    {
+      "epoch": 0.1407541676965752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9363705383129e-05,
+      "loss": 0.9872,
+      "step": 5125
+    },
+    {
+      "epoch": 0.14078163192441845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9363393114288088e-05,
+      "loss": 1.0365,
+      "step": 5126
+    },
+    {
+      "epoch": 0.14080909615226167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9363080771360385e-05,
+      "loss": 1.1971,
+      "step": 5127
+    },
+    {
+      "epoch": 0.14083656038010492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9362768354348362e-05,
+      "loss": 0.9908,
+      "step": 5128
+    },
+    {
+      "epoch": 0.14086402460794814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936245586325449e-05,
+      "loss": 0.9659,
+      "step": 5129
+    },
+    {
+      "epoch": 0.1408914888357914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9362143298081238e-05,
+      "loss": 1.1292,
+      "step": 5130
+    },
+    {
+      "epoch": 0.1409189530636346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936183065883108e-05,
+      "loss": 1.0463,
+      "step": 5131
+    },
+    {
+      "epoch": 0.14094641729147786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936151794550649e-05,
+      "loss": 1.1282,
+      "step": 5132
+    },
+    {
+      "epoch": 0.14097388151932108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936120515810995e-05,
+      "loss": 1.0585,
+      "step": 5133
+    },
+    {
+      "epoch": 0.14100134574716433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9360892296643926e-05,
+      "loss": 0.9356,
+      "step": 5134
+    },
+    {
+      "epoch": 0.14102880997500755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9360579361110896e-05,
+      "loss": 1.0201,
+      "step": 5135
+    },
+    {
+      "epoch": 0.14105627420285077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9360266351513335e-05,
+      "loss": 1.0253,
+      "step": 5136
+    },
+    {
+      "epoch": 0.14108373843069402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935995326785372e-05,
+      "loss": 0.9728,
+      "step": 5137
+    },
+    {
+      "epoch": 0.14111120265853724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359640110134534e-05,
+      "loss": 1.0345,
+      "step": 5138
+    },
+    {
+      "epoch": 0.1411386668863805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359326878358246e-05,
+      "loss": 1.0027,
+      "step": 5139
+    },
+    {
+      "epoch": 0.14116613111422371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359013572527337e-05,
+      "loss": 1.08,
+      "step": 5140
+    },
+    {
+      "epoch": 0.14119359534206696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935870019264429e-05,
+      "loss": 1.0272,
+      "step": 5141
+    },
+    {
+      "epoch": 0.14122105956991018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935838673871158e-05,
+      "loss": 1.0526,
+      "step": 5142
+    },
+    {
+      "epoch": 0.14124852379775343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935807321073169e-05,
+      "loss": 1.0436,
+      "step": 5143
+    },
+    {
+      "epoch": 0.14127598802559665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93577596087071e-05,
+      "loss": 1.0016,
+      "step": 5144
+    },
+    {
+      "epoch": 0.1413034522534399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9357445932640292e-05,
+      "loss": 1.0413,
+      "step": 5145
+    },
+    {
+      "epoch": 0.14133091648128313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9357132182533746e-05,
+      "loss": 1.072,
+      "step": 5146
+    },
+    {
+      "epoch": 0.14135838070912637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9356818358389943e-05,
+      "loss": 1.0215,
+      "step": 5147
+    },
+    {
+      "epoch": 0.1413858449369696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9356504460211375e-05,
+      "loss": 0.9266,
+      "step": 5148
+    },
+    {
+      "epoch": 0.14141330916481284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935619048800051e-05,
+      "loss": 1.0878,
+      "step": 5149
+    },
+    {
+      "epoch": 0.14144077339265607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9355876441759852e-05,
+      "loss": 0.962,
+      "step": 5150
+    },
+    {
+      "epoch": 0.1414682376204993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9355562321491868e-05,
+      "loss": 1.0962,
+      "step": 5151
+    },
+    {
+      "epoch": 0.14149570184834254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9355248127199055e-05,
+      "loss": 1.0789,
+      "step": 5152
+    },
+    {
+      "epoch": 0.14152316607618576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9354933858883894e-05,
+      "loss": 1.0646,
+      "step": 5153
+    },
+    {
+      "epoch": 0.141550630304029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935461951654887e-05,
+      "loss": 1.0453,
+      "step": 5154
+    },
+    {
+      "epoch": 0.14157809453187223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9354305100196475e-05,
+      "loss": 1.0888,
+      "step": 5155
+    },
+    {
+      "epoch": 0.14160555875971548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9353990609829198e-05,
+      "loss": 1.0833,
+      "step": 5156
+    },
+    {
+      "epoch": 0.1416330229875587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9353676045449516e-05,
+      "loss": 1.0357,
+      "step": 5157
+    },
+    {
+      "epoch": 0.14166048721540195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935336140705993e-05,
+      "loss": 1.0623,
+      "step": 5158
+    },
+    {
+      "epoch": 0.14168795144324517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9353046694662926e-05,
+      "loss": 0.9311,
+      "step": 5159
+    },
+    {
+      "epoch": 0.14171541567108842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352731908260995e-05,
+      "loss": 1.0022,
+      "step": 5160
+    },
+    {
+      "epoch": 0.14174287989893164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352417047856624e-05,
+      "loss": 1.0062,
+      "step": 5161
+    },
+    {
+      "epoch": 0.1417703441267749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352102113452304e-05,
+      "loss": 0.9895,
+      "step": 5162
+    },
+    {
+      "epoch": 0.1417978083546181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935178710505053e-05,
+      "loss": 1.0652,
+      "step": 5163
+    },
+    {
+      "epoch": 0.14182527258246133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9351472022653797e-05,
+      "loss": 0.9911,
+      "step": 5164
+    },
+    {
+      "epoch": 0.14185273681030458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9351156866264595e-05,
+      "loss": 0.9089,
+      "step": 5165
+    },
+    {
+      "epoch": 0.1418802010381478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9350841635885418e-05,
+      "loss": 1.0774,
+      "step": 5166
+    },
+    {
+      "epoch": 0.14190766526599105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9350526331518756e-05,
+      "loss": 1.0303,
+      "step": 5167
+    },
+    {
+      "epoch": 0.14193512949383427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935021095316711e-05,
+      "loss": 1.0634,
+      "step": 5168
+    },
+    {
+      "epoch": 0.14196259372167752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9349895500832976e-05,
+      "loss": 1.0408,
+      "step": 5169
+    },
+    {
+      "epoch": 0.14199005794952074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9349579974518844e-05,
+      "loss": 1.0639,
+      "step": 5170
+    },
+    {
+      "epoch": 0.142017522177364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9349264374227212e-05,
+      "loss": 0.9822,
+      "step": 5171
+    },
+    {
+      "epoch": 0.1420449864052072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9348948699960582e-05,
+      "loss": 1.0603,
+      "step": 5172
+    },
+    {
+      "epoch": 0.14207245063305046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9348632951721446e-05,
+      "loss": 0.9813,
+      "step": 5173
+    },
+    {
+      "epoch": 0.14209991486089368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9348317129512306e-05,
+      "loss": 1.0477,
+      "step": 5174
+    },
+    {
+      "epoch": 0.14212737908873693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9348001233335657e-05,
+      "loss": 1.0338,
+      "step": 5175
+    },
+    {
+      "epoch": 0.14215484331658015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347685263194008e-05,
+      "loss": 0.9399,
+      "step": 5176
+    },
+    {
+      "epoch": 0.14218230754442338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347369219089845e-05,
+      "loss": 1.0021,
+      "step": 5177
+    },
+    {
+      "epoch": 0.14220977177226662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934705310102568e-05,
+      "loss": 0.8979,
+      "step": 5178
+    },
+    {
+      "epoch": 0.14223723600010985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9346736909004006e-05,
+      "loss": 0.8994,
+      "step": 5179
+    },
+    {
+      "epoch": 0.1422647002279531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934642064302733e-05,
+      "loss": 1.099,
+      "step": 5180
+    },
+    {
+      "epoch": 0.14229216445579632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934610430309816e-05,
+      "loss": 1.0553,
+      "step": 5181
+    },
+    {
+      "epoch": 0.14231962868363957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345787889218984e-05,
+      "loss": 0.9303,
+      "step": 5182
+    },
+    {
+      "epoch": 0.1423470929114828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934547140139232e-05,
+      "loss": 1.1046,
+      "step": 5183
+    },
+    {
+      "epoch": 0.14237455713932604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345154839620663e-05,
+      "loss": 1.0794,
+      "step": 5184
+    },
+    {
+      "epoch": 0.14240202136716926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934483820390652e-05,
+      "loss": 1.005,
+      "step": 5185
+    },
+    {
+      "epoch": 0.1424294855950125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9344521494252403e-05,
+      "loss": 1.1174,
+      "step": 5186
+    },
+    {
+      "epoch": 0.14245694982285573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9344204710660807e-05,
+      "loss": 1.052,
+      "step": 5187
+    },
+    {
+      "epoch": 0.14248441405069898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9343887853134245e-05,
+      "loss": 1.122,
+      "step": 5188
+    },
+    {
+      "epoch": 0.1425118782785422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9343570921675222e-05,
+      "loss": 1.0297,
+      "step": 5189
+    },
+    {
+      "epoch": 0.14253934250638542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9343253916286246e-05,
+      "loss": 1.072,
+      "step": 5190
+    },
+    {
+      "epoch": 0.14256680673422867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934293683696983e-05,
+      "loss": 0.9628,
+      "step": 5191
+    },
+    {
+      "epoch": 0.1425942709620719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9342619683728473e-05,
+      "loss": 1.0414,
+      "step": 5192
+    },
+    {
+      "epoch": 0.14262173518991514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9342302456564696e-05,
+      "loss": 1.067,
+      "step": 5193
+    },
+    {
+      "epoch": 0.14264919941775836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341985155480996e-05,
+      "loss": 1.0499,
+      "step": 5194
+    },
+    {
+      "epoch": 0.1426766636456016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341667780479896e-05,
+      "loss": 1.0497,
+      "step": 5195
+    },
+    {
+      "epoch": 0.14270412787344483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341350331563902e-05,
+      "loss": 1.1,
+      "step": 5196
+    },
+    {
+      "epoch": 0.14273159210128808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341032808735522e-05,
+      "loss": 0.9996,
+      "step": 5197
+    },
+    {
+      "epoch": 0.1427590563291313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9340715211997273e-05,
+      "loss": 1.1194,
+      "step": 5198
+    },
+    {
+      "epoch": 0.14278652055697455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934039754135167e-05,
+      "loss": 1.0799,
+      "step": 5199
+    },
+    {
+      "epoch": 0.14281398478481777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934007979680122e-05,
+      "loss": 1.0717,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14284144901266102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9339761978348442e-05,
+      "loss": 1.0211,
+      "step": 5201
+    },
+    {
+      "epoch": 0.14286891324050424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933944408599585e-05,
+      "loss": 1.0696,
+      "step": 5202
+    },
+    {
+      "epoch": 0.1428963774683475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933912611974596e-05,
+      "loss": 1.0441,
+      "step": 5203
+    },
+    {
+      "epoch": 0.1429238416961907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9338808079601287e-05,
+      "loss": 1.1337,
+      "step": 5204
+    },
+    {
+      "epoch": 0.14295130592403393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9338489965564344e-05,
+      "loss": 1.1956,
+      "step": 5205
+    },
+    {
+      "epoch": 0.14297877015187718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933817177763765e-05,
+      "loss": 1.0619,
+      "step": 5206
+    },
+    {
+      "epoch": 0.1430062343797204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9337853515823727e-05,
+      "loss": 1.0033,
+      "step": 5207
+    },
+    {
+      "epoch": 0.14303369860756365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9337535180125088e-05,
+      "loss": 1.0002,
+      "step": 5208
+    },
+    {
+      "epoch": 0.14306116283540687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9337216770544255e-05,
+      "loss": 1.1825,
+      "step": 5209
+    },
+    {
+      "epoch": 0.14308862706325012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336898287083744e-05,
+      "loss": 1.0017,
+      "step": 5210
+    },
+    {
+      "epoch": 0.14311609129109334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336579729746077e-05,
+      "loss": 1.0028,
+      "step": 5211
+    },
+    {
+      "epoch": 0.1431435555189366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336261098533774e-05,
+      "loss": 1.0634,
+      "step": 5212
+    },
+    {
+      "epoch": 0.14317101974677982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933594239344936e-05,
+      "loss": 1.0289,
+      "step": 5213
+    },
+    {
+      "epoch": 0.14319848397462306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933562361449535e-05,
+      "loss": 1.0411,
+      "step": 5214
+    },
+    {
+      "epoch": 0.14322594820246629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933530476167427e-05,
+      "loss": 0.986,
+      "step": 5215
+    },
+    {
+      "epoch": 0.14325341243030953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9334985834988646e-05,
+      "loss": 1.0885,
+      "step": 5216
+    },
+    {
+      "epoch": 0.14328087665815276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9334666834440996e-05,
+      "loss": 1.0182,
+      "step": 5217
+    },
+    {
+      "epoch": 0.14330834088599598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9334347760033845e-05,
+      "loss": 0.9649,
+      "step": 5218
+    },
+    {
+      "epoch": 0.14333580511383923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933402861176972e-05,
+      "loss": 1.0473,
+      "step": 5219
+    },
+    {
+      "epoch": 0.14336326934168245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9333709389651147e-05,
+      "loss": 0.979,
+      "step": 5220
+    },
+    {
+      "epoch": 0.1433907335695257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9333390093680648e-05,
+      "loss": 1.0918,
+      "step": 5221
+    },
+    {
+      "epoch": 0.14341819779736892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9333070723860747e-05,
+      "loss": 0.9854,
+      "step": 5222
+    },
+    {
+      "epoch": 0.14344566202521217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933275128019398e-05,
+      "loss": 1.0064,
+      "step": 5223
+    },
+    {
+      "epoch": 0.1434731262530554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9332431762682868e-05,
+      "loss": 1.0297,
+      "step": 5224
+    },
+    {
+      "epoch": 0.14350059048089864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933211217132994e-05,
+      "loss": 1.1664,
+      "step": 5225
+    },
+    {
+      "epoch": 0.14352805470874186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933179250613773e-05,
+      "loss": 1.0285,
+      "step": 5226
+    },
+    {
+      "epoch": 0.1435555189365851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9331472767108758e-05,
+      "loss": 0.8824,
+      "step": 5227
+    },
+    {
+      "epoch": 0.14358298316442833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933115295424556e-05,
+      "loss": 1.1028,
+      "step": 5228
+    },
+    {
+      "epoch": 0.14361044739227158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9330833067550666e-05,
+      "loss": 1.0838,
+      "step": 5229
+    },
+    {
+      "epoch": 0.1436379116201148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9330513107026607e-05,
+      "loss": 0.9664,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14366537584795802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9330193072675912e-05,
+      "loss": 1.1626,
+      "step": 5231
+    },
+    {
+      "epoch": 0.14369284007580127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932987296450112e-05,
+      "loss": 0.9214,
+      "step": 5232
+    },
+    {
+      "epoch": 0.1437203043036445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9329552782504755e-05,
+      "loss": 1.1156,
+      "step": 5233
+    },
+    {
+      "epoch": 0.14374776853148774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9329232526689354e-05,
+      "loss": 1.0732,
+      "step": 5234
+    },
+    {
+      "epoch": 0.14377523275933096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932891219705745e-05,
+      "loss": 1.0945,
+      "step": 5235
+    },
+    {
+      "epoch": 0.1438026969871742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932859179361158e-05,
+      "loss": 0.954,
+      "step": 5236
+    },
+    {
+      "epoch": 0.14383016121501743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9328271316354283e-05,
+      "loss": 1.0859,
+      "step": 5237
+    },
+    {
+      "epoch": 0.14385762544286068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932795076528809e-05,
+      "loss": 1.0199,
+      "step": 5238
+    },
+    {
+      "epoch": 0.1438850896707039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932763014041553e-05,
+      "loss": 0.9742,
+      "step": 5239
+    },
+    {
+      "epoch": 0.14391255389854715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932730944173915e-05,
+      "loss": 1.0859,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14394001812639037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326988669261488e-05,
+      "loss": 1.0097,
+      "step": 5241
+    },
+    {
+      "epoch": 0.14396748235423362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326667822985078e-05,
+      "loss": 1.0391,
+      "step": 5242
+    },
+    {
+      "epoch": 0.14399494658207684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326346902912454e-05,
+      "loss": 1.0996,
+      "step": 5243
+    },
+    {
+      "epoch": 0.1440224108099201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326025909046167e-05,
+      "loss": 0.9881,
+      "step": 5244
+    },
+    {
+      "epoch": 0.14404987503776331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325704841388744e-05,
+      "loss": 0.9943,
+      "step": 5245
+    },
+    {
+      "epoch": 0.14407733926560654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325383699942738e-05,
+      "loss": 1.0736,
+      "step": 5246
+    },
+    {
+      "epoch": 0.14410480349344978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325062484710682e-05,
+      "loss": 0.9867,
+      "step": 5247
+    },
+    {
+      "epoch": 0.144132267721293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9324741195695118e-05,
+      "loss": 0.9502,
+      "step": 5248
+    },
+    {
+      "epoch": 0.14415973194913625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9324419832898593e-05,
+      "loss": 1.0631,
+      "step": 5249
+    },
+    {
+      "epoch": 0.14418719617697948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932409839632364e-05,
+      "loss": 1.0505,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14421466040482273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9323776885972814e-05,
+      "loss": 1.0355,
+      "step": 5251
+    },
+    {
+      "epoch": 0.14424212463266595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9323455301848653e-05,
+      "loss": 0.995,
+      "step": 5252
+    },
+    {
+      "epoch": 0.1442695888605092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9323133643953703e-05,
+      "loss": 1.0513,
+      "step": 5253
+    },
+    {
+      "epoch": 0.14429705308835242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322811912290504e-05,
+      "loss": 0.9934,
+      "step": 5254
+    },
+    {
+      "epoch": 0.14432451731619567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322490106861608e-05,
+      "loss": 1.0324,
+      "step": 5255
+    },
+    {
+      "epoch": 0.1443519815440389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932216822766956e-05,
+      "loss": 1.0693,
+      "step": 5256
+    },
+    {
+      "epoch": 0.14437944577188214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9321846274716906e-05,
+      "loss": 1.059,
+      "step": 5257
+    },
+    {
+      "epoch": 0.14440690999972536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9321524248006192e-05,
+      "loss": 0.9012,
+      "step": 5258
+    },
+    {
+      "epoch": 0.14443437422756858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932120214753997e-05,
+      "loss": 1.0474,
+      "step": 5259
+    },
+    {
+      "epoch": 0.14446183845541183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932087997332078e-05,
+      "loss": 1.0073,
+      "step": 5260
+    },
+    {
+      "epoch": 0.14448930268325505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9320557725351184e-05,
+      "loss": 1.0725,
+      "step": 5261
+    },
+    {
+      "epoch": 0.1445167669110983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932023540363372e-05,
+      "loss": 1.0496,
+      "step": 5262
+    },
+    {
+      "epoch": 0.14454423113894152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9319913008170944e-05,
+      "loss": 1.0433,
+      "step": 5263
+    },
+    {
+      "epoch": 0.14457169536678477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9319590538965407e-05,
+      "loss": 0.9099,
+      "step": 5264
+    },
+    {
+      "epoch": 0.144599159594628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931926799601966e-05,
+      "loss": 1.0435,
+      "step": 5265
+    },
+    {
+      "epoch": 0.14462662382247124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9318945379336257e-05,
+      "loss": 1.0293,
+      "step": 5266
+    },
+    {
+      "epoch": 0.14465408805031446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9318622688917743e-05,
+      "loss": 1.0373,
+      "step": 5267
+    },
+    {
+      "epoch": 0.1446815522781577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931829992476668e-05,
+      "loss": 0.9653,
+      "step": 5268
+    },
+    {
+      "epoch": 0.14470901650600093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931797708688562e-05,
+      "loss": 1.0109,
+      "step": 5269
+    },
+    {
+      "epoch": 0.14473648073384418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931765417527711e-05,
+      "loss": 0.9452,
+      "step": 5270
+    },
+    {
+      "epoch": 0.1447639449616874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9317331189943717e-05,
+      "loss": 1.0092,
+      "step": 5271
+    },
+    {
+      "epoch": 0.14479140918953062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9317008130887986e-05,
+      "loss": 1.0219,
+      "step": 5272
+    },
+    {
+      "epoch": 0.14481887341737387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931668499811248e-05,
+      "loss": 0.9079,
+      "step": 5273
+    },
+    {
+      "epoch": 0.1448463376452171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9316361791619754e-05,
+      "loss": 1.0269,
+      "step": 5274
+    },
+    {
+      "epoch": 0.14487380187306034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9316038511412364e-05,
+      "loss": 0.9924,
+      "step": 5275
+    },
+    {
+      "epoch": 0.14490126610090356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9315715157492864e-05,
+      "loss": 1.0142,
+      "step": 5276
+    },
+    {
+      "epoch": 0.1449287303287468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9315391729863822e-05,
+      "loss": 1.0858,
+      "step": 5277
+    },
+    {
+      "epoch": 0.14495619455659003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931506822852779e-05,
+      "loss": 1.0233,
+      "step": 5278
+    },
+    {
+      "epoch": 0.14498365878443328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9314744653487335e-05,
+      "loss": 0.9749,
+      "step": 5279
+    },
+    {
+      "epoch": 0.1450111230122765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9314421004745007e-05,
+      "loss": 1.0232,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14503858724011975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9314097282303373e-05,
+      "loss": 0.9934,
+      "step": 5281
+    },
+    {
+      "epoch": 0.14506605146796298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9313773486164993e-05,
+      "loss": 0.9987,
+      "step": 5282
+    },
+    {
+      "epoch": 0.14509351569580622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9313449616332432e-05,
+      "loss": 1.0656,
+      "step": 5283
+    },
+    {
+      "epoch": 0.14512097992364945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931312567280825e-05,
+      "loss": 0.9139,
+      "step": 5284
+    },
+    {
+      "epoch": 0.14514844415149267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9312801655595006e-05,
+      "loss": 1.0644,
+      "step": 5285
+    },
+    {
+      "epoch": 0.14517590837933592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931247756469527e-05,
+      "loss": 1.0167,
+      "step": 5286
+    },
+    {
+      "epoch": 0.14520337260717914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93121534001116e-05,
+      "loss": 1.0795,
+      "step": 5287
+    },
+    {
+      "epoch": 0.1452308368350224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9311829161846573e-05,
+      "loss": 0.9719,
+      "step": 5288
+    },
+    {
+      "epoch": 0.1452583010628656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9311504849902743e-05,
+      "loss": 1.0211,
+      "step": 5289
+    },
+    {
+      "epoch": 0.14528576529070886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931118046428268e-05,
+      "loss": 1.0126,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14531322951855208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931085600498895e-05,
+      "loss": 0.96,
+      "step": 5291
+    },
+    {
+      "epoch": 0.14534069374639533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931053147202412e-05,
+      "loss": 0.9203,
+      "step": 5292
+    },
+    {
+      "epoch": 0.14536815797423855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9310206865390757e-05,
+      "loss": 1.0412,
+      "step": 5293
+    },
+    {
+      "epoch": 0.1453956222020818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9309882185091434e-05,
+      "loss": 1.0379,
+      "step": 5294
+    },
+    {
+      "epoch": 0.14542308642992502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9309557431128716e-05,
+      "loss": 1.0261,
+      "step": 5295
+    },
+    {
+      "epoch": 0.14545055065776827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9309232603505173e-05,
+      "loss": 0.9777,
+      "step": 5296
+    },
+    {
+      "epoch": 0.1454780148856115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9308907702223375e-05,
+      "loss": 1.035,
+      "step": 5297
+    },
+    {
+      "epoch": 0.14550547911345474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9308582727285895e-05,
+      "loss": 1.0548,
+      "step": 5298
+    },
+    {
+      "epoch": 0.14553294334129796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93082576786953e-05,
+      "loss": 1.0065,
+      "step": 5299
+    },
+    {
+      "epoch": 0.14556040756914118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9307932556454167e-05,
+      "loss": 1.1354,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14558787179698443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9307607360565067e-05,
+      "loss": 1.082,
+      "step": 5301
+    },
+    {
+      "epoch": 0.14561533602482765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930728209103057e-05,
+      "loss": 1.0261,
+      "step": 5302
+    },
+    {
+      "epoch": 0.1456428002526709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9306956747853253e-05,
+      "loss": 1.0544,
+      "step": 5303
+    },
+    {
+      "epoch": 0.14567026448051412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930663133103569e-05,
+      "loss": 1.0292,
+      "step": 5304
+    },
+    {
+      "epoch": 0.14569772870835737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930630584058045e-05,
+      "loss": 0.9836,
+      "step": 5305
+    },
+    {
+      "epoch": 0.1457251929362006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9305980276490117e-05,
+      "loss": 1.0016,
+      "step": 5306
+    },
+    {
+      "epoch": 0.14575265716404384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9305654638767262e-05,
+      "loss": 0.9884,
+      "step": 5307
+    },
+    {
+      "epoch": 0.14578012139188706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9305328927414466e-05,
+      "loss": 0.9757,
+      "step": 5308
+    },
+    {
+      "epoch": 0.1458075856197303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93050031424343e-05,
+      "loss": 0.8736,
+      "step": 5309
+    },
+    {
+      "epoch": 0.14583504984757353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930467728382934e-05,
+      "loss": 0.9694,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14586251407541678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930435135160218e-05,
+      "loss": 0.9177,
+      "step": 5311
+    },
+    {
+      "epoch": 0.14588997830326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930402534575538e-05,
+      "loss": 1.0252,
+      "step": 5312
+    },
+    {
+      "epoch": 0.14591744253110323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930369926629153e-05,
+      "loss": 1.0281,
+      "step": 5313
+    },
+    {
+      "epoch": 0.14594490675894647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9303373113213206e-05,
+      "loss": 0.9993,
+      "step": 5314
+    },
+    {
+      "epoch": 0.1459723709867897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9303046886522993e-05,
+      "loss": 1.02,
+      "step": 5315
+    },
+    {
+      "epoch": 0.14599983521463294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9302720586223468e-05,
+      "loss": 1.1231,
+      "step": 5316
+    },
+    {
+      "epoch": 0.14602729944247617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9302394212317214e-05,
+      "loss": 0.9,
+      "step": 5317
+    },
+    {
+      "epoch": 0.14605476367031942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9302067764806813e-05,
+      "loss": 1.1188,
+      "step": 5318
+    },
+    {
+      "epoch": 0.14608222789816264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930174124369485e-05,
+      "loss": 1.0174,
+      "step": 5319
+    },
+    {
+      "epoch": 0.14610969212600589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9301414648983904e-05,
+      "loss": 1.1331,
+      "step": 5320
+    },
+    {
+      "epoch": 0.1461371563538491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9301087980676567e-05,
+      "loss": 1.0534,
+      "step": 5321
+    },
+    {
+      "epoch": 0.14616462058169236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9300761238775413e-05,
+      "loss": 1.0434,
+      "step": 5322
+    },
+    {
+      "epoch": 0.14619208480953558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930043442328304e-05,
+      "loss": 1.1598,
+      "step": 5323
+    },
+    {
+      "epoch": 0.14621954903737883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930010753420202e-05,
+      "loss": 1.036,
+      "step": 5324
+    },
+    {
+      "epoch": 0.14624701326522205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9299780571534952e-05,
+      "loss": 1.014,
+      "step": 5325
+    },
+    {
+      "epoch": 0.14627447749306527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9299453535284418e-05,
+      "loss": 0.994,
+      "step": 5326
+    },
+    {
+      "epoch": 0.14630194172090852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9299126425453005e-05,
+      "loss": 1.0367,
+      "step": 5327
+    },
+    {
+      "epoch": 0.14632940594875174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9298799242043296e-05,
+      "loss": 1.176,
+      "step": 5328
+    },
+    {
+      "epoch": 0.146356870176595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9298471985057893e-05,
+      "loss": 1.0607,
+      "step": 5329
+    },
+    {
+      "epoch": 0.1463843344044382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9298144654499375e-05,
+      "loss": 1.0696,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14641179863228146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9297817250370335e-05,
+      "loss": 1.1383,
+      "step": 5331
+    },
+    {
+      "epoch": 0.14643926286012468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929748977267336e-05,
+      "loss": 1.0204,
+      "step": 5332
+    },
+    {
+      "epoch": 0.14646672708796793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9297162221411046e-05,
+      "loss": 0.9571,
+      "step": 5333
+    },
+    {
+      "epoch": 0.14649419131581115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296834596585987e-05,
+      "loss": 1.0592,
+      "step": 5334
+    },
+    {
+      "epoch": 0.1465216555436544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296506898200768e-05,
+      "loss": 0.9749,
+      "step": 5335
+    },
+    {
+      "epoch": 0.14654911977149762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296179126257985e-05,
+      "loss": 1.1176,
+      "step": 5336
+    },
+    {
+      "epoch": 0.14657658399934087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295851280760233e-05,
+      "loss": 1.1108,
+      "step": 5337
+    },
+    {
+      "epoch": 0.1466040482271841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295523361710104e-05,
+      "loss": 1.0602,
+      "step": 5338
+    },
+    {
+      "epoch": 0.14663151245502734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295195369110193e-05,
+      "loss": 0.9833,
+      "step": 5339
+    },
+    {
+      "epoch": 0.14665897668287056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9294867302963096e-05,
+      "loss": 1.0595,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14668644091071378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929453916327141e-05,
+      "loss": 1.0465,
+      "step": 5341
+    },
+    {
+      "epoch": 0.14671390513855703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9294210950037727e-05,
+      "loss": 1.0284,
+      "step": 5342
+    },
+    {
+      "epoch": 0.14674136936640025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929388266326465e-05,
+      "loss": 1.1328,
+      "step": 5343
+    },
+    {
+      "epoch": 0.1467688335942435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9293554302954773e-05,
+      "loss": 1.0692,
+      "step": 5344
+    },
+    {
+      "epoch": 0.14679629782208672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9293225869110694e-05,
+      "loss": 0.9617,
+      "step": 5345
+    },
+    {
+      "epoch": 0.14682376204992997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9292897361735007e-05,
+      "loss": 1.0573,
+      "step": 5346
+    },
+    {
+      "epoch": 0.1468512262777732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9292568780830324e-05,
+      "loss": 1.1345,
+      "step": 5347
+    },
+    {
+      "epoch": 0.14687869050561644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929224012639923e-05,
+      "loss": 1.0447,
+      "step": 5348
+    },
+    {
+      "epoch": 0.14690615473345967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929191139844434e-05,
+      "loss": 1.0809,
+      "step": 5349
+    },
+    {
+      "epoch": 0.14693361896130291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9291582596968243e-05,
+      "loss": 1.0006,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14696108318914614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9291253721973545e-05,
+      "loss": 0.9879,
+      "step": 5351
+    },
+    {
+      "epoch": 0.14698854741698938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9290924773462848e-05,
+      "loss": 1.0131,
+      "step": 5352
+    },
+    {
+      "epoch": 0.1470160116448326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9290595751438758e-05,
+      "loss": 0.9712,
+      "step": 5353
+    },
+    {
+      "epoch": 0.14704347587267583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9290266655903874e-05,
+      "loss": 0.9328,
+      "step": 5354
+    },
+    {
+      "epoch": 0.14707094010051908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92899374868608e-05,
+      "loss": 1.0031,
+      "step": 5355
+    },
+    {
+      "epoch": 0.1470984043283623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9289608244312147e-05,
+      "loss": 1.1071,
+      "step": 5356
+    },
+    {
+      "epoch": 0.14712586855620555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928927892826051e-05,
+      "loss": 0.9793,
+      "step": 5357
+    },
+    {
+      "epoch": 0.14715333278404877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9288949538708504e-05,
+      "loss": 0.9613,
+      "step": 5358
+    },
+    {
+      "epoch": 0.14718079701189202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928862007565873e-05,
+      "loss": 1.0474,
+      "step": 5359
+    },
+    {
+      "epoch": 0.14720826123973524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9288290539113793e-05,
+      "loss": 1.0391,
+      "step": 5360
+    },
+    {
+      "epoch": 0.1472357254675785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9287960929076304e-05,
+      "loss": 1.1629,
+      "step": 5361
+    },
+    {
+      "epoch": 0.1472631896954217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928763124554887e-05,
+      "loss": 1.0198,
+      "step": 5362
+    },
+    {
+      "epoch": 0.14729065392326496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9287301488534106e-05,
+      "loss": 0.9559,
+      "step": 5363
+    },
+    {
+      "epoch": 0.14731811815110818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928697165803461e-05,
+      "loss": 1.1619,
+      "step": 5364
+    },
+    {
+      "epoch": 0.14734558237895143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9286641754052997e-05,
+      "loss": 1.0474,
+      "step": 5365
+    },
+    {
+      "epoch": 0.14737304660679465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9286311776591877e-05,
+      "loss": 0.927,
+      "step": 5366
+    },
+    {
+      "epoch": 0.14740051083463787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928598172565386e-05,
+      "loss": 0.9852,
+      "step": 5367
+    },
+    {
+      "epoch": 0.14742797506248112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928565160124156e-05,
+      "loss": 1.0921,
+      "step": 5368
+    },
+    {
+      "epoch": 0.14745543929032434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9285321403357588e-05,
+      "loss": 1.0869,
+      "step": 5369
+    },
+    {
+      "epoch": 0.1474829035181676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9284991132004557e-05,
+      "loss": 1.1411,
+      "step": 5370
+    },
+    {
+      "epoch": 0.1475103677460108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9284660787185078e-05,
+      "loss": 1.096,
+      "step": 5371
+    },
+    {
+      "epoch": 0.14753783197385406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9284330368901764e-05,
+      "loss": 1.0465,
+      "step": 5372
+    },
+    {
+      "epoch": 0.14756529620169728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283999877157238e-05,
+      "loss": 1.0103,
+      "step": 5373
+    },
+    {
+      "epoch": 0.14759276042954053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283669311954105e-05,
+      "loss": 1.0098,
+      "step": 5374
+    },
+    {
+      "epoch": 0.14762022465738375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283338673294983e-05,
+      "loss": 1.0413,
+      "step": 5375
+    },
+    {
+      "epoch": 0.147647688885227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928300796118249e-05,
+      "loss": 1.0355,
+      "step": 5376
+    },
+    {
+      "epoch": 0.14767515311307022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282677175619242e-05,
+      "loss": 1.0135,
+      "step": 5377
+    },
+    {
+      "epoch": 0.14770261734091347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282346316607857e-05,
+      "loss": 0.9295,
+      "step": 5378
+    },
+    {
+      "epoch": 0.1477300815687567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282015384150953e-05,
+      "loss": 1.0552,
+      "step": 5379
+    },
+    {
+      "epoch": 0.14775754579659992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928168437825115e-05,
+      "loss": 0.9854,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14778501002444316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9281353298911062e-05,
+      "loss": 1.083,
+      "step": 5381
+    },
+    {
+      "epoch": 0.14781247425228639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928102214613331e-05,
+      "loss": 1.0328,
+      "step": 5382
+    },
+    {
+      "epoch": 0.14783993848012963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9280690919920518e-05,
+      "loss": 1.0468,
+      "step": 5383
+    },
+    {
+      "epoch": 0.14786740270797286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9280359620275302e-05,
+      "loss": 1.0464,
+      "step": 5384
+    },
+    {
+      "epoch": 0.1478948669358161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9280028247200293e-05,
+      "loss": 0.9425,
+      "step": 5385
+    },
+    {
+      "epoch": 0.14792233116365933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9279696800698097e-05,
+      "loss": 1.0218,
+      "step": 5386
+    },
+    {
+      "epoch": 0.14794979539150258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927936528077135e-05,
+      "loss": 1.0846,
+      "step": 5387
+    },
+    {
+      "epoch": 0.1479772596193458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9279033687422666e-05,
+      "loss": 1.054,
+      "step": 5388
+    },
+    {
+      "epoch": 0.14800472384718905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9278702020654676e-05,
+      "loss": 1.0618,
+      "step": 5389
+    },
+    {
+      "epoch": 0.14803218807503227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9278370280470003e-05,
+      "loss": 1.1361,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14805965230287552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9278038466871267e-05,
+      "loss": 1.0602,
+      "step": 5391
+    },
+    {
+      "epoch": 0.14808711653071874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92777065798611e-05,
+      "loss": 1.0057,
+      "step": 5392
+    },
+    {
+      "epoch": 0.148114580758562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9277374619442125e-05,
+      "loss": 1.0229,
+      "step": 5393
+    },
+    {
+      "epoch": 0.1481420449864052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9277042585616964e-05,
+      "loss": 1.047,
+      "step": 5394
+    },
+    {
+      "epoch": 0.14816950921424843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9276710478388253e-05,
+      "loss": 1.0819,
+      "step": 5395
+    },
+    {
+      "epoch": 0.14819697344209168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9276378297758614e-05,
+      "loss": 0.9246,
+      "step": 5396
+    },
+    {
+      "epoch": 0.1482244376699349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9276046043730674e-05,
+      "loss": 1.0117,
+      "step": 5397
+    },
+    {
+      "epoch": 0.14825190189777815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9275713716307067e-05,
+      "loss": 1.0837,
+      "step": 5398
+    },
+    {
+      "epoch": 0.14827936612562137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9275381315490418e-05,
+      "loss": 1.108,
+      "step": 5399
+    },
+    {
+      "epoch": 0.14830683035346462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9275048841283358e-05,
+      "loss": 1.0061,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14833429458130784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9274716293688526e-05,
+      "loss": 0.9062,
+      "step": 5401
+    },
+    {
+      "epoch": 0.1483617588091511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927438367270854e-05,
+      "loss": 1.0118,
+      "step": 5402
+    },
+    {
+      "epoch": 0.1483892230369943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9274050978346038e-05,
+      "loss": 0.9643,
+      "step": 5403
+    },
+    {
+      "epoch": 0.14841668726483756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927371821060365e-05,
+      "loss": 0.9778,
+      "step": 5404
+    },
+    {
+      "epoch": 0.14844415149268078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9273385369484018e-05,
+      "loss": 0.9944,
+      "step": 5405
+    },
+    {
+      "epoch": 0.14847161572052403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9273052454989764e-05,
+      "loss": 1.0185,
+      "step": 5406
+    },
+    {
+      "epoch": 0.14849907994836725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927271946712353e-05,
+      "loss": 1.1123,
+      "step": 5407
+    },
+    {
+      "epoch": 0.14852654417621047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9272386405887946e-05,
+      "loss": 1.0122,
+      "step": 5408
+    },
+    {
+      "epoch": 0.14855400840405372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927205327128565e-05,
+      "loss": 1.0909,
+      "step": 5409
+    },
+    {
+      "epoch": 0.14858147263189694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9271720063319272e-05,
+      "loss": 0.9994,
+      "step": 5410
+    },
+    {
+      "epoch": 0.1486089368597402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9271386781991457e-05,
+      "loss": 0.953,
+      "step": 5411
+    },
+    {
+      "epoch": 0.14863640108758341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927105342730484e-05,
+      "loss": 1.1719,
+      "step": 5412
+    },
+    {
+      "epoch": 0.14866386531542666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9270719999262054e-05,
+      "loss": 1.0012,
+      "step": 5413
+    },
+    {
+      "epoch": 0.14869132954326988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9270386497865742e-05,
+      "loss": 1.1079,
+      "step": 5414
+    },
+    {
+      "epoch": 0.14871879377111313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927005292311854e-05,
+      "loss": 0.9795,
+      "step": 5415
+    },
+    {
+      "epoch": 0.14874625799895635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926971927502309e-05,
+      "loss": 0.9465,
+      "step": 5416
+    },
+    {
+      "epoch": 0.1487737222267996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926938555358203e-05,
+      "loss": 1.0253,
+      "step": 5417
+    },
+    {
+      "epoch": 0.14880118645464283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9269051758797997e-05,
+      "loss": 0.9495,
+      "step": 5418
+    },
+    {
+      "epoch": 0.14882865068248607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268717890673644e-05,
+      "loss": 0.9991,
+      "step": 5419
+    },
+    {
+      "epoch": 0.1488561149103293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92683839492116e-05,
+      "loss": 1.0721,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14888357913817252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926804993441451e-05,
+      "loss": 1.0344,
+      "step": 5421
+    },
+    {
+      "epoch": 0.14891104336601577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9267715846285023e-05,
+      "loss": 1.1204,
+      "step": 5422
+    },
+    {
+      "epoch": 0.148938507593859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9267381684825775e-05,
+      "loss": 1.1257,
+      "step": 5423
+    },
+    {
+      "epoch": 0.14896597182170224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9267047450039412e-05,
+      "loss": 1.0808,
+      "step": 5424
+    },
+    {
+      "epoch": 0.14899343604954546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9266713141928585e-05,
+      "loss": 1.0247,
+      "step": 5425
+    },
+    {
+      "epoch": 0.1490209002773887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926637876049593e-05,
+      "loss": 1.0091,
+      "step": 5426
+    },
+    {
+      "epoch": 0.14904836450523193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92660443057441e-05,
+      "loss": 1.0934,
+      "step": 5427
+    },
+    {
+      "epoch": 0.14907582873307518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9265709777675733e-05,
+      "loss": 0.943,
+      "step": 5428
+    },
+    {
+      "epoch": 0.1491032929609184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9265375176293488e-05,
+      "loss": 1.0668,
+      "step": 5429
+    },
+    {
+      "epoch": 0.14913075718876165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9265040501599998e-05,
+      "loss": 1.1304,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14915822141660487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9264705753597924e-05,
+      "loss": 0.9948,
+      "step": 5431
+    },
+    {
+      "epoch": 0.14918568564444812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9264370932289906e-05,
+      "loss": 1.0289,
+      "step": 5432
+    },
+    {
+      "epoch": 0.14921314987229134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92640360376786e-05,
+      "loss": 1.064,
+      "step": 5433
+    },
+    {
+      "epoch": 0.14924061410013456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926370106976665e-05,
+      "loss": 1.0789,
+      "step": 5434
+    },
+    {
+      "epoch": 0.1492680783279778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926336602855671e-05,
+      "loss": 1.0338,
+      "step": 5435
+    },
+    {
+      "epoch": 0.14929554255582103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9263030914051424e-05,
+      "loss": 1.0767,
+      "step": 5436
+    },
+    {
+      "epoch": 0.14932300678366428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262695726253456e-05,
+      "loss": 1.0804,
+      "step": 5437
+    },
+    {
+      "epoch": 0.1493504710115075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926236046516545e-05,
+      "loss": 1.0568,
+      "step": 5438
+    },
+    {
+      "epoch": 0.14937793523935075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926202513079006e-05,
+      "loss": 1.0592,
+      "step": 5439
+    },
+    {
+      "epoch": 0.14940539946719397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926168972312994e-05,
+      "loss": 1.0133,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14943286369503722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9261354242187742e-05,
+      "loss": 0.9486,
+      "step": 5441
+    },
+    {
+      "epoch": 0.14946032792288044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9261018687966123e-05,
+      "loss": 1.0952,
+      "step": 5442
+    },
+    {
+      "epoch": 0.1494877921507237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9260683060467734e-05,
+      "loss": 1.1097,
+      "step": 5443
+    },
+    {
+      "epoch": 0.1495152563785669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9260347359695236e-05,
+      "loss": 1.0544,
+      "step": 5444
+    },
+    {
+      "epoch": 0.14954272060641016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926001158565128e-05,
+      "loss": 1.0192,
+      "step": 5445
+    },
+    {
+      "epoch": 0.14957018483425338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259675738338528e-05,
+      "loss": 1.129,
+      "step": 5446
+    },
+    {
+      "epoch": 0.14959764906209663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259339817759635e-05,
+      "loss": 1.0691,
+      "step": 5447
+    },
+    {
+      "epoch": 0.14962511328993985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259003823917253e-05,
+      "loss": 1.0936,
+      "step": 5448
+    },
+    {
+      "epoch": 0.14965257751778308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9258667756814054e-05,
+      "loss": 1.0241,
+      "step": 5449
+    },
+    {
+      "epoch": 0.14968004174562632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9258331616452686e-05,
+      "loss": 1.0879,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14970750597346955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257995402835812e-05,
+      "loss": 0.9994,
+      "step": 5451
+    },
+    {
+      "epoch": 0.1497349702013128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257659115966093e-05,
+      "loss": 0.9601,
+      "step": 5452
+    },
+    {
+      "epoch": 0.14976243442915602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257322755846188e-05,
+      "loss": 0.9597,
+      "step": 5453
+    },
+    {
+      "epoch": 0.14978989865699927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9256986322478755e-05,
+      "loss": 1.0475,
+      "step": 5454
+    },
+    {
+      "epoch": 0.1498173628848425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9256649815866467e-05,
+      "loss": 1.0839,
+      "step": 5455
+    },
+    {
+      "epoch": 0.14984482711268574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9256313236011978e-05,
+      "loss": 0.9885,
+      "step": 5456
+    },
+    {
+      "epoch": 0.14987229134052896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925597658291795e-05,
+      "loss": 0.9729,
+      "step": 5457
+    },
+    {
+      "epoch": 0.1498997555683722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9255639856587054e-05,
+      "loss": 0.9878,
+      "step": 5458
+    },
+    {
+      "epoch": 0.14992721979621543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925530305702195e-05,
+      "loss": 1.0668,
+      "step": 5459
+    },
+    {
+      "epoch": 0.14995468402405868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92549661842253e-05,
+      "loss": 1.0269,
+      "step": 5460
+    },
+    {
+      "epoch": 0.1499821482519019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9254629238199776e-05,
+      "loss": 1.0171,
+      "step": 5461
+    },
+    {
+      "epoch": 0.15000961247974512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925429221894804e-05,
+      "loss": 1.018,
+      "step": 5462
+    },
+    {
+      "epoch": 0.15003707670758837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925395512647276e-05,
+      "loss": 1.0119,
+      "step": 5463
+    },
+    {
+      "epoch": 0.1500645409354316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92536179607766e-05,
+      "loss": 0.9895,
+      "step": 5464
+    },
+    {
+      "epoch": 0.15009200516327484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925328072186223e-05,
+      "loss": 0.9724,
+      "step": 5465
+    },
+    {
+      "epoch": 0.15011946939111806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9252943409732318e-05,
+      "loss": 1.0938,
+      "step": 5466
+    },
+    {
+      "epoch": 0.1501469336189613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9252606024389537e-05,
+      "loss": 1.0164,
+      "step": 5467
+    },
+    {
+      "epoch": 0.15017439784680453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925226856583655e-05,
+      "loss": 1.1785,
+      "step": 5468
+    },
+    {
+      "epoch": 0.15020186207464778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925193103407603e-05,
+      "loss": 1.1003,
+      "step": 5469
+    },
+    {
+      "epoch": 0.150229326302491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925159342911065e-05,
+      "loss": 1.0331,
+      "step": 5470
+    },
+    {
+      "epoch": 0.15025679053033425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9251255750943077e-05,
+      "loss": 1.1137,
+      "step": 5471
+    },
+    {
+      "epoch": 0.15028425475817747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9250917999575987e-05,
+      "loss": 0.997,
+      "step": 5472
+    },
+    {
+      "epoch": 0.15031171898602072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9250580175012047e-05,
+      "loss": 1.1223,
+      "step": 5473
+    },
+    {
+      "epoch": 0.15033918321386394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9250242277253934e-05,
+      "loss": 1.0302,
+      "step": 5474
+    },
+    {
+      "epoch": 0.15036664744170716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9249904306304323e-05,
+      "loss": 1.1017,
+      "step": 5475
+    },
+    {
+      "epoch": 0.1503941116695504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9249566262165886e-05,
+      "loss": 0.9734,
+      "step": 5476
+    },
+    {
+      "epoch": 0.15042157589739363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9249228144841298e-05,
+      "loss": 0.9628,
+      "step": 5477
+    },
+    {
+      "epoch": 0.15044904012523688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9248889954333232e-05,
+      "loss": 1.1224,
+      "step": 5478
+    },
+    {
+      "epoch": 0.1504765043530801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9248551690644366e-05,
+      "loss": 1.0548,
+      "step": 5479
+    },
+    {
+      "epoch": 0.15050396858092335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924821335377738e-05,
+      "loss": 0.9262,
+      "step": 5480
+    },
+    {
+      "epoch": 0.15053143280876657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9247874943734946e-05,
+      "loss": 1.0445,
+      "step": 5481
+    },
+    {
+      "epoch": 0.15055889703660982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924753646051974e-05,
+      "loss": 1.0889,
+      "step": 5482
+    },
+    {
+      "epoch": 0.15058636126445304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9247197904134446e-05,
+      "loss": 1.0703,
+      "step": 5483
+    },
+    {
+      "epoch": 0.1506138254922963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924685927458174e-05,
+      "loss": 0.9827,
+      "step": 5484
+    },
+    {
+      "epoch": 0.15064128972013952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9246520571864304e-05,
+      "loss": 0.9417,
+      "step": 5485
+    },
+    {
+      "epoch": 0.15066875394798276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924618179598481e-05,
+      "loss": 0.986,
+      "step": 5486
+    },
+    {
+      "epoch": 0.15069621817582599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245842946945948e-05,
+      "loss": 0.9886,
+      "step": 5487
+    },
+    {
+      "epoch": 0.15072368240366923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245504024750392e-05,
+      "loss": 1.0652,
+      "step": 5488
+    },
+    {
+      "epoch": 0.15075114663151246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245165029400833e-05,
+      "loss": 1.0834,
+      "step": 5489
+    },
+    {
+      "epoch": 0.15077861085935568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924482596089994e-05,
+      "loss": 1.1194,
+      "step": 5490
+    },
+    {
+      "epoch": 0.15080607508719893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9244486819250406e-05,
+      "loss": 1.0213,
+      "step": 5491
+    },
+    {
+      "epoch": 0.15083353931504215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924414760445491e-05,
+      "loss": 1.0933,
+      "step": 5492
+    },
+    {
+      "epoch": 0.1508610035428854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924380831651614e-05,
+      "loss": 0.9135,
+      "step": 5493
+    },
+    {
+      "epoch": 0.15088846777072862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9243468955436775e-05,
+      "loss": 0.9711,
+      "step": 5494
+    },
+    {
+      "epoch": 0.15091593199857187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9243129521219506e-05,
+      "loss": 1.143,
+      "step": 5495
+    },
+    {
+      "epoch": 0.1509433962264151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9242790013867014e-05,
+      "loss": 1.0253,
+      "step": 5496
+    },
+    {
+      "epoch": 0.15097086045425834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9242450433381986e-05,
+      "loss": 1.1166,
+      "step": 5497
+    },
+    {
+      "epoch": 0.15099832468210156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924211077976711e-05,
+      "loss": 0.9834,
+      "step": 5498
+    },
+    {
+      "epoch": 0.1510257889099448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241771053025076e-05,
+      "loss": 1.0108,
+      "step": 5499
+    },
+    {
+      "epoch": 0.15105325313778803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241431253158568e-05,
+      "loss": 1.1354,
+      "step": 5500
+    },
+    {
+      "epoch": 0.15108071736563128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241091380170277e-05,
+      "loss": 1.1072,
+      "step": 5501
+    },
+    {
+      "epoch": 0.1511081815934745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924075143406289e-05,
+      "loss": 1.0433,
+      "step": 5502
+    },
+    {
+      "epoch": 0.15113564582131772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92404114148391e-05,
+      "loss": 1.1036,
+      "step": 5503
+    },
+    {
+      "epoch": 0.15116311004916097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9240071322501594e-05,
+      "loss": 1.1456,
+      "step": 5504
+    },
+    {
+      "epoch": 0.1511905742770042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9239731157053064e-05,
+      "loss": 1.1091,
+      "step": 5505
+    },
+    {
+      "epoch": 0.15121803850484744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9239390918496202e-05,
+      "loss": 0.9693,
+      "step": 5506
+    },
+    {
+      "epoch": 0.15124550273269066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92390506068337e-05,
+      "loss": 1.001,
+      "step": 5507
+    },
+    {
+      "epoch": 0.1512729669605339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9238710222068254e-05,
+      "loss": 0.9421,
+      "step": 5508
+    },
+    {
+      "epoch": 0.15130043118837713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923836976420255e-05,
+      "loss": 1.0469,
+      "step": 5509
+    },
+    {
+      "epoch": 0.15132789541622038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9238029233239288e-05,
+      "loss": 1.0743,
+      "step": 5510
+    },
+    {
+      "epoch": 0.1513553596440636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923768862918116e-05,
+      "loss": 0.9569,
+      "step": 5511
+    },
+    {
+      "epoch": 0.15138282387190685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923734795203086e-05,
+      "loss": 1.022,
+      "step": 5512
+    },
+    {
+      "epoch": 0.15141028809975007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9237007201791086e-05,
+      "loss": 1.0118,
+      "step": 5513
+    },
+    {
+      "epoch": 0.15143775232759332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923666637846453e-05,
+      "loss": 0.9954,
+      "step": 5514
+    },
+    {
+      "epoch": 0.15146521655543654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9236325482053897e-05,
+      "loss": 0.9573,
+      "step": 5515
+    },
+    {
+      "epoch": 0.15149268078327977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9235984512561872e-05,
+      "loss": 1.0811,
+      "step": 5516
+    },
+    {
+      "epoch": 0.15152014501112301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9235643469991166e-05,
+      "loss": 1.0299,
+      "step": 5517
+    },
+    {
+      "epoch": 0.15154760923896624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923530235434447e-05,
+      "loss": 0.9945,
+      "step": 5518
+    },
+    {
+      "epoch": 0.15157507346680948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9234961165624484e-05,
+      "loss": 1.0602,
+      "step": 5519
+    },
+    {
+      "epoch": 0.1516025376946527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9234619903833905e-05,
+      "loss": 1.082,
+      "step": 5520
+    },
+    {
+      "epoch": 0.15163000192249595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9234278568975442e-05,
+      "loss": 1.0286,
+      "step": 5521
+    },
+    {
+      "epoch": 0.15165746615033918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923393716105179e-05,
+      "loss": 0.9604,
+      "step": 5522
+    },
+    {
+      "epoch": 0.15168493037818243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923359568006564e-05,
+      "loss": 1.0922,
+      "step": 5523
+    },
+    {
+      "epoch": 0.15171239460602565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9233254126019717e-05,
+      "loss": 0.975,
+      "step": 5524
+    },
+    {
+      "epoch": 0.1517398588338689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9232912498916703e-05,
+      "loss": 1.1182,
+      "step": 5525
+    },
+    {
+      "epoch": 0.15176732306171212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923257079875931e-05,
+      "loss": 1.0315,
+      "step": 5526
+    },
+    {
+      "epoch": 0.15179478728955537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9232229025550242e-05,
+      "loss": 1.0743,
+      "step": 5527
+    },
+    {
+      "epoch": 0.1518222515173986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231887179292198e-05,
+      "loss": 0.9708,
+      "step": 5528
+    },
+    {
+      "epoch": 0.1518497157452418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231545259987893e-05,
+      "loss": 1.0673,
+      "step": 5529
+    },
+    {
+      "epoch": 0.15187717997308506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923120326764002e-05,
+      "loss": 1.0849,
+      "step": 5530
+    },
+    {
+      "epoch": 0.15190464420092828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9230861202251297e-05,
+      "loss": 1.069,
+      "step": 5531
+    },
+    {
+      "epoch": 0.15193210842877153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9230519063824417e-05,
+      "loss": 1.0057,
+      "step": 5532
+    },
+    {
+      "epoch": 0.15195957265661475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9230176852362098e-05,
+      "loss": 1.076,
+      "step": 5533
+    },
+    {
+      "epoch": 0.151987036884458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9229834567867046e-05,
+      "loss": 1.0359,
+      "step": 5534
+    },
+    {
+      "epoch": 0.15201450111230122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9229492210341968e-05,
+      "loss": 1.0134,
+      "step": 5535
+    },
+    {
+      "epoch": 0.15204196534014447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9229149779789572e-05,
+      "loss": 1.0193,
+      "step": 5536
+    },
+    {
+      "epoch": 0.1520694295679877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9228807276212566e-05,
+      "loss": 1.074,
+      "step": 5537
+    },
+    {
+      "epoch": 0.15209689379583094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922846469961366e-05,
+      "loss": 1.1212,
+      "step": 5538
+    },
+    {
+      "epoch": 0.15212435802367416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922812204999557e-05,
+      "loss": 0.9742,
+      "step": 5539
+    },
+    {
+      "epoch": 0.1521518222515174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9227779327361004e-05,
+      "loss": 1.0796,
+      "step": 5540
+    },
+    {
+      "epoch": 0.15217928647936063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9227436531712674e-05,
+      "loss": 0.9447,
+      "step": 5541
+    },
+    {
+      "epoch": 0.15220675070720388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922709366305329e-05,
+      "loss": 1.0674,
+      "step": 5542
+    },
+    {
+      "epoch": 0.1522342149350471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9226750721385564e-05,
+      "loss": 1.1113,
+      "step": 5543
+    },
+    {
+      "epoch": 0.15226167916289032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9226407706712214e-05,
+      "loss": 0.9193,
+      "step": 5544
+    },
+    {
+      "epoch": 0.15228914339073357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9226064619035954e-05,
+      "loss": 1.0679,
+      "step": 5545
+    },
+    {
+      "epoch": 0.1523166076185768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9225721458359498e-05,
+      "loss": 0.9885,
+      "step": 5546
+    },
+    {
+      "epoch": 0.15234407184642004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922537822468556e-05,
+      "loss": 0.9817,
+      "step": 5547
+    },
+    {
+      "epoch": 0.15237153607426326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9225034918016856e-05,
+      "loss": 1.0968,
+      "step": 5548
+    },
+    {
+      "epoch": 0.1523990003021065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9224691538356097e-05,
+      "loss": 1.1115,
+      "step": 5549
+    },
+    {
+      "epoch": 0.15242646452994973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9224348085706012e-05,
+      "loss": 1.0167,
+      "step": 5550
+    },
+    {
+      "epoch": 0.15245392875779298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9224004560069306e-05,
+      "loss": 1.0858,
+      "step": 5551
+    },
+    {
+      "epoch": 0.1524813929856362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922366096144871e-05,
+      "loss": 0.8678,
+      "step": 5552
+    },
+    {
+      "epoch": 0.15250885721347945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922331728984693e-05,
+      "loss": 1.1356,
+      "step": 5553
+    },
+    {
+      "epoch": 0.15253632144132268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922297354526669e-05,
+      "loss": 1.0309,
+      "step": 5554
+    },
+    {
+      "epoch": 0.15256378566916592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9222629727710712e-05,
+      "loss": 1.0029,
+      "step": 5555
+    },
+    {
+      "epoch": 0.15259124989700915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9222285837181716e-05,
+      "loss": 1.0948,
+      "step": 5556
+    },
+    {
+      "epoch": 0.15261871412485237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9221941873682422e-05,
+      "loss": 1.0428,
+      "step": 5557
+    },
+    {
+      "epoch": 0.15264617835269562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922159783721555e-05,
+      "loss": 1.0297,
+      "step": 5558
+    },
+    {
+      "epoch": 0.15267364258053884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9221253727783827e-05,
+      "loss": 1.0562,
+      "step": 5559
+    },
+    {
+      "epoch": 0.1527011068083821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922090954538997e-05,
+      "loss": 1.0813,
+      "step": 5560
+    },
+    {
+      "epoch": 0.1527285710362253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9220565290036707e-05,
+      "loss": 0.9195,
+      "step": 5561
+    },
+    {
+      "epoch": 0.15275603526406856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9220220961726755e-05,
+      "loss": 1.0847,
+      "step": 5562
+    },
+    {
+      "epoch": 0.15278349949191178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9219876560462846e-05,
+      "loss": 1.0236,
+      "step": 5563
+    },
+    {
+      "epoch": 0.15281096371975503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9219532086247702e-05,
+      "loss": 0.9967,
+      "step": 5564
+    },
+    {
+      "epoch": 0.15283842794759825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921918753908405e-05,
+      "loss": 1.1475,
+      "step": 5565
+    },
+    {
+      "epoch": 0.1528658921754415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9218842918974618e-05,
+      "loss": 1.0404,
+      "step": 5566
+    },
+    {
+      "epoch": 0.15289335640328472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9218498225922126e-05,
+      "loss": 1.1074,
+      "step": 5567
+    },
+    {
+      "epoch": 0.15292082063112797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9218153459929305e-05,
+      "loss": 1.021,
+      "step": 5568
+    },
+    {
+      "epoch": 0.1529482848589712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9217808620998884e-05,
+      "loss": 1.1077,
+      "step": 5569
+    },
+    {
+      "epoch": 0.1529757490868144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921746370913359e-05,
+      "loss": 1.0405,
+      "step": 5570
+    },
+    {
+      "epoch": 0.15300321331465766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921711872433615e-05,
+      "loss": 0.9884,
+      "step": 5571
+    },
+    {
+      "epoch": 0.15303067754250088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92167736666093e-05,
+      "loss": 1.0234,
+      "step": 5572
+    },
+    {
+      "epoch": 0.15305814177034413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9216428535955765e-05,
+      "loss": 1.0664,
+      "step": 5573
+    },
+    {
+      "epoch": 0.15308560599818735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921608333237828e-05,
+      "loss": 1.0835,
+      "step": 5574
+    },
+    {
+      "epoch": 0.1531130702260306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921573805587957e-05,
+      "loss": 1.1026,
+      "step": 5575
+    },
+    {
+      "epoch": 0.15314053445387382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9215392706462372e-05,
+      "loss": 1.0926,
+      "step": 5576
+    },
+    {
+      "epoch": 0.15316799868171707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9215047284129416e-05,
+      "loss": 1.1181,
+      "step": 5577
+    },
+    {
+      "epoch": 0.1531954629095603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921470178888344e-05,
+      "loss": 1.1403,
+      "step": 5578
+    },
+    {
+      "epoch": 0.15322292713740354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9214356220727174e-05,
+      "loss": 1.076,
+      "step": 5579
+    },
+    {
+      "epoch": 0.15325039136524676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921401057966335e-05,
+      "loss": 1.1213,
+      "step": 5580
+    },
+    {
+      "epoch": 0.15327785559309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213664865694703e-05,
+      "loss": 1.058,
+      "step": 5581
+    },
+    {
+      "epoch": 0.15330531982093323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213319078823974e-05,
+      "loss": 1.0099,
+      "step": 5582
+    },
+    {
+      "epoch": 0.15333278404877645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921297321905389e-05,
+      "loss": 1.0616,
+      "step": 5583
+    },
+    {
+      "epoch": 0.1533602482766197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9212627286387203e-05,
+      "loss": 1.1274,
+      "step": 5584
+    },
+    {
+      "epoch": 0.15338771250446293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9212281280826632e-05,
+      "loss": 1.0421,
+      "step": 5585
+    },
+    {
+      "epoch": 0.15341517673230617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9211935202374925e-05,
+      "loss": 1.0778,
+      "step": 5586
+    },
+    {
+      "epoch": 0.1534426409601494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921158905103482e-05,
+      "loss": 1.0981,
+      "step": 5587
+    },
+    {
+      "epoch": 0.15347010518799264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9211242826809055e-05,
+      "loss": 1.0038,
+      "step": 5588
+    },
+    {
+      "epoch": 0.15349756941583587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9210896529700363e-05,
+      "loss": 0.9883,
+      "step": 5589
+    },
+    {
+      "epoch": 0.15352503364367912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9210550159711493e-05,
+      "loss": 0.9742,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15355249787152234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9210203716845185e-05,
+      "loss": 1.0341,
+      "step": 5591
+    },
+    {
+      "epoch": 0.15357996209936559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9209857201104175e-05,
+      "loss": 1.0834,
+      "step": 5592
+    },
+    {
+      "epoch": 0.1536074263272088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920951061249121e-05,
+      "loss": 1.058,
+      "step": 5593
+    },
+    {
+      "epoch": 0.15363489055505206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9209163951009024e-05,
+      "loss": 1.0915,
+      "step": 5594
+    },
+    {
+      "epoch": 0.15366235478289528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920881721666037e-05,
+      "loss": 1.0583,
+      "step": 5595
+    },
+    {
+      "epoch": 0.15368981901073853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9208470409447987e-05,
+      "loss": 1.048,
+      "step": 5596
+    },
+    {
+      "epoch": 0.15371728323858175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9208123529374616e-05,
+      "loss": 1.0974,
+      "step": 5597
+    },
+    {
+      "epoch": 0.15374474746642497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9207776576443007e-05,
+      "loss": 0.965,
+      "step": 5598
+    },
+    {
+      "epoch": 0.15377221169426822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9207429550655903e-05,
+      "loss": 1.0456,
+      "step": 5599
+    },
+    {
+      "epoch": 0.15379967592211144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9207082452016048e-05,
+      "loss": 1.0779,
+      "step": 5600
+    },
+    {
+      "epoch": 0.1538271401499547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9206735280526195e-05,
+      "loss": 0.9422,
+      "step": 5601
+    },
+    {
+      "epoch": 0.1538546043777979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920638803618908e-05,
+      "loss": 0.9741,
+      "step": 5602
+    },
+    {
+      "epoch": 0.15388206860564116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920604071900746e-05,
+      "loss": 1.0761,
+      "step": 5603
+    },
+    {
+      "epoch": 0.15390953283348438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920569332898408e-05,
+      "loss": 1.0747,
+      "step": 5604
+    },
+    {
+      "epoch": 0.15393699706132763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920534586612169e-05,
+      "loss": 1.0757,
+      "step": 5605
+    },
+    {
+      "epoch": 0.15396446128917085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9204998330423034e-05,
+      "loss": 1.0197,
+      "step": 5606
+    },
+    {
+      "epoch": 0.1539919255170141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9204650721890866e-05,
+      "loss": 1.0486,
+      "step": 5607
+    },
+    {
+      "epoch": 0.15401938974485732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9204303040527935e-05,
+      "loss": 1.0477,
+      "step": 5608
+    },
+    {
+      "epoch": 0.15404685397270057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9203955286336996e-05,
+      "loss": 1.1115,
+      "step": 5609
+    },
+    {
+      "epoch": 0.1540743182005438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9203607459320793e-05,
+      "loss": 1.0454,
+      "step": 5610
+    },
+    {
+      "epoch": 0.154101782428387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9203259559482088e-05,
+      "loss": 1.0562,
+      "step": 5611
+    },
+    {
+      "epoch": 0.15412924665623026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920291158682363e-05,
+      "loss": 1.0778,
+      "step": 5612
+    },
+    {
+      "epoch": 0.15415671088407348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9202563541348162e-05,
+      "loss": 1.0154,
+      "step": 5613
+    },
+    {
+      "epoch": 0.15418417511191673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920221542305845e-05,
+      "loss": 1.1611,
+      "step": 5614
+    },
+    {
+      "epoch": 0.15421163933975995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9201867231957248e-05,
+      "loss": 1.0451,
+      "step": 5615
+    },
+    {
+      "epoch": 0.1542391035676032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9201518968047306e-05,
+      "loss": 1.0238,
+      "step": 5616
+    },
+    {
+      "epoch": 0.15426656779544642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9201170631331382e-05,
+      "loss": 1.0768,
+      "step": 5617
+    },
+    {
+      "epoch": 0.15429403202328967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920082222181223e-05,
+      "loss": 1.0408,
+      "step": 5618
+    },
+    {
+      "epoch": 0.1543214962511329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920047373949261e-05,
+      "loss": 1.0951,
+      "step": 5619
+    },
+    {
+      "epoch": 0.15434896047897614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920012518437528e-05,
+      "loss": 1.0415,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15437642470681937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9199776556462994e-05,
+      "loss": 1.0845,
+      "step": 5621
+    },
+    {
+      "epoch": 0.15440388893466261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919942785575851e-05,
+      "loss": 1.1295,
+      "step": 5622
+    },
+    {
+      "epoch": 0.15443135316250584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9199079082264592e-05,
+      "loss": 1.0141,
+      "step": 5623
+    },
+    {
+      "epoch": 0.15445881739034906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9198730235983998e-05,
+      "loss": 1.0605,
+      "step": 5624
+    },
+    {
+      "epoch": 0.1544862816181923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9198381316919485e-05,
+      "loss": 1.0633,
+      "step": 5625
+    },
+    {
+      "epoch": 0.15451374584603553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9198032325073816e-05,
+      "loss": 1.1157,
+      "step": 5626
+    },
+    {
+      "epoch": 0.15454121007387878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919768326044975e-05,
+      "loss": 1.047,
+      "step": 5627
+    },
+    {
+      "epoch": 0.154568674301722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9197334123050055e-05,
+      "loss": 0.9313,
+      "step": 5628
+    },
+    {
+      "epoch": 0.15459613852956525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9196984912877486e-05,
+      "loss": 0.9904,
+      "step": 5629
+    },
+    {
+      "epoch": 0.15462360275740847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9196635629934816e-05,
+      "loss": 1.1445,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15465106698525172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91962862742248e-05,
+      "loss": 1.0026,
+      "step": 5631
+    },
+    {
+      "epoch": 0.15467853121309494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9195936845750203e-05,
+      "loss": 1.0052,
+      "step": 5632
+    },
+    {
+      "epoch": 0.1547059954409382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919558734451379e-05,
+      "loss": 1.0051,
+      "step": 5633
+    },
+    {
+      "epoch": 0.1547334596687814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919523777051833e-05,
+      "loss": 1.0087,
+      "step": 5634
+    },
+    {
+      "epoch": 0.15476092389662466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919488812376659e-05,
+      "loss": 1.1432,
+      "step": 5635
+    },
+    {
+      "epoch": 0.15478838812446788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194538404261326e-05,
+      "loss": 1.0121,
+      "step": 5636
+    },
+    {
+      "epoch": 0.15481585235231113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194188612005317e-05,
+      "loss": 0.9477,
+      "step": 5637
+    },
+    {
+      "epoch": 0.15484331658015435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9193838747001327e-05,
+      "loss": 1.0642,
+      "step": 5638
+    },
+    {
+      "epoch": 0.15487078080799757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9193488809252122e-05,
+      "loss": 1.0311,
+      "step": 5639
+    },
+    {
+      "epoch": 0.15489824503584082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919313879876047e-05,
+      "loss": 1.1447,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15492570926368404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919278871552915e-05,
+      "loss": 1.1677,
+      "step": 5641
+    },
+    {
+      "epoch": 0.1549531734915273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919243855956092e-05,
+      "loss": 1.0448,
+      "step": 5642
+    },
+    {
+      "epoch": 0.1549806377193705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9192088330858553e-05,
+      "loss": 1.093,
+      "step": 5643
+    },
+    {
+      "epoch": 0.15500810194721376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9191738029424825e-05,
+      "loss": 1.0165,
+      "step": 5644
+    },
+    {
+      "epoch": 0.15503556617505698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9191387655262504e-05,
+      "loss": 1.138,
+      "step": 5645
+    },
+    {
+      "epoch": 0.15506303040290023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9191037208374365e-05,
+      "loss": 0.9954,
+      "step": 5646
+    },
+    {
+      "epoch": 0.15509049463074345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9190686688763178e-05,
+      "loss": 1.08,
+      "step": 5647
+    },
+    {
+      "epoch": 0.1551179588585867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919033609643172e-05,
+      "loss": 1.1107,
+      "step": 5648
+    },
+    {
+      "epoch": 0.15514542308642992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918998543138276e-05,
+      "loss": 1.023,
+      "step": 5649
+    },
+    {
+      "epoch": 0.15517288731427317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9189634693619078e-05,
+      "loss": 1.0563,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1552003515421164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9189283883143445e-05,
+      "loss": 1.1427,
+      "step": 5651
+    },
+    {
+      "epoch": 0.15522781576995962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918893299995864e-05,
+      "loss": 1.0738,
+      "step": 5652
+    },
+    {
+      "epoch": 0.15525527999780286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9188582044067433e-05,
+      "loss": 0.9669,
+      "step": 5653
+    },
+    {
+      "epoch": 0.15528274422564609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918823101547261e-05,
+      "loss": 1.0651,
+      "step": 5654
+    },
+    {
+      "epoch": 0.15531020845348933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9187879914176943e-05,
+      "loss": 0.9226,
+      "step": 5655
+    },
+    {
+      "epoch": 0.15533767268133256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9187528740183213e-05,
+      "loss": 0.9703,
+      "step": 5656
+    },
+    {
+      "epoch": 0.1553651369091758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9187177493494194e-05,
+      "loss": 1.0915,
+      "step": 5657
+    },
+    {
+      "epoch": 0.15539260113701903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918682617411267e-05,
+      "loss": 0.9871,
+      "step": 5658
+    },
+    {
+      "epoch": 0.15542006536486228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918647478204142e-05,
+      "loss": 0.948,
+      "step": 5659
+    },
+    {
+      "epoch": 0.1554475295927055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9186123317283216e-05,
+      "loss": 1.0506,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15547499382054875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918577177984085e-05,
+      "loss": 0.9846,
+      "step": 5661
+    },
+    {
+      "epoch": 0.15550245804839197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91854201697171e-05,
+      "loss": 0.9743,
+      "step": 5662
+    },
+    {
+      "epoch": 0.15552992227623522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9185068486914748e-05,
+      "loss": 1.0638,
+      "step": 5663
+    },
+    {
+      "epoch": 0.15555738650407844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9184716731436575e-05,
+      "loss": 0.9311,
+      "step": 5664
+    },
+    {
+      "epoch": 0.15558485073192166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9184364903285368e-05,
+      "loss": 1.0864,
+      "step": 5665
+    },
+    {
+      "epoch": 0.1556123149597649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9184013002463904e-05,
+      "loss": 1.0862,
+      "step": 5666
+    },
+    {
+      "epoch": 0.15563977918760813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9183661028974975e-05,
+      "loss": 0.9903,
+      "step": 5667
+    },
+    {
+      "epoch": 0.15566724341545138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918330898282136e-05,
+      "loss": 1.1011,
+      "step": 5668
+    },
+    {
+      "epoch": 0.1556947076432946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918295686400585e-05,
+      "loss": 0.9727,
+      "step": 5669
+    },
+    {
+      "epoch": 0.15572217187113785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9182604672531223e-05,
+      "loss": 0.9995,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15574963609898107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9182252408400276e-05,
+      "loss": 1.0814,
+      "step": 5671
+    },
+    {
+      "epoch": 0.15577710032682432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918190007161579e-05,
+      "loss": 1.0211,
+      "step": 5672
+    },
+    {
+      "epoch": 0.15580456455466754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9181547662180554e-05,
+      "loss": 1.036,
+      "step": 5673
+    },
+    {
+      "epoch": 0.1558320287825108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9181195180097353e-05,
+      "loss": 1.0035,
+      "step": 5674
+    },
+    {
+      "epoch": 0.155859493010354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918084262536898e-05,
+      "loss": 0.8196,
+      "step": 5675
+    },
+    {
+      "epoch": 0.15588695723819726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9180489997998227e-05,
+      "loss": 0.9425,
+      "step": 5676
+    },
+    {
+      "epoch": 0.15591442146604048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918013729798788e-05,
+      "loss": 1.1799,
+      "step": 5677
+    },
+    {
+      "epoch": 0.1559418856938837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917978452534073e-05,
+      "loss": 0.9679,
+      "step": 5678
+    },
+    {
+      "epoch": 0.15596934992172695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917943168005957e-05,
+      "loss": 1.2133,
+      "step": 5679
+    },
+    {
+      "epoch": 0.15599681414957017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917907876214719e-05,
+      "loss": 1.0248,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15602427837741342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9178725771606383e-05,
+      "loss": 0.9781,
+      "step": 5681
+    },
+    {
+      "epoch": 0.15605174260525664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9178372708439942e-05,
+      "loss": 1.0295,
+      "step": 5682
+    },
+    {
+      "epoch": 0.1560792068330999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917801957265066e-05,
+      "loss": 1.0708,
+      "step": 5683
+    },
+    {
+      "epoch": 0.15610667106094311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9177666364241333e-05,
+      "loss": 1.0884,
+      "step": 5684
+    },
+    {
+      "epoch": 0.15613413528878636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9177313083214755e-05,
+      "loss": 0.9618,
+      "step": 5685
+    },
+    {
+      "epoch": 0.15616159951662958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917695972957372e-05,
+      "loss": 0.9821,
+      "step": 5686
+    },
+    {
+      "epoch": 0.15618906374447283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917660630332103e-05,
+      "loss": 0.9933,
+      "step": 5687
+    },
+    {
+      "epoch": 0.15621652797231605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917625280445947e-05,
+      "loss": 0.9951,
+      "step": 5688
+    },
+    {
+      "epoch": 0.1562439922001593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9175899232991843e-05,
+      "loss": 1.0816,
+      "step": 5689
+    },
+    {
+      "epoch": 0.15627145642800253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917554558892095e-05,
+      "loss": 1.0723,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15629892065584577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9175191872249583e-05,
+      "loss": 1.0626,
+      "step": 5691
+    },
+    {
+      "epoch": 0.156326384883689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9174838082980547e-05,
+      "loss": 1.1123,
+      "step": 5692
+    },
+    {
+      "epoch": 0.15635384911153222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9174484221116637e-05,
+      "loss": 1.0422,
+      "step": 5693
+    },
+    {
+      "epoch": 0.15638131333937547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9174130286660654e-05,
+      "loss": 1.1108,
+      "step": 5694
+    },
+    {
+      "epoch": 0.1564087775672187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91737762796154e-05,
+      "loss": 0.9467,
+      "step": 5695
+    },
+    {
+      "epoch": 0.15643624179506194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9173422199983668e-05,
+      "loss": 0.9653,
+      "step": 5696
+    },
+    {
+      "epoch": 0.15646370602290516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917306804776827e-05,
+      "loss": 1.0613,
+      "step": 5697
+    },
+    {
+      "epoch": 0.1564911702507484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9172713822972007e-05,
+      "loss": 1.0697,
+      "step": 5698
+    },
+    {
+      "epoch": 0.15651863447859163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9172359525597677e-05,
+      "loss": 1.0863,
+      "step": 5699
+    },
+    {
+      "epoch": 0.15654609870643488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9172005155648087e-05,
+      "loss": 0.9825,
+      "step": 5700
+    },
+    {
+      "epoch": 0.1565735629342781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917165071312604e-05,
+      "loss": 1.0385,
+      "step": 5701
+    },
+    {
+      "epoch": 0.15660102716212135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9171296198034334e-05,
+      "loss": 1.0271,
+      "step": 5702
+    },
+    {
+      "epoch": 0.15662849138996457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9170941610375782e-05,
+      "loss": 1.0848,
+      "step": 5703
+    },
+    {
+      "epoch": 0.15665595561780782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9170586950153192e-05,
+      "loss": 0.892,
+      "step": 5704
+    },
+    {
+      "epoch": 0.15668341984565104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917023221736936e-05,
+      "loss": 1.0345,
+      "step": 5705
+    },
+    {
+      "epoch": 0.15671088407349426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9169877412027103e-05,
+      "loss": 1.0776,
+      "step": 5706
+    },
+    {
+      "epoch": 0.1567383483013375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9169522534129222e-05,
+      "loss": 1.0392,
+      "step": 5707
+    },
+    {
+      "epoch": 0.15676581252918073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9169167583678524e-05,
+      "loss": 1.061,
+      "step": 5708
+    },
+    {
+      "epoch": 0.15679327675702398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168812560677825e-05,
+      "loss": 1.0846,
+      "step": 5709
+    },
+    {
+      "epoch": 0.1568207409848672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916845746512993e-05,
+      "loss": 1.109,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15684820521271045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168102297037642e-05,
+      "loss": 1.1002,
+      "step": 5711
+    },
+    {
+      "epoch": 0.15687566944055367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916774705640378e-05,
+      "loss": 1.125,
+      "step": 5712
+    },
+    {
+      "epoch": 0.15690313366839692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9167391743231154e-05,
+      "loss": 1.1112,
+      "step": 5713
+    },
+    {
+      "epoch": 0.15693059789624014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916703635752257e-05,
+      "loss": 1.0124,
+      "step": 5714
+    },
+    {
+      "epoch": 0.1569580621240834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9166680899280846e-05,
+      "loss": 1.0926,
+      "step": 5715
+    },
+    {
+      "epoch": 0.1569855263519266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9166325368508794e-05,
+      "loss": 1.1075,
+      "step": 5716
+    },
+    {
+      "epoch": 0.15701299057976986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9165969765209223e-05,
+      "loss": 1.0155,
+      "step": 5717
+    },
+    {
+      "epoch": 0.15704045480761308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9165614089384944e-05,
+      "loss": 0.9,
+      "step": 5718
+    },
+    {
+      "epoch": 0.1570679190354563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916525834103878e-05,
+      "loss": 1.0558,
+      "step": 5719
+    },
+    {
+      "epoch": 0.15709538326329955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9164902520173543e-05,
+      "loss": 1.0999,
+      "step": 5720
+    },
+    {
+      "epoch": 0.15712284749114278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9164546626792047e-05,
+      "loss": 0.9756,
+      "step": 5721
+    },
+    {
+      "epoch": 0.15715031171898602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9164190660897106e-05,
+      "loss": 0.9293,
+      "step": 5722
+    },
+    {
+      "epoch": 0.15717777594682925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916383462249154e-05,
+      "loss": 1.0978,
+      "step": 5723
+    },
+    {
+      "epoch": 0.1572052401746725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9163478511578163e-05,
+      "loss": 1.0297,
+      "step": 5724
+    },
+    {
+      "epoch": 0.15723270440251572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9163122328159795e-05,
+      "loss": 1.0556,
+      "step": 5725
+    },
+    {
+      "epoch": 0.15726016863035897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9162766072239255e-05,
+      "loss": 1.0484,
+      "step": 5726
+    },
+    {
+      "epoch": 0.1572876328582022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9162409743819363e-05,
+      "loss": 1.0293,
+      "step": 5727
+    },
+    {
+      "epoch": 0.15731509708604544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916205334290293e-05,
+      "loss": 1.0376,
+      "step": 5728
+    },
+    {
+      "epoch": 0.15734256131388866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9161696869492784e-05,
+      "loss": 1.0741,
+      "step": 5729
+    },
+    {
+      "epoch": 0.1573700255417319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9161340323591745e-05,
+      "loss": 0.9729,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15739748976957513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9160983705202632e-05,
+      "loss": 1.0853,
+      "step": 5731
+    },
+    {
+      "epoch": 0.15742495399741835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916062701432827e-05,
+      "loss": 0.9554,
+      "step": 5732
+    },
+    {
+      "epoch": 0.1574524182252616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9160270250971473e-05,
+      "loss": 0.9846,
+      "step": 5733
+    },
+    {
+      "epoch": 0.15747988245310482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915991341513507e-05,
+      "loss": 1.0831,
+      "step": 5734
+    },
+    {
+      "epoch": 0.15750734668094807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9159556506821888e-05,
+      "loss": 1.0231,
+      "step": 5735
+    },
+    {
+      "epoch": 0.1575348109087913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9159199526034743e-05,
+      "loss": 1.0887,
+      "step": 5736
+    },
+    {
+      "epoch": 0.15756227513663454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9158842472776466e-05,
+      "loss": 1.1121,
+      "step": 5737
+    },
+    {
+      "epoch": 0.15758973936447776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915848534704988e-05,
+      "loss": 1.0779,
+      "step": 5738
+    },
+    {
+      "epoch": 0.157617203592321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915812814885781e-05,
+      "loss": 1.1001,
+      "step": 5739
+    },
+    {
+      "epoch": 0.15764466782016423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9157770878203078e-05,
+      "loss": 1.0532,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15767213204800748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915741353508852e-05,
+      "loss": 1.0414,
+      "step": 5741
+    },
+    {
+      "epoch": 0.1576995962758507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9157056119516958e-05,
+      "loss": 1.0333,
+      "step": 5742
+    },
+    {
+      "epoch": 0.15772706050369395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915669863149122e-05,
+      "loss": 1.114,
+      "step": 5743
+    },
+    {
+      "epoch": 0.15775452473153717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9156341071014138e-05,
+      "loss": 1.0603,
+      "step": 5744
+    },
+    {
+      "epoch": 0.15778198895938042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9155983438088538e-05,
+      "loss": 1.0339,
+      "step": 5745
+    },
+    {
+      "epoch": 0.15780945318722364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9155625732717247e-05,
+      "loss": 1.0146,
+      "step": 5746
+    },
+    {
+      "epoch": 0.15783691741506686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91552679549031e-05,
+      "loss": 1.0844,
+      "step": 5747
+    },
+    {
+      "epoch": 0.1578643816429101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915491010464893e-05,
+      "loss": 0.996,
+      "step": 5748
+    },
+    {
+      "epoch": 0.15789184587075333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915455218195756e-05,
+      "loss": 1.0357,
+      "step": 5749
+    },
+    {
+      "epoch": 0.15791931009859658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915419418683183e-05,
+      "loss": 1.0142,
+      "step": 5750
+    },
+    {
+      "epoch": 0.1579467743264398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915383611927457e-05,
+      "loss": 1.0147,
+      "step": 5751
+    },
+    {
+      "epoch": 0.15797423855428305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915347797928861e-05,
+      "loss": 0.8563,
+      "step": 5752
+    },
+    {
+      "epoch": 0.15800170278212627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915311976687679e-05,
+      "loss": 1.1345,
+      "step": 5753
+    },
+    {
+      "epoch": 0.15802916700996952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9152761482041938e-05,
+      "loss": 0.9978,
+      "step": 5754
+    },
+    {
+      "epoch": 0.15805663123781274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915240312478689e-05,
+      "loss": 1.014,
+      "step": 5755
+    },
+    {
+      "epoch": 0.158084095465656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9152044695114487e-05,
+      "loss": 0.9344,
+      "step": 5756
+    },
+    {
+      "epoch": 0.15811155969349922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9151686193027563e-05,
+      "loss": 1.0547,
+      "step": 5757
+    },
+    {
+      "epoch": 0.15813902392134246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9151327618528948e-05,
+      "loss": 1.1443,
+      "step": 5758
+    },
+    {
+      "epoch": 0.15816648814918569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9150968971621488e-05,
+      "loss": 1.1422,
+      "step": 5759
+    },
+    {
+      "epoch": 0.1581939523770289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915061025230801e-05,
+      "loss": 0.963,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15822141660487216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9150251460591368e-05,
+      "loss": 1.066,
+      "step": 5761
+    },
+    {
+      "epoch": 0.15824888083271538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9149892596474387e-05,
+      "loss": 0.9648,
+      "step": 5762
+    },
+    {
+      "epoch": 0.15827634506055863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914953365995991e-05,
+      "loss": 1.0867,
+      "step": 5763
+    },
+    {
+      "epoch": 0.15830380928840185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9149174651050782e-05,
+      "loss": 1.0375,
+      "step": 5764
+    },
+    {
+      "epoch": 0.1583312735162451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914881556974984e-05,
+      "loss": 1.0517,
+      "step": 5765
+    },
+    {
+      "epoch": 0.15835873774408832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9148456416059925e-05,
+      "loss": 1.0239,
+      "step": 5766
+    },
+    {
+      "epoch": 0.15838620197193157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9148097189983875e-05,
+      "loss": 0.9822,
+      "step": 5767
+    },
+    {
+      "epoch": 0.1584136661997748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914773789152454e-05,
+      "loss": 1.1195,
+      "step": 5768
+    },
+    {
+      "epoch": 0.15844113042761804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914737852068476e-05,
+      "loss": 1.0461,
+      "step": 5769
+    },
+    {
+      "epoch": 0.15846859465546126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147019077467377e-05,
+      "loss": 1.1013,
+      "step": 5770
+    },
+    {
+      "epoch": 0.1584960588833045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9146659561875235e-05,
+      "loss": 0.9696,
+      "step": 5771
+    },
+    {
+      "epoch": 0.15852352311114773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914629997391118e-05,
+      "loss": 1.0592,
+      "step": 5772
+    },
+    {
+      "epoch": 0.15855098733899095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9145940313578057e-05,
+      "loss": 1.0717,
+      "step": 5773
+    },
+    {
+      "epoch": 0.1585784515668342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9145580580878712e-05,
+      "loss": 1.0447,
+      "step": 5774
+    },
+    {
+      "epoch": 0.15860591579467742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9145220775815988e-05,
+      "loss": 1.0467,
+      "step": 5775
+    },
+    {
+      "epoch": 0.15863338002252067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9144860898392737e-05,
+      "loss": 1.0883,
+      "step": 5776
+    },
+    {
+      "epoch": 0.1586608442503639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9144500948611804e-05,
+      "loss": 0.9432,
+      "step": 5777
+    },
+    {
+      "epoch": 0.15868830847820714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9144140926476037e-05,
+      "loss": 1.0031,
+      "step": 5778
+    },
+    {
+      "epoch": 0.15871577270605036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9143780831988287e-05,
+      "loss": 0.9719,
+      "step": 5779
+    },
+    {
+      "epoch": 0.1587432369338936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9143420665151397e-05,
+      "loss": 1.014,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15877070116173683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9143060425968224e-05,
+      "loss": 0.9533,
+      "step": 5781
+    },
+    {
+      "epoch": 0.15879816538958008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9142700114441614e-05,
+      "loss": 1.0703,
+      "step": 5782
+    },
+    {
+      "epoch": 0.1588256296174233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9142339730574416e-05,
+      "loss": 1.0135,
+      "step": 5783
+    },
+    {
+      "epoch": 0.15885309384526655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914197927436949e-05,
+      "loss": 1.0289,
+      "step": 5784
+    },
+    {
+      "epoch": 0.15888055807310977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914161874582968e-05,
+      "loss": 0.9379,
+      "step": 5785
+    },
+    {
+      "epoch": 0.15890802230095302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914125814495784e-05,
+      "loss": 1.0641,
+      "step": 5786
+    },
+    {
+      "epoch": 0.15893548652879624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914089747175683e-05,
+      "loss": 1.0839,
+      "step": 5787
+    },
+    {
+      "epoch": 0.15896295075663947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9140536726229494e-05,
+      "loss": 1.0161,
+      "step": 5788
+    },
+    {
+      "epoch": 0.15899041498448271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914017590837869e-05,
+      "loss": 1.0104,
+      "step": 5789
+    },
+    {
+      "epoch": 0.15901787921232594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9139815018207275e-05,
+      "loss": 1.0914,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15904534344016918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9139454055718105e-05,
+      "loss": 0.9914,
+      "step": 5791
+    },
+    {
+      "epoch": 0.1590728076680124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913909302091403e-05,
+      "loss": 0.9668,
+      "step": 5792
+    },
+    {
+      "epoch": 0.15910027189585565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9138731913797915e-05,
+      "loss": 1.0724,
+      "step": 5793
+    },
+    {
+      "epoch": 0.15912773612369888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9138370734372613e-05,
+      "loss": 0.9386,
+      "step": 5794
+    },
+    {
+      "epoch": 0.15915520035154213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913800948264098e-05,
+      "loss": 1.0848,
+      "step": 5795
+    },
+    {
+      "epoch": 0.15918266457938535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913764815860588e-05,
+      "loss": 1.1593,
+      "step": 5796
+    },
+    {
+      "epoch": 0.1592101288072286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9137286762270163e-05,
+      "loss": 0.9698,
+      "step": 5797
+    },
+    {
+      "epoch": 0.15923759303507182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9136925293636695e-05,
+      "loss": 1.0569,
+      "step": 5798
+    },
+    {
+      "epoch": 0.15926505726291507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9136563752708335e-05,
+      "loss": 1.0401,
+      "step": 5799
+    },
+    {
+      "epoch": 0.1592925214907583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9136202139487943e-05,
+      "loss": 1.0499,
+      "step": 5800
+    },
+    {
+      "epoch": 0.1593199857186015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9135840453978384e-05,
+      "loss": 1.0862,
+      "step": 5801
+    },
+    {
+      "epoch": 0.15934744994644476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913547869618251e-05,
+      "loss": 1.0433,
+      "step": 5802
+    },
+    {
+      "epoch": 0.15937491417428798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9135116866103196e-05,
+      "loss": 1.0742,
+      "step": 5803
+    },
+    {
+      "epoch": 0.15940237840213123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9134754963743296e-05,
+      "loss": 0.9278,
+      "step": 5804
+    },
+    {
+      "epoch": 0.15942984262997445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9134392989105678e-05,
+      "loss": 1.049,
+      "step": 5805
+    },
+    {
+      "epoch": 0.1594573068578177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91340309421932e-05,
+      "loss": 1.1564,
+      "step": 5806
+    },
+    {
+      "epoch": 0.15948477108566092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9133668823008737e-05,
+      "loss": 1.0958,
+      "step": 5807
+    },
+    {
+      "epoch": 0.15951223531350417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9133306631555143e-05,
+      "loss": 1.0046,
+      "step": 5808
+    },
+    {
+      "epoch": 0.1595396995413474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913294436783529e-05,
+      "loss": 1.1187,
+      "step": 5809
+    },
+    {
+      "epoch": 0.15956716376919064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913258203185205e-05,
+      "loss": 1.1111,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15959462799703386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9132219623608273e-05,
+      "loss": 0.9656,
+      "step": 5811
+    },
+    {
+      "epoch": 0.1596220922248771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9131857143106845e-05,
+      "loss": 1.0472,
+      "step": 5812
+    },
+    {
+      "epoch": 0.15964955645272033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9131494590350622e-05,
+      "loss": 1.0543,
+      "step": 5813
+    },
+    {
+      "epoch": 0.15967702068056355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9131131965342476e-05,
+      "loss": 0.9734,
+      "step": 5814
+    },
+    {
+      "epoch": 0.1597044849084068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913076926808528e-05,
+      "loss": 0.9925,
+      "step": 5815
+    },
+    {
+      "epoch": 0.15973194913625002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9130406498581898e-05,
+      "loss": 1.097,
+      "step": 5816
+    },
+    {
+      "epoch": 0.15975941336409327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9130043656835205e-05,
+      "loss": 1.0005,
+      "step": 5817
+    },
+    {
+      "epoch": 0.1597868775919365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9129680742848067e-05,
+      "loss": 0.909,
+      "step": 5818
+    },
+    {
+      "epoch": 0.15981434181977974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912931775662336e-05,
+      "loss": 1.0137,
+      "step": 5819
+    },
+    {
+      "epoch": 0.15984180604762296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9128954698163956e-05,
+      "loss": 1.0944,
+      "step": 5820
+    },
+    {
+      "epoch": 0.1598692702754662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9128591567472725e-05,
+      "loss": 0.9763,
+      "step": 5821
+    },
+    {
+      "epoch": 0.15989673450330943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9128228364552543e-05,
+      "loss": 1.0004,
+      "step": 5822
+    },
+    {
+      "epoch": 0.15992419873115268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912786508940628e-05,
+      "loss": 1.0295,
+      "step": 5823
+    },
+    {
+      "epoch": 0.1599516629589959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9127501742036812e-05,
+      "loss": 1.0958,
+      "step": 5824
+    },
+    {
+      "epoch": 0.15997912718683915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9127138322447018e-05,
+      "loss": 1.0009,
+      "step": 5825
+    },
+    {
+      "epoch": 0.16000659141468238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9126774830639766e-05,
+      "loss": 0.996,
+      "step": 5826
+    },
+    {
+      "epoch": 0.1600340556425256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912641126661794e-05,
+      "loss": 1.1247,
+      "step": 5827
+    },
+    {
+      "epoch": 0.16006151987036885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912604763038441e-05,
+      "loss": 1.036,
+      "step": 5828
+    },
+    {
+      "epoch": 0.16008898409821207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9125683921942057e-05,
+      "loss": 1.0522,
+      "step": 5829
+    },
+    {
+      "epoch": 0.16011644832605532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912532014129376e-05,
+      "loss": 1.1985,
+      "step": 5830
+    },
+    {
+      "epoch": 0.16014391255389854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912495628844239e-05,
+      "loss": 1.0468,
+      "step": 5831
+    },
+    {
+      "epoch": 0.1601713767817418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9124592363390834e-05,
+      "loss": 1.007,
+      "step": 5832
+    },
+    {
+      "epoch": 0.160198841009585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912422836614197e-05,
+      "loss": 1.1132,
+      "step": 5833
+    },
+    {
+      "epoch": 0.16022630523742826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123864296698677e-05,
+      "loss": 1.1214,
+      "step": 5834
+    },
+    {
+      "epoch": 0.16025376946527148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123500155063836e-05,
+      "loss": 1.037,
+      "step": 5835
+    },
+    {
+      "epoch": 0.16028123369311473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123135941240323e-05,
+      "loss": 0.9332,
+      "step": 5836
+    },
+    {
+      "epoch": 0.16030869792095795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9122771655231027e-05,
+      "loss": 1.0539,
+      "step": 5837
+    },
+    {
+      "epoch": 0.1603361621488012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912240729703883e-05,
+      "loss": 1.0158,
+      "step": 5838
+    },
+    {
+      "epoch": 0.16036362637664442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9122042866666608e-05,
+      "loss": 0.9904,
+      "step": 5839
+    },
+    {
+      "epoch": 0.16039109060448767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9121678364117255e-05,
+      "loss": 1.0466,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1604185548323309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9121313789393644e-05,
+      "loss": 1.0119,
+      "step": 5841
+    },
+    {
+      "epoch": 0.1604460190601741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912094914249867e-05,
+      "loss": 1.0191,
+      "step": 5842
+    },
+    {
+      "epoch": 0.16047348328801736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912058442343521e-05,
+      "loss": 0.9108,
+      "step": 5843
+    },
+    {
+      "epoch": 0.16050094751586058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9120219632206153e-05,
+      "loss": 1.0922,
+      "step": 5844
+    },
+    {
+      "epoch": 0.16052841174370383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9119854768814388e-05,
+      "loss": 1.0506,
+      "step": 5845
+    },
+    {
+      "epoch": 0.16055587597154705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9119489833262796e-05,
+      "loss": 0.9942,
+      "step": 5846
+    },
+    {
+      "epoch": 0.1605833401993903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911912482555427e-05,
+      "loss": 0.8902,
+      "step": 5847
+    },
+    {
+      "epoch": 0.16061080442723352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9118759745691698e-05,
+      "loss": 1.0602,
+      "step": 5848
+    },
+    {
+      "epoch": 0.16063826865507677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9118394593677963e-05,
+      "loss": 1.118,
+      "step": 5849
+    },
+    {
+      "epoch": 0.16066573288292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9118029369515956e-05,
+      "loss": 1.0522,
+      "step": 5850
+    },
+    {
+      "epoch": 0.16069319711076324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9117664073208576e-05,
+      "loss": 1.0125,
+      "step": 5851
+    },
+    {
+      "epoch": 0.16072066133860646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91172987047587e-05,
+      "loss": 1.0034,
+      "step": 5852
+    },
+    {
+      "epoch": 0.1607481255664497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9116933264169227e-05,
+      "loss": 0.973,
+      "step": 5853
+    },
+    {
+      "epoch": 0.16077558979429293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9116567751443045e-05,
+      "loss": 1.086,
+      "step": 5854
+    },
+    {
+      "epoch": 0.16080305402213615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911620216658305e-05,
+      "loss": 1.1246,
+      "step": 5855
+    },
+    {
+      "epoch": 0.1608305182499794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911583650959213e-05,
+      "loss": 0.9086,
+      "step": 5856
+    },
+    {
+      "epoch": 0.16085798247782263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911547078047318e-05,
+      "loss": 1.0123,
+      "step": 5857
+    },
+    {
+      "epoch": 0.16088544670566587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91151049792291e-05,
+      "loss": 1.0306,
+      "step": 5858
+    },
+    {
+      "epoch": 0.1609129109335091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9114739105862773e-05,
+      "loss": 1.0924,
+      "step": 5859
+    },
+    {
+      "epoch": 0.16094037516135234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9114373160377097e-05,
+      "loss": 0.9827,
+      "step": 5860
+    },
+    {
+      "epoch": 0.16096783938919557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9114007142774975e-05,
+      "loss": 1.0443,
+      "step": 5861
+    },
+    {
+      "epoch": 0.16099530361703882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9113641053059297e-05,
+      "loss": 1.0896,
+      "step": 5862
+    },
+    {
+      "epoch": 0.16102276784488204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911327489123296e-05,
+      "loss": 1.0297,
+      "step": 5863
+    },
+    {
+      "epoch": 0.16105023207272529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9112908657298865e-05,
+      "loss": 1.0177,
+      "step": 5864
+    },
+    {
+      "epoch": 0.1610776963005685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9112542351259903e-05,
+      "loss": 1.072,
+      "step": 5865
+    },
+    {
+      "epoch": 0.16110516052841176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9112175973118977e-05,
+      "loss": 1.0021,
+      "step": 5866
+    },
+    {
+      "epoch": 0.16113262475625498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9111809522878985e-05,
+      "loss": 1.0047,
+      "step": 5867
+    },
+    {
+      "epoch": 0.1611600889840982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9111443000542825e-05,
+      "loss": 1.0577,
+      "step": 5868
+    },
+    {
+      "epoch": 0.16118755321194145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91110764061134e-05,
+      "loss": 1.0394,
+      "step": 5869
+    },
+    {
+      "epoch": 0.16121501743978467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911070973959361e-05,
+      "loss": 1.0365,
+      "step": 5870
+    },
+    {
+      "epoch": 0.16124248166762792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9110343000986353e-05,
+      "loss": 1.1059,
+      "step": 5871
+    },
+    {
+      "epoch": 0.16126994589547114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9109976190294533e-05,
+      "loss": 0.9691,
+      "step": 5872
+    },
+    {
+      "epoch": 0.1612974101233144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9109609307521057e-05,
+      "loss": 1.0685,
+      "step": 5873
+    },
+    {
+      "epoch": 0.1613248743511576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910924235266882e-05,
+      "loss": 0.99,
+      "step": 5874
+    },
+    {
+      "epoch": 0.16135233857900086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9108875325740727e-05,
+      "loss": 1.0244,
+      "step": 5875
+    },
+    {
+      "epoch": 0.16137980280684408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9108508226739692e-05,
+      "loss": 0.9844,
+      "step": 5876
+    },
+    {
+      "epoch": 0.16140726703468733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9108141055668604e-05,
+      "loss": 1.0999,
+      "step": 5877
+    },
+    {
+      "epoch": 0.16143473126253055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9107773812530378e-05,
+      "loss": 1.0531,
+      "step": 5878
+    },
+    {
+      "epoch": 0.1614621954903738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910740649732792e-05,
+      "loss": 1.0991,
+      "step": 5879
+    },
+    {
+      "epoch": 0.16148965971821702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910703911006413e-05,
+      "loss": 1.013,
+      "step": 5880
+    },
+    {
+      "epoch": 0.16151712394606027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910667165074192e-05,
+      "loss": 0.9918,
+      "step": 5881
+    },
+    {
+      "epoch": 0.1615445881739035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9106304119364196e-05,
+      "loss": 0.9431,
+      "step": 5882
+    },
+    {
+      "epoch": 0.1615720524017467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9105936515933865e-05,
+      "loss": 0.9995,
+      "step": 5883
+    },
+    {
+      "epoch": 0.16159951662958996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9105568840453837e-05,
+      "loss": 1.0368,
+      "step": 5884
+    },
+    {
+      "epoch": 0.16162698085743318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9105201092927027e-05,
+      "loss": 1.081,
+      "step": 5885
+    },
+    {
+      "epoch": 0.16165444508527643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9104833273356334e-05,
+      "loss": 1.1735,
+      "step": 5886
+    },
+    {
+      "epoch": 0.16168190931311965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9104465381744675e-05,
+      "loss": 0.9639,
+      "step": 5887
+    },
+    {
+      "epoch": 0.1617093735409629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910409741809496e-05,
+      "loss": 0.9634,
+      "step": 5888
+    },
+    {
+      "epoch": 0.16173683776880612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9103729382410098e-05,
+      "loss": 1.0589,
+      "step": 5889
+    },
+    {
+      "epoch": 0.16176430199664937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9103361274693002e-05,
+      "loss": 1.079,
+      "step": 5890
+    },
+    {
+      "epoch": 0.1617917662244926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9102993094946583e-05,
+      "loss": 0.9503,
+      "step": 5891
+    },
+    {
+      "epoch": 0.16181923045233584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910262484317376e-05,
+      "loss": 1.0945,
+      "step": 5892
+    },
+    {
+      "epoch": 0.16184669468017907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9102256519377443e-05,
+      "loss": 1.106,
+      "step": 5893
+    },
+    {
+      "epoch": 0.16187415890802231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9101888123560546e-05,
+      "loss": 1.0223,
+      "step": 5894
+    },
+    {
+      "epoch": 0.16190162313586554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9101519655725988e-05,
+      "loss": 1.0971,
+      "step": 5895
+    },
+    {
+      "epoch": 0.16192908736370876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9101151115876674e-05,
+      "loss": 1.0648,
+      "step": 5896
+    },
+    {
+      "epoch": 0.161956551591552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9100782504015533e-05,
+      "loss": 1.0441,
+      "step": 5897
+    },
+    {
+      "epoch": 0.16198401581939523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9100413820145475e-05,
+      "loss": 1.0274,
+      "step": 5898
+    },
+    {
+      "epoch": 0.16201148004723848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910004506426941e-05,
+      "loss": 0.957,
+      "step": 5899
+    },
+    {
+      "epoch": 0.1620389442750817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909967623639027e-05,
+      "loss": 1.0708,
+      "step": 5900
+    },
+    {
+      "epoch": 0.16206640850292495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9099307336510968e-05,
+      "loss": 1.0829,
+      "step": 5901
+    },
+    {
+      "epoch": 0.16209387273076817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909893836463442e-05,
+      "loss": 1.1168,
+      "step": 5902
+    },
+    {
+      "epoch": 0.16212133695861142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9098569320763546e-05,
+      "loss": 0.8955,
+      "step": 5903
+    },
+    {
+      "epoch": 0.16214880118645464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909820020490127e-05,
+      "loss": 1.0936,
+      "step": 5904
+    },
+    {
+      "epoch": 0.1621762654142979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9097831017050506e-05,
+      "loss": 1.048,
+      "step": 5905
+    },
+    {
+      "epoch": 0.1622037296421411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9097461757214178e-05,
+      "loss": 1.039,
+      "step": 5906
+    },
+    {
+      "epoch": 0.16223119386998436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9097092425395215e-05,
+      "loss": 1.072,
+      "step": 5907
+    },
+    {
+      "epoch": 0.16225865809782758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9096723021596527e-05,
+      "loss": 0.9993,
+      "step": 5908
+    },
+    {
+      "epoch": 0.1622861223256708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9096353545821045e-05,
+      "loss": 0.9979,
+      "step": 5909
+    },
+    {
+      "epoch": 0.16231358655351405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9095983998071686e-05,
+      "loss": 1.027,
+      "step": 5910
+    },
+    {
+      "epoch": 0.16234105078135727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9095614378351385e-05,
+      "loss": 1.1152,
+      "step": 5911
+    },
+    {
+      "epoch": 0.16236851500920052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9095244686663058e-05,
+      "loss": 1.1649,
+      "step": 5912
+    },
+    {
+      "epoch": 0.16239597923704374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909487492300963e-05,
+      "loss": 1.0266,
+      "step": 5913
+    },
+    {
+      "epoch": 0.162423443464887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9094505087394026e-05,
+      "loss": 1.0634,
+      "step": 5914
+    },
+    {
+      "epoch": 0.1624509076927302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909413517981918e-05,
+      "loss": 0.9919,
+      "step": 5915
+    },
+    {
+      "epoch": 0.16247837192057346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909376520028801e-05,
+      "loss": 1.0005,
+      "step": 5916
+    },
+    {
+      "epoch": 0.16250583614841668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9093395148803447e-05,
+      "loss": 1.0396,
+      "step": 5917
+    },
+    {
+      "epoch": 0.16253330037625993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9093025025368423e-05,
+      "loss": 1.1389,
+      "step": 5918
+    },
+    {
+      "epoch": 0.16256076460410315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909265482998586e-05,
+      "loss": 1.0074,
+      "step": 5919
+    },
+    {
+      "epoch": 0.1625882288319464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909228456265869e-05,
+      "loss": 1.0966,
+      "step": 5920
+    },
+    {
+      "epoch": 0.16261569305978962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9091914223389843e-05,
+      "loss": 1.067,
+      "step": 5921
+    },
+    {
+      "epoch": 0.16264315728763284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9091543812182247e-05,
+      "loss": 1.0684,
+      "step": 5922
+    },
+    {
+      "epoch": 0.1626706215154761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909117332903884e-05,
+      "loss": 1.06,
+      "step": 5923
+    },
+    {
+      "epoch": 0.16269808574331932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909080277396254e-05,
+      "loss": 1.074,
+      "step": 5924
+    },
+    {
+      "epoch": 0.16272554997116256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9090432146956293e-05,
+      "loss": 0.9692,
+      "step": 5925
+    },
+    {
+      "epoch": 0.16275301419900579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9090061448023026e-05,
+      "loss": 0.8912,
+      "step": 5926
+    },
+    {
+      "epoch": 0.16278047842684903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9089690677165667e-05,
+      "loss": 1.0694,
+      "step": 5927
+    },
+    {
+      "epoch": 0.16280794265469226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9089319834387158e-05,
+      "loss": 1.0587,
+      "step": 5928
+    },
+    {
+      "epoch": 0.1628354068825355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9088948919690428e-05,
+      "loss": 1.0792,
+      "step": 5929
+    },
+    {
+      "epoch": 0.16286287111037873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9088577933078412e-05,
+      "loss": 0.9817,
+      "step": 5930
+    },
+    {
+      "epoch": 0.16289033533822198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908820687455405e-05,
+      "loss": 1.0814,
+      "step": 5931
+    },
+    {
+      "epoch": 0.1629177995660652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9087835744120277e-05,
+      "loss": 0.92,
+      "step": 5932
+    },
+    {
+      "epoch": 0.16294526379390845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908746454178002e-05,
+      "loss": 1.1049,
+      "step": 5933
+    },
+    {
+      "epoch": 0.16297272802175167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9087093267536227e-05,
+      "loss": 1.1685,
+      "step": 5934
+    },
+    {
+      "epoch": 0.16300019224959492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9086721921391834e-05,
+      "loss": 1.0131,
+      "step": 5935
+    },
+    {
+      "epoch": 0.16302765647743814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9086350503349777e-05,
+      "loss": 1.0377,
+      "step": 5936
+    },
+    {
+      "epoch": 0.16305512070528136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9085979013412993e-05,
+      "loss": 0.9989,
+      "step": 5937
+    },
+    {
+      "epoch": 0.1630825849331246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9085607451584423e-05,
+      "loss": 0.9467,
+      "step": 5938
+    },
+    {
+      "epoch": 0.16311004916096783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9085235817867008e-05,
+      "loss": 0.998,
+      "step": 5939
+    },
+    {
+      "epoch": 0.16313751338881108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908486411226369e-05,
+      "loss": 1.0046,
+      "step": 5940
+    },
+    {
+      "epoch": 0.1631649776166543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9084492334777404e-05,
+      "loss": 1.0723,
+      "step": 5941
+    },
+    {
+      "epoch": 0.16319244184449755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90841204854111e-05,
+      "loss": 1.0791,
+      "step": 5942
+    },
+    {
+      "epoch": 0.16321990607234077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9083748564167714e-05,
+      "loss": 1.0422,
+      "step": 5943
+    },
+    {
+      "epoch": 0.16324737030018402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908337657105019e-05,
+      "loss": 1.1089,
+      "step": 5944
+    },
+    {
+      "epoch": 0.16327483452802724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9083004506061472e-05,
+      "loss": 0.9013,
+      "step": 5945
+    },
+    {
+      "epoch": 0.1633022987558705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9082632369204504e-05,
+      "loss": 1.0526,
+      "step": 5946
+    },
+    {
+      "epoch": 0.1633297629837137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9082260160482233e-05,
+      "loss": 1.0944,
+      "step": 5947
+    },
+    {
+      "epoch": 0.16335722721155696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9081887879897597e-05,
+      "loss": 0.994,
+      "step": 5948
+    },
+    {
+      "epoch": 0.16338469143940018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908151552745355e-05,
+      "loss": 0.9894,
+      "step": 5949
+    },
+    {
+      "epoch": 0.1634121556672434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908114310315303e-05,
+      "loss": 1.0514,
+      "step": 5950
+    },
+    {
+      "epoch": 0.16343961989508665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9080770606998996e-05,
+      "loss": 1.077,
+      "step": 5951
+    },
+    {
+      "epoch": 0.16346708412292987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908039803899438e-05,
+      "loss": 1.0519,
+      "step": 5952
+    },
+    {
+      "epoch": 0.16349454835077312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908002539914214e-05,
+      "loss": 1.0213,
+      "step": 5953
+    },
+    {
+      "epoch": 0.16352201257861634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9079652687445223e-05,
+      "loss": 0.9854,
+      "step": 5954
+    },
+    {
+      "epoch": 0.1635494768064596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9079279903906573e-05,
+      "loss": 1.1005,
+      "step": 5955
+    },
+    {
+      "epoch": 0.16357694103430281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9078907048529144e-05,
+      "loss": 1.0756,
+      "step": 5956
+    },
+    {
+      "epoch": 0.16360440526214606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9078534121315887e-05,
+      "loss": 1.0662,
+      "step": 5957
+    },
+    {
+      "epoch": 0.16363186948998928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9078161122269755e-05,
+      "loss": 1.0179,
+      "step": 5958
+    },
+    {
+      "epoch": 0.16365933371783253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907778805139369e-05,
+      "loss": 1.0359,
+      "step": 5959
+    },
+    {
+      "epoch": 0.16368679794567575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9077414908690654e-05,
+      "loss": 1.0207,
+      "step": 5960
+    },
+    {
+      "epoch": 0.163714262173519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907704169416359e-05,
+      "loss": 1.1345,
+      "step": 5961
+    },
+    {
+      "epoch": 0.16374172640136223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9076668407815463e-05,
+      "loss": 0.9648,
+      "step": 5962
+    },
+    {
+      "epoch": 0.16376919062920545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9076295049649214e-05,
+      "loss": 0.9729,
+      "step": 5963
+    },
+    {
+      "epoch": 0.1637966548570487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9075921619667802e-05,
+      "loss": 1.0702,
+      "step": 5964
+    },
+    {
+      "epoch": 0.16382411908489192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9075548117874184e-05,
+      "loss": 1.0216,
+      "step": 5965
+    },
+    {
+      "epoch": 0.16385158331273517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9075174544271316e-05,
+      "loss": 1.0416,
+      "step": 5966
+    },
+    {
+      "epoch": 0.1638790475405784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907480089886215e-05,
+      "loss": 0.9914,
+      "step": 5967
+    },
+    {
+      "epoch": 0.16390651176842164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9074427181649646e-05,
+      "loss": 1.0115,
+      "step": 5968
+    },
+    {
+      "epoch": 0.16393397599626486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9074053392636755e-05,
+      "loss": 1.045,
+      "step": 5969
+    },
+    {
+      "epoch": 0.1639614402241081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907367953182644e-05,
+      "loss": 1.1144,
+      "step": 5970
+    },
+    {
+      "epoch": 0.16398890445195133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907330559922166e-05,
+      "loss": 1.0213,
+      "step": 5971
+    },
+    {
+      "epoch": 0.16401636867979458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907293159482537e-05,
+      "loss": 1.0242,
+      "step": 5972
+    },
+    {
+      "epoch": 0.1640438329076378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907255751864053e-05,
+      "loss": 1.0523,
+      "step": 5973
+    },
+    {
+      "epoch": 0.16407129713548105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9072183370670102e-05,
+      "loss": 1.0016,
+      "step": 5974
+    },
+    {
+      "epoch": 0.16409876136332427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9071809150917043e-05,
+      "loss": 1.022,
+      "step": 5975
+    },
+    {
+      "epoch": 0.1641262255911675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9071434859384315e-05,
+      "loss": 1.0552,
+      "step": 5976
+    },
+    {
+      "epoch": 0.16415368981901074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9071060496074886e-05,
+      "loss": 1.0826,
+      "step": 5977
+    },
+    {
+      "epoch": 0.16418115404685396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9070686060991707e-05,
+      "loss": 0.9908,
+      "step": 5978
+    },
+    {
+      "epoch": 0.1642086182746972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907031155413775e-05,
+      "loss": 0.9379,
+      "step": 5979
+    },
+    {
+      "epoch": 0.16423608250254043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906993697551597e-05,
+      "loss": 0.9786,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16426354673038368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9069562325129342e-05,
+      "loss": 1.0508,
+      "step": 5981
+    },
+    {
+      "epoch": 0.1642910109582269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9069187602980815e-05,
+      "loss": 1.0165,
+      "step": 5982
+    },
+    {
+      "epoch": 0.16431847518607015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9068812809073368e-05,
+      "loss": 0.9906,
+      "step": 5983
+    },
+    {
+      "epoch": 0.16434593941391337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9068437943409962e-05,
+      "loss": 1.0355,
+      "step": 5984
+    },
+    {
+      "epoch": 0.16437340364175662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906806300599356e-05,
+      "loss": 1.0101,
+      "step": 5985
+    },
+    {
+      "epoch": 0.16440086786959984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906768799682713e-05,
+      "loss": 1.1035,
+      "step": 5986
+    },
+    {
+      "epoch": 0.1644283320974431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9067312915913642e-05,
+      "loss": 1.1009,
+      "step": 5987
+    },
+    {
+      "epoch": 0.1644557963252863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9066937763256063e-05,
+      "loss": 1.0996,
+      "step": 5988
+    },
+    {
+      "epoch": 0.16448326055312956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906656253885736e-05,
+      "loss": 1.0628,
+      "step": 5989
+    },
+    {
+      "epoch": 0.16451072478097278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90661872427205e-05,
+      "loss": 0.9876,
+      "step": 5990
+    },
+    {
+      "epoch": 0.164538189008816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9065811874848455e-05,
+      "loss": 1.0166,
+      "step": 5991
+    },
+    {
+      "epoch": 0.16456565323665925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9065436435244194e-05,
+      "loss": 0.9769,
+      "step": 5992
+    },
+    {
+      "epoch": 0.16459311746450248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906506092391069e-05,
+      "loss": 0.9166,
+      "step": 5993
+    },
+    {
+      "epoch": 0.16462058169234572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9064685340850913e-05,
+      "loss": 1.1465,
+      "step": 5994
+    },
+    {
+      "epoch": 0.16464804592018895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906430968606783e-05,
+      "loss": 1.0035,
+      "step": 5995
+    },
+    {
+      "epoch": 0.1646755101480322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9063933959564424e-05,
+      "loss": 0.9865,
+      "step": 5996
+    },
+    {
+      "epoch": 0.16470297437587542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906355816134366e-05,
+      "loss": 0.9928,
+      "step": 5997
+    },
+    {
+      "epoch": 0.16473043860371867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906318229140851e-05,
+      "loss": 1.1053,
+      "step": 5998
+    },
+    {
+      "epoch": 0.1647579028315619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906280634976195e-05,
+      "loss": 1.1034,
+      "step": 5999
+    },
+    {
+      "epoch": 0.16478536705940514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906243033640696e-05,
+      "loss": 1.0128,
+      "step": 6000
+    },
+    {
+      "epoch": 0.16481283128724836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9062054251346506e-05,
+      "loss": 0.9686,
+      "step": 6001
+    },
+    {
+      "epoch": 0.1648402955150916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9061678094583572e-05,
+      "loss": 1.0479,
+      "step": 6002
+    },
+    {
+      "epoch": 0.16486775974293483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9061301866121132e-05,
+      "loss": 0.995,
+      "step": 6003
+    },
+    {
+      "epoch": 0.16489522397077805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9060925565962158e-05,
+      "loss": 1.0577,
+      "step": 6004
+    },
+    {
+      "epoch": 0.1649226881986213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9060549194109633e-05,
+      "loss": 0.9602,
+      "step": 6005
+    },
+    {
+      "epoch": 0.16495015242646452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906017275056653e-05,
+      "loss": 1.1005,
+      "step": 6006
+    },
+    {
+      "epoch": 0.16497761665430777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9059796235335835e-05,
+      "loss": 1.0481,
+      "step": 6007
+    },
+    {
+      "epoch": 0.165005080882151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905941964842052e-05,
+      "loss": 0.8976,
+      "step": 6008
+    },
+    {
+      "epoch": 0.16503254510999424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905904298982357e-05,
+      "loss": 0.9618,
+      "step": 6009
+    },
+    {
+      "epoch": 0.16506000933783746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905866625954796e-05,
+      "loss": 1.0482,
+      "step": 6010
+    },
+    {
+      "epoch": 0.1650874735656807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905828945759667e-05,
+      "loss": 1.144,
+      "step": 6011
+    },
+    {
+      "epoch": 0.16511493779352393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905791258397269e-05,
+      "loss": 1.0381,
+      "step": 6012
+    },
+    {
+      "epoch": 0.16514240202136718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9057535638678998e-05,
+      "loss": 1.0239,
+      "step": 6013
+    },
+    {
+      "epoch": 0.1651698662492104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9057158621718574e-05,
+      "loss": 0.9675,
+      "step": 6014
+    },
+    {
+      "epoch": 0.16519733047705365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9056781533094404e-05,
+      "loss": 1.0504,
+      "step": 6015
+    },
+    {
+      "epoch": 0.16522479470489687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905640437280947e-05,
+      "loss": 1.1124,
+      "step": 6016
+    },
+    {
+      "epoch": 0.1652522589327401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9056027140866752e-05,
+      "loss": 1.1495,
+      "step": 6017
+    },
+    {
+      "epoch": 0.16527972316058334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9055649837269244e-05,
+      "loss": 1.0911,
+      "step": 6018
+    },
+    {
+      "epoch": 0.16530718738842656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9055272462019925e-05,
+      "loss": 1.0623,
+      "step": 6019
+    },
+    {
+      "epoch": 0.1653346516162698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054895015121784e-05,
+      "loss": 0.9424,
+      "step": 6020
+    },
+    {
+      "epoch": 0.16536211584411303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054517496577804e-05,
+      "loss": 0.9337,
+      "step": 6021
+    },
+    {
+      "epoch": 0.16538958007195628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054139906390978e-05,
+      "loss": 1.0994,
+      "step": 6022
+    },
+    {
+      "epoch": 0.1654170442997995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9053762244564283e-05,
+      "loss": 1.0608,
+      "step": 6023
+    },
+    {
+      "epoch": 0.16544450852764275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905338451110072e-05,
+      "loss": 1.0214,
+      "step": 6024
+    },
+    {
+      "epoch": 0.16547197275548597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905300670600327e-05,
+      "loss": 1.049,
+      "step": 6025
+    },
+    {
+      "epoch": 0.16549943698332922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9052628829274923e-05,
+      "loss": 1.0342,
+      "step": 6026
+    },
+    {
+      "epoch": 0.16552690121117244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905225088091867e-05,
+      "loss": 1.0709,
+      "step": 6027
+    },
+    {
+      "epoch": 0.1655543654390157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9051872860937504e-05,
+      "loss": 0.985,
+      "step": 6028
+    },
+    {
+      "epoch": 0.16558182966685892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905149476933441e-05,
+      "loss": 0.9875,
+      "step": 6029
+    },
+    {
+      "epoch": 0.16560929389470216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9051116606112388e-05,
+      "loss": 0.9824,
+      "step": 6030
+    },
+    {
+      "epoch": 0.16563675812254539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9050738371274423e-05,
+      "loss": 0.9948,
+      "step": 6031
+    },
+    {
+      "epoch": 0.1656642223503886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9050360064823506e-05,
+      "loss": 1.1194,
+      "step": 6032
+    },
+    {
+      "epoch": 0.16569168657823186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9049981686762642e-05,
+      "loss": 0.9566,
+      "step": 6033
+    },
+    {
+      "epoch": 0.16571915080607508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904960323709481e-05,
+      "loss": 1.0087,
+      "step": 6034
+    },
+    {
+      "epoch": 0.16574661503391833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9049224715823017e-05,
+      "loss": 1.0694,
+      "step": 6035
+    },
+    {
+      "epoch": 0.16577407926176155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904884612295025e-05,
+      "loss": 1.0527,
+      "step": 6036
+    },
+    {
+      "epoch": 0.1658015434896048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9048467458479505e-05,
+      "loss": 0.9778,
+      "step": 6037
+    },
+    {
+      "epoch": 0.16582900771744802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9048088722413786e-05,
+      "loss": 1.1996,
+      "step": 6038
+    },
+    {
+      "epoch": 0.16585647194529127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904770991475608e-05,
+      "loss": 1.0578,
+      "step": 6039
+    },
+    {
+      "epoch": 0.1658839361731345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9047331035509394e-05,
+      "loss": 1.0491,
+      "step": 6040
+    },
+    {
+      "epoch": 0.16591140040097774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9046952084676716e-05,
+      "loss": 1.0359,
+      "step": 6041
+    },
+    {
+      "epoch": 0.16593886462882096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9046573062261046e-05,
+      "loss": 1.0514,
+      "step": 6042
+    },
+    {
+      "epoch": 0.1659663288566642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904619396826539e-05,
+      "loss": 0.9889,
+      "step": 6043
+    },
+    {
+      "epoch": 0.16599379308450743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9045814802692742e-05,
+      "loss": 1.0387,
+      "step": 6044
+    },
+    {
+      "epoch": 0.16602125731235065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9045435565546102e-05,
+      "loss": 1.0582,
+      "step": 6045
+    },
+    {
+      "epoch": 0.1660487215401939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9045056256828475e-05,
+      "loss": 0.8449,
+      "step": 6046
+    },
+    {
+      "epoch": 0.16607618576803712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9044676876542855e-05,
+      "loss": 1.0621,
+      "step": 6047
+    },
+    {
+      "epoch": 0.16610364999588037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904429742469225e-05,
+      "loss": 1.054,
+      "step": 6048
+    },
+    {
+      "epoch": 0.1661311142237236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904391790127966e-05,
+      "loss": 1.0205,
+      "step": 6049
+    },
+    {
+      "epoch": 0.16615857845156684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904353830630809e-05,
+      "loss": 1.0222,
+      "step": 6050
+    },
+    {
+      "epoch": 0.16618604267941006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904315863978054e-05,
+      "loss": 1.017,
+      "step": 6051
+    },
+    {
+      "epoch": 0.1662135069072533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9042778901700016e-05,
+      "loss": 1.0927,
+      "step": 6052
+    },
+    {
+      "epoch": 0.16624097113509653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904239909206952e-05,
+      "loss": 1.0789,
+      "step": 6053
+    },
+    {
+      "epoch": 0.16626843536293978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9042019210892065e-05,
+      "loss": 1.0104,
+      "step": 6054
+    },
+    {
+      "epoch": 0.166295899590783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904163925817065e-05,
+      "loss": 1.0998,
+      "step": 6055
+    },
+    {
+      "epoch": 0.16632336381862625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9041259233908282e-05,
+      "loss": 0.954,
+      "step": 6056
+    },
+    {
+      "epoch": 0.16635082804646947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9040879138107965e-05,
+      "loss": 1.0372,
+      "step": 6057
+    },
+    {
+      "epoch": 0.1663782922743127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9040498970772717e-05,
+      "loss": 0.9563,
+      "step": 6058
+    },
+    {
+      "epoch": 0.16640575650215594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9040118731905534e-05,
+      "loss": 0.9404,
+      "step": 6059
+    },
+    {
+      "epoch": 0.16643322072999917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903973842150943e-05,
+      "loss": 0.9819,
+      "step": 6060
+    },
+    {
+      "epoch": 0.16646068495784241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9039358039587417e-05,
+      "loss": 1.0291,
+      "step": 6061
+    },
+    {
+      "epoch": 0.16648814918568564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90389775861425e-05,
+      "loss": 1.0022,
+      "step": 6062
+    },
+    {
+      "epoch": 0.16651561341352888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903859706117769e-05,
+      "loss": 0.9886,
+      "step": 6063
+    },
+    {
+      "epoch": 0.1665430776413721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9038216464696e-05,
+      "loss": 1.0605,
+      "step": 6064
+    },
+    {
+      "epoch": 0.16657054186921535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9037835796700442e-05,
+      "loss": 1.0797,
+      "step": 6065
+    },
+    {
+      "epoch": 0.16659800609705858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903745505719402e-05,
+      "loss": 1.1124,
+      "step": 6066
+    },
+    {
+      "epoch": 0.16662547032490183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9037074246179762e-05,
+      "loss": 1.0617,
+      "step": 6067
+    },
+    {
+      "epoch": 0.16665293455274505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9036693363660668e-05,
+      "loss": 1.0884,
+      "step": 6068
+    },
+    {
+      "epoch": 0.1666803987805883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9036312409639755e-05,
+      "loss": 1.0137,
+      "step": 6069
+    },
+    {
+      "epoch": 0.16670786300843152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9035931384120036e-05,
+      "loss": 1.1341,
+      "step": 6070
+    },
+    {
+      "epoch": 0.16673532723627474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9035550287104532e-05,
+      "loss": 0.9814,
+      "step": 6071
+    },
+    {
+      "epoch": 0.166762791464118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903516911859625e-05,
+      "loss": 1.0625,
+      "step": 6072
+    },
+    {
+      "epoch": 0.1667902556919612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903478787859821e-05,
+      "loss": 0.9377,
+      "step": 6073
+    },
+    {
+      "epoch": 0.16681771991980446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903440656711343e-05,
+      "loss": 1.0329,
+      "step": 6074
+    },
+    {
+      "epoch": 0.16684518414764768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9034025184144925e-05,
+      "loss": 1.0201,
+      "step": 6075
+    },
+    {
+      "epoch": 0.16687264837549093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9033643729695717e-05,
+      "loss": 1.088,
+      "step": 6076
+    },
+    {
+      "epoch": 0.16690011260333415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9033262203768818e-05,
+      "loss": 1.1022,
+      "step": 6077
+    },
+    {
+      "epoch": 0.1669275768311774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903288060636725e-05,
+      "loss": 1.0221,
+      "step": 6078
+    },
+    {
+      "epoch": 0.16695504105902062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903249893749403e-05,
+      "loss": 1.0959,
+      "step": 6079
+    },
+    {
+      "epoch": 0.16698250528686387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903211719715218e-05,
+      "loss": 1.0472,
+      "step": 6080
+    },
+    {
+      "epoch": 0.1670099695147071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903173538534472e-05,
+      "loss": 0.9311,
+      "step": 6081
+    },
+    {
+      "epoch": 0.16703743374255034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903135350207467e-05,
+      "loss": 0.973,
+      "step": 6082
+    },
+    {
+      "epoch": 0.16706489797039356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030971547345056e-05,
+      "loss": 1.0818,
+      "step": 6083
+    },
+    {
+      "epoch": 0.1670923621982368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903058952115889e-05,
+      "loss": 1.0878,
+      "step": 6084
+    },
+    {
+      "epoch": 0.16711982642608003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903020742351921e-05,
+      "loss": 1.0604,
+      "step": 6085
+    },
+    {
+      "epoch": 0.16714729065392325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9029825254429026e-05,
+      "loss": 0.9108,
+      "step": 6086
+    },
+    {
+      "epoch": 0.1671747548817665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9029443013891366e-05,
+      "loss": 0.9068,
+      "step": 6087
+    },
+    {
+      "epoch": 0.16720221910960972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902906070190926e-05,
+      "loss": 0.9196,
+      "step": 6088
+    },
+    {
+      "epoch": 0.16722968333745297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9028678318485726e-05,
+      "loss": 0.8692,
+      "step": 6089
+    },
+    {
+      "epoch": 0.1672571475652962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902829586362379e-05,
+      "loss": 1.0507,
+      "step": 6090
+    },
+    {
+      "epoch": 0.16728461179313944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902791333732648e-05,
+      "loss": 1.0205,
+      "step": 6091
+    },
+    {
+      "epoch": 0.16731207602098266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9027530739596826e-05,
+      "loss": 1.113,
+      "step": 6092
+    },
+    {
+      "epoch": 0.1673395402488259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9027148070437848e-05,
+      "loss": 1.0984,
+      "step": 6093
+    },
+    {
+      "epoch": 0.16736700447666913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902676532985258e-05,
+      "loss": 0.9524,
+      "step": 6094
+    },
+    {
+      "epoch": 0.16739446870451238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9026382517844047e-05,
+      "loss": 0.9438,
+      "step": 6095
+    },
+    {
+      "epoch": 0.1674219329323556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902599963441528e-05,
+      "loss": 1.0533,
+      "step": 6096
+    },
+    {
+      "epoch": 0.16744939716019885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9025616679569306e-05,
+      "loss": 1.0153,
+      "step": 6097
+    },
+    {
+      "epoch": 0.16747686138804208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902523365330916e-05,
+      "loss": 0.9135,
+      "step": 6098
+    },
+    {
+      "epoch": 0.1675043256158853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9024850555637863e-05,
+      "loss": 1.0298,
+      "step": 6099
+    },
+    {
+      "epoch": 0.16753178984372855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9024467386558458e-05,
+      "loss": 1.0989,
+      "step": 6100
+    },
+    {
+      "epoch": 0.16755925407157177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902408414607397e-05,
+      "loss": 1.0295,
+      "step": 6101
+    },
+    {
+      "epoch": 0.16758671829941502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902370083418743e-05,
+      "loss": 1.0254,
+      "step": 6102
+    },
+    {
+      "epoch": 0.16761418252725824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9023317450901877e-05,
+      "loss": 1.0365,
+      "step": 6103
+    },
+    {
+      "epoch": 0.1676416467551015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902293399622034e-05,
+      "loss": 1.0105,
+      "step": 6104
+    },
+    {
+      "epoch": 0.1676691109829447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9022550470145854e-05,
+      "loss": 1.1334,
+      "step": 6105
+    },
+    {
+      "epoch": 0.16769657521078796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9022166872681453e-05,
+      "loss": 1.0378,
+      "step": 6106
+    },
+    {
+      "epoch": 0.16772403943863118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9021783203830173e-05,
+      "loss": 1.0053,
+      "step": 6107
+    },
+    {
+      "epoch": 0.16775150366647443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902139946359505e-05,
+      "loss": 0.9961,
+      "step": 6108
+    },
+    {
+      "epoch": 0.16777896789431765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902101565197912e-05,
+      "loss": 1.1313,
+      "step": 6109
+    },
+    {
+      "epoch": 0.1678064321221609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902063176898542e-05,
+      "loss": 1.1127,
+      "step": 6110
+    },
+    {
+      "epoch": 0.16783389635000412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902024781461699e-05,
+      "loss": 1.0062,
+      "step": 6111
+    },
+    {
+      "epoch": 0.16786136057784734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019863788876857e-05,
+      "loss": 1.0256,
+      "step": 6112
+    },
+    {
+      "epoch": 0.1678888248056906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019479691768073e-05,
+      "loss": 1.0763,
+      "step": 6113
+    },
+    {
+      "epoch": 0.1679162890335338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019095523293674e-05,
+      "loss": 1.0046,
+      "step": 6114
+    },
+    {
+      "epoch": 0.16794375326137706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9018711283456694e-05,
+      "loss": 1.145,
+      "step": 6115
+    },
+    {
+      "epoch": 0.16797121748922028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9018326972260177e-05,
+      "loss": 1.0433,
+      "step": 6116
+    },
+    {
+      "epoch": 0.16799868171706353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9017942589707168e-05,
+      "loss": 1.0232,
+      "step": 6117
+    },
+    {
+      "epoch": 0.16802614594490675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9017558135800697e-05,
+      "loss": 0.975,
+      "step": 6118
+    },
+    {
+      "epoch": 0.16805361017275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901717361054382e-05,
+      "loss": 1.0066,
+      "step": 6119
+    },
+    {
+      "epoch": 0.16808107440059322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901678901393957e-05,
+      "loss": 1.0023,
+      "step": 6120
+    },
+    {
+      "epoch": 0.16810853862843647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901640434599099e-05,
+      "loss": 0.9747,
+      "step": 6121
+    },
+    {
+      "epoch": 0.1681360028562797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901601960670113e-05,
+      "loss": 1.1282,
+      "step": 6122
+    },
+    {
+      "epoch": 0.16816346708412294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9015634796073027e-05,
+      "loss": 1.0314,
+      "step": 6123
+    },
+    {
+      "epoch": 0.16819093131196616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9015249914109728e-05,
+      "loss": 0.9253,
+      "step": 6124
+    },
+    {
+      "epoch": 0.16821839553980938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9014864960814285e-05,
+      "loss": 1.0974,
+      "step": 6125
+    },
+    {
+      "epoch": 0.16824585976765263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9014479936189733e-05,
+      "loss": 0.9757,
+      "step": 6126
+    },
+    {
+      "epoch": 0.16827332399549585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9014094840239125e-05,
+      "loss": 1.1042,
+      "step": 6127
+    },
+    {
+      "epoch": 0.1683007882233391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901370967296551e-05,
+      "loss": 1.0515,
+      "step": 6128
+    },
+    {
+      "epoch": 0.16832825245118233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9013324434371928e-05,
+      "loss": 1.0432,
+      "step": 6129
+    },
+    {
+      "epoch": 0.16835571667902557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9012939124461434e-05,
+      "loss": 1.0381,
+      "step": 6130
+    },
+    {
+      "epoch": 0.1683831809068688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9012553743237076e-05,
+      "loss": 0.965,
+      "step": 6131
+    },
+    {
+      "epoch": 0.16841064513471204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9012168290701904e-05,
+      "loss": 1.0377,
+      "step": 6132
+    },
+    {
+      "epoch": 0.16843810936255527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901178276685896e-05,
+      "loss": 0.9582,
+      "step": 6133
+    },
+    {
+      "epoch": 0.16846557359039852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90113971717113e-05,
+      "loss": 1.0361,
+      "step": 6134
+    },
+    {
+      "epoch": 0.16849303781824174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9011011505261975e-05,
+      "loss": 1.0017,
+      "step": 6135
+    },
+    {
+      "epoch": 0.16852050204608499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901062576751404e-05,
+      "loss": 1.1176,
+      "step": 6136
+    },
+    {
+      "epoch": 0.1685479662739282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9010239958470543e-05,
+      "loss": 1.0345,
+      "step": 6137
+    },
+    {
+      "epoch": 0.16857543050177146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9009854078134534e-05,
+      "loss": 1.1051,
+      "step": 6138
+    },
+    {
+      "epoch": 0.16860289472961468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900946812650907e-05,
+      "loss": 1.013,
+      "step": 6139
+    },
+    {
+      "epoch": 0.1686303589574579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9009082103597206e-05,
+      "loss": 1.0658,
+      "step": 6140
+    },
+    {
+      "epoch": 0.16865782318530115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9008696009401993e-05,
+      "loss": 0.9798,
+      "step": 6141
+    },
+    {
+      "epoch": 0.16868528741314437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9008309843926488e-05,
+      "loss": 0.9751,
+      "step": 6142
+    },
+    {
+      "epoch": 0.16871275164098762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900792360717375e-05,
+      "loss": 1.0841,
+      "step": 6143
+    },
+    {
+      "epoch": 0.16874021586883084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9007537299146824e-05,
+      "loss": 1.0421,
+      "step": 6144
+    },
+    {
+      "epoch": 0.1687676800966741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900715091984878e-05,
+      "loss": 1.0175,
+      "step": 6145
+    },
+    {
+      "epoch": 0.1687951443245173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9006764469282667e-05,
+      "loss": 1.1023,
+      "step": 6146
+    },
+    {
+      "epoch": 0.16882260855236056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9006377947451544e-05,
+      "loss": 0.9664,
+      "step": 6147
+    },
+    {
+      "epoch": 0.16885007278020378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900599135435847e-05,
+      "loss": 1.0382,
+      "step": 6148
+    },
+    {
+      "epoch": 0.16887753700804703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9005604690006504e-05,
+      "loss": 1.0815,
+      "step": 6149
+    },
+    {
+      "epoch": 0.16890500123589025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9005217954398705e-05,
+      "loss": 0.988,
+      "step": 6150
+    },
+    {
+      "epoch": 0.1689324654637335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9004831147538133e-05,
+      "loss": 1.0511,
+      "step": 6151
+    },
+    {
+      "epoch": 0.16895992969157672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9004444269427846e-05,
+      "loss": 0.9821,
+      "step": 6152
+    },
+    {
+      "epoch": 0.16898739391941994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9004057320070914e-05,
+      "loss": 1.0819,
+      "step": 6153
+    },
+    {
+      "epoch": 0.1690148581472632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900367029947039e-05,
+      "loss": 1.0627,
+      "step": 6154
+    },
+    {
+      "epoch": 0.1690423223751064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9003283207629336e-05,
+      "loss": 1.0255,
+      "step": 6155
+    },
+    {
+      "epoch": 0.16906978660294966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900289604455082e-05,
+      "loss": 0.9377,
+      "step": 6156
+    },
+    {
+      "epoch": 0.16909725083079288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9002508810237905e-05,
+      "loss": 0.95,
+      "step": 6157
+    },
+    {
+      "epoch": 0.16912471505863613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900212150469365e-05,
+      "loss": 1.0257,
+      "step": 6158
+    },
+    {
+      "epoch": 0.16915217928647935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9001734127921126e-05,
+      "loss": 0.9956,
+      "step": 6159
+    },
+    {
+      "epoch": 0.1691796435143226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9001346679923392e-05,
+      "loss": 0.9995,
+      "step": 6160
+    },
+    {
+      "epoch": 0.16920710774216582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000959160703517e-05,
+      "loss": 1.1201,
+      "step": 6161
+    },
+    {
+      "epoch": 0.16923457197000907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000571570264565e-05,
+      "loss": 1.0634,
+      "step": 6162
+    },
+    {
+      "epoch": 0.1692620361978523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000183908609605e-05,
+      "loss": 0.9667,
+      "step": 6163
+    },
+    {
+      "epoch": 0.16928950042569554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8999796175741708e-05,
+      "loss": 1.0584,
+      "step": 6164
+    },
+    {
+      "epoch": 0.16931696465353877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8999408371663934e-05,
+      "loss": 0.9923,
+      "step": 6165
+    },
+    {
+      "epoch": 0.169344428881382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8999020496379353e-05,
+      "loss": 0.9672,
+      "step": 6166
+    },
+    {
+      "epoch": 0.16937189310922524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8998632549891037e-05,
+      "loss": 1.0268,
+      "step": 6167
+    },
+    {
+      "epoch": 0.16939935733706846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8998244532202054e-05,
+      "loss": 1.0366,
+      "step": 6168
+    },
+    {
+      "epoch": 0.1694268215649117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8997856443315475e-05,
+      "loss": 1.1081,
+      "step": 6169
+    },
+    {
+      "epoch": 0.16945428579275493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899746828323437e-05,
+      "loss": 0.996,
+      "step": 6170
+    },
+    {
+      "epoch": 0.16948175002059818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8997080051961812e-05,
+      "loss": 1.1118,
+      "step": 6171
+    },
+    {
+      "epoch": 0.1695092142484414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8996691749500874e-05,
+      "loss": 1.0913,
+      "step": 6172
+    },
+    {
+      "epoch": 0.16953667847628465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8996303375854622e-05,
+      "loss": 1.01,
+      "step": 6173
+    },
+    {
+      "epoch": 0.16956414270412787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8995914931026137e-05,
+      "loss": 1.0338,
+      "step": 6174
+    },
+    {
+      "epoch": 0.16959160693197112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8995526415018483e-05,
+      "loss": 0.979,
+      "step": 6175
+    },
+    {
+      "epoch": 0.16961907115981434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899513782783474e-05,
+      "loss": 0.894,
+      "step": 6176
+    },
+    {
+      "epoch": 0.1696465353876576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8994749169477988e-05,
+      "loss": 1.1252,
+      "step": 6177
+    },
+    {
+      "epoch": 0.1696739996155008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899436043995129e-05,
+      "loss": 1.054,
+      "step": 6178
+    },
+    {
+      "epoch": 0.16970146384334406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993971639257733e-05,
+      "loss": 1.0487,
+      "step": 6179
+    },
+    {
+      "epoch": 0.16972892807118728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993582767400386e-05,
+      "loss": 1.005,
+      "step": 6180
+    },
+    {
+      "epoch": 0.1697563922990305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993193824382325e-05,
+      "loss": 1.0312,
+      "step": 6181
+    },
+    {
+      "epoch": 0.16978385652687375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8992804810206634e-05,
+      "loss": 0.9341,
+      "step": 6182
+    },
+    {
+      "epoch": 0.16981132075471697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899241572487639e-05,
+      "loss": 0.9895,
+      "step": 6183
+    },
+    {
+      "epoch": 0.16983878498256022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8992026568394668e-05,
+      "loss": 1.1165,
+      "step": 6184
+    },
+    {
+      "epoch": 0.16986624921040344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8991637340764547e-05,
+      "loss": 0.9858,
+      "step": 6185
+    },
+    {
+      "epoch": 0.1698937134382467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8991248041989107e-05,
+      "loss": 1.0395,
+      "step": 6186
+    },
+    {
+      "epoch": 0.1699211776660899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8990858672071434e-05,
+      "loss": 0.9326,
+      "step": 6187
+    },
+    {
+      "epoch": 0.16994864189393316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89904692310146e-05,
+      "loss": 1.002,
+      "step": 6188
+    },
+    {
+      "epoch": 0.16997610612177638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8990079718821696e-05,
+      "loss": 1.0569,
+      "step": 6189
+    },
+    {
+      "epoch": 0.17000357034961963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8989690135495796e-05,
+      "loss": 0.9945,
+      "step": 6190
+    },
+    {
+      "epoch": 0.17003103457746285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8989300481039987e-05,
+      "loss": 1.0084,
+      "step": 6191
+    },
+    {
+      "epoch": 0.1700584988053061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8988910755457347e-05,
+      "loss": 1.0374,
+      "step": 6192
+    },
+    {
+      "epoch": 0.17008596303314932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8988520958750966e-05,
+      "loss": 1.0542,
+      "step": 6193
+    },
+    {
+      "epoch": 0.17011342726099254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8988131090923922e-05,
+      "loss": 1.0825,
+      "step": 6194
+    },
+    {
+      "epoch": 0.1701408914888358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898774115197931e-05,
+      "loss": 1.1025,
+      "step": 6195
+    },
+    {
+      "epoch": 0.17016835571667902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8987351141920204e-05,
+      "loss": 0.9897,
+      "step": 6196
+    },
+    {
+      "epoch": 0.17019581994452226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898696106074969e-05,
+      "loss": 1.0227,
+      "step": 6197
+    },
+    {
+      "epoch": 0.17022328417236549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8986570908470866e-05,
+      "loss": 0.9922,
+      "step": 6198
+    },
+    {
+      "epoch": 0.17025074840020873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898618068508681e-05,
+      "loss": 1.0117,
+      "step": 6199
+    },
+    {
+      "epoch": 0.17027821262805196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898579039060061e-05,
+      "loss": 1.061,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1703056768558952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8985400025015356e-05,
+      "loss": 1.0911,
+      "step": 6201
+    },
+    {
+      "epoch": 0.17033314108373843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8985009588334137e-05,
+      "loss": 1.0746,
+      "step": 6202
+    },
+    {
+      "epoch": 0.17036060531158168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898461908056004e-05,
+      "loss": 1.0851,
+      "step": 6203
+    },
+    {
+      "epoch": 0.1703880695394249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8984228501696158e-05,
+      "loss": 1.1082,
+      "step": 6204
+    },
+    {
+      "epoch": 0.17041553376726815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8983837851745582e-05,
+      "loss": 0.9185,
+      "step": 6205
+    },
+    {
+      "epoch": 0.17044299799511137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8983447130711397e-05,
+      "loss": 1.1254,
+      "step": 6206
+    },
+    {
+      "epoch": 0.1704704622229546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8983056338596702e-05,
+      "loss": 0.9634,
+      "step": 6207
+    },
+    {
+      "epoch": 0.17049792645079784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8982665475404585e-05,
+      "loss": 1.0285,
+      "step": 6208
+    },
+    {
+      "epoch": 0.17052539067864106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8982274541138137e-05,
+      "loss": 1.0337,
+      "step": 6209
+    },
+    {
+      "epoch": 0.1705528549064843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8981883535800458e-05,
+      "loss": 1.1036,
+      "step": 6210
+    },
+    {
+      "epoch": 0.17058031913432753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898149245939463e-05,
+      "loss": 1.0006,
+      "step": 6211
+    },
+    {
+      "epoch": 0.17060778336217078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898110131192376e-05,
+      "loss": 1.0581,
+      "step": 6212
+    },
+    {
+      "epoch": 0.170635247590014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8980710093390936e-05,
+      "loss": 1.085,
+      "step": 6213
+    },
+    {
+      "epoch": 0.17066271181785725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8980318803799255e-05,
+      "loss": 1.0371,
+      "step": 6214
+    },
+    {
+      "epoch": 0.17069017604570047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897992744315181e-05,
+      "loss": 1.002,
+      "step": 6215
+    },
+    {
+      "epoch": 0.17071764027354372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8979536011451704e-05,
+      "loss": 0.9832,
+      "step": 6216
+    },
+    {
+      "epoch": 0.17074510450138694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897914450870203e-05,
+      "loss": 1.0449,
+      "step": 6217
+    },
+    {
+      "epoch": 0.1707725687292302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8978752934905887e-05,
+      "loss": 1.0486,
+      "step": 6218
+    },
+    {
+      "epoch": 0.1708000329570734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8978361290066373e-05,
+      "loss": 0.9714,
+      "step": 6219
+    },
+    {
+      "epoch": 0.17082749718491663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8977969574186585e-05,
+      "loss": 1.0389,
+      "step": 6220
+    },
+    {
+      "epoch": 0.17085496141275988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8977577787269626e-05,
+      "loss": 1.1331,
+      "step": 6221
+    },
+    {
+      "epoch": 0.1708824256406031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897718592931859e-05,
+      "loss": 1.0912,
+      "step": 6222
+    },
+    {
+      "epoch": 0.17090988986844635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8976794000336585e-05,
+      "loss": 1.0084,
+      "step": 6223
+    },
+    {
+      "epoch": 0.17093735409628957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897640200032671e-05,
+      "loss": 1.0329,
+      "step": 6224
+    },
+    {
+      "epoch": 0.17096481832413282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8976009929292065e-05,
+      "loss": 1.006,
+      "step": 6225
+    },
+    {
+      "epoch": 0.17099228255197604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8975617787235747e-05,
+      "loss": 1.108,
+      "step": 6226
+    },
+    {
+      "epoch": 0.1710197467798193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8975225574160866e-05,
+      "loss": 0.9417,
+      "step": 6227
+    },
+    {
+      "epoch": 0.17104721100766251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8974833290070526e-05,
+      "loss": 1.0237,
+      "step": 6228
+    },
+    {
+      "epoch": 0.17107467523550576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897444093496783e-05,
+      "loss": 0.9873,
+      "step": 6229
+    },
+    {
+      "epoch": 0.17110213946334898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897404850885588e-05,
+      "loss": 1.088,
+      "step": 6230
+    },
+    {
+      "epoch": 0.17112960369119223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8973656011737777e-05,
+      "loss": 0.9571,
+      "step": 6231
+    },
+    {
+      "epoch": 0.17115706791903545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897326344361664e-05,
+      "loss": 1.0147,
+      "step": 6232
+    },
+    {
+      "epoch": 0.1711845321468787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897287080449556e-05,
+      "loss": 1.1055,
+      "step": 6233
+    },
+    {
+      "epoch": 0.17121199637472193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8972478094377653e-05,
+      "loss": 1.0381,
+      "step": 6234
+    },
+    {
+      "epoch": 0.17123946060256515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8972085313266025e-05,
+      "loss": 1.0165,
+      "step": 6235
+    },
+    {
+      "epoch": 0.1712669248304084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897169246116378e-05,
+      "loss": 1.0604,
+      "step": 6236
+    },
+    {
+      "epoch": 0.17129438905825162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8971299538074027e-05,
+      "loss": 1.019,
+      "step": 6237
+    },
+    {
+      "epoch": 0.17132185328609487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897090654399988e-05,
+      "loss": 1.0559,
+      "step": 6238
+    },
+    {
+      "epoch": 0.1713493175139381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897051347894445e-05,
+      "loss": 0.9912,
+      "step": 6239
+    },
+    {
+      "epoch": 0.17137678174178134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8970120342910834e-05,
+      "loss": 1.0769,
+      "step": 6240
+    },
+    {
+      "epoch": 0.17140424596962456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8969727135902153e-05,
+      "loss": 1.0178,
+      "step": 6241
+    },
+    {
+      "epoch": 0.1714317101974678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896933385792152e-05,
+      "loss": 0.9843,
+      "step": 6242
+    },
+    {
+      "epoch": 0.17145917442531103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896894050897204e-05,
+      "loss": 1.0538,
+      "step": 6243
+    },
+    {
+      "epoch": 0.17148663865315428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896854708905683e-05,
+      "loss": 0.9439,
+      "step": 6244
+    },
+    {
+      "epoch": 0.1715141028809975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8968153598179e-05,
+      "loss": 1.0908,
+      "step": 6245
+    },
+    {
+      "epoch": 0.17154156710884075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896776003634167e-05,
+      "loss": 0.8583,
+      "step": 6246
+    },
+    {
+      "epoch": 0.17156903133668397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896736640354794e-05,
+      "loss": 1.0874,
+      "step": 6247
+    },
+    {
+      "epoch": 0.1715964955645272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896697269980094e-05,
+      "loss": 1.0019,
+      "step": 6248
+    },
+    {
+      "epoch": 0.17162395979237044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8966578925103776e-05,
+      "loss": 1.0305,
+      "step": 6249
+    },
+    {
+      "epoch": 0.17165142402021366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8966185079459568e-05,
+      "loss": 1.0663,
+      "step": 6250
+    },
+    {
+      "epoch": 0.1716788882480569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896579116287143e-05,
+      "loss": 1.0756,
+      "step": 6251
+    },
+    {
+      "epoch": 0.17170635247590013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896539717534248e-05,
+      "loss": 1.1353,
+      "step": 6252
+    },
+    {
+      "epoch": 0.17173381670374338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8965003116875832e-05,
+      "loss": 0.9997,
+      "step": 6253
+    },
+    {
+      "epoch": 0.1717612809315866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964608987474608e-05,
+      "loss": 1.0107,
+      "step": 6254
+    },
+    {
+      "epoch": 0.17178874515942985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964214787141926e-05,
+      "loss": 0.9824,
+      "step": 6255
+    },
+    {
+      "epoch": 0.17181620938727307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89638205158809e-05,
+      "loss": 0.9802,
+      "step": 6256
+    },
+    {
+      "epoch": 0.17184367361511632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896342617369466e-05,
+      "loss": 0.9204,
+      "step": 6257
+    },
+    {
+      "epoch": 0.17187113784295954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8963031760586316e-05,
+      "loss": 1.0478,
+      "step": 6258
+    },
+    {
+      "epoch": 0.1718986020708028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8962637276558996e-05,
+      "loss": 0.9726,
+      "step": 6259
+    },
+    {
+      "epoch": 0.171926066298646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8962242721615816e-05,
+      "loss": 1.0237,
+      "step": 6260
+    },
+    {
+      "epoch": 0.17195353052648923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8961848095759897e-05,
+      "loss": 1.0511,
+      "step": 6261
+    },
+    {
+      "epoch": 0.17198099475433248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8961453398994368e-05,
+      "loss": 1.017,
+      "step": 6262
+    },
+    {
+      "epoch": 0.1720084589821757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8961058631322345e-05,
+      "loss": 0.9579,
+      "step": 6263
+    },
+    {
+      "epoch": 0.17203592321001895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896066379274696e-05,
+      "loss": 1.0466,
+      "step": 6264
+    },
+    {
+      "epoch": 0.17206338743786218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896026888327133e-05,
+      "loss": 0.9789,
+      "step": 6265
+    },
+    {
+      "epoch": 0.17209085166570542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8959873902898582e-05,
+      "loss": 0.9967,
+      "step": 6266
+    },
+    {
+      "epoch": 0.17211831589354865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8959478851631838e-05,
+      "loss": 1.1627,
+      "step": 6267
+    },
+    {
+      "epoch": 0.1721457801213919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895908372947423e-05,
+      "loss": 0.9631,
+      "step": 6268
+    },
+    {
+      "epoch": 0.17217324434923512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8958688536428882e-05,
+      "loss": 1.0484,
+      "step": 6269
+    },
+    {
+      "epoch": 0.17220070857707837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8958293272498917e-05,
+      "loss": 0.9644,
+      "step": 6270
+    },
+    {
+      "epoch": 0.1722281728049216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8957897937687467e-05,
+      "loss": 1.0224,
+      "step": 6271
+    },
+    {
+      "epoch": 0.17225563703276484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895750253199766e-05,
+      "loss": 0.9933,
+      "step": 6272
+    },
+    {
+      "epoch": 0.17228310126060806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895710705543262e-05,
+      "loss": 1.059,
+      "step": 6273
+    },
+    {
+      "epoch": 0.1723105654884513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895671150799548e-05,
+      "loss": 1.0094,
+      "step": 6274
+    },
+    {
+      "epoch": 0.17233802971629453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8956315889689372e-05,
+      "loss": 1.0166,
+      "step": 6275
+    },
+    {
+      "epoch": 0.17236549394413775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895592020051742e-05,
+      "loss": 0.9635,
+      "step": 6276
+    },
+    {
+      "epoch": 0.172392958171981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895552444048276e-05,
+      "loss": 1.1114,
+      "step": 6277
+    },
+    {
+      "epoch": 0.17242042239982422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895512860958852e-05,
+      "loss": 1.0434,
+      "step": 6278
+    },
+    {
+      "epoch": 0.17244788662766747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8954732707837835e-05,
+      "loss": 0.9809,
+      "step": 6279
+    },
+    {
+      "epoch": 0.1724753508555107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8954336735233835e-05,
+      "loss": 1.1127,
+      "step": 6280
+    },
+    {
+      "epoch": 0.17250281508335394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8953940691779653e-05,
+      "loss": 0.9756,
+      "step": 6281
+    },
+    {
+      "epoch": 0.17253027931119716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8953544577478428e-05,
+      "loss": 1.0518,
+      "step": 6282
+    },
+    {
+      "epoch": 0.1725577435390404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8953148392333285e-05,
+      "loss": 0.9507,
+      "step": 6283
+    },
+    {
+      "epoch": 0.17258520776688363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895275213634737e-05,
+      "loss": 1.1077,
+      "step": 6284
+    },
+    {
+      "epoch": 0.17261267199472688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8952355809523806e-05,
+      "loss": 0.9861,
+      "step": 6285
+    },
+    {
+      "epoch": 0.1726401362225701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895195941186574e-05,
+      "loss": 0.9388,
+      "step": 6286
+    },
+    {
+      "epoch": 0.17266760045041335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89515629433763e-05,
+      "loss": 1.014,
+      "step": 6287
+    },
+    {
+      "epoch": 0.17269506467825657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8951166404058625e-05,
+      "loss": 1.0633,
+      "step": 6288
+    },
+    {
+      "epoch": 0.1727225289060998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8950769793915857e-05,
+      "loss": 1.1298,
+      "step": 6289
+    },
+    {
+      "epoch": 0.17274999313394304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8950373112951132e-05,
+      "loss": 0.9886,
+      "step": 6290
+    },
+    {
+      "epoch": 0.17277745736178626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8949976361167585e-05,
+      "loss": 0.9752,
+      "step": 6291
+    },
+    {
+      "epoch": 0.1728049215896295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894957953856836e-05,
+      "loss": 1.0233,
+      "step": 6292
+    },
+    {
+      "epoch": 0.17283238581747273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8949182645156595e-05,
+      "loss": 1.0133,
+      "step": 6293
+    },
+    {
+      "epoch": 0.17285985004531598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894878568093543e-05,
+      "loss": 0.9709,
+      "step": 6294
+    },
+    {
+      "epoch": 0.1728873142731592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894838864590801e-05,
+      "loss": 1.0855,
+      "step": 6295
+    },
+    {
+      "epoch": 0.17291477850100245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8947991540077464e-05,
+      "loss": 0.9205,
+      "step": 6296
+    },
+    {
+      "epoch": 0.17294224272884567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8947594363446952e-05,
+      "loss": 0.9999,
+      "step": 6297
+    },
+    {
+      "epoch": 0.17296970695668892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8947197116019603e-05,
+      "loss": 0.9184,
+      "step": 6298
+    },
+    {
+      "epoch": 0.17299717118453214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8946799797798564e-05,
+      "loss": 1.2562,
+      "step": 6299
+    },
+    {
+      "epoch": 0.1730246354123754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8946402408786978e-05,
+      "loss": 1.0483,
+      "step": 6300
+    },
+    {
+      "epoch": 0.17305209964021862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8946004948987995e-05,
+      "loss": 1.0528,
+      "step": 6301
+    },
+    {
+      "epoch": 0.17307956386806184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894560741840475e-05,
+      "loss": 1.037,
+      "step": 6302
+    },
+    {
+      "epoch": 0.17310702809590509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89452098170404e-05,
+      "loss": 0.94,
+      "step": 6303
+    },
+    {
+      "epoch": 0.1731344923237483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8944812144898082e-05,
+      "loss": 1.1579,
+      "step": 6304
+    },
+    {
+      "epoch": 0.17316195655159156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8944414401980945e-05,
+      "loss": 0.971,
+      "step": 6305
+    },
+    {
+      "epoch": 0.17318942077943478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8944016588292138e-05,
+      "loss": 0.9229,
+      "step": 6306
+    },
+    {
+      "epoch": 0.17321688500727803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8943618703834806e-05,
+      "loss": 1.0688,
+      "step": 6307
+    },
+    {
+      "epoch": 0.17324434923512125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89432207486121e-05,
+      "loss": 1.0742,
+      "step": 6308
+    },
+    {
+      "epoch": 0.1732718134629645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8942822722627166e-05,
+      "loss": 1.0719,
+      "step": 6309
+    },
+    {
+      "epoch": 0.17329927769080772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8942424625883157e-05,
+      "loss": 1.1171,
+      "step": 6310
+    },
+    {
+      "epoch": 0.17332674191865097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8942026458383216e-05,
+      "loss": 1.054,
+      "step": 6311
+    },
+    {
+      "epoch": 0.1733542061464942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89416282201305e-05,
+      "loss": 0.9645,
+      "step": 6312
+    },
+    {
+      "epoch": 0.17338167037433744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8941229911128158e-05,
+      "loss": 0.9052,
+      "step": 6313
+    },
+    {
+      "epoch": 0.17340913460218066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8940831531379338e-05,
+      "loss": 0.9815,
+      "step": 6314
+    },
+    {
+      "epoch": 0.17343659883002388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89404330808872e-05,
+      "loss": 0.9897,
+      "step": 6315
+    },
+    {
+      "epoch": 0.17346406305786713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8940034559654893e-05,
+      "loss": 1.0521,
+      "step": 6316
+    },
+    {
+      "epoch": 0.17349152728571035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8939635967685564e-05,
+      "loss": 1.0404,
+      "step": 6317
+    },
+    {
+      "epoch": 0.1735189915135536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8939237304982376e-05,
+      "loss": 1.0453,
+      "step": 6318
+    },
+    {
+      "epoch": 0.17354645574139682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8938838571548478e-05,
+      "loss": 0.9705,
+      "step": 6319
+    },
+    {
+      "epoch": 0.17357391996924007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893843976738703e-05,
+      "loss": 0.9287,
+      "step": 6320
+    },
+    {
+      "epoch": 0.1736013841970833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893804089250118e-05,
+      "loss": 1.0123,
+      "step": 6321
+    },
+    {
+      "epoch": 0.17362884842492654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893764194689409e-05,
+      "loss": 1.005,
+      "step": 6322
+    },
+    {
+      "epoch": 0.17365631265276976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893724293056892e-05,
+      "loss": 1.0105,
+      "step": 6323
+    },
+    {
+      "epoch": 0.173683776880613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8936843843528815e-05,
+      "loss": 0.9947,
+      "step": 6324
+    },
+    {
+      "epoch": 0.17371124110845623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8936444685776943e-05,
+      "loss": 1.0232,
+      "step": 6325
+    },
+    {
+      "epoch": 0.17373870533629948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8936045457316455e-05,
+      "loss": 0.9592,
+      "step": 6326
+    },
+    {
+      "epoch": 0.1737661695641427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893564615815052e-05,
+      "loss": 1.0452,
+      "step": 6327
+    },
+    {
+      "epoch": 0.17379363379198595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8935246788282285e-05,
+      "loss": 1.0509,
+      "step": 6328
+    },
+    {
+      "epoch": 0.17382109801982917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893484734771492e-05,
+      "loss": 0.8775,
+      "step": 6329
+    },
+    {
+      "epoch": 0.1738485622476724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8934447836451583e-05,
+      "loss": 1.107,
+      "step": 6330
+    },
+    {
+      "epoch": 0.17387602647551564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893404825449543e-05,
+      "loss": 1.0512,
+      "step": 6331
+    },
+    {
+      "epoch": 0.17390349070335887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8933648601849627e-05,
+      "loss": 1.0566,
+      "step": 6332
+    },
+    {
+      "epoch": 0.17393095493120211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893324887851734e-05,
+      "loss": 1.0485,
+      "step": 6333
+    },
+    {
+      "epoch": 0.17395841915904534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893284908450172e-05,
+      "loss": 1.0259,
+      "step": 6334
+    },
+    {
+      "epoch": 0.17398588338688858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8932449219805944e-05,
+      "loss": 0.9754,
+      "step": 6335
+    },
+    {
+      "epoch": 0.1740133476147318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893204928443317e-05,
+      "loss": 1.0576,
+      "step": 6336
+    },
+    {
+      "epoch": 0.17404081184257505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8931649278386557e-05,
+      "loss": 1.011,
+      "step": 6337
+    },
+    {
+      "epoch": 0.17406827607041828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8931249201669278e-05,
+      "loss": 1.0019,
+      "step": 6338
+    },
+    {
+      "epoch": 0.17409574029826153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8930849054284498e-05,
+      "loss": 0.9922,
+      "step": 6339
+    },
+    {
+      "epoch": 0.17412320452610475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8930448836235377e-05,
+      "loss": 1.0512,
+      "step": 6340
+    },
+    {
+      "epoch": 0.174150668753948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8930048547525087e-05,
+      "loss": 1.0204,
+      "step": 6341
+    },
+    {
+      "epoch": 0.17417813298179122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8929648188156796e-05,
+      "loss": 1.0758,
+      "step": 6342
+    },
+    {
+      "epoch": 0.17420559720963444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8929247758133665e-05,
+      "loss": 1.0369,
+      "step": 6343
+    },
+    {
+      "epoch": 0.1742330614374777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8928847257458872e-05,
+      "loss": 1.0836,
+      "step": 6344
+    },
+    {
+      "epoch": 0.1742605256653209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892844668613558e-05,
+      "loss": 1.028,
+      "step": 6345
+    },
+    {
+      "epoch": 0.17428798989316416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8928046044166956e-05,
+      "loss": 1.033,
+      "step": 6346
+    },
+    {
+      "epoch": 0.17431545412100738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892764533155617e-05,
+      "loss": 0.958,
+      "step": 6347
+    },
+    {
+      "epoch": 0.17434291834885063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8927244548306403e-05,
+      "loss": 0.8536,
+      "step": 6348
+    },
+    {
+      "epoch": 0.17437038257669385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8926843694420816e-05,
+      "loss": 1.08,
+      "step": 6349
+    },
+    {
+      "epoch": 0.1743978468045371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8926442769902585e-05,
+      "loss": 1.0165,
+      "step": 6350
+    },
+    {
+      "epoch": 0.17442531103238032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892604177475488e-05,
+      "loss": 1.0009,
+      "step": 6351
+    },
+    {
+      "epoch": 0.17445277526022357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8925640708980878e-05,
+      "loss": 0.9328,
+      "step": 6352
+    },
+    {
+      "epoch": 0.1744802394880668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8925239572583742e-05,
+      "loss": 1.0615,
+      "step": 6353
+    },
+    {
+      "epoch": 0.17450770371591004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892483836556666e-05,
+      "loss": 1.0156,
+      "step": 6354
+    },
+    {
+      "epoch": 0.17453516794375326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8924437087932797e-05,
+      "loss": 1.086,
+      "step": 6355
+    },
+    {
+      "epoch": 0.17456263217159648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892403573968533e-05,
+      "loss": 0.9839,
+      "step": 6356
+    },
+    {
+      "epoch": 0.17459009639943973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8923634320827436e-05,
+      "loss": 1.0835,
+      "step": 6357
+    },
+    {
+      "epoch": 0.17461756062728295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8923232831362292e-05,
+      "loss": 0.9324,
+      "step": 6358
+    },
+    {
+      "epoch": 0.1746450248551262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892283127129307e-05,
+      "loss": 0.9625,
+      "step": 6359
+    },
+    {
+      "epoch": 0.17467248908296942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8922429640622954e-05,
+      "loss": 1.0966,
+      "step": 6360
+    },
+    {
+      "epoch": 0.17469995331081267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892202793935512e-05,
+      "loss": 1.1099,
+      "step": 6361
+    },
+    {
+      "epoch": 0.1747274175386559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8921626167492742e-05,
+      "loss": 0.9404,
+      "step": 6362
+    },
+    {
+      "epoch": 0.17475488176649914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8921224325039e-05,
+      "loss": 0.9711,
+      "step": 6363
+    },
+    {
+      "epoch": 0.17478234599434236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920822411997077e-05,
+      "loss": 1.0057,
+      "step": 6364
+    },
+    {
+      "epoch": 0.1748098102221856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920420428370153e-05,
+      "loss": 1.048,
+      "step": 6365
+    },
+    {
+      "epoch": 0.17483727445002883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920018374161406e-05,
+      "loss": 1.0589,
+      "step": 6366
+    },
+    {
+      "epoch": 0.17486473867787208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891961624937402e-05,
+      "loss": 1.0083,
+      "step": 6367
+    },
+    {
+      "epoch": 0.1748922029057153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8919214054011173e-05,
+      "loss": 1.0807,
+      "step": 6368
+    },
+    {
+      "epoch": 0.17491966713355853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8918811788076052e-05,
+      "loss": 0.9894,
+      "step": 6369
+    },
+    {
+      "epoch": 0.17494713136140178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8918409451571834e-05,
+      "loss": 0.947,
+      "step": 6370
+    },
+    {
+      "epoch": 0.174974595589245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891800704450171e-05,
+      "loss": 1.1057,
+      "step": 6371
+    },
+    {
+      "epoch": 0.17500205981708825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8917604566868856e-05,
+      "loss": 0.9357,
+      "step": 6372
+    },
+    {
+      "epoch": 0.17502952404493147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8917202018676458e-05,
+      "loss": 1.0385,
+      "step": 6373
+    },
+    {
+      "epoch": 0.17505698827277472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891679939992771e-05,
+      "loss": 0.9695,
+      "step": 6374
+    },
+    {
+      "epoch": 0.17508445250061794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8916396710625785e-05,
+      "loss": 1.0415,
+      "step": 6375
+    },
+    {
+      "epoch": 0.1751119167284612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8915993950773877e-05,
+      "loss": 0.9061,
+      "step": 6376
+    },
+    {
+      "epoch": 0.1751393809563044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8915591120375174e-05,
+      "loss": 1.0384,
+      "step": 6377
+    },
+    {
+      "epoch": 0.17516684518414766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891518821943286e-05,
+      "loss": 1.1138,
+      "step": 6378
+    },
+    {
+      "epoch": 0.17519430941199088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891478524795012e-05,
+      "loss": 0.9917,
+      "step": 6379
+    },
+    {
+      "epoch": 0.17522177363983413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891438220593015e-05,
+      "loss": 0.9705,
+      "step": 6380
+    },
+    {
+      "epoch": 0.17524923786767735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891397909337613e-05,
+      "loss": 0.9688,
+      "step": 6381
+    },
+    {
+      "epoch": 0.1752767020955206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891357591029126e-05,
+      "loss": 1.0249,
+      "step": 6382
+    },
+    {
+      "epoch": 0.17530416632336382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891317265667872e-05,
+      "loss": 1.016,
+      "step": 6383
+    },
+    {
+      "epoch": 0.17533163055120704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8912769332541708e-05,
+      "loss": 1.0356,
+      "step": 6384
+    },
+    {
+      "epoch": 0.1753590947790503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891236593788341e-05,
+      "loss": 1.0291,
+      "step": 6385
+    },
+    {
+      "epoch": 0.1753865590068935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891196247270702e-05,
+      "loss": 1.0704,
+      "step": 6386
+    },
+    {
+      "epoch": 0.17541402323473676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8911558937015736e-05,
+      "loss": 0.9722,
+      "step": 6387
+    },
+    {
+      "epoch": 0.17544148746257998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8911155330812742e-05,
+      "loss": 1.0903,
+      "step": 6388
+    },
+    {
+      "epoch": 0.17546895169042323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891075165410124e-05,
+      "loss": 1.0641,
+      "step": 6389
+    },
+    {
+      "epoch": 0.17549641591826645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8910347906884413e-05,
+      "loss": 1.0374,
+      "step": 6390
+    },
+    {
+      "epoch": 0.1755238801461097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890994408916546e-05,
+      "loss": 1.0208,
+      "step": 6391
+    },
+    {
+      "epoch": 0.17555134437395292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890954020094758e-05,
+      "loss": 1.0627,
+      "step": 6392
+    },
+    {
+      "epoch": 0.17557880860179617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8909136242233974e-05,
+      "loss": 0.9873,
+      "step": 6393
+    },
+    {
+      "epoch": 0.1756062728296394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8908732213027826e-05,
+      "loss": 1.084,
+      "step": 6394
+    },
+    {
+      "epoch": 0.17563373705748264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8908328113332334e-05,
+      "loss": 1.0289,
+      "step": 6395
+    },
+    {
+      "epoch": 0.17566120128532586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8907923943150704e-05,
+      "loss": 1.0707,
+      "step": 6396
+    },
+    {
+      "epoch": 0.17568866551316908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8907519702486128e-05,
+      "loss": 1.035,
+      "step": 6397
+    },
+    {
+      "epoch": 0.17571612974101233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8907115391341807e-05,
+      "loss": 1.1301,
+      "step": 6398
+    },
+    {
+      "epoch": 0.17574359396885555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8906711009720936e-05,
+      "loss": 0.9849,
+      "step": 6399
+    },
+    {
+      "epoch": 0.1757710581966988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8906306557626718e-05,
+      "loss": 1.0896,
+      "step": 6400
+    },
+    {
+      "epoch": 0.17579852242454203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890590203506235e-05,
+      "loss": 1.0514,
+      "step": 6401
+    },
+    {
+      "epoch": 0.17582598665238527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890549744203104e-05,
+      "loss": 1.0439,
+      "step": 6402
+    },
+    {
+      "epoch": 0.1758534508802285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8905092778535982e-05,
+      "loss": 0.9856,
+      "step": 6403
+    },
+    {
+      "epoch": 0.17588091510807174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8904688044580383e-05,
+      "loss": 1.0162,
+      "step": 6404
+    },
+    {
+      "epoch": 0.17590837933591497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8904283240167442e-05,
+      "loss": 1.101,
+      "step": 6405
+    },
+    {
+      "epoch": 0.17593584356375822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890387836530036e-05,
+      "loss": 0.9528,
+      "step": 6406
+    },
+    {
+      "epoch": 0.17596330779160144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8903473419982346e-05,
+      "loss": 0.9735,
+      "step": 6407
+    },
+    {
+      "epoch": 0.17599077201944469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8903068404216602e-05,
+      "loss": 1.1134,
+      "step": 6408
+    },
+    {
+      "epoch": 0.1760182362472879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890266331800633e-05,
+      "loss": 1.0484,
+      "step": 6409
+    },
+    {
+      "epoch": 0.17604570047513113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8902258161354737e-05,
+      "loss": 1.0256,
+      "step": 6410
+    },
+    {
+      "epoch": 0.17607316470297438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890185293426503e-05,
+      "loss": 1.0117,
+      "step": 6411
+    },
+    {
+      "epoch": 0.1761006289308176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8901447636740416e-05,
+      "loss": 0.9792,
+      "step": 6412
+    },
+    {
+      "epoch": 0.17612809315866085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89010422687841e-05,
+      "loss": 1.0221,
+      "step": 6413
+    },
+    {
+      "epoch": 0.17615555738650407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8900636830399285e-05,
+      "loss": 0.9367,
+      "step": 6414
+    },
+    {
+      "epoch": 0.17618302161434732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8900231321589188e-05,
+      "loss": 0.9577,
+      "step": 6415
+    },
+    {
+      "epoch": 0.17621048584219054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8899825742357012e-05,
+      "loss": 0.966,
+      "step": 6416
+    },
+    {
+      "epoch": 0.1762379500700338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8899420092705966e-05,
+      "loss": 0.9955,
+      "step": 6417
+    },
+    {
+      "epoch": 0.176265414297877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8899014372639265e-05,
+      "loss": 0.9553,
+      "step": 6418
+    },
+    {
+      "epoch": 0.17629287852572026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889860858216011e-05,
+      "loss": 0.9335,
+      "step": 6419
+    },
+    {
+      "epoch": 0.17632034275356348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889820272127172e-05,
+      "loss": 1.031,
+      "step": 6420
+    },
+    {
+      "epoch": 0.17634780698140673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8897796789977302e-05,
+      "loss": 1.0604,
+      "step": 6421
+    },
+    {
+      "epoch": 0.17637527120924995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889739078828007e-05,
+      "loss": 0.9061,
+      "step": 6422
+    },
+    {
+      "epoch": 0.1764027354370932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8896984716183235e-05,
+      "loss": 0.9875,
+      "step": 6423
+    },
+    {
+      "epoch": 0.17643019966493642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889657857369001e-05,
+      "loss": 0.9259,
+      "step": 6424
+    },
+    {
+      "epoch": 0.17645766389277964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889617236080361e-05,
+      "loss": 1.0629,
+      "step": 6425
+    },
+    {
+      "epoch": 0.1764851281206229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889576607752725e-05,
+      "loss": 0.9009,
+      "step": 6426
+    },
+    {
+      "epoch": 0.1765125923484661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8895359723864138e-05,
+      "loss": 1.0676,
+      "step": 6427
+    },
+    {
+      "epoch": 0.17654005657630936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88949532998175e-05,
+      "loss": 1.1731,
+      "step": 6428
+    },
+    {
+      "epoch": 0.17656752080415258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889454680539054e-05,
+      "loss": 0.912,
+      "step": 6429
+    },
+    {
+      "epoch": 0.17659498503199583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8894140240586483e-05,
+      "loss": 1.1449,
+      "step": 6430
+    },
+    {
+      "epoch": 0.17662244925983905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8893733605408548e-05,
+      "loss": 1.1179,
+      "step": 6431
+    },
+    {
+      "epoch": 0.1766499134876823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889332689985994e-05,
+      "loss": 1.0112,
+      "step": 6432
+    },
+    {
+      "epoch": 0.17667737771552552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889292012394389e-05,
+      "loss": 1.0571,
+      "step": 6433
+    },
+    {
+      "epoch": 0.17670484194336877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8892513277663605e-05,
+      "loss": 1.0046,
+      "step": 6434
+    },
+    {
+      "epoch": 0.176732306171212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8892106361022316e-05,
+      "loss": 1.016,
+      "step": 6435
+    },
+    {
+      "epoch": 0.17675977039905524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8891699374023235e-05,
+      "loss": 0.94,
+      "step": 6436
+    },
+    {
+      "epoch": 0.17678723462689847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8891292316669584e-05,
+      "loss": 0.9654,
+      "step": 6437
+    },
+    {
+      "epoch": 0.1768146988547417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8890885188964585e-05,
+      "loss": 1.0485,
+      "step": 6438
+    },
+    {
+      "epoch": 0.17684216308258494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8890477990911456e-05,
+      "loss": 1.0794,
+      "step": 6439
+    },
+    {
+      "epoch": 0.17686962731042816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8890070722513424e-05,
+      "loss": 1.097,
+      "step": 6440
+    },
+    {
+      "epoch": 0.1768970915382714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888966338377371e-05,
+      "loss": 1.1772,
+      "step": 6441
+    },
+    {
+      "epoch": 0.17692455576611463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8889255974695533e-05,
+      "loss": 1.0243,
+      "step": 6442
+    },
+    {
+      "epoch": 0.17695201999395788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888884849528212e-05,
+      "loss": 1.0858,
+      "step": 6443
+    },
+    {
+      "epoch": 0.1769794842218011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8888440945536695e-05,
+      "loss": 0.9703,
+      "step": 6444
+    },
+    {
+      "epoch": 0.17700694844964435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8888033325462482e-05,
+      "loss": 1.0961,
+      "step": 6445
+    },
+    {
+      "epoch": 0.17703441267748757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8887625635062705e-05,
+      "loss": 0.9985,
+      "step": 6446
+    },
+    {
+      "epoch": 0.17706187690533082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8887217874340592e-05,
+      "loss": 1.0143,
+      "step": 6447
+    },
+    {
+      "epoch": 0.17708934113317404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888681004329937e-05,
+      "loss": 1.1205,
+      "step": 6448
+    },
+    {
+      "epoch": 0.1771168053610173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8886402141942262e-05,
+      "loss": 0.9532,
+      "step": 6449
+    },
+    {
+      "epoch": 0.1771442695888605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88859941702725e-05,
+      "loss": 1.1059,
+      "step": 6450
+    },
+    {
+      "epoch": 0.17717173381670373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888558612829331e-05,
+      "loss": 0.9928,
+      "step": 6451
+    },
+    {
+      "epoch": 0.17719919804454698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888517801600792e-05,
+      "loss": 1.0331,
+      "step": 6452
+    },
+    {
+      "epoch": 0.1772266622723902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888476983341956e-05,
+      "loss": 1.0229,
+      "step": 6453
+    },
+    {
+      "epoch": 0.17725412650023345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884361580531457e-05,
+      "loss": 0.9814,
+      "step": 6454
+    },
+    {
+      "epoch": 0.17728159072807667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8883953257346847e-05,
+      "loss": 1.0797,
+      "step": 6455
+    },
+    {
+      "epoch": 0.17730905495591992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888354486386896e-05,
+      "loss": 1.0182,
+      "step": 6456
+    },
+    {
+      "epoch": 0.17733651918376314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888313640010102e-05,
+      "loss": 1.1272,
+      "step": 6457
+    },
+    {
+      "epoch": 0.1773639834116064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8882727866046264e-05,
+      "loss": 1.0515,
+      "step": 6458
+    },
+    {
+      "epoch": 0.1773914476394496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8882319261707928e-05,
+      "loss": 1.0033,
+      "step": 6459
+    },
+    {
+      "epoch": 0.17741891186729286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8881910587089236e-05,
+      "loss": 1.0441,
+      "step": 6460
+    },
+    {
+      "epoch": 0.17744637609513608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888150184219343e-05,
+      "loss": 1.1014,
+      "step": 6461
+    },
+    {
+      "epoch": 0.17747384032297933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888109302702374e-05,
+      "loss": 0.9227,
+      "step": 6462
+    },
+    {
+      "epoch": 0.17750130455082255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8880684141583404e-05,
+      "loss": 0.9707,
+      "step": 6463
+    },
+    {
+      "epoch": 0.17752876877866577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8880275185875655e-05,
+      "loss": 1.062,
+      "step": 6464
+    },
+    {
+      "epoch": 0.17755623300650902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8879866159903726e-05,
+      "loss": 1.1586,
+      "step": 6465
+    },
+    {
+      "epoch": 0.17758369723435224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8879457063670857e-05,
+      "loss": 0.996,
+      "step": 6466
+    },
+    {
+      "epoch": 0.1776111614621955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8879047897180288e-05,
+      "loss": 1.0064,
+      "step": 6467
+    },
+    {
+      "epoch": 0.17763862569003872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8878638660435248e-05,
+      "loss": 1.0717,
+      "step": 6468
+    },
+    {
+      "epoch": 0.17766608991788196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887822935343898e-05,
+      "loss": 0.9746,
+      "step": 6469
+    },
+    {
+      "epoch": 0.17769355414572519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8877819976194724e-05,
+      "loss": 0.9848,
+      "step": 6470
+    },
+    {
+      "epoch": 0.17772101837356843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8877410528705718e-05,
+      "loss": 1.0002,
+      "step": 6471
+    },
+    {
+      "epoch": 0.17774848260141166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8877001010975197e-05,
+      "loss": 1.1798,
+      "step": 6472
+    },
+    {
+      "epoch": 0.1777759468292549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887659142300641e-05,
+      "loss": 1.1611,
+      "step": 6473
+    },
+    {
+      "epoch": 0.17780341105709813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887618176480259e-05,
+      "loss": 1.0814,
+      "step": 6474
+    },
+    {
+      "epoch": 0.17783087528494138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8875772036366983e-05,
+      "loss": 0.8464,
+      "step": 6475
+    },
+    {
+      "epoch": 0.1778583395127846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887536223770283e-05,
+      "loss": 0.9956,
+      "step": 6476
+    },
+    {
+      "epoch": 0.17788580374062785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887495236881337e-05,
+      "loss": 1.0605,
+      "step": 6477
+    },
+    {
+      "epoch": 0.17791326796847107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8874542429701856e-05,
+      "loss": 0.9959,
+      "step": 6478
+    },
+    {
+      "epoch": 0.1779407321963143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887413242037152e-05,
+      "loss": 1.0483,
+      "step": 6479
+    },
+    {
+      "epoch": 0.17796819642415754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887372234082561e-05,
+      "loss": 1.0481,
+      "step": 6480
+    },
+    {
+      "epoch": 0.17799566065200076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8873312191067374e-05,
+      "loss": 1.0589,
+      "step": 6481
+    },
+    {
+      "epoch": 0.178023124879844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887290197110005e-05,
+      "loss": 1.0116,
+      "step": 6482
+    },
+    {
+      "epoch": 0.17805058910768723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887249168092689e-05,
+      "loss": 0.9908,
+      "step": 6483
+    },
+    {
+      "epoch": 0.17807805333553048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887208132055114e-05,
+      "loss": 0.9533,
+      "step": 6484
+    },
+    {
+      "epoch": 0.1781055175633737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887167088997605e-05,
+      "loss": 1.1333,
+      "step": 6485
+    },
+    {
+      "epoch": 0.17813298179121695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887126038920486e-05,
+      "loss": 1.1243,
+      "step": 6486
+    },
+    {
+      "epoch": 0.17816044601906017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887084981824082e-05,
+      "loss": 1.0757,
+      "step": 6487
+    },
+    {
+      "epoch": 0.17818791024690342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8870439177087184e-05,
+      "loss": 1.063,
+      "step": 6488
+    },
+    {
+      "epoch": 0.17821537447474664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8870028465747192e-05,
+      "loss": 1.0099,
+      "step": 6489
+    },
+    {
+      "epoch": 0.1782428387025899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88696176842241e-05,
+      "loss": 0.9968,
+      "step": 6490
+    },
+    {
+      "epoch": 0.1782703029304331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886920683252116e-05,
+      "loss": 1.1021,
+      "step": 6491
+    },
+    {
+      "epoch": 0.17829776715827633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8868795910641616e-05,
+      "loss": 1.0507,
+      "step": 6492
+    },
+    {
+      "epoch": 0.17832523138611958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886838491858873e-05,
+      "loss": 1.0438,
+      "step": 6493
+    },
+    {
+      "epoch": 0.1783526956139628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8867973856365738e-05,
+      "loss": 1.0543,
+      "step": 6494
+    },
+    {
+      "epoch": 0.17838015984180605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886756272397591e-05,
+      "loss": 1.017,
+      "step": 6495
+    },
+    {
+      "epoch": 0.17840762406964927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886715152142249e-05,
+      "loss": 1.0635,
+      "step": 6496
+    },
+    {
+      "epoch": 0.17843508829749252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8866740248708727e-05,
+      "loss": 0.9643,
+      "step": 6497
+    },
+    {
+      "epoch": 0.17846255252533574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8866328905837884e-05,
+      "loss": 1.0766,
+      "step": 6498
+    },
+    {
+      "epoch": 0.178490016753179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8865917492813213e-05,
+      "loss": 0.9112,
+      "step": 6499
+    },
+    {
+      "epoch": 0.17851748098102221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8865506009637965e-05,
+      "loss": 1.0477,
+      "step": 6500
+    },
+    {
+      "epoch": 0.17854494520886546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88650944563154e-05,
+      "loss": 0.9937,
+      "step": 6501
+    },
+    {
+      "epoch": 0.17857240943670868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8864682832848777e-05,
+      "loss": 1.0104,
+      "step": 6502
+    },
+    {
+      "epoch": 0.17859987366455193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8864271139241346e-05,
+      "loss": 0.9817,
+      "step": 6503
+    },
+    {
+      "epoch": 0.17862733789239515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886385937549637e-05,
+      "loss": 1.1007,
+      "step": 6504
+    },
+    {
+      "epoch": 0.17865480212023838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8863447541617103e-05,
+      "loss": 1.0742,
+      "step": 6505
+    },
+    {
+      "epoch": 0.17868226634808163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8863035637606807e-05,
+      "loss": 0.8489,
+      "step": 6506
+    },
+    {
+      "epoch": 0.17870973057592485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886262366346874e-05,
+      "loss": 0.9664,
+      "step": 6507
+    },
+    {
+      "epoch": 0.1787371948037681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8862211619206164e-05,
+      "loss": 1.0556,
+      "step": 6508
+    },
+    {
+      "epoch": 0.17876465903161132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8861799504822333e-05,
+      "loss": 1.0356,
+      "step": 6509
+    },
+    {
+      "epoch": 0.17879212325945457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8861387320320515e-05,
+      "loss": 1.0124,
+      "step": 6510
+    },
+    {
+      "epoch": 0.1788195874872978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8860975065703966e-05,
+      "loss": 1.1118,
+      "step": 6511
+    },
+    {
+      "epoch": 0.17884705171514104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8860562740975948e-05,
+      "loss": 1.0651,
+      "step": 6512
+    },
+    {
+      "epoch": 0.17887451594298426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886015034613973e-05,
+      "loss": 1.0806,
+      "step": 6513
+    },
+    {
+      "epoch": 0.1789019801708275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8859737881198566e-05,
+      "loss": 1.0198,
+      "step": 6514
+    },
+    {
+      "epoch": 0.17892944439867073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8859325346155727e-05,
+      "loss": 1.0385,
+      "step": 6515
+    },
+    {
+      "epoch": 0.17895690862651398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885891274101447e-05,
+      "loss": 0.9573,
+      "step": 6516
+    },
+    {
+      "epoch": 0.1789843728543572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8858500065778067e-05,
+      "loss": 1.0467,
+      "step": 6517
+    },
+    {
+      "epoch": 0.17901183708220042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8858087320449782e-05,
+      "loss": 1.1185,
+      "step": 6518
+    },
+    {
+      "epoch": 0.17903930131004367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885767450503287e-05,
+      "loss": 1.0253,
+      "step": 6519
+    },
+    {
+      "epoch": 0.1790667655378869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8857261619530618e-05,
+      "loss": 1.0476,
+      "step": 6520
+    },
+    {
+      "epoch": 0.17909422976573014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8856848663946274e-05,
+      "loss": 0.9276,
+      "step": 6521
+    },
+    {
+      "epoch": 0.17912169399357336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8856435638283114e-05,
+      "loss": 1.0114,
+      "step": 6522
+    },
+    {
+      "epoch": 0.1791491582214166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8856022542544404e-05,
+      "loss": 1.093,
+      "step": 6523
+    },
+    {
+      "epoch": 0.17917662244925983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8855609376733415e-05,
+      "loss": 0.9576,
+      "step": 6524
+    },
+    {
+      "epoch": 0.17920408667710308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8855196140853415e-05,
+      "loss": 0.9856,
+      "step": 6525
+    },
+    {
+      "epoch": 0.1792315509049463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885478283490767e-05,
+      "loss": 0.9457,
+      "step": 6526
+    },
+    {
+      "epoch": 0.17925901513278955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8854369458899454e-05,
+      "loss": 1.0653,
+      "step": 6527
+    },
+    {
+      "epoch": 0.17928647936063277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8853956012832037e-05,
+      "loss": 1.0848,
+      "step": 6528
+    },
+    {
+      "epoch": 0.17931394358847602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885354249670869e-05,
+      "loss": 1.1043,
+      "step": 6529
+    },
+    {
+      "epoch": 0.17934140781631924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8853128910532684e-05,
+      "loss": 1.0338,
+      "step": 6530
+    },
+    {
+      "epoch": 0.1793688720441625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8852715254307298e-05,
+      "loss": 1.0788,
+      "step": 6531
+    },
+    {
+      "epoch": 0.1793963362720057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885230152803579e-05,
+      "loss": 1.0465,
+      "step": 6532
+    },
+    {
+      "epoch": 0.17942380049984893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8851887731721456e-05,
+      "loss": 0.9755,
+      "step": 6533
+    },
+    {
+      "epoch": 0.17945126472769218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8851473865367548e-05,
+      "loss": 0.9739,
+      "step": 6534
+    },
+    {
+      "epoch": 0.1794787289555354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885105992897735e-05,
+      "loss": 1.0024,
+      "step": 6535
+    },
+    {
+      "epoch": 0.17950619318337865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8850645922554142e-05,
+      "loss": 1.1754,
+      "step": 6536
+    },
+    {
+      "epoch": 0.17953365741122188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8850231846101193e-05,
+      "loss": 1.0223,
+      "step": 6537
+    },
+    {
+      "epoch": 0.17956112163906512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884981769962178e-05,
+      "loss": 1.0185,
+      "step": 6538
+    },
+    {
+      "epoch": 0.17958858586690835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8849403483119185e-05,
+      "loss": 0.9991,
+      "step": 6539
+    },
+    {
+      "epoch": 0.1796160500947516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8848989196596678e-05,
+      "loss": 1.0393,
+      "step": 6540
+    },
+    {
+      "epoch": 0.17964351432259482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8848574840057543e-05,
+      "loss": 1.0999,
+      "step": 6541
+    },
+    {
+      "epoch": 0.17967097855043807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8848160413505057e-05,
+      "loss": 1.0187,
+      "step": 6542
+    },
+    {
+      "epoch": 0.1796984427782813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8847745916942495e-05,
+      "loss": 1.0582,
+      "step": 6543
+    },
+    {
+      "epoch": 0.17972590700612454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8847331350373142e-05,
+      "loss": 0.9548,
+      "step": 6544
+    },
+    {
+      "epoch": 0.17975337123396776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8846916713800277e-05,
+      "loss": 1.0608,
+      "step": 6545
+    },
+    {
+      "epoch": 0.17978083546181098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884650200722718e-05,
+      "loss": 1.0201,
+      "step": 6546
+    },
+    {
+      "epoch": 0.17980829968965423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8846087230657128e-05,
+      "loss": 1.1155,
+      "step": 6547
+    },
+    {
+      "epoch": 0.17983576391749745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884567238409341e-05,
+      "loss": 0.9987,
+      "step": 6548
+    },
+    {
+      "epoch": 0.1798632281453407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884525746753931e-05,
+      "loss": 1.1362,
+      "step": 6549
+    },
+    {
+      "epoch": 0.17989069237318392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8844842480998102e-05,
+      "loss": 1.0712,
+      "step": 6550
+    },
+    {
+      "epoch": 0.17991815660102717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8844427424473075e-05,
+      "loss": 1.0031,
+      "step": 6551
+    },
+    {
+      "epoch": 0.1799456208288704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8844012297967513e-05,
+      "loss": 1.0318,
+      "step": 6552
+    },
+    {
+      "epoch": 0.17997308505671364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8843597101484697e-05,
+      "loss": 0.9316,
+      "step": 6553
+    },
+    {
+      "epoch": 0.18000054928455686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8843181835027916e-05,
+      "loss": 1.12,
+      "step": 6554
+    },
+    {
+      "epoch": 0.1800280135124001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8842766498600458e-05,
+      "loss": 1.0691,
+      "step": 6555
+    },
+    {
+      "epoch": 0.18005547774024333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88423510922056e-05,
+      "loss": 1.1068,
+      "step": 6556
+    },
+    {
+      "epoch": 0.18008294196808658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884193561584664e-05,
+      "loss": 0.9331,
+      "step": 6557
+    },
+    {
+      "epoch": 0.1801104061959298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884152006952686e-05,
+      "loss": 1.0306,
+      "step": 6558
+    },
+    {
+      "epoch": 0.18013787042377302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884110445324955e-05,
+      "loss": 0.9682,
+      "step": 6559
+    },
+    {
+      "epoch": 0.18016533465161627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8840688767017995e-05,
+      "loss": 1.0736,
+      "step": 6560
+    },
+    {
+      "epoch": 0.1801927988794595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8840273010835482e-05,
+      "loss": 1.0178,
+      "step": 6561
+    },
+    {
+      "epoch": 0.18022026310730274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8839857184705307e-05,
+      "loss": 0.986,
+      "step": 6562
+    },
+    {
+      "epoch": 0.18024772733514596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883944128863076e-05,
+      "loss": 1.0312,
+      "step": 6563
+    },
+    {
+      "epoch": 0.1802751915629892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8839025322615126e-05,
+      "loss": 0.9939,
+      "step": 6564
+    },
+    {
+      "epoch": 0.18030265579083243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8838609286661703e-05,
+      "loss": 0.9776,
+      "step": 6565
+    },
+    {
+      "epoch": 0.18033012001867568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8838193180773778e-05,
+      "loss": 0.9246,
+      "step": 6566
+    },
+    {
+      "epoch": 0.1803575842465189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8837777004954646e-05,
+      "loss": 0.9589,
+      "step": 6567
+    },
+    {
+      "epoch": 0.18038504847436215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88373607592076e-05,
+      "loss": 1.0124,
+      "step": 6568
+    },
+    {
+      "epoch": 0.18041251270220537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883694444353593e-05,
+      "loss": 1.0558,
+      "step": 6569
+    },
+    {
+      "epoch": 0.18043997693004862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8836528057942934e-05,
+      "loss": 0.967,
+      "step": 6570
+    },
+    {
+      "epoch": 0.18046744115789184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88361116024319e-05,
+      "loss": 1.1533,
+      "step": 6571
+    },
+    {
+      "epoch": 0.1804949053857351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8835695077006137e-05,
+      "loss": 1.0546,
+      "step": 6572
+    },
+    {
+      "epoch": 0.18052236961357832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8835278481668923e-05,
+      "loss": 1.0342,
+      "step": 6573
+    },
+    {
+      "epoch": 0.18054983384142154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883486181642357e-05,
+      "loss": 1.0966,
+      "step": 6574
+    },
+    {
+      "epoch": 0.18057729806926479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8834445081273364e-05,
+      "loss": 1.0099,
+      "step": 6575
+    },
+    {
+      "epoch": 0.180604762297108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883402827622161e-05,
+      "loss": 1.0787,
+      "step": 6576
+    },
+    {
+      "epoch": 0.18063222652495126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88336114012716e-05,
+      "loss": 1.036,
+      "step": 6577
+    },
+    {
+      "epoch": 0.18065969075279448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8833194456426636e-05,
+      "loss": 1.0432,
+      "step": 6578
+    },
+    {
+      "epoch": 0.18068715498063773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8832777441690014e-05,
+      "loss": 1.0348,
+      "step": 6579
+    },
+    {
+      "epoch": 0.18071461920848095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8832360357065035e-05,
+      "loss": 1.0214,
+      "step": 6580
+    },
+    {
+      "epoch": 0.1807420834363242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8831943202555e-05,
+      "loss": 0.9867,
+      "step": 6581
+    },
+    {
+      "epoch": 0.18076954766416742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883152597816321e-05,
+      "loss": 1.0594,
+      "step": 6582
+    },
+    {
+      "epoch": 0.18079701189201067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8831108683892964e-05,
+      "loss": 1.0321,
+      "step": 6583
+    },
+    {
+      "epoch": 0.1808244761198539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883069131974757e-05,
+      "loss": 1.0135,
+      "step": 6584
+    },
+    {
+      "epoch": 0.18085194034769714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883027388573032e-05,
+      "loss": 1.0517,
+      "step": 6585
+    },
+    {
+      "epoch": 0.18087940457554036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829856381844527e-05,
+      "loss": 0.9953,
+      "step": 6586
+    },
+    {
+      "epoch": 0.18090686880338358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829438808093484e-05,
+      "loss": 1.0613,
+      "step": 6587
+    },
+    {
+      "epoch": 0.18093433303122683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829021164480506e-05,
+      "loss": 0.9739,
+      "step": 6588
+    },
+    {
+      "epoch": 0.18096179725907005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8828603451008893e-05,
+      "loss": 1.1323,
+      "step": 6589
+    },
+    {
+      "epoch": 0.1809892614869133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8828185667681946e-05,
+      "loss": 0.814,
+      "step": 6590
+    },
+    {
+      "epoch": 0.18101672571475652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8827767814502978e-05,
+      "loss": 0.9241,
+      "step": 6591
+    },
+    {
+      "epoch": 0.18104418994259977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8827349891475288e-05,
+      "loss": 1.044,
+      "step": 6592
+    },
+    {
+      "epoch": 0.181071654170443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882693189860219e-05,
+      "loss": 1.0896,
+      "step": 6593
+    },
+    {
+      "epoch": 0.18109911839828624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8826513835886985e-05,
+      "loss": 1.1302,
+      "step": 6594
+    },
+    {
+      "epoch": 0.18112658262612946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882609570333298e-05,
+      "loss": 1.0091,
+      "step": 6595
+    },
+    {
+      "epoch": 0.1811540468539727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8825677500943493e-05,
+      "loss": 1.0762,
+      "step": 6596
+    },
+    {
+      "epoch": 0.18118151108181593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8825259228721825e-05,
+      "loss": 0.947,
+      "step": 6597
+    },
+    {
+      "epoch": 0.18120897530965918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8824840886671286e-05,
+      "loss": 0.9313,
+      "step": 6598
+    },
+    {
+      "epoch": 0.1812364395375024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882442247479519e-05,
+      "loss": 1.0219,
+      "step": 6599
+    },
+    {
+      "epoch": 0.18126390376534562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8824003993096843e-05,
+      "loss": 0.9763,
+      "step": 6600
+    },
+    {
+      "epoch": 0.18129136799318887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8823585441579556e-05,
+      "loss": 1.0027,
+      "step": 6601
+    },
+    {
+      "epoch": 0.1813188322210321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882316682024665e-05,
+      "loss": 1.0295,
+      "step": 6602
+    },
+    {
+      "epoch": 0.18134629644887534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8822748129101424e-05,
+      "loss": 1.0525,
+      "step": 6603
+    },
+    {
+      "epoch": 0.18137376067671857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8822329368147197e-05,
+      "loss": 1.0109,
+      "step": 6604
+    },
+    {
+      "epoch": 0.18140122490456181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8821910537387288e-05,
+      "loss": 1.0477,
+      "step": 6605
+    },
+    {
+      "epoch": 0.18142868913240504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8821491636825002e-05,
+      "loss": 1.044,
+      "step": 6606
+    },
+    {
+      "epoch": 0.18145615336024828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882107266646366e-05,
+      "loss": 1.0699,
+      "step": 6607
+    },
+    {
+      "epoch": 0.1814836175880915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8820653626306572e-05,
+      "loss": 1.0117,
+      "step": 6608
+    },
+    {
+      "epoch": 0.18151108181593475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882023451635705e-05,
+      "loss": 1.0085,
+      "step": 6609
+    },
+    {
+      "epoch": 0.18153854604377798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8819815336618425e-05,
+      "loss": 1.006,
+      "step": 6610
+    },
+    {
+      "epoch": 0.18156601027162123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8819396087094e-05,
+      "loss": 0.9772,
+      "step": 6611
+    },
+    {
+      "epoch": 0.18159347449946445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88189767677871e-05,
+      "loss": 1.0508,
+      "step": 6612
+    },
+    {
+      "epoch": 0.18162093872730767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8818557378701037e-05,
+      "loss": 0.9841,
+      "step": 6613
+    },
+    {
+      "epoch": 0.18164840295515092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8818137919839132e-05,
+      "loss": 0.9631,
+      "step": 6614
+    },
+    {
+      "epoch": 0.18167586718299414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8817718391204705e-05,
+      "loss": 1.0128,
+      "step": 6615
+    },
+    {
+      "epoch": 0.1817033314108374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8817298792801075e-05,
+      "loss": 0.9614,
+      "step": 6616
+    },
+    {
+      "epoch": 0.1817307956386806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881687912463156e-05,
+      "loss": 0.9944,
+      "step": 6617
+    },
+    {
+      "epoch": 0.18175825986652386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881645938669948e-05,
+      "loss": 0.9886,
+      "step": 6618
+    },
+    {
+      "epoch": 0.18178572409436708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8816039579008162e-05,
+      "loss": 1.0637,
+      "step": 6619
+    },
+    {
+      "epoch": 0.18181318832221033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8815619701560922e-05,
+      "loss": 1.0894,
+      "step": 6620
+    },
+    {
+      "epoch": 0.18184065255005355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8815199754361085e-05,
+      "loss": 0.898,
+      "step": 6621
+    },
+    {
+      "epoch": 0.1818681167778968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8814779737411973e-05,
+      "loss": 1.1121,
+      "step": 6622
+    },
+    {
+      "epoch": 0.18189558100574002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881435965071691e-05,
+      "loss": 0.9622,
+      "step": 6623
+    },
+    {
+      "epoch": 0.18192304523358327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813939494279215e-05,
+      "loss": 1.033,
+      "step": 6624
+    },
+    {
+      "epoch": 0.1819505094614265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813519268102222e-05,
+      "loss": 1.0774,
+      "step": 6625
+    },
+    {
+      "epoch": 0.18197797368926974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813098972189247e-05,
+      "loss": 1.0546,
+      "step": 6626
+    },
+    {
+      "epoch": 0.18200543791711296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881267860654362e-05,
+      "loss": 1.1039,
+      "step": 6627
+    },
+    {
+      "epoch": 0.18203290214495618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8812258171168663e-05,
+      "loss": 0.9388,
+      "step": 6628
+    },
+    {
+      "epoch": 0.18206036637279943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8811837666067712e-05,
+      "loss": 1.0226,
+      "step": 6629
+    },
+    {
+      "epoch": 0.18208783060064265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881141709124408e-05,
+      "loss": 1.0957,
+      "step": 6630
+    },
+    {
+      "epoch": 0.1821152948284859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881099644670111e-05,
+      "loss": 1.0804,
+      "step": 6631
+    },
+    {
+      "epoch": 0.18214275905632912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881057573244212e-05,
+      "loss": 0.9124,
+      "step": 6632
+    },
+    {
+      "epoch": 0.18217022328417237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8810154948470444e-05,
+      "loss": 1.0676,
+      "step": 6633
+    },
+    {
+      "epoch": 0.1821976875120156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8809734094789407e-05,
+      "loss": 1.0219,
+      "step": 6634
+    },
+    {
+      "epoch": 0.18222515173985884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880931317140234e-05,
+      "loss": 1.0276,
+      "step": 6635
+    },
+    {
+      "epoch": 0.18225261596770206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880889217831258e-05,
+      "loss": 1.1302,
+      "step": 6636
+    },
+    {
+      "epoch": 0.1822800801955453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880847111552345e-05,
+      "loss": 1.0742,
+      "step": 6637
+    },
+    {
+      "epoch": 0.18230754442338853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8808049983038284e-05,
+      "loss": 1.0511,
+      "step": 6638
+    },
+    {
+      "epoch": 0.18233500865123178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8807628780860415e-05,
+      "loss": 0.9688,
+      "step": 6639
+    },
+    {
+      "epoch": 0.182362472879075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880720750899317e-05,
+      "loss": 0.9781,
+      "step": 6640
+    },
+    {
+      "epoch": 0.18238993710691823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8806786167439897e-05,
+      "loss": 0.9451,
+      "step": 6641
+    },
+    {
+      "epoch": 0.18241740133476148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8806364756203915e-05,
+      "loss": 1.0122,
+      "step": 6642
+    },
+    {
+      "epoch": 0.1824448655626047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8805943275288564e-05,
+      "loss": 1.0881,
+      "step": 6643
+    },
+    {
+      "epoch": 0.18247232979044795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880552172469718e-05,
+      "loss": 1.0424,
+      "step": 6644
+    },
+    {
+      "epoch": 0.18249979401829117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8805100104433098e-05,
+      "loss": 1.0643,
+      "step": 6645
+    },
+    {
+      "epoch": 0.18252725824613442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880467841449965e-05,
+      "loss": 0.9671,
+      "step": 6646
+    },
+    {
+      "epoch": 0.18255472247397764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880425665490018e-05,
+      "loss": 1.106,
+      "step": 6647
+    },
+    {
+      "epoch": 0.1825821867018209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803834825638017e-05,
+      "loss": 1.0094,
+      "step": 6648
+    },
+    {
+      "epoch": 0.1826096509296641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803412926716508e-05,
+      "loss": 1.0225,
+      "step": 6649
+    },
+    {
+      "epoch": 0.18263711515750736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880299095813898e-05,
+      "loss": 1.0675,
+      "step": 6650
+    },
+    {
+      "epoch": 0.18266457938535058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880256891990878e-05,
+      "loss": 1.0816,
+      "step": 6651
+    },
+    {
+      "epoch": 0.18269204361319383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8802146812029246e-05,
+      "loss": 1.0123,
+      "step": 6652
+    },
+    {
+      "epoch": 0.18271950784103705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8801724634503716e-05,
+      "loss": 0.9982,
+      "step": 6653
+    },
+    {
+      "epoch": 0.18274697206888027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880130238733553e-05,
+      "loss": 1.0723,
+      "step": 6654
+    },
+    {
+      "epoch": 0.18277443629672352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8800880070528032e-05,
+      "loss": 1.0918,
+      "step": 6655
+    },
+    {
+      "epoch": 0.18280190052456674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8800457684084557e-05,
+      "loss": 0.9,
+      "step": 6656
+    },
+    {
+      "epoch": 0.18282936475241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8800035228008458e-05,
+      "loss": 1.0123,
+      "step": 6657
+    },
+    {
+      "epoch": 0.1828568289802532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879961270230307e-05,
+      "loss": 0.9813,
+      "step": 6658
+    },
+    {
+      "epoch": 0.18288429320809646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8799190106971734e-05,
+      "loss": 1.0089,
+      "step": 6659
+    },
+    {
+      "epoch": 0.18291175743593968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87987674420178e-05,
+      "loss": 0.9929,
+      "step": 6660
+    },
+    {
+      "epoch": 0.18293922166378293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879834470744461e-05,
+      "loss": 1.0045,
+      "step": 6661
+    },
+    {
+      "epoch": 0.18296668589162615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8797921903255507e-05,
+      "loss": 1.0458,
+      "step": 6662
+    },
+    {
+      "epoch": 0.1829941501194694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8797499029453837e-05,
+      "loss": 0.9864,
+      "step": 6663
+    },
+    {
+      "epoch": 0.18302161434731262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8797076086042953e-05,
+      "loss": 1.0654,
+      "step": 6664
+    },
+    {
+      "epoch": 0.18304907857515587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879665307302619e-05,
+      "loss": 0.9694,
+      "step": 6665
+    },
+    {
+      "epoch": 0.1830765428029991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87962299904069e-05,
+      "loss": 0.9835,
+      "step": 6666
+    },
+    {
+      "epoch": 0.18310400703084231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8795806838188433e-05,
+      "loss": 0.9332,
+      "step": 6667
+    },
+    {
+      "epoch": 0.18313147125868556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8795383616374136e-05,
+      "loss": 1.0629,
+      "step": 6668
+    },
+    {
+      "epoch": 0.18315893548652878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8794960324967354e-05,
+      "loss": 0.9537,
+      "step": 6669
+    },
+    {
+      "epoch": 0.18318639971437203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8794536963971442e-05,
+      "loss": 0.9642,
+      "step": 6670
+    },
+    {
+      "epoch": 0.18321386394221525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8794113533389744e-05,
+      "loss": 1.0217,
+      "step": 6671
+    },
+    {
+      "epoch": 0.1832413281700585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8793690033225615e-05,
+      "loss": 1.0108,
+      "step": 6672
+    },
+    {
+      "epoch": 0.18326879239790173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8793266463482407e-05,
+      "loss": 1.0324,
+      "step": 6673
+    },
+    {
+      "epoch": 0.18329625662574497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8792842824163463e-05,
+      "loss": 1.0276,
+      "step": 6674
+    },
+    {
+      "epoch": 0.1833237208535882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8792419115272143e-05,
+      "loss": 1.1329,
+      "step": 6675
+    },
+    {
+      "epoch": 0.18335118508143144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8791995336811798e-05,
+      "loss": 0.9614,
+      "step": 6676
+    },
+    {
+      "epoch": 0.18337864930927467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879157148878578e-05,
+      "loss": 1.0032,
+      "step": 6677
+    },
+    {
+      "epoch": 0.18340611353711792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8791147571197443e-05,
+      "loss": 1.0496,
+      "step": 6678
+    },
+    {
+      "epoch": 0.18343357776496114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8790723584050138e-05,
+      "loss": 1.0254,
+      "step": 6679
+    },
+    {
+      "epoch": 0.18346104199280439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8790299527347225e-05,
+      "loss": 0.9349,
+      "step": 6680
+    },
+    {
+      "epoch": 0.1834885062206476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878987540109206e-05,
+      "loss": 1.0273,
+      "step": 6681
+    },
+    {
+      "epoch": 0.18351597044849083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878945120528799e-05,
+      "loss": 1.0783,
+      "step": 6682
+    },
+    {
+      "epoch": 0.18354343467633408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878902693993838e-05,
+      "loss": 1.1301,
+      "step": 6683
+    },
+    {
+      "epoch": 0.1835708989041773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8788602605046585e-05,
+      "loss": 0.9545,
+      "step": 6684
+    },
+    {
+      "epoch": 0.18359836313202055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8788178200615963e-05,
+      "loss": 1.0528,
+      "step": 6685
+    },
+    {
+      "epoch": 0.18362582735986377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787753726649868e-05,
+      "loss": 0.9659,
+      "step": 6686
+    },
+    {
+      "epoch": 0.18365329158770702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787329183151662e-05,
+      "loss": 0.9431,
+      "step": 6687
+    },
+    {
+      "epoch": 0.18368075581555024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8786904570124706e-05,
+      "loss": 1.0737,
+      "step": 6688
+    },
+    {
+      "epoch": 0.1837082200433935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8786479887572354e-05,
+      "loss": 1.0745,
+      "step": 6689
+    },
+    {
+      "epoch": 0.1837356842712367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8786055135497973e-05,
+      "loss": 1.0767,
+      "step": 6690
+    },
+    {
+      "epoch": 0.18376314849907996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878563031390492e-05,
+      "loss": 1.0565,
+      "step": 6691
+    },
+    {
+      "epoch": 0.18379061272692318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8785205422796555e-05,
+      "loss": 1.0476,
+      "step": 6692
+    },
+    {
+      "epoch": 0.18381807695476643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878478046217624e-05,
+      "loss": 1.1106,
+      "step": 6693
+    },
+    {
+      "epoch": 0.18384554118260965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8784355432047342e-05,
+      "loss": 1.1318,
+      "step": 6694
+    },
+    {
+      "epoch": 0.18387300541045287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8783930332413222e-05,
+      "loss": 1.0586,
+      "step": 6695
+    },
+    {
+      "epoch": 0.18390046963829612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8783505163277236e-05,
+      "loss": 0.9149,
+      "step": 6696
+    },
+    {
+      "epoch": 0.18392793386613934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8783079924642764e-05,
+      "loss": 1.0273,
+      "step": 6697
+    },
+    {
+      "epoch": 0.1839553980939826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8782654616513154e-05,
+      "loss": 0.9766,
+      "step": 6698
+    },
+    {
+      "epoch": 0.1839828623218258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8782229238891782e-05,
+      "loss": 1.1065,
+      "step": 6699
+    },
+    {
+      "epoch": 0.18401032654966906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878180379178201e-05,
+      "loss": 1.0999,
+      "step": 6700
+    },
+    {
+      "epoch": 0.18403779077751228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8781378275187204e-05,
+      "loss": 0.9775,
+      "step": 6701
+    },
+    {
+      "epoch": 0.18406525500535553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780952689110733e-05,
+      "loss": 0.9989,
+      "step": 6702
+    },
+    {
+      "epoch": 0.18409271923319875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780527033555964e-05,
+      "loss": 1.124,
+      "step": 6703
+    },
+    {
+      "epoch": 0.184120183461042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780101308526258e-05,
+      "loss": 1.057,
+      "step": 6704
+    },
+    {
+      "epoch": 0.18414764768888522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8779675514024994e-05,
+      "loss": 1.0071,
+      "step": 6705
+    },
+    {
+      "epoch": 0.18417511191672847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877924965005554e-05,
+      "loss": 1.1813,
+      "step": 6706
+    },
+    {
+      "epoch": 0.1842025761445717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8778823716621254e-05,
+      "loss": 1.0146,
+      "step": 6707
+    },
+    {
+      "epoch": 0.18423004037241492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877839771372552e-05,
+      "loss": 1.0481,
+      "step": 6708
+    },
+    {
+      "epoch": 0.18425750460025817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87779716413717e-05,
+      "loss": 1.1167,
+      "step": 6709
+    },
+    {
+      "epoch": 0.1842849688281014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877754549956317e-05,
+      "loss": 1.1218,
+      "step": 6710
+    },
+    {
+      "epoch": 0.18431243305594464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87771192883033e-05,
+      "loss": 1.0116,
+      "step": 6711
+    },
+    {
+      "epoch": 0.18433989728378786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877669300759546e-05,
+      "loss": 0.9589,
+      "step": 6712
+    },
+    {
+      "epoch": 0.1843673615116311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8776266657443022e-05,
+      "loss": 0.9072,
+      "step": 6713
+    },
+    {
+      "epoch": 0.18439482573947433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877584023784937e-05,
+      "loss": 1.1299,
+      "step": 6714
+    },
+    {
+      "epoch": 0.18442228996731758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8775413748817866e-05,
+      "loss": 1.0319,
+      "step": 6715
+    },
+    {
+      "epoch": 0.1844497541951608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877498719035189e-05,
+      "loss": 1.0687,
+      "step": 6716
+    },
+    {
+      "epoch": 0.18447721842300405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877456056245482e-05,
+      "loss": 1.0852,
+      "step": 6717
+    },
+    {
+      "epoch": 0.18450468265084727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8774133865130024e-05,
+      "loss": 0.9993,
+      "step": 6718
+    },
+    {
+      "epoch": 0.18453214687869052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877370709838088e-05,
+      "loss": 0.9546,
+      "step": 6719
+    },
+    {
+      "epoch": 0.18455961110653374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877328026221077e-05,
+      "loss": 0.9725,
+      "step": 6720
+    },
+    {
+      "epoch": 0.184587075334377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8772853356623066e-05,
+      "loss": 1.0305,
+      "step": 6721
+    },
+    {
+      "epoch": 0.1846145395622202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877242638162115e-05,
+      "loss": 1.0117,
+      "step": 6722
+    },
+    {
+      "epoch": 0.18464200379006343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87719993372084e-05,
+      "loss": 0.981,
+      "step": 6723
+    },
+    {
+      "epoch": 0.18466946801790668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877157222338819e-05,
+      "loss": 1.0859,
+      "step": 6724
+    },
+    {
+      "epoch": 0.1846969322457499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8771145040163904e-05,
+      "loss": 1.0289,
+      "step": 6725
+    },
+    {
+      "epoch": 0.18472439647359315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877071778753892e-05,
+      "loss": 1.1173,
+      "step": 6726
+    },
+    {
+      "epoch": 0.18475186070143637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877029046551662e-05,
+      "loss": 1.0303,
+      "step": 6727
+    },
+    {
+      "epoch": 0.18477932492927962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8769863074100387e-05,
+      "loss": 1.0196,
+      "step": 6728
+    },
+    {
+      "epoch": 0.18480678915712284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8769435613293598e-05,
+      "loss": 1.0104,
+      "step": 6729
+    },
+    {
+      "epoch": 0.1848342533849661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876900808309964e-05,
+      "loss": 0.9891,
+      "step": 6730
+    },
+    {
+      "epoch": 0.1848617176128093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876858048352189e-05,
+      "loss": 0.8989,
+      "step": 6731
+    },
+    {
+      "epoch": 0.18488918184065256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8768152814563735e-05,
+      "loss": 1.041,
+      "step": 6732
+    },
+    {
+      "epoch": 0.18491664606849578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876772507622856e-05,
+      "loss": 1.0516,
+      "step": 6733
+    },
+    {
+      "epoch": 0.18494411029633903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8767297268519745e-05,
+      "loss": 1.0013,
+      "step": 6734
+    },
+    {
+      "epoch": 0.18497157452418225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8766869391440678e-05,
+      "loss": 1.054,
+      "step": 6735
+    },
+    {
+      "epoch": 0.18499903875202547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8766441444994748e-05,
+      "loss": 1.1508,
+      "step": 6736
+    },
+    {
+      "epoch": 0.18502650297986872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8766013429185335e-05,
+      "loss": 1.099,
+      "step": 6737
+    },
+    {
+      "epoch": 0.18505396720771194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876558534401583e-05,
+      "loss": 0.8519,
+      "step": 6738
+    },
+    {
+      "epoch": 0.1850814314355552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8765157189489615e-05,
+      "loss": 1.0131,
+      "step": 6739
+    },
+    {
+      "epoch": 0.18510889566339842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8764728965610082e-05,
+      "loss": 0.9736,
+      "step": 6740
+    },
+    {
+      "epoch": 0.18513635989124166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876430067238062e-05,
+      "loss": 1.0063,
+      "step": 6741
+    },
+    {
+      "epoch": 0.18516382411908489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876387230980462e-05,
+      "loss": 1.0656,
+      "step": 6742
+    },
+    {
+      "epoch": 0.18519128834692813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876344387788546e-05,
+      "loss": 1.0085,
+      "step": 6743
+    },
+    {
+      "epoch": 0.18521875257477136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876301537662654e-05,
+      "loss": 1.1179,
+      "step": 6744
+    },
+    {
+      "epoch": 0.1852462168026146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8762586806031247e-05,
+      "loss": 1.0449,
+      "step": 6745
+    },
+    {
+      "epoch": 0.18527368103045783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8762158166102974e-05,
+      "loss": 1.049,
+      "step": 6746
+    },
+    {
+      "epoch": 0.18530114525830108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8761729456845112e-05,
+      "loss": 1.007,
+      "step": 6747
+    },
+    {
+      "epoch": 0.1853286094861443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8761300678261053e-05,
+      "loss": 1.066,
+      "step": 6748
+    },
+    {
+      "epoch": 0.18535607371398752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8760871830354187e-05,
+      "loss": 0.9599,
+      "step": 6749
+    },
+    {
+      "epoch": 0.18538353794183077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8760442913127913e-05,
+      "loss": 1.0362,
+      "step": 6750
+    },
+    {
+      "epoch": 0.185411002169674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876001392658562e-05,
+      "loss": 0.9391,
+      "step": 6751
+    },
+    {
+      "epoch": 0.18543846639751724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8759584870730702e-05,
+      "loss": 1.0126,
+      "step": 6752
+    },
+    {
+      "epoch": 0.18546593062536046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8759155745566554e-05,
+      "loss": 1.0616,
+      "step": 6753
+    },
+    {
+      "epoch": 0.1854933948532037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8758726551096574e-05,
+      "loss": 0.9698,
+      "step": 6754
+    },
+    {
+      "epoch": 0.18552085908104693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8758297287324158e-05,
+      "loss": 1.0051,
+      "step": 6755
+    },
+    {
+      "epoch": 0.18554832330889018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87578679542527e-05,
+      "loss": 0.9457,
+      "step": 6756
+    },
+    {
+      "epoch": 0.1855757875367334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87574385518856e-05,
+      "loss": 1.0607,
+      "step": 6757
+    },
+    {
+      "epoch": 0.18560325176457665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8757009080226253e-05,
+      "loss": 0.9788,
+      "step": 6758
+    },
+    {
+      "epoch": 0.18563071599241987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8756579539278056e-05,
+      "loss": 0.9081,
+      "step": 6759
+    },
+    {
+      "epoch": 0.18565818022026312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875614992904441e-05,
+      "loss": 0.9723,
+      "step": 6760
+    },
+    {
+      "epoch": 0.18568564444810634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875572024952872e-05,
+      "loss": 0.9876,
+      "step": 6761
+    },
+    {
+      "epoch": 0.18571310867594956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8755290500734375e-05,
+      "loss": 0.9246,
+      "step": 6762
+    },
+    {
+      "epoch": 0.1857405729037928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875486068266478e-05,
+      "loss": 1.0338,
+      "step": 6763
+    },
+    {
+      "epoch": 0.18576803713163603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8754430795323334e-05,
+      "loss": 1.0101,
+      "step": 6764
+    },
+    {
+      "epoch": 0.18579550135947928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875400083871344e-05,
+      "loss": 1.0052,
+      "step": 6765
+    },
+    {
+      "epoch": 0.1858229655873225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8753570812838506e-05,
+      "loss": 1.1562,
+      "step": 6766
+    },
+    {
+      "epoch": 0.18585042981516575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8753140717701924e-05,
+      "loss": 1.1071,
+      "step": 6767
+    },
+    {
+      "epoch": 0.18587789404300897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8752710553307107e-05,
+      "loss": 0.9916,
+      "step": 6768
+    },
+    {
+      "epoch": 0.18590535827085222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8752280319657446e-05,
+      "loss": 1.0398,
+      "step": 6769
+    },
+    {
+      "epoch": 0.18593282249869544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875185001675636e-05,
+      "loss": 1.1066,
+      "step": 6770
+    },
+    {
+      "epoch": 0.1859602867265387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8751419644607245e-05,
+      "loss": 1.0557,
+      "step": 6771
+    },
+    {
+      "epoch": 0.18598775095438191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8750989203213504e-05,
+      "loss": 1.2135,
+      "step": 6772
+    },
+    {
+      "epoch": 0.18601521518222516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875055869257855e-05,
+      "loss": 1.1378,
+      "step": 6773
+    },
+    {
+      "epoch": 0.18604267941006838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875012811270579e-05,
+      "loss": 1.0512,
+      "step": 6774
+    },
+    {
+      "epoch": 0.18607014363791163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8749697463598618e-05,
+      "loss": 1.0197,
+      "step": 6775
+    },
+    {
+      "epoch": 0.18609760786575485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8749266745260455e-05,
+      "loss": 1.0323,
+      "step": 6776
+    },
+    {
+      "epoch": 0.18612507209359808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8748835957694703e-05,
+      "loss": 1.0526,
+      "step": 6777
+    },
+    {
+      "epoch": 0.18615253632144133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8748405100904774e-05,
+      "loss": 0.9623,
+      "step": 6778
+    },
+    {
+      "epoch": 0.18618000054928455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.874797417489407e-05,
+      "loss": 0.9361,
+      "step": 6779
+    },
+    {
+      "epoch": 0.1862074647771278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8747543179666012e-05,
+      "loss": 0.9174,
+      "step": 6780
+    },
+    {
+      "epoch": 0.18623492900497102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8747112115224e-05,
+      "loss": 0.9219,
+      "step": 6781
+    },
+    {
+      "epoch": 0.18626239323281427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8746680981571452e-05,
+      "loss": 1.0126,
+      "step": 6782
+    },
+    {
+      "epoch": 0.1862898574606575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8746249778711772e-05,
+      "loss": 1.0403,
+      "step": 6783
+    },
+    {
+      "epoch": 0.18631732168850074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.874581850664838e-05,
+      "loss": 1.0103,
+      "step": 6784
+    },
+    {
+      "epoch": 0.18634478591634396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.874538716538468e-05,
+      "loss": 1.0367,
+      "step": 6785
+    },
+    {
+      "epoch": 0.1863722501441872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8744955754924095e-05,
+      "loss": 1.0672,
+      "step": 6786
+    },
+    {
+      "epoch": 0.18639971437203043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8744524275270026e-05,
+      "loss": 1.0239,
+      "step": 6787
+    },
+    {
+      "epoch": 0.18642717859987368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8744092726425895e-05,
+      "loss": 1.0641,
+      "step": 6788
+    },
+    {
+      "epoch": 0.1864546428277169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8743661108395116e-05,
+      "loss": 1.1691,
+      "step": 6789
+    },
+    {
+      "epoch": 0.18648210705556012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8743229421181105e-05,
+      "loss": 1.0158,
+      "step": 6790
+    },
+    {
+      "epoch": 0.18650957128340337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8742797664787278e-05,
+      "loss": 1.0262,
+      "step": 6791
+    },
+    {
+      "epoch": 0.1865370355112466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8742365839217042e-05,
+      "loss": 0.9673,
+      "step": 6792
+    },
+    {
+      "epoch": 0.18656449973908984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741933944473826e-05,
+      "loss": 1.0212,
+      "step": 6793
+    },
+    {
+      "epoch": 0.18659196396693306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741501980561043e-05,
+      "loss": 1.0729,
+      "step": 6794
+    },
+    {
+      "epoch": 0.1866194281947763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741069947482108e-05,
+      "loss": 1.0745,
+      "step": 6795
+    },
+    {
+      "epoch": 0.18664689242261953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8740637845240442e-05,
+      "loss": 0.9922,
+      "step": 6796
+    },
+    {
+      "epoch": 0.18667435665046278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8740205673839464e-05,
+      "loss": 1.0187,
+      "step": 6797
+    },
+    {
+      "epoch": 0.186701820878306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8739773433282593e-05,
+      "loss": 1.062,
+      "step": 6798
+    },
+    {
+      "epoch": 0.18672928510614925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8739341123573248e-05,
+      "loss": 0.9362,
+      "step": 6799
+    },
+    {
+      "epoch": 0.18675674933399247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8738908744714853e-05,
+      "loss": 0.8943,
+      "step": 6800
+    },
+    {
+      "epoch": 0.18678421356183572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8738476296710825e-05,
+      "loss": 1.1034,
+      "step": 6801
+    },
+    {
+      "epoch": 0.18681167778967894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873804377956459e-05,
+      "loss": 0.9964,
+      "step": 6802
+    },
+    {
+      "epoch": 0.18683914201752216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8737611193279564e-05,
+      "loss": 0.9073,
+      "step": 6803
+    },
+    {
+      "epoch": 0.1868666062453654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8737178537859176e-05,
+      "loss": 0.9436,
+      "step": 6804
+    },
+    {
+      "epoch": 0.18689407047320863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8736745813306846e-05,
+      "loss": 1.0787,
+      "step": 6805
+    },
+    {
+      "epoch": 0.18692153470105188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8736313019626002e-05,
+      "loss": 0.9732,
+      "step": 6806
+    },
+    {
+      "epoch": 0.1869489989288951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8735880156820063e-05,
+      "loss": 1.0918,
+      "step": 6807
+    },
+    {
+      "epoch": 0.18697646315673835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8735447224892454e-05,
+      "loss": 1.0777,
+      "step": 6808
+    },
+    {
+      "epoch": 0.18700392738458158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8735014223846604e-05,
+      "loss": 0.9783,
+      "step": 6809
+    },
+    {
+      "epoch": 0.18703139161242482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873458115368594e-05,
+      "loss": 0.9942,
+      "step": 6810
+    },
+    {
+      "epoch": 0.18705885584026805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8734148014413884e-05,
+      "loss": 0.9558,
+      "step": 6811
+    },
+    {
+      "epoch": 0.1870863200681113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8733714806033867e-05,
+      "loss": 1.1082,
+      "step": 6812
+    },
+    {
+      "epoch": 0.18711378429595452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8733281528549315e-05,
+      "loss": 1.0547,
+      "step": 6813
+    },
+    {
+      "epoch": 0.18714124852379777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8732848181963656e-05,
+      "loss": 1.073,
+      "step": 6814
+    },
+    {
+      "epoch": 0.187168712751641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873241476628032e-05,
+      "loss": 1.1039,
+      "step": 6815
+    },
+    {
+      "epoch": 0.18719617697948424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8731981281502734e-05,
+      "loss": 1.0964,
+      "step": 6816
+    },
+    {
+      "epoch": 0.18722364120732746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8731547727634333e-05,
+      "loss": 1.0301,
+      "step": 6817
+    },
+    {
+      "epoch": 0.18725110543517068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8731114104678543e-05,
+      "loss": 1.0849,
+      "step": 6818
+    },
+    {
+      "epoch": 0.18727856966301393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8730680412638794e-05,
+      "loss": 1.0055,
+      "step": 6819
+    },
+    {
+      "epoch": 0.18730603389085715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873024665151852e-05,
+      "loss": 1.0324,
+      "step": 6820
+    },
+    {
+      "epoch": 0.1873334981187004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8729812821321155e-05,
+      "loss": 1.0308,
+      "step": 6821
+    },
+    {
+      "epoch": 0.18736096234654362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872937892205013e-05,
+      "loss": 1.146,
+      "step": 6822
+    },
+    {
+      "epoch": 0.18738842657438687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8728944953708873e-05,
+      "loss": 0.9669,
+      "step": 6823
+    },
+    {
+      "epoch": 0.1874158908022301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8728510916300828e-05,
+      "loss": 1.017,
+      "step": 6824
+    },
+    {
+      "epoch": 0.18744335503007334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872807680982942e-05,
+      "loss": 1.0203,
+      "step": 6825
+    },
+    {
+      "epoch": 0.18747081925791656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872764263429809e-05,
+      "loss": 0.9984,
+      "step": 6826
+    },
+    {
+      "epoch": 0.1874982834857598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872720838971027e-05,
+      "loss": 1.0614,
+      "step": 6827
+    },
+    {
+      "epoch": 0.18752574771360303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872677407606939e-05,
+      "loss": 1.0288,
+      "step": 6828
+    },
+    {
+      "epoch": 0.18755321194144628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8726339693378898e-05,
+      "loss": 0.956,
+      "step": 6829
+    },
+    {
+      "epoch": 0.1875806761692895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725905241642227e-05,
+      "loss": 0.9318,
+      "step": 6830
+    },
+    {
+      "epoch": 0.18760814039713272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725470720862816e-05,
+      "loss": 1.0931,
+      "step": 6831
+    },
+    {
+      "epoch": 0.18763560462497597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725036131044098e-05,
+      "loss": 1.1438,
+      "step": 6832
+    },
+    {
+      "epoch": 0.1876630688528192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872460147218951e-05,
+      "loss": 1.0833,
+      "step": 6833
+    },
+    {
+      "epoch": 0.18769053308066244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87241667443025e-05,
+      "loss": 1.0935,
+      "step": 6834
+    },
+    {
+      "epoch": 0.18771799730850566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8723731947386507e-05,
+      "loss": 0.9186,
+      "step": 6835
+    },
+    {
+      "epoch": 0.1877454615363489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872329708144496e-05,
+      "loss": 1.0126,
+      "step": 6836
+    },
+    {
+      "epoch": 0.18777292576419213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8722862146481313e-05,
+      "loss": 0.922,
+      "step": 6837
+    },
+    {
+      "epoch": 0.18780038999203538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8722427142498998e-05,
+      "loss": 0.989,
+      "step": 6838
+    },
+    {
+      "epoch": 0.1878278542198786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8721992069501462e-05,
+      "loss": 1.0022,
+      "step": 6839
+    },
+    {
+      "epoch": 0.18785531844772185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872155692749215e-05,
+      "loss": 1.0833,
+      "step": 6840
+    },
+    {
+      "epoch": 0.18788278267556507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8721121716474497e-05,
+      "loss": 1.0491,
+      "step": 6841
+    },
+    {
+      "epoch": 0.18791024690340832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872068643645195e-05,
+      "loss": 1.045,
+      "step": 6842
+    },
+    {
+      "epoch": 0.18793771113125154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8720251087427956e-05,
+      "loss": 0.954,
+      "step": 6843
+    },
+    {
+      "epoch": 0.18796517535909477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8719815669405955e-05,
+      "loss": 1.0249,
+      "step": 6844
+    },
+    {
+      "epoch": 0.18799263958693802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87193801823894e-05,
+      "loss": 1.0696,
+      "step": 6845
+    },
+    {
+      "epoch": 0.18802010381478124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8718944626381727e-05,
+      "loss": 1.0685,
+      "step": 6846
+    },
+    {
+      "epoch": 0.18804756804262449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8718509001386388e-05,
+      "loss": 1.1104,
+      "step": 6847
+    },
+    {
+      "epoch": 0.1880750322704677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8718073307406826e-05,
+      "loss": 1.0939,
+      "step": 6848
+    },
+    {
+      "epoch": 0.18810249649831096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8717637544446496e-05,
+      "loss": 1.0978,
+      "step": 6849
+    },
+    {
+      "epoch": 0.18812996072615418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8717201712508835e-05,
+      "loss": 1.0766,
+      "step": 6850
+    },
+    {
+      "epoch": 0.18815742495399743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87167658115973e-05,
+      "loss": 1.0142,
+      "step": 6851
+    },
+    {
+      "epoch": 0.18818488918184065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871632984171534e-05,
+      "loss": 1.0477,
+      "step": 6852
+    },
+    {
+      "epoch": 0.1882123534096839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87158938028664e-05,
+      "loss": 0.9457,
+      "step": 6853
+    },
+    {
+      "epoch": 0.18823981763752712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871545769505393e-05,
+      "loss": 1.1093,
+      "step": 6854
+    },
+    {
+      "epoch": 0.18826728186537037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8715021518281385e-05,
+      "loss": 1.1859,
+      "step": 6855
+    },
+    {
+      "epoch": 0.1882947460932136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8714585272552215e-05,
+      "loss": 0.9728,
+      "step": 6856
+    },
+    {
+      "epoch": 0.1883222103210568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8714148957869867e-05,
+      "loss": 1.0176,
+      "step": 6857
+    },
+    {
+      "epoch": 0.18834967454890006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87137125742378e-05,
+      "loss": 0.9998,
+      "step": 6858
+    },
+    {
+      "epoch": 0.18837713877674328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871327612165946e-05,
+      "loss": 0.9454,
+      "step": 6859
+    },
+    {
+      "epoch": 0.18840460300458653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8712839600138308e-05,
+      "loss": 1.018,
+      "step": 6860
+    },
+    {
+      "epoch": 0.18843206723242975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8712403009677794e-05,
+      "loss": 1.1039,
+      "step": 6861
+    },
+    {
+      "epoch": 0.188459531460273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871196635028137e-05,
+      "loss": 1.0555,
+      "step": 6862
+    },
+    {
+      "epoch": 0.18848699568811622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8711529621952496e-05,
+      "loss": 1.03,
+      "step": 6863
+    },
+    {
+      "epoch": 0.18851445991595947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8711092824694624e-05,
+      "loss": 0.9931,
+      "step": 6864
+    },
+    {
+      "epoch": 0.1885419241438027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871065595851121e-05,
+      "loss": 1.0557,
+      "step": 6865
+    },
+    {
+      "epoch": 0.18856938837164594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8710219023405714e-05,
+      "loss": 0.9153,
+      "step": 6866
+    },
+    {
+      "epoch": 0.18859685259948916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8709782019381587e-05,
+      "loss": 1.0514,
+      "step": 6867
+    },
+    {
+      "epoch": 0.1886243168273324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8709344946442297e-05,
+      "loss": 1.0673,
+      "step": 6868
+    },
+    {
+      "epoch": 0.18865178105517563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708907804591293e-05,
+      "loss": 1.0747,
+      "step": 6869
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708470593832035e-05,
+      "loss": 0.8712,
+      "step": 6870
+    },
+    {
+      "epoch": 0.1887067095108621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708033314167988e-05,
+      "loss": 1.0392,
+      "step": 6871
+    },
+    {
+      "epoch": 0.18873417373870532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8707595965602605e-05,
+      "loss": 0.9491,
+      "step": 6872
+    },
+    {
+      "epoch": 0.18876163796654857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870715854813935e-05,
+      "loss": 1.0904,
+      "step": 6873
+    },
+    {
+      "epoch": 0.1887891021943918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8706721061781684e-05,
+      "loss": 0.8936,
+      "step": 6874
+    },
+    {
+      "epoch": 0.18881656642223504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8706283506533073e-05,
+      "loss": 1.0443,
+      "step": 6875
+    },
+    {
+      "epoch": 0.18884403065007827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870584588239697e-05,
+      "loss": 1.1034,
+      "step": 6876
+    },
+    {
+      "epoch": 0.18887149487792151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8705408189376842e-05,
+      "loss": 0.9688,
+      "step": 6877
+    },
+    {
+      "epoch": 0.18889895910576474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8704970427476152e-05,
+      "loss": 1.047,
+      "step": 6878
+    },
+    {
+      "epoch": 0.18892642333360798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8704532596698365e-05,
+      "loss": 1.0777,
+      "step": 6879
+    },
+    {
+      "epoch": 0.1889538875614512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8704094697046942e-05,
+      "loss": 1.1516,
+      "step": 6880
+    },
+    {
+      "epoch": 0.18898135178929445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8703656728525355e-05,
+      "loss": 1.0237,
+      "step": 6881
+    },
+    {
+      "epoch": 0.18900881601713768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8703218691137062e-05,
+      "loss": 0.9877,
+      "step": 6882
+    },
+    {
+      "epoch": 0.18903628024498093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8702780584885532e-05,
+      "loss": 0.8592,
+      "step": 6883
+    },
+    {
+      "epoch": 0.18906374447282415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870234240977423e-05,
+      "loss": 1.09,
+      "step": 6884
+    },
+    {
+      "epoch": 0.18909120870066737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8701904165806624e-05,
+      "loss": 1.0475,
+      "step": 6885
+    },
+    {
+      "epoch": 0.18911867292851062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8701465852986184e-05,
+      "loss": 0.9302,
+      "step": 6886
+    },
+    {
+      "epoch": 0.18914613715635384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8701027471316372e-05,
+      "loss": 1.0417,
+      "step": 6887
+    },
+    {
+      "epoch": 0.1891736013841971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8700589020800658e-05,
+      "loss": 1.0197,
+      "step": 6888
+    },
+    {
+      "epoch": 0.1892010656120403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8700150501442516e-05,
+      "loss": 0.9658,
+      "step": 6889
+    },
+    {
+      "epoch": 0.18922852983988356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8699711913245414e-05,
+      "loss": 1.0659,
+      "step": 6890
+    },
+    {
+      "epoch": 0.18925599406772678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869927325621282e-05,
+      "loss": 1.0857,
+      "step": 6891
+    },
+    {
+      "epoch": 0.18928345829557003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869883453034821e-05,
+      "loss": 1.0464,
+      "step": 6892
+    },
+    {
+      "epoch": 0.18931092252341325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8698395735655047e-05,
+      "loss": 0.9883,
+      "step": 6893
+    },
+    {
+      "epoch": 0.1893383867512565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869795687213681e-05,
+      "loss": 0.9922,
+      "step": 6894
+    },
+    {
+      "epoch": 0.18936585097909972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869751793979697e-05,
+      "loss": 0.9834,
+      "step": 6895
+    },
+    {
+      "epoch": 0.18939331520694297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8697078938638995e-05,
+      "loss": 1.1479,
+      "step": 6896
+    },
+    {
+      "epoch": 0.1894207794347862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8696639868666365e-05,
+      "loss": 0.9838,
+      "step": 6897
+    },
+    {
+      "epoch": 0.1894482436626294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8696200729882554e-05,
+      "loss": 0.9517,
+      "step": 6898
+    },
+    {
+      "epoch": 0.18947570789047266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869576152229103e-05,
+      "loss": 1.0035,
+      "step": 6899
+    },
+    {
+      "epoch": 0.18950317211831588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8695322245895275e-05,
+      "loss": 1.033,
+      "step": 6900
+    },
+    {
+      "epoch": 0.18953063634615913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869488290069876e-05,
+      "loss": 1.0725,
+      "step": 6901
+    },
+    {
+      "epoch": 0.18955810057400235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8694443486704965e-05,
+      "loss": 0.886,
+      "step": 6902
+    },
+    {
+      "epoch": 0.1895855648018456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8694004003917367e-05,
+      "loss": 0.9788,
+      "step": 6903
+    },
+    {
+      "epoch": 0.18961302902968882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8693564452339438e-05,
+      "loss": 1.0686,
+      "step": 6904
+    },
+    {
+      "epoch": 0.18964049325753207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869312483197466e-05,
+      "loss": 1.0423,
+      "step": 6905
+    },
+    {
+      "epoch": 0.1896679574853753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8692685142826513e-05,
+      "loss": 1.0586,
+      "step": 6906
+    },
+    {
+      "epoch": 0.18969542171321854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869224538489847e-05,
+      "loss": 1.0516,
+      "step": 6907
+    },
+    {
+      "epoch": 0.18972288594106176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8691805558194018e-05,
+      "loss": 1.0146,
+      "step": 6908
+    },
+    {
+      "epoch": 0.189750350168905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8691365662716636e-05,
+      "loss": 1.1158,
+      "step": 6909
+    },
+    {
+      "epoch": 0.18977781439674823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8690925698469798e-05,
+      "loss": 0.9566,
+      "step": 6910
+    },
+    {
+      "epoch": 0.18980527862459146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869048566545699e-05,
+      "loss": 0.8789,
+      "step": 6911
+    },
+    {
+      "epoch": 0.1898327428524347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8690045563681692e-05,
+      "loss": 0.9697,
+      "step": 6912
+    },
+    {
+      "epoch": 0.18986020708027793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868960539314739e-05,
+      "loss": 1.1193,
+      "step": 6913
+    },
+    {
+      "epoch": 0.18988767130812118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8689165153857563e-05,
+      "loss": 0.9832,
+      "step": 6914
+    },
+    {
+      "epoch": 0.1899151355359644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8688724845815695e-05,
+      "loss": 0.9889,
+      "step": 6915
+    },
+    {
+      "epoch": 0.18994259976380765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868828446902527e-05,
+      "loss": 1.0152,
+      "step": 6916
+    },
+    {
+      "epoch": 0.18997006399165087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8687844023489773e-05,
+      "loss": 1.0031,
+      "step": 6917
+    },
+    {
+      "epoch": 0.18999752821949412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868740350921269e-05,
+      "loss": 0.9692,
+      "step": 6918
+    },
+    {
+      "epoch": 0.19002499244733734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8686962926197502e-05,
+      "loss": 1.0628,
+      "step": 6919
+    },
+    {
+      "epoch": 0.1900524566751806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8686522274447702e-05,
+      "loss": 1.093,
+      "step": 6920
+    },
+    {
+      "epoch": 0.1900799209030238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868608155396677e-05,
+      "loss": 1.0516,
+      "step": 6921
+    },
+    {
+      "epoch": 0.19010738513086706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8685640764758196e-05,
+      "loss": 0.8796,
+      "step": 6922
+    },
+    {
+      "epoch": 0.19013484935871028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8685199906825473e-05,
+      "loss": 1.0099,
+      "step": 6923
+    },
+    {
+      "epoch": 0.19016231358655353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868475898017208e-05,
+      "loss": 1.057,
+      "step": 6924
+    },
+    {
+      "epoch": 0.19018977781439675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868431798480151e-05,
+      "loss": 1.0295,
+      "step": 6925
+    },
+    {
+      "epoch": 0.19021724204223997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868387692071725e-05,
+      "loss": 0.9752,
+      "step": 6926
+    },
+    {
+      "epoch": 0.19024470627008322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8683435787922798e-05,
+      "loss": 0.9595,
+      "step": 6927
+    },
+    {
+      "epoch": 0.19027217049792644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682994586421635e-05,
+      "loss": 1.0304,
+      "step": 6928
+    },
+    {
+      "epoch": 0.1902996347257697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682553316217255e-05,
+      "loss": 1.0557,
+      "step": 6929
+    },
+    {
+      "epoch": 0.1903270989536129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682111977313146e-05,
+      "loss": 1.0495,
+      "step": 6930
+    },
+    {
+      "epoch": 0.19035456318145616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868167056971281e-05,
+      "loss": 0.9547,
+      "step": 6931
+    },
+    {
+      "epoch": 0.19038202740929938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868122909341973e-05,
+      "loss": 1.0792,
+      "step": 6932
+    },
+    {
+      "epoch": 0.19040949163714263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8680787548437408e-05,
+      "loss": 1.0327,
+      "step": 6933
+    },
+    {
+      "epoch": 0.19043695586498585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8680345934769322e-05,
+      "loss": 1.1365,
+      "step": 6934
+    },
+    {
+      "epoch": 0.1904644200928291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8679904252418984e-05,
+      "loss": 0.9825,
+      "step": 6935
+    },
+    {
+      "epoch": 0.19049188432067232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8679462501389883e-05,
+      "loss": 1.0615,
+      "step": 6936
+    },
+    {
+      "epoch": 0.19051934854851557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867902068168551e-05,
+      "loss": 1.0846,
+      "step": 6937
+    },
+    {
+      "epoch": 0.1905468127763588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867857879330936e-05,
+      "loss": 0.9035,
+      "step": 6938
+    },
+    {
+      "epoch": 0.19057427700420201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8678136836264938e-05,
+      "loss": 0.9716,
+      "step": 6939
+    },
+    {
+      "epoch": 0.19060174123204526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8677694810555734e-05,
+      "loss": 1.007,
+      "step": 6940
+    },
+    {
+      "epoch": 0.19062920545988848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867725271618525e-05,
+      "loss": 0.9241,
+      "step": 6941
+    },
+    {
+      "epoch": 0.19065666968773173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8676810553156975e-05,
+      "loss": 1.0257,
+      "step": 6942
+    },
+    {
+      "epoch": 0.19068413391557495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8676368321474417e-05,
+      "loss": 0.9355,
+      "step": 6943
+    },
+    {
+      "epoch": 0.1907115981434182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867592602114107e-05,
+      "loss": 1.01,
+      "step": 6944
+    },
+    {
+      "epoch": 0.19073906237126143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867548365216044e-05,
+      "loss": 1.062,
+      "step": 6945
+    },
+    {
+      "epoch": 0.19076652659910467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867504121453602e-05,
+      "loss": 1.0043,
+      "step": 6946
+    },
+    {
+      "epoch": 0.1907939908269479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8674598708271312e-05,
+      "loss": 1.0516,
+      "step": 6947
+    },
+    {
+      "epoch": 0.19082145505479114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867415613336982e-05,
+      "loss": 1.0823,
+      "step": 6948
+    },
+    {
+      "epoch": 0.19084891928263437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867371348983505e-05,
+      "loss": 0.9039,
+      "step": 6949
+    },
+    {
+      "epoch": 0.19087638351047762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867327077767049e-05,
+      "loss": 1.0635,
+      "step": 6950
+    },
+    {
+      "epoch": 0.19090384773832084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8672827996879658e-05,
+      "loss": 1.0536,
+      "step": 6951
+    },
+    {
+      "epoch": 0.19093131196616406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867238514746605e-05,
+      "loss": 1.1114,
+      "step": 6952
+    },
+    {
+      "epoch": 0.1909587761940073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867194222943317e-05,
+      "loss": 1.0435,
+      "step": 6953
+    },
+    {
+      "epoch": 0.19098624042185053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8671499242784525e-05,
+      "loss": 1.107,
+      "step": 6954
+    },
+    {
+      "epoch": 0.19101370464969378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867105618752362e-05,
+      "loss": 0.9392,
+      "step": 6955
+    },
+    {
+      "epoch": 0.191041168877537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8670613063653956e-05,
+      "loss": 0.9524,
+      "step": 6956
+    },
+    {
+      "epoch": 0.19106863310538025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8670169871179046e-05,
+      "loss": 1.0623,
+      "step": 6957
+    },
+    {
+      "epoch": 0.19109609733322347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8669726610102392e-05,
+      "loss": 1.0275,
+      "step": 6958
+    },
+    {
+      "epoch": 0.19112356156106672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8669283280427504e-05,
+      "loss": 0.982,
+      "step": 6959
+    },
+    {
+      "epoch": 0.19115102578890994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668839882157886e-05,
+      "loss": 1.0088,
+      "step": 6960
+    },
+    {
+      "epoch": 0.1911784900167532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668396415297054e-05,
+      "loss": 0.9222,
+      "step": 6961
+    },
+    {
+      "epoch": 0.1912059542445964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8667952879848505e-05,
+      "loss": 1.0793,
+      "step": 6962
+    },
+    {
+      "epoch": 0.19123341847243966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866750927581576e-05,
+      "loss": 0.935,
+      "step": 6963
+    },
+    {
+      "epoch": 0.19126088270028288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8667065603202323e-05,
+      "loss": 1.0147,
+      "step": 6964
+    },
+    {
+      "epoch": 0.19128834692812613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8666621862011704e-05,
+      "loss": 1.0399,
+      "step": 6965
+    },
+    {
+      "epoch": 0.19131581115596935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866617805224742e-05,
+      "loss": 0.9934,
+      "step": 6966
+    },
+    {
+      "epoch": 0.19134327538381257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8665734173912974e-05,
+      "loss": 1.0185,
+      "step": 6967
+    },
+    {
+      "epoch": 0.19137073961165582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8665290227011882e-05,
+      "loss": 1.0031,
+      "step": 6968
+    },
+    {
+      "epoch": 0.19139820383949904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866484621154766e-05,
+      "loss": 1.0265,
+      "step": 6969
+    },
+    {
+      "epoch": 0.1914256680673423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8664402127523818e-05,
+      "loss": 0.9915,
+      "step": 6970
+    },
+    {
+      "epoch": 0.1914531322951855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866395797494387e-05,
+      "loss": 1.0104,
+      "step": 6971
+    },
+    {
+      "epoch": 0.19148059652302876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8663513753811327e-05,
+      "loss": 1.0588,
+      "step": 6972
+    },
+    {
+      "epoch": 0.19150806075087198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866306946412971e-05,
+      "loss": 1.0487,
+      "step": 6973
+    },
+    {
+      "epoch": 0.19153552497871523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8662625105902532e-05,
+      "loss": 0.9662,
+      "step": 6974
+    },
+    {
+      "epoch": 0.19156298920655845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866218067913331e-05,
+      "loss": 1.0404,
+      "step": 6975
+    },
+    {
+      "epoch": 0.1915904534344017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8661736183825556e-05,
+      "loss": 1.0099,
+      "step": 6976
+    },
+    {
+      "epoch": 0.19161791766224492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8661291619982793e-05,
+      "loss": 0.9606,
+      "step": 6977
+    },
+    {
+      "epoch": 0.19164538189008817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8660846987608533e-05,
+      "loss": 0.958,
+      "step": 6978
+    },
+    {
+      "epoch": 0.1916728461179314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8660402286706298e-05,
+      "loss": 0.9401,
+      "step": 6979
+    },
+    {
+      "epoch": 0.19170031034577462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8659957517279602e-05,
+      "loss": 0.9757,
+      "step": 6980
+    },
+    {
+      "epoch": 0.19172777457361787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8659512679331973e-05,
+      "loss": 0.9194,
+      "step": 6981
+    },
+    {
+      "epoch": 0.1917552388014611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8659067772866924e-05,
+      "loss": 1.0176,
+      "step": 6982
+    },
+    {
+      "epoch": 0.19178270302930434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8658622797887975e-05,
+      "loss": 0.9214,
+      "step": 6983
+    },
+    {
+      "epoch": 0.19181016725714756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8658177754398647e-05,
+      "loss": 1.0086,
+      "step": 6984
+    },
+    {
+      "epoch": 0.1918376314849908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8657732642402466e-05,
+      "loss": 0.9962,
+      "step": 6985
+    },
+    {
+      "epoch": 0.19186509571283403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865728746190295e-05,
+      "loss": 0.9608,
+      "step": 6986
+    },
+    {
+      "epoch": 0.19189255994067728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865684221290362e-05,
+      "loss": 0.9534,
+      "step": 6987
+    },
+    {
+      "epoch": 0.1919200241685205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8656396895408004e-05,
+      "loss": 0.977,
+      "step": 6988
+    },
+    {
+      "epoch": 0.19194748839636375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865595150941962e-05,
+      "loss": 0.9968,
+      "step": 6989
+    },
+    {
+      "epoch": 0.19197495262420697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8655506054941997e-05,
+      "loss": 1.0234,
+      "step": 6990
+    },
+    {
+      "epoch": 0.19200241685205022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8655060531978657e-05,
+      "loss": 1.0385,
+      "step": 6991
+    },
+    {
+      "epoch": 0.19202988107989344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8654614940533124e-05,
+      "loss": 1.0238,
+      "step": 6992
+    },
+    {
+      "epoch": 0.19205734530773666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8654169280608924e-05,
+      "loss": 0.9768,
+      "step": 6993
+    },
+    {
+      "epoch": 0.1920848095355799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8653723552209588e-05,
+      "loss": 0.9353,
+      "step": 6994
+    },
+    {
+      "epoch": 0.19211227376342313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8653277755338636e-05,
+      "loss": 1.1482,
+      "step": 6995
+    },
+    {
+      "epoch": 0.19213973799126638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86528318899996e-05,
+      "loss": 1.0872,
+      "step": 6996
+    },
+    {
+      "epoch": 0.1921672022191096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8652385956196006e-05,
+      "loss": 1.0484,
+      "step": 6997
+    },
+    {
+      "epoch": 0.19219466644695285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8651939953931382e-05,
+      "loss": 1.062,
+      "step": 6998
+    },
+    {
+      "epoch": 0.19222213067479607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865149388320926e-05,
+      "loss": 1.0775,
+      "step": 6999
+    },
+    {
+      "epoch": 0.19224959490263932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8651047744033165e-05,
+      "loss": 0.9907,
+      "step": 7000
+    },
+    {
+      "epoch": 0.19227705913048254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8650601536406626e-05,
+      "loss": 1.0695,
+      "step": 7001
+    },
+    {
+      "epoch": 0.1923045233583258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865015526033318e-05,
+      "loss": 0.9568,
+      "step": 7002
+    },
+    {
+      "epoch": 0.192331987586169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8649708915816357e-05,
+      "loss": 1.0533,
+      "step": 7003
+    },
+    {
+      "epoch": 0.19235945181401226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8649262502859683e-05,
+      "loss": 0.9817,
+      "step": 7004
+    },
+    {
+      "epoch": 0.19238691604185548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8648816021466695e-05,
+      "loss": 1.0365,
+      "step": 7005
+    },
+    {
+      "epoch": 0.1924143802696987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8648369471640926e-05,
+      "loss": 1.0396,
+      "step": 7006
+    },
+    {
+      "epoch": 0.19244184449754195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8647922853385906e-05,
+      "loss": 1.0766,
+      "step": 7007
+    },
+    {
+      "epoch": 0.19246930872538517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8647476166705173e-05,
+      "loss": 1.0836,
+      "step": 7008
+    },
+    {
+      "epoch": 0.19249677295322842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8647029411602256e-05,
+      "loss": 0.9838,
+      "step": 7009
+    },
+    {
+      "epoch": 0.19252423718107164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8646582588080694e-05,
+      "loss": 1.0504,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1925517014089149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864613569614402e-05,
+      "loss": 1.0022,
+      "step": 7011
+    },
+    {
+      "epoch": 0.19257916563675812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864568873579577e-05,
+      "loss": 0.9788,
+      "step": 7012
+    },
+    {
+      "epoch": 0.19260662986460136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8645241707039483e-05,
+      "loss": 0.9415,
+      "step": 7013
+    },
+    {
+      "epoch": 0.19263409409244459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8644794609878696e-05,
+      "loss": 1.0773,
+      "step": 7014
+    },
+    {
+      "epoch": 0.19266155832028783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8644347444316945e-05,
+      "loss": 1.0494,
+      "step": 7015
+    },
+    {
+      "epoch": 0.19268902254813106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864390021035777e-05,
+      "loss": 1.1196,
+      "step": 7016
+    },
+    {
+      "epoch": 0.1927164867759743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8643452908004704e-05,
+      "loss": 0.9097,
+      "step": 7017
+    },
+    {
+      "epoch": 0.19274395100381753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864300553726129e-05,
+      "loss": 0.9854,
+      "step": 7018
+    },
+    {
+      "epoch": 0.19277141523166078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8642558098131073e-05,
+      "loss": 1.052,
+      "step": 7019
+    },
+    {
+      "epoch": 0.192798879459504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8642110590617588e-05,
+      "loss": 1.0135,
+      "step": 7020
+    },
+    {
+      "epoch": 0.19282634368734722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8641663014724374e-05,
+      "loss": 1.0321,
+      "step": 7021
+    },
+    {
+      "epoch": 0.19285380791519047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8641215370454974e-05,
+      "loss": 1.0704,
+      "step": 7022
+    },
+    {
+      "epoch": 0.1928812721430337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8640767657812935e-05,
+      "loss": 1.0408,
+      "step": 7023
+    },
+    {
+      "epoch": 0.19290873637087694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8640319876801792e-05,
+      "loss": 1.182,
+      "step": 7024
+    },
+    {
+      "epoch": 0.19293620059872016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863987202742509e-05,
+      "loss": 0.9919,
+      "step": 7025
+    },
+    {
+      "epoch": 0.1929636648265634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8639424109686378e-05,
+      "loss": 0.9922,
+      "step": 7026
+    },
+    {
+      "epoch": 0.19299112905440663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8638976123589195e-05,
+      "loss": 0.9862,
+      "step": 7027
+    },
+    {
+      "epoch": 0.19301859328224988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863852806913708e-05,
+      "loss": 1.0284,
+      "step": 7028
+    },
+    {
+      "epoch": 0.1930460575100931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863807994633359e-05,
+      "loss": 1.0616,
+      "step": 7029
+    },
+    {
+      "epoch": 0.19307352173793635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8637631755182267e-05,
+      "loss": 0.8828,
+      "step": 7030
+    },
+    {
+      "epoch": 0.19310098596577957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8637183495686654e-05,
+      "loss": 1.0606,
+      "step": 7031
+    },
+    {
+      "epoch": 0.19312845019362282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8636735167850298e-05,
+      "loss": 1.0673,
+      "step": 7032
+    },
+    {
+      "epoch": 0.19315591442146604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8636286771676747e-05,
+      "loss": 0.9857,
+      "step": 7033
+    },
+    {
+      "epoch": 0.19318337864930926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8635838307169552e-05,
+      "loss": 1.0104,
+      "step": 7034
+    },
+    {
+      "epoch": 0.1932108428771525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863538977433226e-05,
+      "loss": 1.0286,
+      "step": 7035
+    },
+    {
+      "epoch": 0.19323830710499573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8634941173168418e-05,
+      "loss": 0.9717,
+      "step": 7036
+    },
+    {
+      "epoch": 0.19326577133283898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8634492503681577e-05,
+      "loss": 0.925,
+      "step": 7037
+    },
+    {
+      "epoch": 0.1932932355606822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8634043765875284e-05,
+      "loss": 1.0161,
+      "step": 7038
+    },
+    {
+      "epoch": 0.19332069978852545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8633594959753094e-05,
+      "loss": 1.0781,
+      "step": 7039
+    },
+    {
+      "epoch": 0.19334816401636867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8633146085318558e-05,
+      "loss": 0.9776,
+      "step": 7040
+    },
+    {
+      "epoch": 0.19337562824421192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8632697142575226e-05,
+      "loss": 1.0964,
+      "step": 7041
+    },
+    {
+      "epoch": 0.19340309247205514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8632248131526648e-05,
+      "loss": 0.9751,
+      "step": 7042
+    },
+    {
+      "epoch": 0.1934305566998984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8631799052176376e-05,
+      "loss": 1.0081,
+      "step": 7043
+    },
+    {
+      "epoch": 0.19345802092774161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863134990452797e-05,
+      "loss": 1.0115,
+      "step": 7044
+    },
+    {
+      "epoch": 0.19348548515558486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8630900688584984e-05,
+      "loss": 0.9762,
+      "step": 7045
+    },
+    {
+      "epoch": 0.19351294938342808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863045140435096e-05,
+      "loss": 1.0637,
+      "step": 7046
+    },
+    {
+      "epoch": 0.1935404136112713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8630002051829467e-05,
+      "loss": 1.0997,
+      "step": 7047
+    },
+    {
+      "epoch": 0.19356787783911455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8629552631024052e-05,
+      "loss": 1.0107,
+      "step": 7048
+    },
+    {
+      "epoch": 0.19359534206695778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862910314193828e-05,
+      "loss": 0.9904,
+      "step": 7049
+    },
+    {
+      "epoch": 0.19362280629480103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8628653584575694e-05,
+      "loss": 1.0888,
+      "step": 7050
+    },
+    {
+      "epoch": 0.19365027052264425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8628203958939864e-05,
+      "loss": 1.0188,
+      "step": 7051
+    },
+    {
+      "epoch": 0.1936777347504875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8627754265034338e-05,
+      "loss": 1.0443,
+      "step": 7052
+    },
+    {
+      "epoch": 0.19370519897833072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862730450286268e-05,
+      "loss": 1.0637,
+      "step": 7053
+    },
+    {
+      "epoch": 0.19373266320617397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8626854672428446e-05,
+      "loss": 1.1153,
+      "step": 7054
+    },
+    {
+      "epoch": 0.1937601274340172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8626404773735194e-05,
+      "loss": 0.9879,
+      "step": 7055
+    },
+    {
+      "epoch": 0.19378759166186044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8625954806786487e-05,
+      "loss": 1.0801,
+      "step": 7056
+    },
+    {
+      "epoch": 0.19381505588970366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8625504771585887e-05,
+      "loss": 0.9092,
+      "step": 7057
+    },
+    {
+      "epoch": 0.1938425201175469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862505466813695e-05,
+      "loss": 1.0864,
+      "step": 7058
+    },
+    {
+      "epoch": 0.19386998434539013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862460449644324e-05,
+      "loss": 0.9444,
+      "step": 7059
+    },
+    {
+      "epoch": 0.19389744857323335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862415425650832e-05,
+      "loss": 1.0684,
+      "step": 7060
+    },
+    {
+      "epoch": 0.1939249128010766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8623703948335747e-05,
+      "loss": 1.0948,
+      "step": 7061
+    },
+    {
+      "epoch": 0.19395237702891982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862325357192909e-05,
+      "loss": 1.0376,
+      "step": 7062
+    },
+    {
+      "epoch": 0.19397984125676307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8622803127291914e-05,
+      "loss": 1.0003,
+      "step": 7063
+    },
+    {
+      "epoch": 0.1940073054846063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8622352614427775e-05,
+      "loss": 1.0115,
+      "step": 7064
+    },
+    {
+      "epoch": 0.19403476971244954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862190203334024e-05,
+      "loss": 0.9079,
+      "step": 7065
+    },
+    {
+      "epoch": 0.19406223394029276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862145138403288e-05,
+      "loss": 1.082,
+      "step": 7066
+    },
+    {
+      "epoch": 0.194089698168136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8621000666509257e-05,
+      "loss": 1.0399,
+      "step": 7067
+    },
+    {
+      "epoch": 0.19411716239597923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8620549880772936e-05,
+      "loss": 0.906,
+      "step": 7068
+    },
+    {
+      "epoch": 0.19414462662382248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8620099026827487e-05,
+      "loss": 1.0939,
+      "step": 7069
+    },
+    {
+      "epoch": 0.1941720908516657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8619648104676472e-05,
+      "loss": 1.2291,
+      "step": 7070
+    },
+    {
+      "epoch": 0.19419955507950895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8619197114323466e-05,
+      "loss": 1.0162,
+      "step": 7071
+    },
+    {
+      "epoch": 0.19422701930735217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861874605577203e-05,
+      "loss": 1.0946,
+      "step": 7072
+    },
+    {
+      "epoch": 0.19425448353519542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861829492902574e-05,
+      "loss": 0.9966,
+      "step": 7073
+    },
+    {
+      "epoch": 0.19428194776303864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8617843734088162e-05,
+      "loss": 1.1378,
+      "step": 7074
+    },
+    {
+      "epoch": 0.19430941199088186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8617392470962865e-05,
+      "loss": 1.0478,
+      "step": 7075
+    },
+    {
+      "epoch": 0.1943368762187251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8616941139653422e-05,
+      "loss": 1.1025,
+      "step": 7076
+    },
+    {
+      "epoch": 0.19436434044656833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86164897401634e-05,
+      "loss": 1.1212,
+      "step": 7077
+    },
+    {
+      "epoch": 0.19439180467441158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8616038272496374e-05,
+      "loss": 1.0499,
+      "step": 7078
+    },
+    {
+      "epoch": 0.1944192689022548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8615586736655918e-05,
+      "loss": 1.094,
+      "step": 7079
+    },
+    {
+      "epoch": 0.19444673313009805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86151351326456e-05,
+      "loss": 1.0383,
+      "step": 7080
+    },
+    {
+      "epoch": 0.19447419735794128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8614683460468997e-05,
+      "loss": 0.9147,
+      "step": 7081
+    },
+    {
+      "epoch": 0.19450166158578452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8614231720129682e-05,
+      "loss": 0.9615,
+      "step": 7082
+    },
+    {
+      "epoch": 0.19452912581362775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8613779911631228e-05,
+      "loss": 1.0831,
+      "step": 7083
+    },
+    {
+      "epoch": 0.194556590041471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861332803497721e-05,
+      "loss": 1.0855,
+      "step": 7084
+    },
+    {
+      "epoch": 0.19458405426931422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8612876090171204e-05,
+      "loss": 0.9845,
+      "step": 7085
+    },
+    {
+      "epoch": 0.19461151849715747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8612424077216788e-05,
+      "loss": 1.0944,
+      "step": 7086
+    },
+    {
+      "epoch": 0.1946389827250007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8611971996117534e-05,
+      "loss": 0.9972,
+      "step": 7087
+    },
+    {
+      "epoch": 0.1946664469528439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8611519846877024e-05,
+      "loss": 1.0503,
+      "step": 7088
+    },
+    {
+      "epoch": 0.19469391118068716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8611067629498832e-05,
+      "loss": 1.0819,
+      "step": 7089
+    },
+    {
+      "epoch": 0.19472137540853038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8610615343986536e-05,
+      "loss": 1.0734,
+      "step": 7090
+    },
+    {
+      "epoch": 0.19474883963637363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8610162990343718e-05,
+      "loss": 1.052,
+      "step": 7091
+    },
+    {
+      "epoch": 0.19477630386421685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8609710568573956e-05,
+      "loss": 0.9866,
+      "step": 7092
+    },
+    {
+      "epoch": 0.1948037680920601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8609258078680826e-05,
+      "loss": 1.0983,
+      "step": 7093
+    },
+    {
+      "epoch": 0.19483123231990332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860880552066791e-05,
+      "loss": 1.1075,
+      "step": 7094
+    },
+    {
+      "epoch": 0.19485869654774657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8608352894538795e-05,
+      "loss": 1.1226,
+      "step": 7095
+    },
+    {
+      "epoch": 0.1948861607755898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8607900200297053e-05,
+      "loss": 0.9262,
+      "step": 7096
+    },
+    {
+      "epoch": 0.19491362500343304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8607447437946272e-05,
+      "loss": 0.9393,
+      "step": 7097
+    },
+    {
+      "epoch": 0.19494108923127626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606994607490034e-05,
+      "loss": 1.0238,
+      "step": 7098
+    },
+    {
+      "epoch": 0.1949685534591195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860654170893192e-05,
+      "loss": 1.1241,
+      "step": 7099
+    },
+    {
+      "epoch": 0.19499601768696273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606088742275516e-05,
+      "loss": 0.9062,
+      "step": 7100
+    },
+    {
+      "epoch": 0.19502348191480595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86056357075244e-05,
+      "loss": 1.0467,
+      "step": 7101
+    },
+    {
+      "epoch": 0.1950509461426492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860518260468216e-05,
+      "loss": 1.0825,
+      "step": 7102
+    },
+    {
+      "epoch": 0.19507841037049242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8604729433752387e-05,
+      "loss": 1.0091,
+      "step": 7103
+    },
+    {
+      "epoch": 0.19510587459833567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8604276194738657e-05,
+      "loss": 1.104,
+      "step": 7104
+    },
+    {
+      "epoch": 0.1951333388261789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8603822887644565e-05,
+      "loss": 0.9418,
+      "step": 7105
+    },
+    {
+      "epoch": 0.19516080305402214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860336951247369e-05,
+      "loss": 0.9715,
+      "step": 7106
+    },
+    {
+      "epoch": 0.19518826728186536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8602916069229624e-05,
+      "loss": 1.0018,
+      "step": 7107
+    },
+    {
+      "epoch": 0.1952157315097086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8602462557915953e-05,
+      "loss": 0.9946,
+      "step": 7108
+    },
+    {
+      "epoch": 0.19524319573755183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8602008978536267e-05,
+      "loss": 0.918,
+      "step": 7109
+    },
+    {
+      "epoch": 0.19527065996539508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8601555331094156e-05,
+      "loss": 0.9966,
+      "step": 7110
+    },
+    {
+      "epoch": 0.1952981241932383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8601101615593205e-05,
+      "loss": 1.1054,
+      "step": 7111
+    },
+    {
+      "epoch": 0.19532558842108155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8600647832037003e-05,
+      "loss": 1.0193,
+      "step": 7112
+    },
+    {
+      "epoch": 0.19535305264892477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860019398042915e-05,
+      "loss": 1.0473,
+      "step": 7113
+    },
+    {
+      "epoch": 0.19538051687676802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8599740060773225e-05,
+      "loss": 1.0283,
+      "step": 7114
+    },
+    {
+      "epoch": 0.19540798110461124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859928607307283e-05,
+      "loss": 0.9789,
+      "step": 7115
+    },
+    {
+      "epoch": 0.19543544533245447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859883201733155e-05,
+      "loss": 1.0704,
+      "step": 7116
+    },
+    {
+      "epoch": 0.19546290956029772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859837789355298e-05,
+      "loss": 1.0307,
+      "step": 7117
+    },
+    {
+      "epoch": 0.19549037378814094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8597923701740718e-05,
+      "loss": 0.9991,
+      "step": 7118
+    },
+    {
+      "epoch": 0.19551783801598419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8597469441898347e-05,
+      "loss": 1.1685,
+      "step": 7119
+    },
+    {
+      "epoch": 0.1955453022438274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8597015114029472e-05,
+      "loss": 1.2692,
+      "step": 7120
+    },
+    {
+      "epoch": 0.19557276647167066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859656071813768e-05,
+      "loss": 1.0455,
+      "step": 7121
+    },
+    {
+      "epoch": 0.19560023069951388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8596106254226568e-05,
+      "loss": 1.0016,
+      "step": 7122
+    },
+    {
+      "epoch": 0.19562769492735713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8595651722299738e-05,
+      "loss": 1.027,
+      "step": 7123
+    },
+    {
+      "epoch": 0.19565515915520035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859519712236078e-05,
+      "loss": 1.0292,
+      "step": 7124
+    },
+    {
+      "epoch": 0.1956826233830436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8594742454413293e-05,
+      "loss": 1.0097,
+      "step": 7125
+    },
+    {
+      "epoch": 0.19571008761088682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8594287718460875e-05,
+      "loss": 0.9785,
+      "step": 7126
+    },
+    {
+      "epoch": 0.19573755183873007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8593832914507123e-05,
+      "loss": 0.9833,
+      "step": 7127
+    },
+    {
+      "epoch": 0.1957650160665733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8593378042555637e-05,
+      "loss": 1.0706,
+      "step": 7128
+    },
+    {
+      "epoch": 0.1957924802944165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8592923102610014e-05,
+      "loss": 1.0651,
+      "step": 7129
+    },
+    {
+      "epoch": 0.19581994452225976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8592468094673856e-05,
+      "loss": 0.936,
+      "step": 7130
+    },
+    {
+      "epoch": 0.19584740875010298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8592013018750758e-05,
+      "loss": 0.9316,
+      "step": 7131
+    },
+    {
+      "epoch": 0.19587487297794623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8591557874844328e-05,
+      "loss": 0.9731,
+      "step": 7132
+    },
+    {
+      "epoch": 0.19590233720578945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8591102662958164e-05,
+      "loss": 0.8645,
+      "step": 7133
+    },
+    {
+      "epoch": 0.1959298014336327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8590647383095868e-05,
+      "loss": 0.977,
+      "step": 7134
+    },
+    {
+      "epoch": 0.19595726566147592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859019203526104e-05,
+      "loss": 0.951,
+      "step": 7135
+    },
+    {
+      "epoch": 0.19598472988931917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858973661945729e-05,
+      "loss": 1.0691,
+      "step": 7136
+    },
+    {
+      "epoch": 0.1960121941171624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8589281135688213e-05,
+      "loss": 1.0658,
+      "step": 7137
+    },
+    {
+      "epoch": 0.19603965834500564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8588825583957418e-05,
+      "loss": 0.9895,
+      "step": 7138
+    },
+    {
+      "epoch": 0.19606712257284886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8588369964268506e-05,
+      "loss": 1.0385,
+      "step": 7139
+    },
+    {
+      "epoch": 0.1960945868006921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8587914276625084e-05,
+      "loss": 1.0067,
+      "step": 7140
+    },
+    {
+      "epoch": 0.19612205102853533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858745852103076e-05,
+      "loss": 1.048,
+      "step": 7141
+    },
+    {
+      "epoch": 0.19614951525637855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8587002697489137e-05,
+      "loss": 1.0847,
+      "step": 7142
+    },
+    {
+      "epoch": 0.1961769794842218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858654680600382e-05,
+      "loss": 1.0593,
+      "step": 7143
+    },
+    {
+      "epoch": 0.19620444371206502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8586090846578423e-05,
+      "loss": 0.9987,
+      "step": 7144
+    },
+    {
+      "epoch": 0.19623190793990827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8585634819216545e-05,
+      "loss": 1.0018,
+      "step": 7145
+    },
+    {
+      "epoch": 0.1962593721677515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85851787239218e-05,
+      "loss": 0.9903,
+      "step": 7146
+    },
+    {
+      "epoch": 0.19628683639559474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8584722560697798e-05,
+      "loss": 0.964,
+      "step": 7147
+    },
+    {
+      "epoch": 0.19631430062343797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858426632954814e-05,
+      "loss": 1.0189,
+      "step": 7148
+    },
+    {
+      "epoch": 0.19634176485128121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8583810030476446e-05,
+      "loss": 0.9825,
+      "step": 7149
+    },
+    {
+      "epoch": 0.19636922907912444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858335366348632e-05,
+      "loss": 0.9687,
+      "step": 7150
+    },
+    {
+      "epoch": 0.19639669330696768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8582897228581375e-05,
+      "loss": 0.9846,
+      "step": 7151
+    },
+    {
+      "epoch": 0.1964241575348109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8582440725765222e-05,
+      "loss": 1.0146,
+      "step": 7152
+    },
+    {
+      "epoch": 0.19645162176265415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8581984155041475e-05,
+      "loss": 0.8933,
+      "step": 7153
+    },
+    {
+      "epoch": 0.19647908599049738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8581527516413747e-05,
+      "loss": 1.1195,
+      "step": 7154
+    },
+    {
+      "epoch": 0.1965065502183406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8581070809885645e-05,
+      "loss": 1.0782,
+      "step": 7155
+    },
+    {
+      "epoch": 0.19653401444618385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8580614035460788e-05,
+      "loss": 0.97,
+      "step": 7156
+    },
+    {
+      "epoch": 0.19656147867402707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8580157193142788e-05,
+      "loss": 1.0363,
+      "step": 7157
+    },
+    {
+      "epoch": 0.19658894290187032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8579700282935263e-05,
+      "loss": 0.9667,
+      "step": 7158
+    },
+    {
+      "epoch": 0.19661640712971354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8579243304841826e-05,
+      "loss": 0.8981,
+      "step": 7159
+    },
+    {
+      "epoch": 0.1966438713575568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8578786258866094e-05,
+      "loss": 1.0237,
+      "step": 7160
+    },
+    {
+      "epoch": 0.1966713355854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857832914501168e-05,
+      "loss": 0.912,
+      "step": 7161
+    },
+    {
+      "epoch": 0.19669879981324326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85778719632822e-05,
+      "loss": 0.9923,
+      "step": 7162
+    },
+    {
+      "epoch": 0.19672626404108648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8577414713681275e-05,
+      "loss": 1.0764,
+      "step": 7163
+    },
+    {
+      "epoch": 0.19675372826892973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576957396212523e-05,
+      "loss": 0.9543,
+      "step": 7164
+    },
+    {
+      "epoch": 0.19678119249677295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576500010879566e-05,
+      "loss": 1.0357,
+      "step": 7165
+    },
+    {
+      "epoch": 0.1968086567246162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576042557686013e-05,
+      "loss": 1.018,
+      "step": 7166
+    },
+    {
+      "epoch": 0.19683612095245942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8575585036635494e-05,
+      "loss": 1.1467,
+      "step": 7167
+    },
+    {
+      "epoch": 0.19686358518030267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8575127447731623e-05,
+      "loss": 1.0124,
+      "step": 7168
+    },
+    {
+      "epoch": 0.1968910494081459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857466979097802e-05,
+      "loss": 0.9635,
+      "step": 7169
+    },
+    {
+      "epoch": 0.1969185136359891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857421206637831e-05,
+      "loss": 0.986,
+      "step": 7170
+    },
+    {
+      "epoch": 0.19694597786383236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8573754273936112e-05,
+      "loss": 0.9685,
+      "step": 7171
+    },
+    {
+      "epoch": 0.19697344209167558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8573296413655048e-05,
+      "loss": 1.0161,
+      "step": 7172
+    },
+    {
+      "epoch": 0.19700090631951883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8572838485538743e-05,
+      "loss": 1.1317,
+      "step": 7173
+    },
+    {
+      "epoch": 0.19702837054736205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8572380489590817e-05,
+      "loss": 1.1864,
+      "step": 7174
+    },
+    {
+      "epoch": 0.1970558347752053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85719224258149e-05,
+      "loss": 0.9757,
+      "step": 7175
+    },
+    {
+      "epoch": 0.19708329900304852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857146429421461e-05,
+      "loss": 1.0585,
+      "step": 7176
+    },
+    {
+      "epoch": 0.19711076323089177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8571006094793573e-05,
+      "loss": 1.1302,
+      "step": 7177
+    },
+    {
+      "epoch": 0.197138227458735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8570547827555418e-05,
+      "loss": 1.0345,
+      "step": 7178
+    },
+    {
+      "epoch": 0.19716569168657824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8570089492503767e-05,
+      "loss": 0.9176,
+      "step": 7179
+    },
+    {
+      "epoch": 0.19719315591442146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856963108964225e-05,
+      "loss": 0.97,
+      "step": 7180
+    },
+    {
+      "epoch": 0.1972206201422647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8569172618974488e-05,
+      "loss": 1.0822,
+      "step": 7181
+    },
+    {
+      "epoch": 0.19724808437010793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8568714080504118e-05,
+      "loss": 1.038,
+      "step": 7182
+    },
+    {
+      "epoch": 0.19727554859795116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8568255474234763e-05,
+      "loss": 0.9943,
+      "step": 7183
+    },
+    {
+      "epoch": 0.1973030128257944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8567796800170048e-05,
+      "loss": 0.97,
+      "step": 7184
+    },
+    {
+      "epoch": 0.19733047705363763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8567338058313605e-05,
+      "loss": 1.047,
+      "step": 7185
+    },
+    {
+      "epoch": 0.19735794128148088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856687924866907e-05,
+      "loss": 1.149,
+      "step": 7186
+    },
+    {
+      "epoch": 0.1973854055093241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8566420371240064e-05,
+      "loss": 0.9843,
+      "step": 7187
+    },
+    {
+      "epoch": 0.19741286973716735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856596142603022e-05,
+      "loss": 0.9565,
+      "step": 7188
+    },
+    {
+      "epoch": 0.19744033396501057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8565502413043175e-05,
+      "loss": 1.0382,
+      "step": 7189
+    },
+    {
+      "epoch": 0.19746779819285382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856504333228256e-05,
+      "loss": 1.0738,
+      "step": 7190
+    },
+    {
+      "epoch": 0.19749526242069704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8564584183752e-05,
+      "loss": 1.0537,
+      "step": 7191
+    },
+    {
+      "epoch": 0.1975227266485403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856412496745513e-05,
+      "loss": 1.139,
+      "step": 7192
+    },
+    {
+      "epoch": 0.1975501908763835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8563665683395588e-05,
+      "loss": 0.9394,
+      "step": 7193
+    },
+    {
+      "epoch": 0.19757765510422676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8563206331577005e-05,
+      "loss": 0.9872,
+      "step": 7194
+    },
+    {
+      "epoch": 0.19760511933206998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856274691200302e-05,
+      "loss": 1.0167,
+      "step": 7195
+    },
+    {
+      "epoch": 0.1976325835599132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8562287424677262e-05,
+      "loss": 1.0155,
+      "step": 7196
+    },
+    {
+      "epoch": 0.19766004778775645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856182786960337e-05,
+      "loss": 0.9782,
+      "step": 7197
+    },
+    {
+      "epoch": 0.19768751201559967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8561368246784978e-05,
+      "loss": 0.9712,
+      "step": 7198
+    },
+    {
+      "epoch": 0.19771497624344292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8560908556225726e-05,
+      "loss": 0.8938,
+      "step": 7199
+    },
+    {
+      "epoch": 0.19774244047128614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856044879792925e-05,
+      "loss": 1.1297,
+      "step": 7200
+    },
+    {
+      "epoch": 0.1977699046991294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8559988971899182e-05,
+      "loss": 0.9465,
+      "step": 7201
+    },
+    {
+      "epoch": 0.1977973689269726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8559529078139166e-05,
+      "loss": 0.9618,
+      "step": 7202
+    },
+    {
+      "epoch": 0.19782483315481586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8559069116652844e-05,
+      "loss": 1.0604,
+      "step": 7203
+    },
+    {
+      "epoch": 0.19785229738265908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855860908744385e-05,
+      "loss": 0.951,
+      "step": 7204
+    },
+    {
+      "epoch": 0.19787976161050233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8558148990515828e-05,
+      "loss": 0.9994,
+      "step": 7205
+    },
+    {
+      "epoch": 0.19790722583834555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855768882587241e-05,
+      "loss": 0.9638,
+      "step": 7206
+    },
+    {
+      "epoch": 0.1979346900661888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8557228593517252e-05,
+      "loss": 1.0463,
+      "step": 7207
+    },
+    {
+      "epoch": 0.19796215429403202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8556768293453983e-05,
+      "loss": 0.945,
+      "step": 7208
+    },
+    {
+      "epoch": 0.19798961852187527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8556307925686244e-05,
+      "loss": 0.9866,
+      "step": 7209
+    },
+    {
+      "epoch": 0.1980170827497185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8555847490217687e-05,
+      "loss": 1.0621,
+      "step": 7210
+    },
+    {
+      "epoch": 0.19804454697756171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855538698705195e-05,
+      "loss": 1.0753,
+      "step": 7211
+    },
+    {
+      "epoch": 0.19807201120540496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554926416192676e-05,
+      "loss": 1.0242,
+      "step": 7212
+    },
+    {
+      "epoch": 0.19809947543324818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855446577764351e-05,
+      "loss": 0.9694,
+      "step": 7213
+    },
+    {
+      "epoch": 0.19812693966109143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85540050714081e-05,
+      "loss": 0.9221,
+      "step": 7214
+    },
+    {
+      "epoch": 0.19815440388893465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8553544297490085e-05,
+      "loss": 1.1012,
+      "step": 7215
+    },
+    {
+      "epoch": 0.1981818681167779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8553083455893116e-05,
+      "loss": 1.0519,
+      "step": 7216
+    },
+    {
+      "epoch": 0.19820933234462113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8552622546620836e-05,
+      "loss": 1.1481,
+      "step": 7217
+    },
+    {
+      "epoch": 0.19823679657246437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8552161569676893e-05,
+      "loss": 0.9677,
+      "step": 7218
+    },
+    {
+      "epoch": 0.1982642608003076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8551700525064937e-05,
+      "loss": 0.9825,
+      "step": 7219
+    },
+    {
+      "epoch": 0.19829172502815084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8551239412788615e-05,
+      "loss": 0.9498,
+      "step": 7220
+    },
+    {
+      "epoch": 0.19831918925599407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855077823285157e-05,
+      "loss": 1.0424,
+      "step": 7221
+    },
+    {
+      "epoch": 0.19834665348383732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855031698525746e-05,
+      "loss": 1.2288,
+      "step": 7222
+    },
+    {
+      "epoch": 0.19837411771168054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8549855670009926e-05,
+      "loss": 0.9392,
+      "step": 7223
+    },
+    {
+      "epoch": 0.19840158193952376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8549394287112625e-05,
+      "loss": 0.9907,
+      "step": 7224
+    },
+    {
+      "epoch": 0.198429046167367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8548932836569206e-05,
+      "loss": 0.9242,
+      "step": 7225
+    },
+    {
+      "epoch": 0.19845651039521023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8548471318383314e-05,
+      "loss": 1.0085,
+      "step": 7226
+    },
+    {
+      "epoch": 0.19848397462305348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854800973255861e-05,
+      "loss": 1.0612,
+      "step": 7227
+    },
+    {
+      "epoch": 0.1985114388508967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8547548079098738e-05,
+      "loss": 0.9833,
+      "step": 7228
+    },
+    {
+      "epoch": 0.19853890307873995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8547086358007358e-05,
+      "loss": 1.0308,
+      "step": 7229
+    },
+    {
+      "epoch": 0.19856636730658317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8546624569288115e-05,
+      "loss": 0.9709,
+      "step": 7230
+    },
+    {
+      "epoch": 0.19859383153442642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8546162712944672e-05,
+      "loss": 1.1011,
+      "step": 7231
+    },
+    {
+      "epoch": 0.19862129576226964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8545700788980678e-05,
+      "loss": 1.1136,
+      "step": 7232
+    },
+    {
+      "epoch": 0.1986487599901129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8545238797399792e-05,
+      "loss": 1.0137,
+      "step": 7233
+    },
+    {
+      "epoch": 0.1986762242179561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8544776738205663e-05,
+      "loss": 0.9667,
+      "step": 7234
+    },
+    {
+      "epoch": 0.19870368844579936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854431461140195e-05,
+      "loss": 1.0297,
+      "step": 7235
+    },
+    {
+      "epoch": 0.19873115267364258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8543852416992313e-05,
+      "loss": 1.0203,
+      "step": 7236
+    },
+    {
+      "epoch": 0.1987586169014858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8543390154980404e-05,
+      "loss": 1.0417,
+      "step": 7237
+    },
+    {
+      "epoch": 0.19878608112932905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8542927825369882e-05,
+      "loss": 1.1245,
+      "step": 7238
+    },
+    {
+      "epoch": 0.19881354535717227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8542465428164406e-05,
+      "loss": 0.9088,
+      "step": 7239
+    },
+    {
+      "epoch": 0.19884100958501552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8542002963367636e-05,
+      "loss": 0.929,
+      "step": 7240
+    },
+    {
+      "epoch": 0.19886847381285874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8541540430983233e-05,
+      "loss": 0.9823,
+      "step": 7241
+    },
+    {
+      "epoch": 0.198895938040702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8541077831014848e-05,
+      "loss": 1.0494,
+      "step": 7242
+    },
+    {
+      "epoch": 0.1989234022685452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854061516346615e-05,
+      "loss": 1.0538,
+      "step": 7243
+    },
+    {
+      "epoch": 0.19895086649638846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854015242834079e-05,
+      "loss": 0.9274,
+      "step": 7244
+    },
+    {
+      "epoch": 0.19897833072423168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853968962564244e-05,
+      "loss": 0.9608,
+      "step": 7245
+    },
+    {
+      "epoch": 0.19900579495207493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8539226755374758e-05,
+      "loss": 1.0215,
+      "step": 7246
+    },
+    {
+      "epoch": 0.19903325917991815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853876381754141e-05,
+      "loss": 1.0319,
+      "step": 7247
+    },
+    {
+      "epoch": 0.1990607234077614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8538300812146047e-05,
+      "loss": 0.9386,
+      "step": 7248
+    },
+    {
+      "epoch": 0.19908818763560462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853783773919235e-05,
+      "loss": 1.0335,
+      "step": 7249
+    },
+    {
+      "epoch": 0.19911565186344785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8537374598683967e-05,
+      "loss": 1.0452,
+      "step": 7250
+    },
+    {
+      "epoch": 0.1991431160912911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8536911390624574e-05,
+      "loss": 1.0064,
+      "step": 7251
+    },
+    {
+      "epoch": 0.19917058031913432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8536448115017824e-05,
+      "loss": 1.0054,
+      "step": 7252
+    },
+    {
+      "epoch": 0.19919804454697757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8535984771867396e-05,
+      "loss": 1.0514,
+      "step": 7253
+    },
+    {
+      "epoch": 0.1992255087748208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8535521361176946e-05,
+      "loss": 1.0495,
+      "step": 7254
+    },
+    {
+      "epoch": 0.19925297300266404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8535057882950147e-05,
+      "loss": 0.9875,
+      "step": 7255
+    },
+    {
+      "epoch": 0.19928043723050726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8534594337190664e-05,
+      "loss": 0.9594,
+      "step": 7256
+    },
+    {
+      "epoch": 0.1993079014583505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8534130723902165e-05,
+      "loss": 1.0139,
+      "step": 7257
+    },
+    {
+      "epoch": 0.19933536568619373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8533667043088313e-05,
+      "loss": 0.9615,
+      "step": 7258
+    },
+    {
+      "epoch": 0.19936282991403698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8533203294752787e-05,
+      "loss": 1.0299,
+      "step": 7259
+    },
+    {
+      "epoch": 0.1993902941418802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853273947889925e-05,
+      "loss": 0.8716,
+      "step": 7260
+    },
+    {
+      "epoch": 0.19941775836972345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8532275595531373e-05,
+      "loss": 1.0365,
+      "step": 7261
+    },
+    {
+      "epoch": 0.19944522259756667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8531811644652827e-05,
+      "loss": 1.0183,
+      "step": 7262
+    },
+    {
+      "epoch": 0.19947268682540992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853134762626728e-05,
+      "loss": 1.0152,
+      "step": 7263
+    },
+    {
+      "epoch": 0.19950015105325314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8530883540378407e-05,
+      "loss": 1.0743,
+      "step": 7264
+    },
+    {
+      "epoch": 0.19952761528109636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853041938698988e-05,
+      "loss": 1.0431,
+      "step": 7265
+    },
+    {
+      "epoch": 0.1995550795089396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8529955166105367e-05,
+      "loss": 1.0133,
+      "step": 7266
+    },
+    {
+      "epoch": 0.19958254373678283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852949087772855e-05,
+      "loss": 1.0258,
+      "step": 7267
+    },
+    {
+      "epoch": 0.19961000796462608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8529026521863093e-05,
+      "loss": 1.1062,
+      "step": 7268
+    },
+    {
+      "epoch": 0.1996374721924693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8528562098512676e-05,
+      "loss": 1.0327,
+      "step": 7269
+    },
+    {
+      "epoch": 0.19966493642031255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8528097607680974e-05,
+      "loss": 0.9675,
+      "step": 7270
+    },
+    {
+      "epoch": 0.19969240064815577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8527633049371655e-05,
+      "loss": 0.9821,
+      "step": 7271
+    },
+    {
+      "epoch": 0.19971986487599902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85271684235884e-05,
+      "loss": 0.9929,
+      "step": 7272
+    },
+    {
+      "epoch": 0.19974732910384224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8526703730334892e-05,
+      "loss": 1.0073,
+      "step": 7273
+    },
+    {
+      "epoch": 0.1997747933316855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85262389696148e-05,
+      "loss": 0.9865,
+      "step": 7274
+    },
+    {
+      "epoch": 0.1998022575595287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8525774141431795e-05,
+      "loss": 1.0264,
+      "step": 7275
+    },
+    {
+      "epoch": 0.19982972178737196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852530924578957e-05,
+      "loss": 1.0005,
+      "step": 7276
+    },
+    {
+      "epoch": 0.19985718601521518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8524844282691794e-05,
+      "loss": 1.0079,
+      "step": 7277
+    },
+    {
+      "epoch": 0.1998846502430584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8524379252142143e-05,
+      "loss": 1.0361,
+      "step": 7278
+    },
+    {
+      "epoch": 0.19991211447090165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8523914154144308e-05,
+      "loss": 1.0043,
+      "step": 7279
+    },
+    {
+      "epoch": 0.19993957869874487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852344898870196e-05,
+      "loss": 1.0145,
+      "step": 7280
+    },
+    {
+      "epoch": 0.19996704292658812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852298375581878e-05,
+      "loss": 0.9542,
+      "step": 7281
+    },
+    {
+      "epoch": 0.19999450715443134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8522518455498452e-05,
+      "loss": 1.0472,
+      "step": 7282
+    },
+    {
+      "epoch": 0.2000219713822746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852205308774466e-05,
+      "loss": 0.9898,
+      "step": 7283
+    },
+    {
+      "epoch": 0.20004943561011782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852158765256108e-05,
+      "loss": 0.8802,
+      "step": 7284
+    },
+    {
+      "epoch": 0.20007689983796106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8521122149951396e-05,
+      "loss": 0.9521,
+      "step": 7285
+    },
+    {
+      "epoch": 0.20010436406580429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8520656579919295e-05,
+      "loss": 0.9011,
+      "step": 7286
+    },
+    {
+      "epoch": 0.20013182829364753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852019094246846e-05,
+      "loss": 0.9527,
+      "step": 7287
+    },
+    {
+      "epoch": 0.20015929252149076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851972523760257e-05,
+      "loss": 1.0281,
+      "step": 7288
+    },
+    {
+      "epoch": 0.200186756749334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8519259465325317e-05,
+      "loss": 1.0146,
+      "step": 7289
+    },
+    {
+      "epoch": 0.20021422097717723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8518793625640382e-05,
+      "loss": 0.99,
+      "step": 7290
+    },
+    {
+      "epoch": 0.20024168520502045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851832771855145e-05,
+      "loss": 1.1155,
+      "step": 7291
+    },
+    {
+      "epoch": 0.2002691494328637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8517861744062212e-05,
+      "loss": 1.007,
+      "step": 7292
+    },
+    {
+      "epoch": 0.20029661366070692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851739570217635e-05,
+      "loss": 1.057,
+      "step": 7293
+    },
+    {
+      "epoch": 0.20032407788855017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8516929592897555e-05,
+      "loss": 1.0784,
+      "step": 7294
+    },
+    {
+      "epoch": 0.2003515421163934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8516463416229517e-05,
+      "loss": 0.9701,
+      "step": 7295
+    },
+    {
+      "epoch": 0.20037900634423664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851599717217592e-05,
+      "loss": 0.9923,
+      "step": 7296
+    },
+    {
+      "epoch": 0.20040647057207986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851553086074045e-05,
+      "loss": 1.0019,
+      "step": 7297
+    },
+    {
+      "epoch": 0.2004339347999231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8515064481926805e-05,
+      "loss": 1.1212,
+      "step": 7298
+    },
+    {
+      "epoch": 0.20046139902776633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851459803573867e-05,
+      "loss": 1.0299,
+      "step": 7299
+    },
+    {
+      "epoch": 0.20048886325560958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851413152217974e-05,
+      "loss": 1.0488,
+      "step": 7300
+    },
+    {
+      "epoch": 0.2005163274834528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85136649412537e-05,
+      "loss": 0.969,
+      "step": 7301
+    },
+    {
+      "epoch": 0.20054379171129605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8513198292964246e-05,
+      "loss": 0.9862,
+      "step": 7302
+    },
+    {
+      "epoch": 0.20057125593913927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851273157731507e-05,
+      "loss": 0.9337,
+      "step": 7303
+    },
+    {
+      "epoch": 0.2005987201669825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8512264794309863e-05,
+      "loss": 1.0434,
+      "step": 7304
+    },
+    {
+      "epoch": 0.20062618439482574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851179794395232e-05,
+      "loss": 0.9839,
+      "step": 7305
+    },
+    {
+      "epoch": 0.20065364862266896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8511331026246137e-05,
+      "loss": 0.9553,
+      "step": 7306
+    },
+    {
+      "epoch": 0.2006811128505122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8510864041195004e-05,
+      "loss": 1.1043,
+      "step": 7307
+    },
+    {
+      "epoch": 0.20070857707835543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851039698880262e-05,
+      "loss": 1.0097,
+      "step": 7308
+    },
+    {
+      "epoch": 0.20073604130619868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8509929869072677e-05,
+      "loss": 1.123,
+      "step": 7309
+    },
+    {
+      "epoch": 0.2007635055340419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850946268200887e-05,
+      "loss": 0.9709,
+      "step": 7310
+    },
+    {
+      "epoch": 0.20079096976188515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85089954276149e-05,
+      "loss": 1.1092,
+      "step": 7311
+    },
+    {
+      "epoch": 0.20081843398972837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850852810589446e-05,
+      "loss": 1.1107,
+      "step": 7312
+    },
+    {
+      "epoch": 0.20084589821757162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8508060716851255e-05,
+      "loss": 1.0534,
+      "step": 7313
+    },
+    {
+      "epoch": 0.20087336244541484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8507593260488974e-05,
+      "loss": 1.0378,
+      "step": 7314
+    },
+    {
+      "epoch": 0.2009008266732581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850712573681132e-05,
+      "loss": 1.046,
+      "step": 7315
+    },
+    {
+      "epoch": 0.20092829090110131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8506658145821992e-05,
+      "loss": 0.9418,
+      "step": 7316
+    },
+    {
+      "epoch": 0.20095575512894456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850619048752469e-05,
+      "loss": 0.9425,
+      "step": 7317
+    },
+    {
+      "epoch": 0.20098321935678778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8505722761923116e-05,
+      "loss": 1.0206,
+      "step": 7318
+    },
+    {
+      "epoch": 0.201010683584631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8505254969020963e-05,
+      "loss": 1.011,
+      "step": 7319
+    },
+    {
+      "epoch": 0.20103814781247425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8504787108821942e-05,
+      "loss": 1.1085,
+      "step": 7320
+    },
+    {
+      "epoch": 0.20106561204031748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850431918132975e-05,
+      "loss": 1.0112,
+      "step": 7321
+    },
+    {
+      "epoch": 0.20109307626816073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850385118654809e-05,
+      "loss": 1.0013,
+      "step": 7322
+    },
+    {
+      "epoch": 0.20112054049600395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8503383124480667e-05,
+      "loss": 0.9184,
+      "step": 7323
+    },
+    {
+      "epoch": 0.2011480047238472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8502914995131183e-05,
+      "loss": 1.0198,
+      "step": 7324
+    },
+    {
+      "epoch": 0.20117546895169042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850244679850334e-05,
+      "loss": 0.9279,
+      "step": 7325
+    },
+    {
+      "epoch": 0.20120293317953367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8501978534600847e-05,
+      "loss": 1.033,
+      "step": 7326
+    },
+    {
+      "epoch": 0.2012303974073769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8501510203427403e-05,
+      "loss": 0.9515,
+      "step": 7327
+    },
+    {
+      "epoch": 0.20125786163522014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850104180498672e-05,
+      "loss": 1.0421,
+      "step": 7328
+    },
+    {
+      "epoch": 0.20128532586306336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8500573339282496e-05,
+      "loss": 0.9445,
+      "step": 7329
+    },
+    {
+      "epoch": 0.2013127900909066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850010480631845e-05,
+      "loss": 1.0438,
+      "step": 7330
+    },
+    {
+      "epoch": 0.20134025431874983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8499636206098278e-05,
+      "loss": 1.0079,
+      "step": 7331
+    },
+    {
+      "epoch": 0.20136771854659305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8499167538625694e-05,
+      "loss": 1.0915,
+      "step": 7332
+    },
+    {
+      "epoch": 0.2013951827744363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8498698803904402e-05,
+      "loss": 1.0748,
+      "step": 7333
+    },
+    {
+      "epoch": 0.20142264700227952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8498230001938117e-05,
+      "loss": 0.9718,
+      "step": 7334
+    },
+    {
+      "epoch": 0.20145011123012277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8497761132730538e-05,
+      "loss": 0.9867,
+      "step": 7335
+    },
+    {
+      "epoch": 0.201477575457966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8497292196285386e-05,
+      "loss": 0.9585,
+      "step": 7336
+    },
+    {
+      "epoch": 0.20150503968580924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8496823192606363e-05,
+      "loss": 1.0395,
+      "step": 7337
+    },
+    {
+      "epoch": 0.20153250391365246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8496354121697186e-05,
+      "loss": 1.065,
+      "step": 7338
+    },
+    {
+      "epoch": 0.2015599681414957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8495884983561566e-05,
+      "loss": 1.1416,
+      "step": 7339
+    },
+    {
+      "epoch": 0.20158743236933893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.849541577820321e-05,
+      "loss": 1.0063,
+      "step": 7340
+    },
+    {
+      "epoch": 0.20161489659718218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8494946505625836e-05,
+      "loss": 1.0992,
+      "step": 7341
+    },
+    {
+      "epoch": 0.2016423608250254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.849447716583315e-05,
+      "loss": 0.9789,
+      "step": 7342
+    },
+    {
+      "epoch": 0.20166982505286865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8494007758828874e-05,
+      "loss": 1.0458,
+      "step": 7343
+    },
+    {
+      "epoch": 0.20169728928071187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8493538284616717e-05,
+      "loss": 1.0266,
+      "step": 7344
+    },
+    {
+      "epoch": 0.2017247535085551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8493068743200396e-05,
+      "loss": 1.0447,
+      "step": 7345
+    },
+    {
+      "epoch": 0.20175221773639834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8492599134583625e-05,
+      "loss": 0.996,
+      "step": 7346
+    },
+    {
+      "epoch": 0.20177968196424156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8492129458770122e-05,
+      "loss": 1.0593,
+      "step": 7347
+    },
+    {
+      "epoch": 0.2018071461920848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84916597157636e-05,
+      "loss": 0.9791,
+      "step": 7348
+    },
+    {
+      "epoch": 0.20183461041992803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8491189905567772e-05,
+      "loss": 0.9836,
+      "step": 7349
+    },
+    {
+      "epoch": 0.20186207464777128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8490720028186367e-05,
+      "loss": 0.9364,
+      "step": 7350
+    },
+    {
+      "epoch": 0.2018895388756145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8490250083623092e-05,
+      "loss": 0.9308,
+      "step": 7351
+    },
+    {
+      "epoch": 0.20191700310345775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8489780071881668e-05,
+      "loss": 1.0173,
+      "step": 7352
+    },
+    {
+      "epoch": 0.20194446733130098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848930999296582e-05,
+      "loss": 1.0515,
+      "step": 7353
+    },
+    {
+      "epoch": 0.20197193155914422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848883984687926e-05,
+      "loss": 1.0391,
+      "step": 7354
+    },
+    {
+      "epoch": 0.20199939578698745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8488369633625713e-05,
+      "loss": 0.9889,
+      "step": 7355
+    },
+    {
+      "epoch": 0.2020268600148307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8487899353208897e-05,
+      "loss": 1.0382,
+      "step": 7356
+    },
+    {
+      "epoch": 0.20205432424267392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848742900563253e-05,
+      "loss": 0.9964,
+      "step": 7357
+    },
+    {
+      "epoch": 0.20208178847051717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848695859090034e-05,
+      "loss": 1.1274,
+      "step": 7358
+    },
+    {
+      "epoch": 0.2021092526983604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848648810901605e-05,
+      "loss": 0.9884,
+      "step": 7359
+    },
+    {
+      "epoch": 0.2021367169262036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8486017559983373e-05,
+      "loss": 0.9683,
+      "step": 7360
+    },
+    {
+      "epoch": 0.20216418115404686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8485546943806038e-05,
+      "loss": 0.9864,
+      "step": 7361
+    },
+    {
+      "epoch": 0.20219164538189008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8485076260487773e-05,
+      "loss": 0.9252,
+      "step": 7362
+    },
+    {
+      "epoch": 0.20221910960973333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8484605510032293e-05,
+      "loss": 0.9597,
+      "step": 7363
+    },
+    {
+      "epoch": 0.20224657383757655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848413469244333e-05,
+      "loss": 0.9551,
+      "step": 7364
+    },
+    {
+      "epoch": 0.2022740380654198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8483663807724607e-05,
+      "loss": 0.9764,
+      "step": 7365
+    },
+    {
+      "epoch": 0.20230150229326302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8483192855879848e-05,
+      "loss": 0.9555,
+      "step": 7366
+    },
+    {
+      "epoch": 0.20232896652110627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8482721836912783e-05,
+      "loss": 1.0563,
+      "step": 7367
+    },
+    {
+      "epoch": 0.2023564307489495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8482250750827136e-05,
+      "loss": 0.8943,
+      "step": 7368
+    },
+    {
+      "epoch": 0.20238389497679274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8481779597626636e-05,
+      "loss": 1.1157,
+      "step": 7369
+    },
+    {
+      "epoch": 0.20241135920463596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8481308377315008e-05,
+      "loss": 0.9391,
+      "step": 7370
+    },
+    {
+      "epoch": 0.2024388234324792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8480837089895985e-05,
+      "loss": 1.0287,
+      "step": 7371
+    },
+    {
+      "epoch": 0.20246628766032243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8480365735373294e-05,
+      "loss": 0.9821,
+      "step": 7372
+    },
+    {
+      "epoch": 0.20249375188816565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8479894313750663e-05,
+      "loss": 0.9058,
+      "step": 7373
+    },
+    {
+      "epoch": 0.2025212161160089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8479422825031824e-05,
+      "loss": 0.9607,
+      "step": 7374
+    },
+    {
+      "epoch": 0.20254868034385212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8478951269220508e-05,
+      "loss": 1.0502,
+      "step": 7375
+    },
+    {
+      "epoch": 0.20257614457169537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8478479646320446e-05,
+      "loss": 1.0338,
+      "step": 7376
+    },
+    {
+      "epoch": 0.2026036087995386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8478007956335364e-05,
+      "loss": 1.0252,
+      "step": 7377
+    },
+    {
+      "epoch": 0.20263107302738184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8477536199269003e-05,
+      "loss": 1.0043,
+      "step": 7378
+    },
+    {
+      "epoch": 0.20265853725522506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847706437512509e-05,
+      "loss": 1.0687,
+      "step": 7379
+    },
+    {
+      "epoch": 0.2026860014830683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8476592483907358e-05,
+      "loss": 0.9552,
+      "step": 7380
+    },
+    {
+      "epoch": 0.20271346571091153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847612052561955e-05,
+      "loss": 1.0624,
+      "step": 7381
+    },
+    {
+      "epoch": 0.20274092993875478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8475648500265385e-05,
+      "loss": 1.0541,
+      "step": 7382
+    },
+    {
+      "epoch": 0.202768394166598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847517640784861e-05,
+      "loss": 1.0217,
+      "step": 7383
+    },
+    {
+      "epoch": 0.20279585839444125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8474704248372956e-05,
+      "loss": 0.9905,
+      "step": 7384
+    },
+    {
+      "epoch": 0.20282332262228447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847423202184216e-05,
+      "loss": 0.9502,
+      "step": 7385
+    },
+    {
+      "epoch": 0.2028507868501277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8473759728259955e-05,
+      "loss": 1.0363,
+      "step": 7386
+    },
+    {
+      "epoch": 0.20287825107797094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8473287367630083e-05,
+      "loss": 1.001,
+      "step": 7387
+    },
+    {
+      "epoch": 0.20290571530581417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8472814939956275e-05,
+      "loss": 0.941,
+      "step": 7388
+    },
+    {
+      "epoch": 0.20293317953365742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847234244524228e-05,
+      "loss": 0.9465,
+      "step": 7389
+    },
+    {
+      "epoch": 0.20296064376150064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8471869883491824e-05,
+      "loss": 0.9557,
+      "step": 7390
+    },
+    {
+      "epoch": 0.20298810798934389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8471397254708654e-05,
+      "loss": 1.0369,
+      "step": 7391
+    },
+    {
+      "epoch": 0.2030155722171871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8470924558896505e-05,
+      "loss": 1.066,
+      "step": 7392
+    },
+    {
+      "epoch": 0.20304303644503036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8470451796059126e-05,
+      "loss": 1.0008,
+      "step": 7393
+    },
+    {
+      "epoch": 0.20307050067287358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8469978966200245e-05,
+      "loss": 1.0967,
+      "step": 7394
+    },
+    {
+      "epoch": 0.20309796490071683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8469506069323612e-05,
+      "loss": 1.0482,
+      "step": 7395
+    },
+    {
+      "epoch": 0.20312542912856005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8469033105432965e-05,
+      "loss": 0.9814,
+      "step": 7396
+    },
+    {
+      "epoch": 0.2031528933564033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8468560074532047e-05,
+      "loss": 0.9461,
+      "step": 7397
+    },
+    {
+      "epoch": 0.20318035758424652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8468086976624603e-05,
+      "loss": 1.0723,
+      "step": 7398
+    },
+    {
+      "epoch": 0.20320782181208974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8467613811714375e-05,
+      "loss": 0.9813,
+      "step": 7399
+    },
+    {
+      "epoch": 0.203235286039933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8467140579805105e-05,
+      "loss": 1.0705,
+      "step": 7400
+    },
+    {
+      "epoch": 0.2032627502677762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846666728090054e-05,
+      "loss": 1.0012,
+      "step": 7401
+    },
+    {
+      "epoch": 0.20329021449561946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846619391500442e-05,
+      "loss": 1.0223,
+      "step": 7402
+    },
+    {
+      "epoch": 0.20331767872346268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84657204821205e-05,
+      "loss": 0.9973,
+      "step": 7403
+    },
+    {
+      "epoch": 0.20334514295130593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8465246982252518e-05,
+      "loss": 1.0229,
+      "step": 7404
+    },
+    {
+      "epoch": 0.20337260717914915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846477341540422e-05,
+      "loss": 1.0538,
+      "step": 7405
+    },
+    {
+      "epoch": 0.2034000714069924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846429978157936e-05,
+      "loss": 1.0616,
+      "step": 7406
+    },
+    {
+      "epoch": 0.20342753563483562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846382608078168e-05,
+      "loss": 1.0743,
+      "step": 7407
+    },
+    {
+      "epoch": 0.20345499986267887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846335231301493e-05,
+      "loss": 1.0774,
+      "step": 7408
+    },
+    {
+      "epoch": 0.2034824640905221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8462878478282858e-05,
+      "loss": 1.0924,
+      "step": 7409
+    },
+    {
+      "epoch": 0.20350992831836534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846240457658921e-05,
+      "loss": 1.0338,
+      "step": 7410
+    },
+    {
+      "epoch": 0.20353739254620856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846193060793774e-05,
+      "loss": 1.0323,
+      "step": 7411
+    },
+    {
+      "epoch": 0.2035648567740518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8461456572332203e-05,
+      "loss": 0.9164,
+      "step": 7412
+    },
+    {
+      "epoch": 0.20359232100189503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8460982469776337e-05,
+      "loss": 0.9883,
+      "step": 7413
+    },
+    {
+      "epoch": 0.20361978522973825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8460508300273904e-05,
+      "loss": 1.0572,
+      "step": 7414
+    },
+    {
+      "epoch": 0.2036472494575815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846003406382865e-05,
+      "loss": 0.9558,
+      "step": 7415
+    },
+    {
+      "epoch": 0.20367471368542472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845955976044433e-05,
+      "loss": 1.0089,
+      "step": 7416
+    },
+    {
+      "epoch": 0.20370217791326797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8459085390124697e-05,
+      "loss": 1.0178,
+      "step": 7417
+    },
+    {
+      "epoch": 0.2037296421411112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8458610952873502e-05,
+      "loss": 1.0013,
+      "step": 7418
+    },
+    {
+      "epoch": 0.20375710636895444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84581364486945e-05,
+      "loss": 1.0722,
+      "step": 7419
+    },
+    {
+      "epoch": 0.20378457059679767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845766187759145e-05,
+      "loss": 1.054,
+      "step": 7420
+    },
+    {
+      "epoch": 0.20381203482464091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84571872395681e-05,
+      "loss": 1.0195,
+      "step": 7421
+    },
+    {
+      "epoch": 0.20383949905248414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845671253462821e-05,
+      "loss": 0.9999,
+      "step": 7422
+    },
+    {
+      "epoch": 0.20386696328032738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8456237762775537e-05,
+      "loss": 1.1077,
+      "step": 7423
+    },
+    {
+      "epoch": 0.2038944275081706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8455762924013832e-05,
+      "loss": 0.9089,
+      "step": 7424
+    },
+    {
+      "epoch": 0.20392189173601385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8455288018346852e-05,
+      "loss": 0.9498,
+      "step": 7425
+    },
+    {
+      "epoch": 0.20394935596385708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8454813045778364e-05,
+      "loss": 0.9074,
+      "step": 7426
+    },
+    {
+      "epoch": 0.2039768201917003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8454338006312114e-05,
+      "loss": 0.9236,
+      "step": 7427
+    },
+    {
+      "epoch": 0.20400428441954355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8453862899951872e-05,
+      "loss": 1.02,
+      "step": 7428
+    },
+    {
+      "epoch": 0.20403174864738677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845338772670139e-05,
+      "loss": 0.9428,
+      "step": 7429
+    },
+    {
+      "epoch": 0.20405921287523002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8452912486564433e-05,
+      "loss": 1.032,
+      "step": 7430
+    },
+    {
+      "epoch": 0.20408667710307324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8452437179544755e-05,
+      "loss": 0.9852,
+      "step": 7431
+    },
+    {
+      "epoch": 0.2041141413309165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845196180564612e-05,
+      "loss": 0.9549,
+      "step": 7432
+    },
+    {
+      "epoch": 0.2041416055587597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845148636487229e-05,
+      "loss": 0.9514,
+      "step": 7433
+    },
+    {
+      "epoch": 0.20416906978660296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8451010857227023e-05,
+      "loss": 0.9771,
+      "step": 7434
+    },
+    {
+      "epoch": 0.20419653401444618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8450535282714087e-05,
+      "loss": 1.0489,
+      "step": 7435
+    },
+    {
+      "epoch": 0.20422399824228943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845005964133724e-05,
+      "loss": 1.0342,
+      "step": 7436
+    },
+    {
+      "epoch": 0.20425146247013265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8449583933100254e-05,
+      "loss": 1.1472,
+      "step": 7437
+    },
+    {
+      "epoch": 0.2042789266979759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8449108158006882e-05,
+      "loss": 0.9813,
+      "step": 7438
+    },
+    {
+      "epoch": 0.20430639092581912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8448632316060894e-05,
+      "loss": 1.077,
+      "step": 7439
+    },
+    {
+      "epoch": 0.20433385515366234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8448156407266052e-05,
+      "loss": 0.9848,
+      "step": 7440
+    },
+    {
+      "epoch": 0.2043613193815056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8447680431626128e-05,
+      "loss": 1.0474,
+      "step": 7441
+    },
+    {
+      "epoch": 0.2043887836093488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8447204389144886e-05,
+      "loss": 1.088,
+      "step": 7442
+    },
+    {
+      "epoch": 0.20441624783719206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8446728279826086e-05,
+      "loss": 1.1049,
+      "step": 7443
+    },
+    {
+      "epoch": 0.20444371206503528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84462521036735e-05,
+      "loss": 1.044,
+      "step": 7444
+    },
+    {
+      "epoch": 0.20447117629287853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84457758606909e-05,
+      "loss": 0.9539,
+      "step": 7445
+    },
+    {
+      "epoch": 0.20449864052072175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8445299550882045e-05,
+      "loss": 1.0179,
+      "step": 7446
+    },
+    {
+      "epoch": 0.204526104748565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844482317425071e-05,
+      "loss": 1.061,
+      "step": 7447
+    },
+    {
+      "epoch": 0.20455356897640822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8444346730800665e-05,
+      "loss": 0.9839,
+      "step": 7448
+    },
+    {
+      "epoch": 0.20458103320425147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8443870220535676e-05,
+      "loss": 0.9854,
+      "step": 7449
+    },
+    {
+      "epoch": 0.2046084974320947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8443393643459516e-05,
+      "loss": 1.0195,
+      "step": 7450
+    },
+    {
+      "epoch": 0.20463596165993794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8442916999575954e-05,
+      "loss": 0.9912,
+      "step": 7451
+    },
+    {
+      "epoch": 0.20466342588778116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844244028888876e-05,
+      "loss": 0.9974,
+      "step": 7452
+    },
+    {
+      "epoch": 0.20469089011562439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844196351140171e-05,
+      "loss": 1.0857,
+      "step": 7453
+    },
+    {
+      "epoch": 0.20471835434346763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8441486667118577e-05,
+      "loss": 1.0208,
+      "step": 7454
+    },
+    {
+      "epoch": 0.20474581857131086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8441009756043126e-05,
+      "loss": 1.0905,
+      "step": 7455
+    },
+    {
+      "epoch": 0.2047732827991541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844053277817914e-05,
+      "loss": 0.9928,
+      "step": 7456
+    },
+    {
+      "epoch": 0.20480074702699733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844005573353039e-05,
+      "loss": 1.0647,
+      "step": 7457
+    },
+    {
+      "epoch": 0.20482821125484058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843957862210065e-05,
+      "loss": 1.0438,
+      "step": 7458
+    },
+    {
+      "epoch": 0.2048556754826838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843910144389369e-05,
+      "loss": 1.0046,
+      "step": 7459
+    },
+    {
+      "epoch": 0.20488313971052705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8438624198913296e-05,
+      "loss": 0.978,
+      "step": 7460
+    },
+    {
+      "epoch": 0.20491060393837027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8438146887163232e-05,
+      "loss": 0.934,
+      "step": 7461
+    },
+    {
+      "epoch": 0.20493806816621352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8437669508647286e-05,
+      "loss": 0.9794,
+      "step": 7462
+    },
+    {
+      "epoch": 0.20496553239405674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843719206336923e-05,
+      "loss": 1.074,
+      "step": 7463
+    },
+    {
+      "epoch": 0.2049929966219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843671455133284e-05,
+      "loss": 0.9476,
+      "step": 7464
+    },
+    {
+      "epoch": 0.2050204608497432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84362369725419e-05,
+      "loss": 0.9846,
+      "step": 7465
+    },
+    {
+      "epoch": 0.20504792507758646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843575932700018e-05,
+      "loss": 1.1298,
+      "step": 7466
+    },
+    {
+      "epoch": 0.20507538930542968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8435281614711472e-05,
+      "loss": 0.9109,
+      "step": 7467
+    },
+    {
+      "epoch": 0.2051028535332729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8434803835679545e-05,
+      "loss": 1.0574,
+      "step": 7468
+    },
+    {
+      "epoch": 0.20513031776111615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843432598990818e-05,
+      "loss": 0.8705,
+      "step": 7469
+    },
+    {
+      "epoch": 0.20515778198895937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8433848077401164e-05,
+      "loss": 0.9434,
+      "step": 7470
+    },
+    {
+      "epoch": 0.20518524621680262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8433370098162272e-05,
+      "loss": 1.0917,
+      "step": 7471
+    },
+    {
+      "epoch": 0.20521271044464584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843289205219529e-05,
+      "loss": 1.0166,
+      "step": 7472
+    },
+    {
+      "epoch": 0.2052401746724891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8432413939504e-05,
+      "loss": 0.9742,
+      "step": 7473
+    },
+    {
+      "epoch": 0.2052676389003323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8431935760092186e-05,
+      "loss": 1.0676,
+      "step": 7474
+    },
+    {
+      "epoch": 0.20529510312817556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8431457513963628e-05,
+      "loss": 1.04,
+      "step": 7475
+    },
+    {
+      "epoch": 0.20532256735601878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843097920112211e-05,
+      "loss": 0.953,
+      "step": 7476
+    },
+    {
+      "epoch": 0.20535003158386203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8430500821571424e-05,
+      "loss": 1.0529,
+      "step": 7477
+    },
+    {
+      "epoch": 0.20537749581170525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8430022375315347e-05,
+      "loss": 1.0179,
+      "step": 7478
+    },
+    {
+      "epoch": 0.2054049600395485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842954386235767e-05,
+      "loss": 1.0724,
+      "step": 7479
+    },
+    {
+      "epoch": 0.20543242426739172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8429065282702173e-05,
+      "loss": 1.0389,
+      "step": 7480
+    },
+    {
+      "epoch": 0.20545988849523494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8428586636352645e-05,
+      "loss": 0.9718,
+      "step": 7481
+    },
+    {
+      "epoch": 0.2054873527230782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842810792331288e-05,
+      "loss": 1.0204,
+      "step": 7482
+    },
+    {
+      "epoch": 0.20551481695092141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8427629143586656e-05,
+      "loss": 1.0508,
+      "step": 7483
+    },
+    {
+      "epoch": 0.20554228117876466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8427150297177768e-05,
+      "loss": 0.9895,
+      "step": 7484
+    },
+    {
+      "epoch": 0.20556974540660788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842667138409e-05,
+      "loss": 1.0073,
+      "step": 7485
+    },
+    {
+      "epoch": 0.20559720963445113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8426192404327147e-05,
+      "loss": 1.0107,
+      "step": 7486
+    },
+    {
+      "epoch": 0.20562467386229435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8425713357892993e-05,
+      "loss": 0.9903,
+      "step": 7487
+    },
+    {
+      "epoch": 0.2056521380901376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8425234244791333e-05,
+      "loss": 1.0331,
+      "step": 7488
+    },
+    {
+      "epoch": 0.20567960231798083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8424755065025957e-05,
+      "loss": 1.0098,
+      "step": 7489
+    },
+    {
+      "epoch": 0.20570706654582407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8424275818600654e-05,
+      "loss": 0.9993,
+      "step": 7490
+    },
+    {
+      "epoch": 0.2057345307736673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8423796505519214e-05,
+      "loss": 1.0192,
+      "step": 7491
+    },
+    {
+      "epoch": 0.20576199500151054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842331712578544e-05,
+      "loss": 0.8852,
+      "step": 7492
+    },
+    {
+      "epoch": 0.20578945922935377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842283767940311e-05,
+      "loss": 0.9526,
+      "step": 7493
+    },
+    {
+      "epoch": 0.205816923457197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8422358166376034e-05,
+      "loss": 1.0118,
+      "step": 7494
+    },
+    {
+      "epoch": 0.20584438768504024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8421878586707993e-05,
+      "loss": 1.042,
+      "step": 7495
+    },
+    {
+      "epoch": 0.20587185191288346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842139894040279e-05,
+      "loss": 1.009,
+      "step": 7496
+    },
+    {
+      "epoch": 0.2058993161407267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842091922746421e-05,
+      "loss": 1.0358,
+      "step": 7497
+    },
+    {
+      "epoch": 0.20592678036856993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8420439447896058e-05,
+      "loss": 0.9087,
+      "step": 7498
+    },
+    {
+      "epoch": 0.20595424459641318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841995960170213e-05,
+      "loss": 1.0307,
+      "step": 7499
+    },
+    {
+      "epoch": 0.2059817088242564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8419479688886215e-05,
+      "loss": 1.1478,
+      "step": 7500
+    },
+    {
+      "epoch": 0.20600917305209965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841899970945212e-05,
+      "loss": 1.1013,
+      "step": 7501
+    },
+    {
+      "epoch": 0.20603663727994287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8418519663403636e-05,
+      "loss": 0.8854,
+      "step": 7502
+    },
+    {
+      "epoch": 0.20606410150778612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8418039550744565e-05,
+      "loss": 1.0241,
+      "step": 7503
+    },
+    {
+      "epoch": 0.20609156573562934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8417559371478702e-05,
+      "loss": 0.9727,
+      "step": 7504
+    },
+    {
+      "epoch": 0.2061190299634726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8417079125609853e-05,
+      "loss": 1.0852,
+      "step": 7505
+    },
+    {
+      "epoch": 0.2061464941913158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8416598813141808e-05,
+      "loss": 1.0345,
+      "step": 7506
+    },
+    {
+      "epoch": 0.20617395841915906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8416118434078375e-05,
+      "loss": 0.9627,
+      "step": 7507
+    },
+    {
+      "epoch": 0.20620142264700228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841563798842335e-05,
+      "loss": 0.9311,
+      "step": 7508
+    },
+    {
+      "epoch": 0.2062288868748455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8415157476180542e-05,
+      "loss": 1.0735,
+      "step": 7509
+    },
+    {
+      "epoch": 0.20625635110268875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8414676897353744e-05,
+      "loss": 0.9199,
+      "step": 7510
+    },
+    {
+      "epoch": 0.20628381533053197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841419625194677e-05,
+      "loss": 0.9595,
+      "step": 7511
+    },
+    {
+      "epoch": 0.20631127955837522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841371553996341e-05,
+      "loss": 0.9395,
+      "step": 7512
+    },
+    {
+      "epoch": 0.20633874378621844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8413234761407476e-05,
+      "loss": 1.0,
+      "step": 7513
+    },
+    {
+      "epoch": 0.2063662080140617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8412753916282765e-05,
+      "loss": 1.0226,
+      "step": 7514
+    },
+    {
+      "epoch": 0.2063936722419049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841227300459309e-05,
+      "loss": 0.9772,
+      "step": 7515
+    },
+    {
+      "epoch": 0.20642113646974816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841179202634225e-05,
+      "loss": 1.0239,
+      "step": 7516
+    },
+    {
+      "epoch": 0.20644860069759138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841131098153406e-05,
+      "loss": 1.0165,
+      "step": 7517
+    },
+    {
+      "epoch": 0.20647606492543463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8410829870172315e-05,
+      "loss": 0.953,
+      "step": 7518
+    },
+    {
+      "epoch": 0.20650352915327785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8410348692260824e-05,
+      "loss": 0.9223,
+      "step": 7519
+    },
+    {
+      "epoch": 0.2065309933811211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8409867447803396e-05,
+      "loss": 0.9294,
+      "step": 7520
+    },
+    {
+      "epoch": 0.20655845760896432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840938613680384e-05,
+      "loss": 0.9212,
+      "step": 7521
+    },
+    {
+      "epoch": 0.20658592183680755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8408904759265965e-05,
+      "loss": 0.9632,
+      "step": 7522
+    },
+    {
+      "epoch": 0.2066133860646508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8408423315193578e-05,
+      "loss": 0.9708,
+      "step": 7523
+    },
+    {
+      "epoch": 0.20664085029249402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8407941804590487e-05,
+      "loss": 1.0497,
+      "step": 7524
+    },
+    {
+      "epoch": 0.20666831452033727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8407460227460507e-05,
+      "loss": 1.0008,
+      "step": 7525
+    },
+    {
+      "epoch": 0.2066957787481805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840697858380744e-05,
+      "loss": 0.9633,
+      "step": 7526
+    },
+    {
+      "epoch": 0.20672324297602374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8406496873635104e-05,
+      "loss": 0.988,
+      "step": 7527
+    },
+    {
+      "epoch": 0.20675070720386696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840601509694731e-05,
+      "loss": 1.1283,
+      "step": 7528
+    },
+    {
+      "epoch": 0.2067781714317102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8405533253747868e-05,
+      "loss": 1.0226,
+      "step": 7529
+    },
+    {
+      "epoch": 0.20680563565955343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840505134404059e-05,
+      "loss": 1.1017,
+      "step": 7530
+    },
+    {
+      "epoch": 0.20683309988739668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8404569367829287e-05,
+      "loss": 1.0036,
+      "step": 7531
+    },
+    {
+      "epoch": 0.2068605641152399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840408732511778e-05,
+      "loss": 1.0569,
+      "step": 7532
+    },
+    {
+      "epoch": 0.20688802834308315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8403605215909875e-05,
+      "loss": 1.0747,
+      "step": 7533
+    },
+    {
+      "epoch": 0.20691549257092637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8403123040209396e-05,
+      "loss": 1.1053,
+      "step": 7534
+    },
+    {
+      "epoch": 0.2069429567987696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840264079802015e-05,
+      "loss": 1.011,
+      "step": 7535
+    },
+    {
+      "epoch": 0.20697042102661284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8402158489345953e-05,
+      "loss": 0.968,
+      "step": 7536
+    },
+    {
+      "epoch": 0.20699788525445606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8401676114190624e-05,
+      "loss": 0.9461,
+      "step": 7537
+    },
+    {
+      "epoch": 0.2070253494822993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8401193672557977e-05,
+      "loss": 0.9775,
+      "step": 7538
+    },
+    {
+      "epoch": 0.20705281371014253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8400711164451835e-05,
+      "loss": 1.1074,
+      "step": 7539
+    },
+    {
+      "epoch": 0.20708027793798578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840022858987601e-05,
+      "loss": 1.0468,
+      "step": 7540
+    },
+    {
+      "epoch": 0.207107742165829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8399745948834323e-05,
+      "loss": 0.9798,
+      "step": 7541
+    },
+    {
+      "epoch": 0.20713520639367225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8399263241330594e-05,
+      "loss": 1.0543,
+      "step": 7542
+    },
+    {
+      "epoch": 0.20716267062151547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839878046736864e-05,
+      "loss": 1.0662,
+      "step": 7543
+    },
+    {
+      "epoch": 0.20719013484935872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8398297626952278e-05,
+      "loss": 0.9903,
+      "step": 7544
+    },
+    {
+      "epoch": 0.20721759907720194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8397814720085335e-05,
+      "loss": 0.9182,
+      "step": 7545
+    },
+    {
+      "epoch": 0.2072450633050452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8397331746771628e-05,
+      "loss": 1.0212,
+      "step": 7546
+    },
+    {
+      "epoch": 0.2072725275328884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8396848707014984e-05,
+      "loss": 0.9329,
+      "step": 7547
+    },
+    {
+      "epoch": 0.20729999176073163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8396365600819216e-05,
+      "loss": 0.988,
+      "step": 7548
+    },
+    {
+      "epoch": 0.20732745598857488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839588242818815e-05,
+      "loss": 1.0669,
+      "step": 7549
+    },
+    {
+      "epoch": 0.2073549202164181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8395399189125612e-05,
+      "loss": 0.9854,
+      "step": 7550
+    },
+    {
+      "epoch": 0.20738238444426135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8394915883635423e-05,
+      "loss": 1.0384,
+      "step": 7551
+    },
+    {
+      "epoch": 0.20740984867210457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8394432511721408e-05,
+      "loss": 1.134,
+      "step": 7552
+    },
+    {
+      "epoch": 0.20743731289994782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8393949073387393e-05,
+      "loss": 1.1309,
+      "step": 7553
+    },
+    {
+      "epoch": 0.20746477712779104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8393465568637196e-05,
+      "loss": 1.113,
+      "step": 7554
+    },
+    {
+      "epoch": 0.2074922413556343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839298199747465e-05,
+      "loss": 1.0765,
+      "step": 7555
+    },
+    {
+      "epoch": 0.20751970558347752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8392498359903586e-05,
+      "loss": 1.0015,
+      "step": 7556
+    },
+    {
+      "epoch": 0.20754716981132076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8392014655927818e-05,
+      "loss": 0.9496,
+      "step": 7557
+    },
+    {
+      "epoch": 0.20757463403916399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839153088555118e-05,
+      "loss": 1.0368,
+      "step": 7558
+    },
+    {
+      "epoch": 0.20760209826700723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83910470487775e-05,
+      "loss": 1.073,
+      "step": 7559
+    },
+    {
+      "epoch": 0.20762956249485046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8390563145610603e-05,
+      "loss": 1.0795,
+      "step": 7560
+    },
+    {
+      "epoch": 0.2076570267226937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8390079176054322e-05,
+      "loss": 1.0248,
+      "step": 7561
+    },
+    {
+      "epoch": 0.20768449095053693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8389595140112484e-05,
+      "loss": 0.9305,
+      "step": 7562
+    },
+    {
+      "epoch": 0.20771195517838015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838911103778892e-05,
+      "loss": 1.1274,
+      "step": 7563
+    },
+    {
+      "epoch": 0.2077394194062234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838862686908746e-05,
+      "loss": 0.9725,
+      "step": 7564
+    },
+    {
+      "epoch": 0.20776688363406662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8388142634011933e-05,
+      "loss": 1.0106,
+      "step": 7565
+    },
+    {
+      "epoch": 0.20779434786190987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8387658332566177e-05,
+      "loss": 0.9707,
+      "step": 7566
+    },
+    {
+      "epoch": 0.2078218120897531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8387173964754015e-05,
+      "loss": 0.949,
+      "step": 7567
+    },
+    {
+      "epoch": 0.20784927631759634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8386689530579285e-05,
+      "loss": 1.0779,
+      "step": 7568
+    },
+    {
+      "epoch": 0.20787674054543956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8386205030045818e-05,
+      "loss": 0.986,
+      "step": 7569
+    },
+    {
+      "epoch": 0.2079042047732828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8385720463157447e-05,
+      "loss": 1.0797,
+      "step": 7570
+    },
+    {
+      "epoch": 0.20793166900112603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838523582991801e-05,
+      "loss": 1.0099,
+      "step": 7571
+    },
+    {
+      "epoch": 0.20795913322896928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8384751130331337e-05,
+      "loss": 0.9588,
+      "step": 7572
+    },
+    {
+      "epoch": 0.2079865974568125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8384266364401265e-05,
+      "loss": 0.9329,
+      "step": 7573
+    },
+    {
+      "epoch": 0.20801406168465575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8383781532131634e-05,
+      "loss": 0.9955,
+      "step": 7574
+    },
+    {
+      "epoch": 0.20804152591249897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838329663352627e-05,
+      "loss": 1.0258,
+      "step": 7575
+    },
+    {
+      "epoch": 0.2080689901403422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8382811668589017e-05,
+      "loss": 1.0308,
+      "step": 7576
+    },
+    {
+      "epoch": 0.20809645436818544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8382326637323717e-05,
+      "loss": 0.9767,
+      "step": 7577
+    },
+    {
+      "epoch": 0.20812391859602866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8381841539734195e-05,
+      "loss": 1.0482,
+      "step": 7578
+    },
+    {
+      "epoch": 0.2081513828238719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8381356375824297e-05,
+      "loss": 0.9325,
+      "step": 7579
+    },
+    {
+      "epoch": 0.20817884705171513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838087114559786e-05,
+      "loss": 0.9736,
+      "step": 7580
+    },
+    {
+      "epoch": 0.20820631127955838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8380385849058728e-05,
+      "loss": 1.0748,
+      "step": 7581
+    },
+    {
+      "epoch": 0.2082337755074016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8379900486210735e-05,
+      "loss": 1.1172,
+      "step": 7582
+    },
+    {
+      "epoch": 0.20826123973524485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8379415057057725e-05,
+      "loss": 0.9901,
+      "step": 7583
+    },
+    {
+      "epoch": 0.20828870396308807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8378929561603532e-05,
+      "loss": 0.9383,
+      "step": 7584
+    },
+    {
+      "epoch": 0.20831616819093132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8378443999852007e-05,
+      "loss": 0.954,
+      "step": 7585
+    },
+    {
+      "epoch": 0.20834363241877454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8377958371806987e-05,
+      "loss": 0.9929,
+      "step": 7586
+    },
+    {
+      "epoch": 0.2083710966466178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837747267747231e-05,
+      "loss": 0.9453,
+      "step": 7587
+    },
+    {
+      "epoch": 0.20839856087446101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8376986916851832e-05,
+      "loss": 0.9534,
+      "step": 7588
+    },
+    {
+      "epoch": 0.20842602510230424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8376501089949382e-05,
+      "loss": 1.0029,
+      "step": 7589
+    },
+    {
+      "epoch": 0.20845348933014748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8376015196768816e-05,
+      "loss": 1.0765,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2084809535579907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837552923731397e-05,
+      "loss": 1.0366,
+      "step": 7591
+    },
+    {
+      "epoch": 0.20850841778583395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8375043211588695e-05,
+      "loss": 1.1184,
+      "step": 7592
+    },
+    {
+      "epoch": 0.20853588201367718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837455711959683e-05,
+      "loss": 0.8673,
+      "step": 7593
+    },
+    {
+      "epoch": 0.20856334624152043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8374070961342228e-05,
+      "loss": 1.0034,
+      "step": 7594
+    },
+    {
+      "epoch": 0.20859081046936365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8373584736828732e-05,
+      "loss": 1.0317,
+      "step": 7595
+    },
+    {
+      "epoch": 0.2086182746972069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8373098446060188e-05,
+      "loss": 0.9733,
+      "step": 7596
+    },
+    {
+      "epoch": 0.20864573892505012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8372612089040446e-05,
+      "loss": 1.0361,
+      "step": 7597
+    },
+    {
+      "epoch": 0.20867320315289337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837212566577336e-05,
+      "loss": 0.9718,
+      "step": 7598
+    },
+    {
+      "epoch": 0.2087006673807366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8371639176262764e-05,
+      "loss": 0.9835,
+      "step": 7599
+    },
+    {
+      "epoch": 0.20872813160857984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8371152620512522e-05,
+      "loss": 1.0684,
+      "step": 7600
+    },
+    {
+      "epoch": 0.20875559583642306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8370665998526475e-05,
+      "loss": 1.0195,
+      "step": 7601
+    },
+    {
+      "epoch": 0.20878306006426628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8370179310308475e-05,
+      "loss": 1.1234,
+      "step": 7602
+    },
+    {
+      "epoch": 0.20881052429210953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836969255586238e-05,
+      "loss": 0.9262,
+      "step": 7603
+    },
+    {
+      "epoch": 0.20883798851995275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8369205735192032e-05,
+      "loss": 1.0708,
+      "step": 7604
+    },
+    {
+      "epoch": 0.208865452747796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8368718848301283e-05,
+      "loss": 0.9857,
+      "step": 7605
+    },
+    {
+      "epoch": 0.20889291697563922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8368231895193993e-05,
+      "loss": 1.0049,
+      "step": 7606
+    },
+    {
+      "epoch": 0.20892038120348247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8367744875874006e-05,
+      "loss": 0.9865,
+      "step": 7607
+    },
+    {
+      "epoch": 0.2089478454313257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8367257790345184e-05,
+      "loss": 1.0389,
+      "step": 7608
+    },
+    {
+      "epoch": 0.20897530965916894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8366770638611376e-05,
+      "loss": 1.0632,
+      "step": 7609
+    },
+    {
+      "epoch": 0.20900277388701216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836628342067644e-05,
+      "loss": 1.0953,
+      "step": 7610
+    },
+    {
+      "epoch": 0.2090302381148554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8365796136544227e-05,
+      "loss": 0.8605,
+      "step": 7611
+    },
+    {
+      "epoch": 0.20905770234269863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8365308786218594e-05,
+      "loss": 1.0663,
+      "step": 7612
+    },
+    {
+      "epoch": 0.20908516657054188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8364821369703398e-05,
+      "loss": 1.0838,
+      "step": 7613
+    },
+    {
+      "epoch": 0.2091126307983851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8364333887002494e-05,
+      "loss": 1.0543,
+      "step": 7614
+    },
+    {
+      "epoch": 0.20914009502622835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8363846338119745e-05,
+      "loss": 1.1322,
+      "step": 7615
+    },
+    {
+      "epoch": 0.20916755925407157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8363358723058996e-05,
+      "loss": 1.0714,
+      "step": 7616
+    },
+    {
+      "epoch": 0.2091950234819148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8362871041824116e-05,
+      "loss": 0.9594,
+      "step": 7617
+    },
+    {
+      "epoch": 0.20922248770975804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8362383294418965e-05,
+      "loss": 1.0633,
+      "step": 7618
+    },
+    {
+      "epoch": 0.20924995193760126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8361895480847395e-05,
+      "loss": 1.0569,
+      "step": 7619
+    },
+    {
+      "epoch": 0.2092774161654445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8361407601113268e-05,
+      "loss": 0.9846,
+      "step": 7620
+    },
+    {
+      "epoch": 0.20930488039328773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8360919655220447e-05,
+      "loss": 1.0606,
+      "step": 7621
+    },
+    {
+      "epoch": 0.20933234462113098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836043164317279e-05,
+      "loss": 1.0886,
+      "step": 7622
+    },
+    {
+      "epoch": 0.2093598088489742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8359943564974162e-05,
+      "loss": 0.9986,
+      "step": 7623
+    },
+    {
+      "epoch": 0.20938727307681745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8359455420628416e-05,
+      "loss": 1.032,
+      "step": 7624
+    },
+    {
+      "epoch": 0.20941473730466068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8358967210139426e-05,
+      "loss": 0.9969,
+      "step": 7625
+    },
+    {
+      "epoch": 0.20944220153250392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8358478933511047e-05,
+      "loss": 0.8885,
+      "step": 7626
+    },
+    {
+      "epoch": 0.20946966576034715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835799059074715e-05,
+      "loss": 1.0816,
+      "step": 7627
+    },
+    {
+      "epoch": 0.2094971299881904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8357502181851584e-05,
+      "loss": 0.9043,
+      "step": 7628
+    },
+    {
+      "epoch": 0.20952459421603362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835701370682823e-05,
+      "loss": 0.9699,
+      "step": 7629
+    },
+    {
+      "epoch": 0.20955205844387684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8356525165680947e-05,
+      "loss": 0.9974,
+      "step": 7630
+    },
+    {
+      "epoch": 0.2095795226717201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8356036558413596e-05,
+      "loss": 1.1015,
+      "step": 7631
+    },
+    {
+      "epoch": 0.2096069868995633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8355547885030048e-05,
+      "loss": 0.957,
+      "step": 7632
+    },
+    {
+      "epoch": 0.20963445112740656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835505914553417e-05,
+      "loss": 1.0428,
+      "step": 7633
+    },
+    {
+      "epoch": 0.20966191535524978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8354570339929826e-05,
+      "loss": 1.0688,
+      "step": 7634
+    },
+    {
+      "epoch": 0.20968937958309303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835408146822089e-05,
+      "loss": 1.0408,
+      "step": 7635
+    },
+    {
+      "epoch": 0.20971684381093625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835359253041122e-05,
+      "loss": 0.9196,
+      "step": 7636
+    },
+    {
+      "epoch": 0.2097443080387795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8353103526504693e-05,
+      "loss": 0.9364,
+      "step": 7637
+    },
+    {
+      "epoch": 0.20977177226662272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8352614456505174e-05,
+      "loss": 0.8102,
+      "step": 7638
+    },
+    {
+      "epoch": 0.20979923649446597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8352125320416533e-05,
+      "loss": 0.985,
+      "step": 7639
+    },
+    {
+      "epoch": 0.2098267007223092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8351636118242642e-05,
+      "loss": 0.9716,
+      "step": 7640
+    },
+    {
+      "epoch": 0.20985416495015244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835114684998737e-05,
+      "loss": 1.0289,
+      "step": 7641
+    },
+    {
+      "epoch": 0.20988162917799566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835065751565459e-05,
+      "loss": 0.8813,
+      "step": 7642
+    },
+    {
+      "epoch": 0.20990909340583888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8350168115248175e-05,
+      "loss": 1.0393,
+      "step": 7643
+    },
+    {
+      "epoch": 0.20993655763368213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8349678648771992e-05,
+      "loss": 0.9865,
+      "step": 7644
+    },
+    {
+      "epoch": 0.20996402186152535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834918911622992e-05,
+      "loss": 1.0713,
+      "step": 7645
+    },
+    {
+      "epoch": 0.2099914860893686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834869951762583e-05,
+      "loss": 0.9678,
+      "step": 7646
+    },
+    {
+      "epoch": 0.21001895031721182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834820985296359e-05,
+      "loss": 0.9421,
+      "step": 7647
+    },
+    {
+      "epoch": 0.21004641454505507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8347720122247087e-05,
+      "loss": 0.9839,
+      "step": 7648
+    },
+    {
+      "epoch": 0.2100738787728983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8347230325480184e-05,
+      "loss": 0.9831,
+      "step": 7649
+    },
+    {
+      "epoch": 0.21010134300074154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834674046266676e-05,
+      "loss": 1.0236,
+      "step": 7650
+    },
+    {
+      "epoch": 0.21012880722858476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8346250533810698e-05,
+      "loss": 0.9567,
+      "step": 7651
+    },
+    {
+      "epoch": 0.210156271456428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8345760538915863e-05,
+      "loss": 1.0543,
+      "step": 7652
+    },
+    {
+      "epoch": 0.21018373568427123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834527047798614e-05,
+      "loss": 1.1455,
+      "step": 7653
+    },
+    {
+      "epoch": 0.21021119991211448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8344780351025405e-05,
+      "loss": 1.0043,
+      "step": 7654
+    },
+    {
+      "epoch": 0.2102386641399577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8344290158037536e-05,
+      "loss": 0.9722,
+      "step": 7655
+    },
+    {
+      "epoch": 0.21026612836780095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834379989902641e-05,
+      "loss": 1.0262,
+      "step": 7656
+    },
+    {
+      "epoch": 0.21029359259564417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8343309573995908e-05,
+      "loss": 1.0333,
+      "step": 7657
+    },
+    {
+      "epoch": 0.2103210568234874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8342819182949906e-05,
+      "loss": 0.9911,
+      "step": 7658
+    },
+    {
+      "epoch": 0.21034852105133064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834232872589229e-05,
+      "loss": 0.9758,
+      "step": 7659
+    },
+    {
+      "epoch": 0.21037598527917387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8341838202826937e-05,
+      "loss": 0.9868,
+      "step": 7660
+    },
+    {
+      "epoch": 0.21040344950701712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834134761375773e-05,
+      "loss": 1.0126,
+      "step": 7661
+    },
+    {
+      "epoch": 0.21043091373486034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8340856958688548e-05,
+      "loss": 1.0297,
+      "step": 7662
+    },
+    {
+      "epoch": 0.21045837796270359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8340366237623274e-05,
+      "loss": 0.9563,
+      "step": 7663
+    },
+    {
+      "epoch": 0.2104858421905468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8339875450565792e-05,
+      "loss": 0.9087,
+      "step": 7664
+    },
+    {
+      "epoch": 0.21051330641839006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833938459751998e-05,
+      "loss": 1.016,
+      "step": 7665
+    },
+    {
+      "epoch": 0.21054077064623328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8338893678489738e-05,
+      "loss": 1.0682,
+      "step": 7666
+    },
+    {
+      "epoch": 0.21056823487407653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833840269347893e-05,
+      "loss": 0.9091,
+      "step": 7667
+    },
+    {
+      "epoch": 0.21059569910191975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8337911642491454e-05,
+      "loss": 1.1298,
+      "step": 7668
+    },
+    {
+      "epoch": 0.210623163329763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833742052553119e-05,
+      "loss": 1.0506,
+      "step": 7669
+    },
+    {
+      "epoch": 0.21065062755760622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8336929342602025e-05,
+      "loss": 1.057,
+      "step": 7670
+    },
+    {
+      "epoch": 0.21067809178544944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8336438093707848e-05,
+      "loss": 1.0929,
+      "step": 7671
+    },
+    {
+      "epoch": 0.2107055560132927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833594677885254e-05,
+      "loss": 1.0084,
+      "step": 7672
+    },
+    {
+      "epoch": 0.2107330202411359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8335455398039994e-05,
+      "loss": 0.9228,
+      "step": 7673
+    },
+    {
+      "epoch": 0.21076048446897916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8334963951274096e-05,
+      "loss": 1.0024,
+      "step": 7674
+    },
+    {
+      "epoch": 0.21078794869682238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833447243855873e-05,
+      "loss": 0.9976,
+      "step": 7675
+    },
+    {
+      "epoch": 0.21081541292466563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8333980859897795e-05,
+      "loss": 0.9874,
+      "step": 7676
+    },
+    {
+      "epoch": 0.21084287715250885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8333489215295172e-05,
+      "loss": 0.9598,
+      "step": 7677
+    },
+    {
+      "epoch": 0.2108703413803521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833299750475476e-05,
+      "loss": 0.9321,
+      "step": 7678
+    },
+    {
+      "epoch": 0.21089780560819532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8332505728280437e-05,
+      "loss": 0.9881,
+      "step": 7679
+    },
+    {
+      "epoch": 0.21092526983603857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8332013885876105e-05,
+      "loss": 1.0343,
+      "step": 7680
+    },
+    {
+      "epoch": 0.2109527340638818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833152197754565e-05,
+      "loss": 0.9973,
+      "step": 7681
+    },
+    {
+      "epoch": 0.21098019829172504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8331030003292966e-05,
+      "loss": 1.0344,
+      "step": 7682
+    },
+    {
+      "epoch": 0.21100766251956826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8330537963121945e-05,
+      "loss": 1.0065,
+      "step": 7683
+    },
+    {
+      "epoch": 0.21103512674741148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8330045857036483e-05,
+      "loss": 1.0556,
+      "step": 7684
+    },
+    {
+      "epoch": 0.21106259097525473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832955368504047e-05,
+      "loss": 1.0251,
+      "step": 7685
+    },
+    {
+      "epoch": 0.21109005520309795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83290614471378e-05,
+      "loss": 1.0691,
+      "step": 7686
+    },
+    {
+      "epoch": 0.2111175194309412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8328569143332372e-05,
+      "loss": 0.9734,
+      "step": 7687
+    },
+    {
+      "epoch": 0.21114498365878442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8328076773628082e-05,
+      "loss": 1.1449,
+      "step": 7688
+    },
+    {
+      "epoch": 0.21117244788662767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8327584338028813e-05,
+      "loss": 0.9687,
+      "step": 7689
+    },
+    {
+      "epoch": 0.2111999121144709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8327091836538478e-05,
+      "loss": 1.1003,
+      "step": 7690
+    },
+    {
+      "epoch": 0.21122737634231414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8326599269160964e-05,
+      "loss": 1.1422,
+      "step": 7691
+    },
+    {
+      "epoch": 0.21125484057015737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8326106635900173e-05,
+      "loss": 1.1313,
+      "step": 7692
+    },
+    {
+      "epoch": 0.21128230479800061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832561393676e-05,
+      "loss": 0.9499,
+      "step": 7693
+    },
+    {
+      "epoch": 0.21130976902584384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8325121171744345e-05,
+      "loss": 0.9791,
+      "step": 7694
+    },
+    {
+      "epoch": 0.21133723325368708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832462834085711e-05,
+      "loss": 1.0112,
+      "step": 7695
+    },
+    {
+      "epoch": 0.2113646974815303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8324135444102185e-05,
+      "loss": 1.0773,
+      "step": 7696
+    },
+    {
+      "epoch": 0.21139216170937353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832364248148348e-05,
+      "loss": 0.968,
+      "step": 7697
+    },
+    {
+      "epoch": 0.21141962593721678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8323149453004887e-05,
+      "loss": 0.9118,
+      "step": 7698
+    },
+    {
+      "epoch": 0.21144709016506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8322656358670314e-05,
+      "loss": 1.0779,
+      "step": 7699
+    },
+    {
+      "epoch": 0.21147455439290325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832216319848366e-05,
+      "loss": 0.9859,
+      "step": 7700
+    },
+    {
+      "epoch": 0.21150201862074647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832166997244883e-05,
+      "loss": 1.0507,
+      "step": 7701
+    },
+    {
+      "epoch": 0.21152948284858972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832117668056972e-05,
+      "loss": 1.0649,
+      "step": 7702
+    },
+    {
+      "epoch": 0.21155694707643294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8320683322850236e-05,
+      "loss": 1.0873,
+      "step": 7703
+    },
+    {
+      "epoch": 0.2115844113042762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832018989929428e-05,
+      "loss": 0.9179,
+      "step": 7704
+    },
+    {
+      "epoch": 0.2116118755321194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8319696409905768e-05,
+      "loss": 1.0399,
+      "step": 7705
+    },
+    {
+      "epoch": 0.21163933975996266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8319202854688588e-05,
+      "loss": 0.9228,
+      "step": 7706
+    },
+    {
+      "epoch": 0.21166680398780588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8318709233646655e-05,
+      "loss": 0.9458,
+      "step": 7707
+    },
+    {
+      "epoch": 0.21169426821564913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8318215546783872e-05,
+      "loss": 0.9685,
+      "step": 7708
+    },
+    {
+      "epoch": 0.21172173244349235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8317721794104142e-05,
+      "loss": 1.0027,
+      "step": 7709
+    },
+    {
+      "epoch": 0.2117491966713356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8317227975611378e-05,
+      "loss": 1.0663,
+      "step": 7710
+    },
+    {
+      "epoch": 0.21177666089917882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8316734091309484e-05,
+      "loss": 0.9472,
+      "step": 7711
+    },
+    {
+      "epoch": 0.21180412512702204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831624014120237e-05,
+      "loss": 0.9586,
+      "step": 7712
+    },
+    {
+      "epoch": 0.2118315893548653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8315746125293944e-05,
+      "loss": 1.0463,
+      "step": 7713
+    },
+    {
+      "epoch": 0.2118590535827085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831525204358811e-05,
+      "loss": 0.997,
+      "step": 7714
+    },
+    {
+      "epoch": 0.21188651781055176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831475789608878e-05,
+      "loss": 1.0365,
+      "step": 7715
+    },
+    {
+      "epoch": 0.21191398203839498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8314263682799866e-05,
+      "loss": 1.0267,
+      "step": 7716
+    },
+    {
+      "epoch": 0.21194144626623823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8313769403725277e-05,
+      "loss": 0.9418,
+      "step": 7717
+    },
+    {
+      "epoch": 0.21196891049408145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8313275058868923e-05,
+      "loss": 1.1013,
+      "step": 7718
+    },
+    {
+      "epoch": 0.2119963747219247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831278064823472e-05,
+      "loss": 1.0256,
+      "step": 7719
+    },
+    {
+      "epoch": 0.21202383894976792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8312286171826575e-05,
+      "loss": 0.9375,
+      "step": 7720
+    },
+    {
+      "epoch": 0.21205130317761117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83117916296484e-05,
+      "loss": 0.9453,
+      "step": 7721
+    },
+    {
+      "epoch": 0.2120787674054544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8311297021704112e-05,
+      "loss": 1.0558,
+      "step": 7722
+    },
+    {
+      "epoch": 0.21210623163329764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831080234799762e-05,
+      "loss": 1.0103,
+      "step": 7723
+    },
+    {
+      "epoch": 0.21213369586114086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8310307608532846e-05,
+      "loss": 1.0231,
+      "step": 7724
+    },
+    {
+      "epoch": 0.21216116008898409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8309812803313692e-05,
+      "loss": 0.9662,
+      "step": 7725
+    },
+    {
+      "epoch": 0.21218862431682733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8309317932344087e-05,
+      "loss": 0.9594,
+      "step": 7726
+    },
+    {
+      "epoch": 0.21221608854467056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8308822995627934e-05,
+      "loss": 1.006,
+      "step": 7727
+    },
+    {
+      "epoch": 0.2122435527725138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830832799316916e-05,
+      "loss": 1.1011,
+      "step": 7728
+    },
+    {
+      "epoch": 0.21227101700035703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8307832924971673e-05,
+      "loss": 1.0675,
+      "step": 7729
+    },
+    {
+      "epoch": 0.21229848122820028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8307337791039392e-05,
+      "loss": 0.9973,
+      "step": 7730
+    },
+    {
+      "epoch": 0.2123259454560435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306842591376243e-05,
+      "loss": 0.9858,
+      "step": 7731
+    },
+    {
+      "epoch": 0.21235340968388675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306347325986132e-05,
+      "loss": 0.8902,
+      "step": 7732
+    },
+    {
+      "epoch": 0.21238087391172997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8305851994872987e-05,
+      "loss": 1.0314,
+      "step": 7733
+    },
+    {
+      "epoch": 0.21240833813957322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830535659804072e-05,
+      "loss": 0.9757,
+      "step": 7734
+    },
+    {
+      "epoch": 0.21243580236741644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830486113549326e-05,
+      "loss": 1.0883,
+      "step": 7735
+    },
+    {
+      "epoch": 0.2124632665952597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8304365607234513e-05,
+      "loss": 1.0581,
+      "step": 7736
+    },
+    {
+      "epoch": 0.2124907308231029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8303870013268416e-05,
+      "loss": 1.0032,
+      "step": 7737
+    },
+    {
+      "epoch": 0.21251819505094613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8303374353598878e-05,
+      "loss": 1.0005,
+      "step": 7738
+    },
+    {
+      "epoch": 0.21254565927878938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8302878628229826e-05,
+      "loss": 0.9895,
+      "step": 7739
+    },
+    {
+      "epoch": 0.2125731235066326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8302382837165183e-05,
+      "loss": 1.1046,
+      "step": 7740
+    },
+    {
+      "epoch": 0.21260058773447585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8301886980408876e-05,
+      "loss": 1.0874,
+      "step": 7741
+    },
+    {
+      "epoch": 0.21262805196231907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830139105796482e-05,
+      "loss": 1.1488,
+      "step": 7742
+    },
+    {
+      "epoch": 0.21265551619016232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830089506983694e-05,
+      "loss": 1.0118,
+      "step": 7743
+    },
+    {
+      "epoch": 0.21268298041800554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8300399016029166e-05,
+      "loss": 0.9569,
+      "step": 7744
+    },
+    {
+      "epoch": 0.2127104446458488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8299902896545414e-05,
+      "loss": 1.0574,
+      "step": 7745
+    },
+    {
+      "epoch": 0.212737908873692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829940671138962e-05,
+      "loss": 0.9277,
+      "step": 7746
+    },
+    {
+      "epoch": 0.21276537310153526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298910460565708e-05,
+      "loss": 1.0649,
+      "step": 7747
+    },
+    {
+      "epoch": 0.21279283732937848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298414144077593e-05,
+      "loss": 1.0565,
+      "step": 7748
+    },
+    {
+      "epoch": 0.21282030155722173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829791776192922e-05,
+      "loss": 0.9165,
+      "step": 7749
+    },
+    {
+      "epoch": 0.21284776578506495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8297421314124502e-05,
+      "loss": 0.9767,
+      "step": 7750
+    },
+    {
+      "epoch": 0.2128752300129082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8296924800667373e-05,
+      "loss": 0.9921,
+      "step": 7751
+    },
+    {
+      "epoch": 0.21290269424075142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8296428221561765e-05,
+      "loss": 1.105,
+      "step": 7752
+    },
+    {
+      "epoch": 0.21293015846859464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82959315768116e-05,
+      "loss": 1.0208,
+      "step": 7753
+    },
+    {
+      "epoch": 0.2129576226964379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829543486642081e-05,
+      "loss": 1.0004,
+      "step": 7754
+    },
+    {
+      "epoch": 0.21298508692428111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8294938090393327e-05,
+      "loss": 1.0912,
+      "step": 7755
+    },
+    {
+      "epoch": 0.21301255115212436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829444124873308e-05,
+      "loss": 0.9068,
+      "step": 7756
+    },
+    {
+      "epoch": 0.21304001537996758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8293944341444e-05,
+      "loss": 0.9637,
+      "step": 7757
+    },
+    {
+      "epoch": 0.21306747960781083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829344736853002e-05,
+      "loss": 1.0582,
+      "step": 7758
+    },
+    {
+      "epoch": 0.21309494383565405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8292950329995074e-05,
+      "loss": 1.1563,
+      "step": 7759
+    },
+    {
+      "epoch": 0.2131224080634973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8292453225843093e-05,
+      "loss": 1.0461,
+      "step": 7760
+    },
+    {
+      "epoch": 0.21314987229134053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829195605607801e-05,
+      "loss": 0.9937,
+      "step": 7761
+    },
+    {
+      "epoch": 0.21317733651918377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8291458820703756e-05,
+      "loss": 1.0695,
+      "step": 7762
+    },
+    {
+      "epoch": 0.213204800747027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8290961519724266e-05,
+      "loss": 1.0381,
+      "step": 7763
+    },
+    {
+      "epoch": 0.21323226497487024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829046415314348e-05,
+      "loss": 1.0401,
+      "step": 7764
+    },
+    {
+      "epoch": 0.21325972920271347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8289966720965334e-05,
+      "loss": 0.9825,
+      "step": 7765
+    },
+    {
+      "epoch": 0.2132871934305567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8289469223193754e-05,
+      "loss": 1.0722,
+      "step": 7766
+    },
+    {
+      "epoch": 0.21331465765839994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8288971659832684e-05,
+      "loss": 0.9318,
+      "step": 7767
+    },
+    {
+      "epoch": 0.21334212188624316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8288474030886062e-05,
+      "loss": 1.0245,
+      "step": 7768
+    },
+    {
+      "epoch": 0.2133695861140864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828797633635782e-05,
+      "loss": 0.9674,
+      "step": 7769
+    },
+    {
+      "epoch": 0.21339705034192963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82874785762519e-05,
+      "loss": 0.967,
+      "step": 7770
+    },
+    {
+      "epoch": 0.21342451456977288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8286980750572237e-05,
+      "loss": 0.9355,
+      "step": 7771
+    },
+    {
+      "epoch": 0.2134519787976161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8286482859322775e-05,
+      "loss": 1.0284,
+      "step": 7772
+    },
+    {
+      "epoch": 0.21347944302545935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828598490250745e-05,
+      "loss": 0.9679,
+      "step": 7773
+    },
+    {
+      "epoch": 0.21350690725330257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8285486880130202e-05,
+      "loss": 1.002,
+      "step": 7774
+    },
+    {
+      "epoch": 0.21353437148114582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8284988792194976e-05,
+      "loss": 0.9569,
+      "step": 7775
+    },
+    {
+      "epoch": 0.21356183570898904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8284490638705702e-05,
+      "loss": 1.0438,
+      "step": 7776
+    },
+    {
+      "epoch": 0.2135892999368323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8283992419666337e-05,
+      "loss": 1.0227,
+      "step": 7777
+    },
+    {
+      "epoch": 0.2136167641646755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8283494135080812e-05,
+      "loss": 0.9624,
+      "step": 7778
+    },
+    {
+      "epoch": 0.21364422839251873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828299578495307e-05,
+      "loss": 0.9082,
+      "step": 7779
+    },
+    {
+      "epoch": 0.21367169262036198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8282497369287058e-05,
+      "loss": 1.0265,
+      "step": 7780
+    },
+    {
+      "epoch": 0.2136991568482052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828199888808672e-05,
+      "loss": 1.0712,
+      "step": 7781
+    },
+    {
+      "epoch": 0.21372662107604845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8281500341355997e-05,
+      "loss": 0.949,
+      "step": 7782
+    },
+    {
+      "epoch": 0.21375408530389167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828100172909884e-05,
+      "loss": 1.0145,
+      "step": 7783
+    },
+    {
+      "epoch": 0.21378154953173492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8280503051319182e-05,
+      "loss": 1.0837,
+      "step": 7784
+    },
+    {
+      "epoch": 0.21380901375957814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828000430802098e-05,
+      "loss": 0.9922,
+      "step": 7785
+    },
+    {
+      "epoch": 0.2138364779874214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827950549920818e-05,
+      "loss": 1.0651,
+      "step": 7786
+    },
+    {
+      "epoch": 0.2138639422152646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827900662488472e-05,
+      "loss": 0.9888,
+      "step": 7787
+    },
+    {
+      "epoch": 0.21389140644310786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8278507685054556e-05,
+      "loss": 0.9689,
+      "step": 7788
+    },
+    {
+      "epoch": 0.21391887067095108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827800867972163e-05,
+      "loss": 1.0093,
+      "step": 7789
+    },
+    {
+      "epoch": 0.21394633489879433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8277509608889895e-05,
+      "loss": 1.0638,
+      "step": 7790
+    },
+    {
+      "epoch": 0.21397379912663755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8277010472563295e-05,
+      "loss": 1.0573,
+      "step": 7791
+    },
+    {
+      "epoch": 0.21400126335448078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8276511270745784e-05,
+      "loss": 1.0515,
+      "step": 7792
+    },
+    {
+      "epoch": 0.21402872758232402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8276012003441307e-05,
+      "loss": 0.9361,
+      "step": 7793
+    },
+    {
+      "epoch": 0.21405619181016725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827551267065382e-05,
+      "loss": 0.9905,
+      "step": 7794
+    },
+    {
+      "epoch": 0.2140836560380105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8275013272387272e-05,
+      "loss": 1.0193,
+      "step": 7795
+    },
+    {
+      "epoch": 0.21411112026585372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8274513808645612e-05,
+      "loss": 0.8558,
+      "step": 7796
+    },
+    {
+      "epoch": 0.21413858449369697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827401427943279e-05,
+      "loss": 1.157,
+      "step": 7797
+    },
+    {
+      "epoch": 0.2141660487215402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827351468475277e-05,
+      "loss": 0.9539,
+      "step": 7798
+    },
+    {
+      "epoch": 0.21419351294938344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827301502460949e-05,
+      "loss": 1.0951,
+      "step": 7799
+    },
+    {
+      "epoch": 0.21422097717722666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8272515299006916e-05,
+      "loss": 0.9797,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2142484414050699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8272015507948994e-05,
+      "loss": 1.1224,
+      "step": 7801
+    },
+    {
+      "epoch": 0.21427590563291313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827151565143968e-05,
+      "loss": 1.0281,
+      "step": 7802
+    },
+    {
+      "epoch": 0.21430336986075638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8271015729482927e-05,
+      "loss": 1.0316,
+      "step": 7803
+    },
+    {
+      "epoch": 0.2143308340885996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82705157420827e-05,
+      "loss": 0.9927,
+      "step": 7804
+    },
+    {
+      "epoch": 0.21435829831644285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8270015689242944e-05,
+      "loss": 0.9449,
+      "step": 7805
+    },
+    {
+      "epoch": 0.21438576254428607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8269515570967623e-05,
+      "loss": 1.1041,
+      "step": 7806
+    },
+    {
+      "epoch": 0.2144132267721293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826901538726069e-05,
+      "loss": 0.9847,
+      "step": 7807
+    },
+    {
+      "epoch": 0.21444069099997254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8268515138126107e-05,
+      "loss": 0.8588,
+      "step": 7808
+    },
+    {
+      "epoch": 0.21446815522781576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8268014823567824e-05,
+      "loss": 0.9827,
+      "step": 7809
+    },
+    {
+      "epoch": 0.214495619455659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8267514443589812e-05,
+      "loss": 0.9639,
+      "step": 7810
+    },
+    {
+      "epoch": 0.21452308368350223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8267013998196018e-05,
+      "loss": 1.0384,
+      "step": 7811
+    },
+    {
+      "epoch": 0.21455054791134548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826651348739041e-05,
+      "loss": 1.0409,
+      "step": 7812
+    },
+    {
+      "epoch": 0.2145780121391887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8266012911176944e-05,
+      "loss": 1.1047,
+      "step": 7813
+    },
+    {
+      "epoch": 0.21460547636703195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826551226955958e-05,
+      "loss": 1.0421,
+      "step": 7814
+    },
+    {
+      "epoch": 0.21463294059487517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8265011562542284e-05,
+      "loss": 0.9117,
+      "step": 7815
+    },
+    {
+      "epoch": 0.21466040482271842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826451079012901e-05,
+      "loss": 1.0743,
+      "step": 7816
+    },
+    {
+      "epoch": 0.21468786905056164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826400995232373e-05,
+      "loss": 1.0127,
+      "step": 7817
+    },
+    {
+      "epoch": 0.2147153332784049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82635090491304e-05,
+      "loss": 1.0332,
+      "step": 7818
+    },
+    {
+      "epoch": 0.2147427975062481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8263008080552985e-05,
+      "loss": 1.0036,
+      "step": 7819
+    },
+    {
+      "epoch": 0.21477026173409133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8262507046595454e-05,
+      "loss": 1.0645,
+      "step": 7820
+    },
+    {
+      "epoch": 0.21479772596193458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8262005947261762e-05,
+      "loss": 1.0063,
+      "step": 7821
+    },
+    {
+      "epoch": 0.2148251901897778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826150478255588e-05,
+      "loss": 0.9968,
+      "step": 7822
+    },
+    {
+      "epoch": 0.21485265441762105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826100355248177e-05,
+      "loss": 1.052,
+      "step": 7823
+    },
+    {
+      "epoch": 0.21488011864546427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8260502257043404e-05,
+      "loss": 1.0244,
+      "step": 7824
+    },
+    {
+      "epoch": 0.21490758287330752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8260000896244743e-05,
+      "loss": 0.9331,
+      "step": 7825
+    },
+    {
+      "epoch": 0.21493504710115074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8259499470089754e-05,
+      "loss": 1.0579,
+      "step": 7826
+    },
+    {
+      "epoch": 0.214962511328994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8258997978582403e-05,
+      "loss": 1.0336,
+      "step": 7827
+    },
+    {
+      "epoch": 0.21498997555683722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8258496421726666e-05,
+      "loss": 0.9633,
+      "step": 7828
+    },
+    {
+      "epoch": 0.21501743978468046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8257994799526505e-05,
+      "loss": 1.1213,
+      "step": 7829
+    },
+    {
+      "epoch": 0.21504490401252369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8257493111985887e-05,
+      "loss": 0.9553,
+      "step": 7830
+    },
+    {
+      "epoch": 0.21507236824036693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825699135910879e-05,
+      "loss": 0.9434,
+      "step": 7831
+    },
+    {
+      "epoch": 0.21509983246821016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8256489540899177e-05,
+      "loss": 1.0565,
+      "step": 7832
+    },
+    {
+      "epoch": 0.21512729669605338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8255987657361017e-05,
+      "loss": 0.8957,
+      "step": 7833
+    },
+    {
+      "epoch": 0.21515476092389663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8255485708498287e-05,
+      "loss": 0.9775,
+      "step": 7834
+    },
+    {
+      "epoch": 0.21518222515173985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8254983694314957e-05,
+      "loss": 0.9742,
+      "step": 7835
+    },
+    {
+      "epoch": 0.2152096893795831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8254481614815e-05,
+      "loss": 0.9763,
+      "step": 7836
+    },
+    {
+      "epoch": 0.21523715360742632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8253979470002385e-05,
+      "loss": 1.0107,
+      "step": 7837
+    },
+    {
+      "epoch": 0.21526461783526957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8253477259881087e-05,
+      "loss": 1.0436,
+      "step": 7838
+    },
+    {
+      "epoch": 0.2152920820631128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8252974984455076e-05,
+      "loss": 1.0188,
+      "step": 7839
+    },
+    {
+      "epoch": 0.21531954629095604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8252472643728335e-05,
+      "loss": 1.0673,
+      "step": 7840
+    },
+    {
+      "epoch": 0.21534701051879926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8251970237704833e-05,
+      "loss": 0.9627,
+      "step": 7841
+    },
+    {
+      "epoch": 0.2153744747466425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8251467766388546e-05,
+      "loss": 0.8873,
+      "step": 7842
+    },
+    {
+      "epoch": 0.21540193897448573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825096522978345e-05,
+      "loss": 0.9724,
+      "step": 7843
+    },
+    {
+      "epoch": 0.21542940320232898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8250462627893516e-05,
+      "loss": 1.12,
+      "step": 7844
+    },
+    {
+      "epoch": 0.2154568674301722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824995996072273e-05,
+      "loss": 1.0539,
+      "step": 7845
+    },
+    {
+      "epoch": 0.21548433165801542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824945722827507e-05,
+      "loss": 0.9821,
+      "step": 7846
+    },
+    {
+      "epoch": 0.21551179588585867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82489544305545e-05,
+      "loss": 1.0371,
+      "step": 7847
+    },
+    {
+      "epoch": 0.2155392601137019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8248451567565013e-05,
+      "loss": 1.0469,
+      "step": 7848
+    },
+    {
+      "epoch": 0.21556672434154514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824794863931058e-05,
+      "loss": 0.9718,
+      "step": 7849
+    },
+    {
+      "epoch": 0.21559418856938836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8247445645795182e-05,
+      "loss": 1.0263,
+      "step": 7850
+    },
+    {
+      "epoch": 0.2156216527972316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82469425870228e-05,
+      "loss": 0.9976,
+      "step": 7851
+    },
+    {
+      "epoch": 0.21564911702507483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8246439462997413e-05,
+      "loss": 0.9424,
+      "step": 7852
+    },
+    {
+      "epoch": 0.21567658125291808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8245936273723003e-05,
+      "loss": 0.9331,
+      "step": 7853
+    },
+    {
+      "epoch": 0.2157040454807613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8245433019203555e-05,
+      "loss": 1.0278,
+      "step": 7854
+    },
+    {
+      "epoch": 0.21573150970860455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8244929699443038e-05,
+      "loss": 1.0803,
+      "step": 7855
+    },
+    {
+      "epoch": 0.21575897393644777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8244426314445453e-05,
+      "loss": 1.0235,
+      "step": 7856
+    },
+    {
+      "epoch": 0.21578643816429102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8243922864214767e-05,
+      "loss": 1.0563,
+      "step": 7857
+    },
+    {
+      "epoch": 0.21581390239213424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824341934875497e-05,
+      "loss": 0.9826,
+      "step": 7858
+    },
+    {
+      "epoch": 0.2158413666199775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824291576807005e-05,
+      "loss": 0.9551,
+      "step": 7859
+    },
+    {
+      "epoch": 0.21586883084782071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8242412122163987e-05,
+      "loss": 0.9559,
+      "step": 7860
+    },
+    {
+      "epoch": 0.21589629507566394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8241908411040763e-05,
+      "loss": 0.9763,
+      "step": 7861
+    },
+    {
+      "epoch": 0.21592375930350718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8241404634704368e-05,
+      "loss": 1.0664,
+      "step": 7862
+    },
+    {
+      "epoch": 0.2159512235313504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824090079315879e-05,
+      "loss": 0.9807,
+      "step": 7863
+    },
+    {
+      "epoch": 0.21597868775919365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824039688640801e-05,
+      "loss": 1.092,
+      "step": 7864
+    },
+    {
+      "epoch": 0.21600615198703688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8239892914456017e-05,
+      "loss": 0.9638,
+      "step": 7865
+    },
+    {
+      "epoch": 0.21603361621488013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8239388877306803e-05,
+      "loss": 0.8445,
+      "step": 7866
+    },
+    {
+      "epoch": 0.21606108044272335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823888477496435e-05,
+      "loss": 1.0266,
+      "step": 7867
+    },
+    {
+      "epoch": 0.2160885446705666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823838060743265e-05,
+      "loss": 0.9735,
+      "step": 7868
+    },
+    {
+      "epoch": 0.21611600889840982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8237876374715688e-05,
+      "loss": 1.0482,
+      "step": 7869
+    },
+    {
+      "epoch": 0.21614347312625307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823737207681746e-05,
+      "loss": 0.9614,
+      "step": 7870
+    },
+    {
+      "epoch": 0.2161709373540963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8236867713741957e-05,
+      "loss": 0.9372,
+      "step": 7871
+    },
+    {
+      "epoch": 0.21619840158193954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823636328549316e-05,
+      "loss": 1.0221,
+      "step": 7872
+    },
+    {
+      "epoch": 0.21622586580978276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823585879207507e-05,
+      "loss": 0.9839,
+      "step": 7873
+    },
+    {
+      "epoch": 0.21625333003762598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8235354233491675e-05,
+      "loss": 1.0103,
+      "step": 7874
+    },
+    {
+      "epoch": 0.21628079426546923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823484960974696e-05,
+      "loss": 1.0944,
+      "step": 7875
+    },
+    {
+      "epoch": 0.21630825849331245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8234344920844936e-05,
+      "loss": 0.9493,
+      "step": 7876
+    },
+    {
+      "epoch": 0.2163357227211557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8233840166789577e-05,
+      "loss": 1.0719,
+      "step": 7877
+    },
+    {
+      "epoch": 0.21636318694899892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8233335347584892e-05,
+      "loss": 0.9411,
+      "step": 7878
+    },
+    {
+      "epoch": 0.21639065117684217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8232830463234863e-05,
+      "loss": 1.0747,
+      "step": 7879
+    },
+    {
+      "epoch": 0.2164181154046854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823232551374349e-05,
+      "loss": 1.0795,
+      "step": 7880
+    },
+    {
+      "epoch": 0.21644557963252864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8231820499114773e-05,
+      "loss": 1.0091,
+      "step": 7881
+    },
+    {
+      "epoch": 0.21647304386037186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82313154193527e-05,
+      "loss": 1.0061,
+      "step": 7882
+    },
+    {
+      "epoch": 0.2165005080882151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8230810274461273e-05,
+      "loss": 1.0064,
+      "step": 7883
+    },
+    {
+      "epoch": 0.21652797231605833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8230305064444483e-05,
+      "loss": 1.0732,
+      "step": 7884
+    },
+    {
+      "epoch": 0.21655543654390158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8229799789306337e-05,
+      "loss": 0.9367,
+      "step": 7885
+    },
+    {
+      "epoch": 0.2165829007717448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8229294449050826e-05,
+      "loss": 1.0604,
+      "step": 7886
+    },
+    {
+      "epoch": 0.21661036499958802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8228789043681944e-05,
+      "loss": 1.0926,
+      "step": 7887
+    },
+    {
+      "epoch": 0.21663782922743127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8228283573203702e-05,
+      "loss": 1.0786,
+      "step": 7888
+    },
+    {
+      "epoch": 0.2166652934552745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822777803762009e-05,
+      "loss": 0.9838,
+      "step": 7889
+    },
+    {
+      "epoch": 0.21669275768311774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822727243693511e-05,
+      "loss": 1.0202,
+      "step": 7890
+    },
+    {
+      "epoch": 0.21672022191096096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8226766771152764e-05,
+      "loss": 1.0054,
+      "step": 7891
+    },
+    {
+      "epoch": 0.2167476861388042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8226261040277056e-05,
+      "loss": 1.0015,
+      "step": 7892
+    },
+    {
+      "epoch": 0.21677515036664743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822575524431198e-05,
+      "loss": 0.9475,
+      "step": 7893
+    },
+    {
+      "epoch": 0.21680261459449068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822524938326154e-05,
+      "loss": 1.0221,
+      "step": 7894
+    },
+    {
+      "epoch": 0.2168300788223339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822474345712975e-05,
+      "loss": 1.0443,
+      "step": 7895
+    },
+    {
+      "epoch": 0.21685754305017715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82242374659206e-05,
+      "loss": 1.0626,
+      "step": 7896
+    },
+    {
+      "epoch": 0.21688500727802038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8223731409638094e-05,
+      "loss": 0.8917,
+      "step": 7897
+    },
+    {
+      "epoch": 0.21691247150586362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822322528828624e-05,
+      "loss": 0.8947,
+      "step": 7898
+    },
+    {
+      "epoch": 0.21693993573370685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8222719101869044e-05,
+      "loss": 1.049,
+      "step": 7899
+    },
+    {
+      "epoch": 0.2169673999615501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8222212850390507e-05,
+      "loss": 0.9685,
+      "step": 7900
+    },
+    {
+      "epoch": 0.21699486418939332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822170653385464e-05,
+      "loss": 1.0431,
+      "step": 7901
+    },
+    {
+      "epoch": 0.21702232841723654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8221200152265446e-05,
+      "loss": 0.9936,
+      "step": 7902
+    },
+    {
+      "epoch": 0.2170497926450798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822069370562693e-05,
+      "loss": 1.047,
+      "step": 7903
+    },
+    {
+      "epoch": 0.217077256872923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82201871939431e-05,
+      "loss": 1.0349,
+      "step": 7904
+    },
+    {
+      "epoch": 0.21710472110076626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8219680617217967e-05,
+      "loss": 0.9683,
+      "step": 7905
+    },
+    {
+      "epoch": 0.21713218532860948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8219173975455535e-05,
+      "loss": 1.0628,
+      "step": 7906
+    },
+    {
+      "epoch": 0.21715964955645273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821866726865982e-05,
+      "loss": 1.0057,
+      "step": 7907
+    },
+    {
+      "epoch": 0.21718711378429595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821816049683482e-05,
+      "loss": 0.904,
+      "step": 7908
+    },
+    {
+      "epoch": 0.2172145780121392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821765365998455e-05,
+      "loss": 1.0517,
+      "step": 7909
+    },
+    {
+      "epoch": 0.21724204223998242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8217146758113026e-05,
+      "loss": 1.0899,
+      "step": 7910
+    },
+    {
+      "epoch": 0.21726950646782567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821663979122425e-05,
+      "loss": 1.0644,
+      "step": 7911
+    },
+    {
+      "epoch": 0.2172969706956689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821613275932224e-05,
+      "loss": 1.1296,
+      "step": 7912
+    },
+    {
+      "epoch": 0.21732443492351214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8215625662411004e-05,
+      "loss": 0.9431,
+      "step": 7913
+    },
+    {
+      "epoch": 0.21735189915135536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8215118500494554e-05,
+      "loss": 0.9715,
+      "step": 7914
+    },
+    {
+      "epoch": 0.21737936337919858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8214611273576904e-05,
+      "loss": 0.955,
+      "step": 7915
+    },
+    {
+      "epoch": 0.21740682760704183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821410398166207e-05,
+      "loss": 1.0791,
+      "step": 7916
+    },
+    {
+      "epoch": 0.21743429183488505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8213596624754063e-05,
+      "loss": 0.8924,
+      "step": 7917
+    },
+    {
+      "epoch": 0.2174617560627283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8213089202856896e-05,
+      "loss": 1.0477,
+      "step": 7918
+    },
+    {
+      "epoch": 0.21748922029057152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8212581715974586e-05,
+      "loss": 1.0012,
+      "step": 7919
+    },
+    {
+      "epoch": 0.21751668451841477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821207416411115e-05,
+      "loss": 1.1525,
+      "step": 7920
+    },
+    {
+      "epoch": 0.217544148746258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82115665472706e-05,
+      "loss": 1.0023,
+      "step": 7921
+    },
+    {
+      "epoch": 0.21757161297410124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8211058865456955e-05,
+      "loss": 0.9804,
+      "step": 7922
+    },
+    {
+      "epoch": 0.21759907720194446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8210551118674233e-05,
+      "loss": 0.9346,
+      "step": 7923
+    },
+    {
+      "epoch": 0.2176265414297877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8210043306926452e-05,
+      "loss": 0.9431,
+      "step": 7924
+    },
+    {
+      "epoch": 0.21765400565763093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820953543021762e-05,
+      "loss": 1.0828,
+      "step": 7925
+    },
+    {
+      "epoch": 0.21768146988547418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8209027488551774e-05,
+      "loss": 1.01,
+      "step": 7926
+    },
+    {
+      "epoch": 0.2177089341133174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820851948193292e-05,
+      "loss": 1.0539,
+      "step": 7927
+    },
+    {
+      "epoch": 0.21773639834116063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8208011410365073e-05,
+      "loss": 1.0844,
+      "step": 7928
+    },
+    {
+      "epoch": 0.21776386256900387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8207503273852267e-05,
+      "loss": 1.1117,
+      "step": 7929
+    },
+    {
+      "epoch": 0.2177913267968471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8206995072398515e-05,
+      "loss": 1.0023,
+      "step": 7930
+    },
+    {
+      "epoch": 0.21781879102469034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8206486806007836e-05,
+      "loss": 1.0216,
+      "step": 7931
+    },
+    {
+      "epoch": 0.21784625525253357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205978474684255e-05,
+      "loss": 1.0376,
+      "step": 7932
+    },
+    {
+      "epoch": 0.21787371948037682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205470078431797e-05,
+      "loss": 0.9471,
+      "step": 7933
+    },
+    {
+      "epoch": 0.21790118370822004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820496161725448e-05,
+      "loss": 0.9662,
+      "step": 7934
+    },
+    {
+      "epoch": 0.21792864793606329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8204453091156327e-05,
+      "loss": 0.9447,
+      "step": 7935
+    },
+    {
+      "epoch": 0.2179561121639065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8203944500141365e-05,
+      "loss": 1.0503,
+      "step": 7936
+    },
+    {
+      "epoch": 0.21798357639174976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8203435844213613e-05,
+      "loss": 0.9624,
+      "step": 7937
+    },
+    {
+      "epoch": 0.21801104061959298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82029271233771e-05,
+      "loss": 0.94,
+      "step": 7938
+    },
+    {
+      "epoch": 0.21803850484743623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8202418337635854e-05,
+      "loss": 0.8438,
+      "step": 7939
+    },
+    {
+      "epoch": 0.21806596907527945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8201909486993892e-05,
+      "loss": 1.0488,
+      "step": 7940
+    },
+    {
+      "epoch": 0.21809343330312267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820140057145525e-05,
+      "loss": 0.9646,
+      "step": 7941
+    },
+    {
+      "epoch": 0.21812089753096592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8200891591023943e-05,
+      "loss": 1.0139,
+      "step": 7942
+    },
+    {
+      "epoch": 0.21814836175880914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820038254570401e-05,
+      "loss": 1.0073,
+      "step": 7943
+    },
+    {
+      "epoch": 0.2181758259866524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819987343549947e-05,
+      "loss": 0.9271,
+      "step": 7944
+    },
+    {
+      "epoch": 0.2182032902144956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819936426041436e-05,
+      "loss": 0.8884,
+      "step": 7945
+    },
+    {
+      "epoch": 0.21823075444233886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81988550204527e-05,
+      "loss": 0.9361,
+      "step": 7946
+    },
+    {
+      "epoch": 0.21825821867018208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8198345715618524e-05,
+      "loss": 0.9939,
+      "step": 7947
+    },
+    {
+      "epoch": 0.21828568289802533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819783634591586e-05,
+      "loss": 1.0578,
+      "step": 7948
+    },
+    {
+      "epoch": 0.21831314712586855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8197326911348742e-05,
+      "loss": 0.8643,
+      "step": 7949
+    },
+    {
+      "epoch": 0.2183406113537118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8196817411921197e-05,
+      "loss": 1.1148,
+      "step": 7950
+    },
+    {
+      "epoch": 0.21836807558155502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8196307847637258e-05,
+      "loss": 1.0594,
+      "step": 7951
+    },
+    {
+      "epoch": 0.21839553980939827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8195798218500955e-05,
+      "loss": 1.1447,
+      "step": 7952
+    },
+    {
+      "epoch": 0.2184230040372415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8195288524516322e-05,
+      "loss": 0.9542,
+      "step": 7953
+    },
+    {
+      "epoch": 0.21845046826508474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8194778765687393e-05,
+      "loss": 0.9055,
+      "step": 7954
+    },
+    {
+      "epoch": 0.21847793249292796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8194268942018197e-05,
+      "loss": 0.9231,
+      "step": 7955
+    },
+    {
+      "epoch": 0.21850539672077118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8193759053512777e-05,
+      "loss": 1.0284,
+      "step": 7956
+    },
+    {
+      "epoch": 0.21853286094861443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8193249100175156e-05,
+      "loss": 0.9704,
+      "step": 7957
+    },
+    {
+      "epoch": 0.21856032517645765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8192739082009377e-05,
+      "loss": 1.0642,
+      "step": 7958
+    },
+    {
+      "epoch": 0.2185877894043009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819222899901947e-05,
+      "loss": 0.9357,
+      "step": 7959
+    },
+    {
+      "epoch": 0.21861525363214412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8191718851209473e-05,
+      "loss": 0.9886,
+      "step": 7960
+    },
+    {
+      "epoch": 0.21864271785998737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8191208638583426e-05,
+      "loss": 0.9615,
+      "step": 7961
+    },
+    {
+      "epoch": 0.2186701820878306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8190698361145365e-05,
+      "loss": 0.9413,
+      "step": 7962
+    },
+    {
+      "epoch": 0.21869764631567384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8190188018899322e-05,
+      "loss": 0.9332,
+      "step": 7963
+    },
+    {
+      "epoch": 0.21872511054351707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818967761184934e-05,
+      "loss": 0.9815,
+      "step": 7964
+    },
+    {
+      "epoch": 0.21875257477136031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8189167139999457e-05,
+      "loss": 0.9606,
+      "step": 7965
+    },
+    {
+      "epoch": 0.21878003899920354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8188656603353712e-05,
+      "loss": 0.9631,
+      "step": 7966
+    },
+    {
+      "epoch": 0.21880750322704678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8188146001916146e-05,
+      "loss": 1.0063,
+      "step": 7967
+    },
+    {
+      "epoch": 0.21883496745489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8187635335690793e-05,
+      "loss": 1.0534,
+      "step": 7968
+    },
+    {
+      "epoch": 0.21886243168273323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81871246046817e-05,
+      "loss": 0.9648,
+      "step": 7969
+    },
+    {
+      "epoch": 0.21888989591057648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8186613808892902e-05,
+      "loss": 1.0164,
+      "step": 7970
+    },
+    {
+      "epoch": 0.2189173601384197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818610294832845e-05,
+      "loss": 0.985,
+      "step": 7971
+    },
+    {
+      "epoch": 0.21894482436626295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8185592022992376e-05,
+      "loss": 0.9568,
+      "step": 7972
+    },
+    {
+      "epoch": 0.21897228859410617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818508103288873e-05,
+      "loss": 0.9724,
+      "step": 7973
+    },
+    {
+      "epoch": 0.21899975282194942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8184569978021548e-05,
+      "loss": 1.0126,
+      "step": 7974
+    },
+    {
+      "epoch": 0.21902721704979264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818405885839488e-05,
+      "loss": 0.919,
+      "step": 7975
+    },
+    {
+      "epoch": 0.2190546812776359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818354767401277e-05,
+      "loss": 1.0894,
+      "step": 7976
+    },
+    {
+      "epoch": 0.2190821455054791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818303642487926e-05,
+      "loss": 1.1045,
+      "step": 7977
+    },
+    {
+      "epoch": 0.21910960973332236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81825251109984e-05,
+      "loss": 0.9739,
+      "step": 7978
+    },
+    {
+      "epoch": 0.21913707396116558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8182013732374223e-05,
+      "loss": 0.9858,
+      "step": 7979
+    },
+    {
+      "epoch": 0.21916453818900883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818150228901079e-05,
+      "loss": 1.0171,
+      "step": 7980
+    },
+    {
+      "epoch": 0.21919200241685205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818099078091214e-05,
+      "loss": 0.8918,
+      "step": 7981
+    },
+    {
+      "epoch": 0.21921946664469527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8180479208082323e-05,
+      "loss": 1.0561,
+      "step": 7982
+    },
+    {
+      "epoch": 0.21924693087253852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8179967570525386e-05,
+      "loss": 1.0546,
+      "step": 7983
+    },
+    {
+      "epoch": 0.21927439510038174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8179455868245374e-05,
+      "loss": 0.9834,
+      "step": 7984
+    },
+    {
+      "epoch": 0.219301859328225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817894410124634e-05,
+      "loss": 1.0768,
+      "step": 7985
+    },
+    {
+      "epoch": 0.2193293235560682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8178432269532334e-05,
+      "loss": 0.9315,
+      "step": 7986
+    },
+    {
+      "epoch": 0.21935678778391146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81779203731074e-05,
+      "loss": 0.9822,
+      "step": 7987
+    },
+    {
+      "epoch": 0.21938425201175468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8177408411975594e-05,
+      "loss": 0.9165,
+      "step": 7988
+    },
+    {
+      "epoch": 0.21941171623959793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8176896386140966e-05,
+      "loss": 0.9931,
+      "step": 7989
+    },
+    {
+      "epoch": 0.21943918046744115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8176384295607567e-05,
+      "loss": 1.0205,
+      "step": 7990
+    },
+    {
+      "epoch": 0.2194666446952844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8175872140379448e-05,
+      "loss": 0.9331,
+      "step": 7991
+    },
+    {
+      "epoch": 0.21949410892312762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817535992046066e-05,
+      "loss": 1.1091,
+      "step": 7992
+    },
+    {
+      "epoch": 0.21952157315097087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8174847635855256e-05,
+      "loss": 0.9949,
+      "step": 7993
+    },
+    {
+      "epoch": 0.2195490373788141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8174335286567294e-05,
+      "loss": 1.019,
+      "step": 7994
+    },
+    {
+      "epoch": 0.21957650160665731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8173822872600822e-05,
+      "loss": 0.9971,
+      "step": 7995
+    },
+    {
+      "epoch": 0.21960396583450056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81733103939599e-05,
+      "loss": 1.0498,
+      "step": 7996
+    },
+    {
+      "epoch": 0.21963143006234379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8172797850648584e-05,
+      "loss": 1.032,
+      "step": 7997
+    },
+    {
+      "epoch": 0.21965889429018703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817228524267092e-05,
+      "loss": 0.9367,
+      "step": 7998
+    },
+    {
+      "epoch": 0.21968635851803026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817177257003097e-05,
+      "loss": 0.9518,
+      "step": 7999
+    },
+    {
+      "epoch": 0.2197138227458735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817125983273279e-05,
+      "loss": 1.1141,
+      "step": 8000
+    },
+    {
+      "epoch": 0.21974128697371673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8170747030780436e-05,
+      "loss": 0.9363,
+      "step": 8001
+    },
+    {
+      "epoch": 0.21976875120155998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8170234164177968e-05,
+      "loss": 0.9414,
+      "step": 8002
+    },
+    {
+      "epoch": 0.2197962154294032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8169721232929443e-05,
+      "loss": 1.0864,
+      "step": 8003
+    },
+    {
+      "epoch": 0.21982367965724645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816920823703892e-05,
+      "loss": 0.9973,
+      "step": 8004
+    },
+    {
+      "epoch": 0.21985114388508967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8168695176510452e-05,
+      "loss": 1.0226,
+      "step": 8005
+    },
+    {
+      "epoch": 0.21987860811293292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8168182051348107e-05,
+      "loss": 1.0003,
+      "step": 8006
+    },
+    {
+      "epoch": 0.21990607234077614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8167668861555942e-05,
+      "loss": 0.9173,
+      "step": 8007
+    },
+    {
+      "epoch": 0.2199335365686194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8167155607138015e-05,
+      "loss": 1.0038,
+      "step": 8008
+    },
+    {
+      "epoch": 0.2199610007964626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816664228809839e-05,
+      "loss": 0.9353,
+      "step": 8009
+    },
+    {
+      "epoch": 0.21998846502430583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816612890444113e-05,
+      "loss": 0.8974,
+      "step": 8010
+    },
+    {
+      "epoch": 0.22001592925214908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816561545617029e-05,
+      "loss": 0.9191,
+      "step": 8011
+    },
+    {
+      "epoch": 0.2200433934799923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8165101943289944e-05,
+      "loss": 1.0591,
+      "step": 8012
+    },
+    {
+      "epoch": 0.22007085770783555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8164588365804144e-05,
+      "loss": 1.0541,
+      "step": 8013
+    },
+    {
+      "epoch": 0.22009832193567877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816407472371696e-05,
+      "loss": 1.0708,
+      "step": 8014
+    },
+    {
+      "epoch": 0.22012578616352202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816356101703245e-05,
+      "loss": 0.9837,
+      "step": 8015
+    },
+    {
+      "epoch": 0.22015325039136524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8163047245754687e-05,
+      "loss": 1.0743,
+      "step": 8016
+    },
+    {
+      "epoch": 0.2201807146192085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8162533409887735e-05,
+      "loss": 1.1412,
+      "step": 8017
+    },
+    {
+      "epoch": 0.2202081788470517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8162019509435652e-05,
+      "loss": 0.9273,
+      "step": 8018
+    },
+    {
+      "epoch": 0.22023564307489496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816150554440251e-05,
+      "loss": 1.0756,
+      "step": 8019
+    },
+    {
+      "epoch": 0.22026310730273818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8160991514792378e-05,
+      "loss": 0.9876,
+      "step": 8020
+    },
+    {
+      "epoch": 0.22029057153058143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816047742060931e-05,
+      "loss": 0.9964,
+      "step": 8021
+    },
+    {
+      "epoch": 0.22031803575842465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8159963261857397e-05,
+      "loss": 1.0701,
+      "step": 8022
+    },
+    {
+      "epoch": 0.22034549998626787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8159449038540685e-05,
+      "loss": 1.0243,
+      "step": 8023
+    },
+    {
+      "epoch": 0.22037296421411112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8158934750663254e-05,
+      "loss": 1.0989,
+      "step": 8024
+    },
+    {
+      "epoch": 0.22040042844195434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815842039822917e-05,
+      "loss": 0.9857,
+      "step": 8025
+    },
+    {
+      "epoch": 0.2204278926697976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8157905981242504e-05,
+      "loss": 0.9826,
+      "step": 8026
+    },
+    {
+      "epoch": 0.22045535689764081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8157391499707326e-05,
+      "loss": 0.8903,
+      "step": 8027
+    },
+    {
+      "epoch": 0.22048282112548406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8156876953627707e-05,
+      "loss": 0.9158,
+      "step": 8028
+    },
+    {
+      "epoch": 0.22051028535332728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8156362343007717e-05,
+      "loss": 0.9879,
+      "step": 8029
+    },
+    {
+      "epoch": 0.22053774958117053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815584766785143e-05,
+      "loss": 0.9549,
+      "step": 8030
+    },
+    {
+      "epoch": 0.22056521380901375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8155332928162918e-05,
+      "loss": 0.9286,
+      "step": 8031
+    },
+    {
+      "epoch": 0.220592678036857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815481812394625e-05,
+      "loss": 1.0271,
+      "step": 8032
+    },
+    {
+      "epoch": 0.22062014226470023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8154303255205503e-05,
+      "loss": 1.0507,
+      "step": 8033
+    },
+    {
+      "epoch": 0.22064760649254347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815378832194475e-05,
+      "loss": 1.0063,
+      "step": 8034
+    },
+    {
+      "epoch": 0.2206750707203867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8153273324168065e-05,
+      "loss": 0.936,
+      "step": 8035
+    },
+    {
+      "epoch": 0.22070253494822992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8152758261879523e-05,
+      "loss": 1.0323,
+      "step": 8036
+    },
+    {
+      "epoch": 0.22072999917607317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81522431350832e-05,
+      "loss": 0.8963,
+      "step": 8037
+    },
+    {
+      "epoch": 0.2207574634039164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8151727943783173e-05,
+      "loss": 1.0958,
+      "step": 8038
+    },
+    {
+      "epoch": 0.22078492763175964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8151212687983514e-05,
+      "loss": 1.0135,
+      "step": 8039
+    },
+    {
+      "epoch": 0.22081239185960286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8150697367688306e-05,
+      "loss": 0.9421,
+      "step": 8040
+    },
+    {
+      "epoch": 0.2208398560874461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815018198290162e-05,
+      "loss": 0.8778,
+      "step": 8041
+    },
+    {
+      "epoch": 0.22086732031528933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814966653362754e-05,
+      "loss": 0.9824,
+      "step": 8042
+    },
+    {
+      "epoch": 0.22089478454313258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8149151019870138e-05,
+      "loss": 0.9876,
+      "step": 8043
+    },
+    {
+      "epoch": 0.2209222487709758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81486354416335e-05,
+      "loss": 1.0371,
+      "step": 8044
+    },
+    {
+      "epoch": 0.22094971299881905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8148119798921698e-05,
+      "loss": 1.1006,
+      "step": 8045
+    },
+    {
+      "epoch": 0.22097717722666227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814760409173882e-05,
+      "loss": 1.006,
+      "step": 8046
+    },
+    {
+      "epoch": 0.22100464145450552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8147088320088938e-05,
+      "loss": 1.0623,
+      "step": 8047
+    },
+    {
+      "epoch": 0.22103210568234874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814657248397614e-05,
+      "loss": 0.9985,
+      "step": 8048
+    },
+    {
+      "epoch": 0.221059569910192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8146056583404505e-05,
+      "loss": 1.0659,
+      "step": 8049
+    },
+    {
+      "epoch": 0.2210870341380352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8145540618378118e-05,
+      "loss": 1.0105,
+      "step": 8050
+    },
+    {
+      "epoch": 0.22111449836587843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814502458890105e-05,
+      "loss": 0.9387,
+      "step": 8051
+    },
+    {
+      "epoch": 0.22114196259372168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8144508494977398e-05,
+      "loss": 0.9717,
+      "step": 8052
+    },
+    {
+      "epoch": 0.2211694268215649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814399233661124e-05,
+      "loss": 1.0004,
+      "step": 8053
+    },
+    {
+      "epoch": 0.22119689104940815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814347611380666e-05,
+      "loss": 0.9876,
+      "step": 8054
+    },
+    {
+      "epoch": 0.22122435527725137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814295982656774e-05,
+      "loss": 1.0443,
+      "step": 8055
+    },
+    {
+      "epoch": 0.22125181950509462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814244347489857e-05,
+      "loss": 0.9927,
+      "step": 8056
+    },
+    {
+      "epoch": 0.22127928373293784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8141927058803233e-05,
+      "loss": 1.0321,
+      "step": 8057
+    },
+    {
+      "epoch": 0.2213067479607811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8141410578285815e-05,
+      "loss": 0.9428,
+      "step": 8058
+    },
+    {
+      "epoch": 0.2213342121886243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8140894033350402e-05,
+      "loss": 0.9196,
+      "step": 8059
+    },
+    {
+      "epoch": 0.22136167641646756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8140377424001082e-05,
+      "loss": 0.9586,
+      "step": 8060
+    },
+    {
+      "epoch": 0.22138914064431078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8139860750241943e-05,
+      "loss": 0.8948,
+      "step": 8061
+    },
+    {
+      "epoch": 0.22141660487215403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813934401207707e-05,
+      "loss": 0.8775,
+      "step": 8062
+    },
+    {
+      "epoch": 0.22144406909999725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8138827209510557e-05,
+      "loss": 0.988,
+      "step": 8063
+    },
+    {
+      "epoch": 0.22147153332784048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813831034254649e-05,
+      "loss": 0.999,
+      "step": 8064
+    },
+    {
+      "epoch": 0.22149899755568372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813779341118896e-05,
+      "loss": 0.9179,
+      "step": 8065
+    },
+    {
+      "epoch": 0.22152646178352695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8137276415442055e-05,
+      "loss": 1.0195,
+      "step": 8066
+    },
+    {
+      "epoch": 0.2215539260113702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8136759355309867e-05,
+      "loss": 0.9155,
+      "step": 8067
+    },
+    {
+      "epoch": 0.22158139023921342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813624223079649e-05,
+      "loss": 0.9928,
+      "step": 8068
+    },
+    {
+      "epoch": 0.22160885446705667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8135725041906008e-05,
+      "loss": 1.0003,
+      "step": 8069
+    },
+    {
+      "epoch": 0.2216363186948999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8135207788642522e-05,
+      "loss": 1.0753,
+      "step": 8070
+    },
+    {
+      "epoch": 0.22166378292274314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8134690471010116e-05,
+      "loss": 0.9935,
+      "step": 8071
+    },
+    {
+      "epoch": 0.22169124715058636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8134173089012893e-05,
+      "loss": 1.0288,
+      "step": 8072
+    },
+    {
+      "epoch": 0.2217187113784296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813365564265494e-05,
+      "loss": 0.971,
+      "step": 8073
+    },
+    {
+      "epoch": 0.22174617560627283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8133138131940352e-05,
+      "loss": 1.1163,
+      "step": 8074
+    },
+    {
+      "epoch": 0.22177363983411608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8132620556873223e-05,
+      "loss": 1.0392,
+      "step": 8075
+    },
+    {
+      "epoch": 0.2218011040619593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8132102917457656e-05,
+      "loss": 0.9933,
+      "step": 8076
+    },
+    {
+      "epoch": 0.22182856828980252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8131585213697732e-05,
+      "loss": 1.0039,
+      "step": 8077
+    },
+    {
+      "epoch": 0.22185603251764577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8131067445597562e-05,
+      "loss": 0.9599,
+      "step": 8078
+    },
+    {
+      "epoch": 0.221883496745489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8130549613161238e-05,
+      "loss": 1.0147,
+      "step": 8079
+    },
+    {
+      "epoch": 0.22191096097333224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8130031716392853e-05,
+      "loss": 0.922,
+      "step": 8080
+    },
+    {
+      "epoch": 0.22193842520117546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8129513755296506e-05,
+      "loss": 1.0204,
+      "step": 8081
+    },
+    {
+      "epoch": 0.2219658894290187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81289957298763e-05,
+      "loss": 1.0087,
+      "step": 8082
+    },
+    {
+      "epoch": 0.22199335365686193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8128477640136332e-05,
+      "loss": 1.084,
+      "step": 8083
+    },
+    {
+      "epoch": 0.22202081788470518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8127959486080697e-05,
+      "loss": 1.0137,
+      "step": 8084
+    },
+    {
+      "epoch": 0.2220482821125484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81274412677135e-05,
+      "loss": 0.946,
+      "step": 8085
+    },
+    {
+      "epoch": 0.22207574634039165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8126922985038837e-05,
+      "loss": 1.0308,
+      "step": 8086
+    },
+    {
+      "epoch": 0.22210321056823487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8126404638060813e-05,
+      "loss": 0.9885,
+      "step": 8087
+    },
+    {
+      "epoch": 0.22213067479607812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8125886226783528e-05,
+      "loss": 1.1065,
+      "step": 8088
+    },
+    {
+      "epoch": 0.22215813902392134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8125367751211082e-05,
+      "loss": 0.8257,
+      "step": 8089
+    },
+    {
+      "epoch": 0.22218560325176456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812484921134758e-05,
+      "loss": 0.9704,
+      "step": 8090
+    },
+    {
+      "epoch": 0.2222130674796078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8124330607197124e-05,
+      "loss": 0.9609,
+      "step": 8091
+    },
+    {
+      "epoch": 0.22224053170745103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8123811938763814e-05,
+      "loss": 1.0535,
+      "step": 8092
+    },
+    {
+      "epoch": 0.22226799593529428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8123293206051763e-05,
+      "loss": 1.0695,
+      "step": 8093
+    },
+    {
+      "epoch": 0.2222954601631375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8122774409065066e-05,
+      "loss": 0.997,
+      "step": 8094
+    },
+    {
+      "epoch": 0.22232292439098075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8122255547807833e-05,
+      "loss": 1.0071,
+      "step": 8095
+    },
+    {
+      "epoch": 0.22235038861882397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8121736622284167e-05,
+      "loss": 1.0872,
+      "step": 8096
+    },
+    {
+      "epoch": 0.22237785284666722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8121217632498172e-05,
+      "loss": 0.9585,
+      "step": 8097
+    },
+    {
+      "epoch": 0.22240531707451044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812069857845396e-05,
+      "loss": 1.0453,
+      "step": 8098
+    },
+    {
+      "epoch": 0.2224327813023537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8120179460155638e-05,
+      "loss": 1.0283,
+      "step": 8099
+    },
+    {
+      "epoch": 0.22246024553019692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8119660277607306e-05,
+      "loss": 0.9837,
+      "step": 8100
+    },
+    {
+      "epoch": 0.22248770975804016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811914103081308e-05,
+      "loss": 1.0978,
+      "step": 8101
+    },
+    {
+      "epoch": 0.22251517398588339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8118621719777062e-05,
+      "loss": 1.031,
+      "step": 8102
+    },
+    {
+      "epoch": 0.22254263821372663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8118102344503364e-05,
+      "loss": 1.0164,
+      "step": 8103
+    },
+    {
+      "epoch": 0.22257010244156986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81175829049961e-05,
+      "loss": 1.029,
+      "step": 8104
+    },
+    {
+      "epoch": 0.22259756666941308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811706340125937e-05,
+      "loss": 1.0343,
+      "step": 8105
+    },
+    {
+      "epoch": 0.22262503089725633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8116543833297294e-05,
+      "loss": 0.9523,
+      "step": 8106
+    },
+    {
+      "epoch": 0.22265249512509955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8116024201113977e-05,
+      "loss": 1.0172,
+      "step": 8107
+    },
+    {
+      "epoch": 0.2226799593529428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8115504504713534e-05,
+      "loss": 1.0966,
+      "step": 8108
+    },
+    {
+      "epoch": 0.22270742358078602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811498474410008e-05,
+      "loss": 1.0186,
+      "step": 8109
+    },
+    {
+      "epoch": 0.22273488780862927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8114464919277714e-05,
+      "loss": 1.0338,
+      "step": 8110
+    },
+    {
+      "epoch": 0.2227623520364725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8113945030250565e-05,
+      "loss": 0.9831,
+      "step": 8111
+    },
+    {
+      "epoch": 0.22278981626431574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8113425077022736e-05,
+      "loss": 0.959,
+      "step": 8112
+    },
+    {
+      "epoch": 0.22281728049215896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8112905059598347e-05,
+      "loss": 0.9652,
+      "step": 8113
+    },
+    {
+      "epoch": 0.2228447447200022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811238497798151e-05,
+      "loss": 0.9054,
+      "step": 8114
+    },
+    {
+      "epoch": 0.22287220894784543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8111864832176342e-05,
+      "loss": 1.0558,
+      "step": 8115
+    },
+    {
+      "epoch": 0.22289967317568868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8111344622186955e-05,
+      "loss": 1.0457,
+      "step": 8116
+    },
+    {
+      "epoch": 0.2229271374035319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8110824348017468e-05,
+      "loss": 1.0627,
+      "step": 8117
+    },
+    {
+      "epoch": 0.22295460163137512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8110304009671997e-05,
+      "loss": 1.0069,
+      "step": 8118
+    },
+    {
+      "epoch": 0.22298206585921837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810978360715466e-05,
+      "loss": 1.0175,
+      "step": 8119
+    },
+    {
+      "epoch": 0.2230095300870616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8109263140469574e-05,
+      "loss": 0.9413,
+      "step": 8120
+    },
+    {
+      "epoch": 0.22303699431490484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8108742609620854e-05,
+      "loss": 0.9586,
+      "step": 8121
+    },
+    {
+      "epoch": 0.22306445854274806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8108222014612623e-05,
+      "loss": 1.1967,
+      "step": 8122
+    },
+    {
+      "epoch": 0.2230919227705913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8107701355449e-05,
+      "loss": 0.9314,
+      "step": 8123
+    },
+    {
+      "epoch": 0.22311938699843453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8107180632134102e-05,
+      "loss": 0.9602,
+      "step": 8124
+    },
+    {
+      "epoch": 0.22314685122627778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810665984467205e-05,
+      "loss": 0.9477,
+      "step": 8125
+    },
+    {
+      "epoch": 0.223174315454121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8106138993066967e-05,
+      "loss": 1.0652,
+      "step": 8126
+    },
+    {
+      "epoch": 0.22320177968196425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8105618077322972e-05,
+      "loss": 1.0967,
+      "step": 8127
+    },
+    {
+      "epoch": 0.22322924390980747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8105097097444184e-05,
+      "loss": 0.9951,
+      "step": 8128
+    },
+    {
+      "epoch": 0.22325670813765072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810457605343473e-05,
+      "loss": 0.9724,
+      "step": 8129
+    },
+    {
+      "epoch": 0.22328417236549394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810405494529873e-05,
+      "loss": 1.0354,
+      "step": 8130
+    },
+    {
+      "epoch": 0.22331163659333716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810353377304031e-05,
+      "loss": 0.9902,
+      "step": 8131
+    },
+    {
+      "epoch": 0.22333910082118041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810301253666359e-05,
+      "loss": 1.0339,
+      "step": 8132
+    },
+    {
+      "epoch": 0.22336656504902364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8102491236172698e-05,
+      "loss": 1.047,
+      "step": 8133
+    },
+    {
+      "epoch": 0.22339402927686688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8101969871571755e-05,
+      "loss": 0.9479,
+      "step": 8134
+    },
+    {
+      "epoch": 0.2234214935047101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810144844286489e-05,
+      "loss": 1.0085,
+      "step": 8135
+    },
+    {
+      "epoch": 0.22344895773255335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8100926950056222e-05,
+      "loss": 1.0534,
+      "step": 8136
+    },
+    {
+      "epoch": 0.22347642196039658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8100405393149886e-05,
+      "loss": 0.9847,
+      "step": 8137
+    },
+    {
+      "epoch": 0.22350388618823983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8099883772150003e-05,
+      "loss": 1.0521,
+      "step": 8138
+    },
+    {
+      "epoch": 0.22353135041608305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8099362087060704e-05,
+      "loss": 0.9519,
+      "step": 8139
+    },
+    {
+      "epoch": 0.2235588146439263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8098840337886115e-05,
+      "loss": 1.0316,
+      "step": 8140
+    },
+    {
+      "epoch": 0.22358627887176952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8098318524630363e-05,
+      "loss": 0.9643,
+      "step": 8141
+    },
+    {
+      "epoch": 0.22361374309961277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8097796647297578e-05,
+      "loss": 0.9674,
+      "step": 8142
+    },
+    {
+      "epoch": 0.223641207327456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8097274705891887e-05,
+      "loss": 1.0216,
+      "step": 8143
+    },
+    {
+      "epoch": 0.22366867155529924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8096752700417424e-05,
+      "loss": 1.0251,
+      "step": 8144
+    },
+    {
+      "epoch": 0.22369613578314246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8096230630878315e-05,
+      "loss": 1.0031,
+      "step": 8145
+    },
+    {
+      "epoch": 0.22372360001098568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8095708497278693e-05,
+      "loss": 0.9933,
+      "step": 8146
+    },
+    {
+      "epoch": 0.22375106423882893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8095186299622693e-05,
+      "loss": 1.0203,
+      "step": 8147
+    },
+    {
+      "epoch": 0.22377852846667215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809466403791444e-05,
+      "loss": 0.9693,
+      "step": 8148
+    },
+    {
+      "epoch": 0.2238059926945154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8094141712158074e-05,
+      "loss": 1.0097,
+      "step": 8149
+    },
+    {
+      "epoch": 0.22383345692235862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8093619322357722e-05,
+      "loss": 0.923,
+      "step": 8150
+    },
+    {
+      "epoch": 0.22386092115020187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8093096868517515e-05,
+      "loss": 0.956,
+      "step": 8151
+    },
+    {
+      "epoch": 0.2238883853780451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8092574350641593e-05,
+      "loss": 1.0056,
+      "step": 8152
+    },
+    {
+      "epoch": 0.22391584960588834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809205176873409e-05,
+      "loss": 1.0844,
+      "step": 8153
+    },
+    {
+      "epoch": 0.22394331383373156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8091529122799137e-05,
+      "loss": 0.9386,
+      "step": 8154
+    },
+    {
+      "epoch": 0.2239707780615748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8091006412840867e-05,
+      "loss": 1.0755,
+      "step": 8155
+    },
+    {
+      "epoch": 0.22399824228941803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809048363886343e-05,
+      "loss": 0.9445,
+      "step": 8156
+    },
+    {
+      "epoch": 0.22402570651726128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808996080087094e-05,
+      "loss": 1.0097,
+      "step": 8157
+    },
+    {
+      "epoch": 0.2240531707451045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8089437898867553e-05,
+      "loss": 1.005,
+      "step": 8158
+    },
+    {
+      "epoch": 0.22408063497294772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8088914932857402e-05,
+      "loss": 1.0145,
+      "step": 8159
+    },
+    {
+      "epoch": 0.22410809920079097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808839190284462e-05,
+      "loss": 0.9966,
+      "step": 8160
+    },
+    {
+      "epoch": 0.2241355634286342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808786880883335e-05,
+      "loss": 0.942,
+      "step": 8161
+    },
+    {
+      "epoch": 0.22416302765647744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8087345650827725e-05,
+      "loss": 0.9949,
+      "step": 8162
+    },
+    {
+      "epoch": 0.22419049188432066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8086822428831895e-05,
+      "loss": 0.9157,
+      "step": 8163
+    },
+    {
+      "epoch": 0.2242179561121639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808629914284999e-05,
+      "loss": 1.0245,
+      "step": 8164
+    },
+    {
+      "epoch": 0.22424542034000713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808577579288615e-05,
+      "loss": 0.9849,
+      "step": 8165
+    },
+    {
+      "epoch": 0.22427288456785038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8085252378944522e-05,
+      "loss": 1.0771,
+      "step": 8166
+    },
+    {
+      "epoch": 0.2243003487956936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8084728901029248e-05,
+      "loss": 0.9475,
+      "step": 8167
+    },
+    {
+      "epoch": 0.22432781302353685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8084205359144466e-05,
+      "loss": 1.0607,
+      "step": 8168
+    },
+    {
+      "epoch": 0.22435527725138008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8083681753294317e-05,
+      "loss": 1.0407,
+      "step": 8169
+    },
+    {
+      "epoch": 0.22438274147922332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808315808348295e-05,
+      "loss": 1.0191,
+      "step": 8170
+    },
+    {
+      "epoch": 0.22441020570706655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80826343497145e-05,
+      "loss": 1.0019,
+      "step": 8171
+    },
+    {
+      "epoch": 0.22443766993490977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8082110551993125e-05,
+      "loss": 1.0835,
+      "step": 8172
+    },
+    {
+      "epoch": 0.22446513416275302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8081586690322953e-05,
+      "loss": 1.0838,
+      "step": 8173
+    },
+    {
+      "epoch": 0.22449259839059624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8081062764708143e-05,
+      "loss": 0.9499,
+      "step": 8174
+    },
+    {
+      "epoch": 0.2245200626184395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8080538775152828e-05,
+      "loss": 0.8902,
+      "step": 8175
+    },
+    {
+      "epoch": 0.2245475268462827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8080014721661163e-05,
+      "loss": 1.0475,
+      "step": 8176
+    },
+    {
+      "epoch": 0.22457499107412596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807949060423729e-05,
+      "loss": 0.9766,
+      "step": 8177
+    },
+    {
+      "epoch": 0.22460245530196918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807896642288536e-05,
+      "loss": 1.0315,
+      "step": 8178
+    },
+    {
+      "epoch": 0.22462991952981243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8078442177609515e-05,
+      "loss": 1.0154,
+      "step": 8179
+    },
+    {
+      "epoch": 0.22465738375765565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807791786841391e-05,
+      "loss": 1.0128,
+      "step": 8180
+    },
+    {
+      "epoch": 0.2246848479854989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8077393495302687e-05,
+      "loss": 1.0142,
+      "step": 8181
+    },
+    {
+      "epoch": 0.22471231221334212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807686905828e-05,
+      "loss": 1.1123,
+      "step": 8182
+    },
+    {
+      "epoch": 0.22473977644118537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8076344557349995e-05,
+      "loss": 0.8495,
+      "step": 8183
+    },
+    {
+      "epoch": 0.2247672406690286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8075819992516822e-05,
+      "loss": 0.9935,
+      "step": 8184
+    },
+    {
+      "epoch": 0.2247947048968718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8075295363784634e-05,
+      "loss": 1.106,
+      "step": 8185
+    },
+    {
+      "epoch": 0.22482216912471506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8074770671157584e-05,
+      "loss": 0.9477,
+      "step": 8186
+    },
+    {
+      "epoch": 0.22484963335255828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8074245914639817e-05,
+      "loss": 1.0428,
+      "step": 8187
+    },
+    {
+      "epoch": 0.22487709758040153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807372109423549e-05,
+      "loss": 1.0064,
+      "step": 8188
+    },
+    {
+      "epoch": 0.22490456180824475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8073196209948755e-05,
+      "loss": 0.9557,
+      "step": 8189
+    },
+    {
+      "epoch": 0.224932026036088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8072671261783763e-05,
+      "loss": 1.0276,
+      "step": 8190
+    },
+    {
+      "epoch": 0.22495949026393122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8072146249744668e-05,
+      "loss": 0.9048,
+      "step": 8191
+    },
+    {
+      "epoch": 0.22498695449177447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8071621173835625e-05,
+      "loss": 1.1149,
+      "step": 8192
+    },
+    {
+      "epoch": 0.2250144187196177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807109603406079e-05,
+      "loss": 1.0141,
+      "step": 8193
+    },
+    {
+      "epoch": 0.22504188294746094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8070570830424316e-05,
+      "loss": 0.9891,
+      "step": 8194
+    },
+    {
+      "epoch": 0.22506934717530416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807004556293036e-05,
+      "loss": 0.9453,
+      "step": 8195
+    },
+    {
+      "epoch": 0.2250968114031474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806952023158308e-05,
+      "loss": 0.9543,
+      "step": 8196
+    },
+    {
+      "epoch": 0.22512427563099063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8068994836386627e-05,
+      "loss": 0.9213,
+      "step": 8197
+    },
+    {
+      "epoch": 0.22515173985883388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8068469377345162e-05,
+      "loss": 0.9968,
+      "step": 8198
+    },
+    {
+      "epoch": 0.2251792040866771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806794385446284e-05,
+      "loss": 0.9869,
+      "step": 8199
+    },
+    {
+      "epoch": 0.22520666831452033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8067418267743823e-05,
+      "loss": 0.8938,
+      "step": 8200
+    },
+    {
+      "epoch": 0.22523413254236357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8066892617192264e-05,
+      "loss": 1.0074,
+      "step": 8201
+    },
+    {
+      "epoch": 0.2252615967702068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806636690281233e-05,
+      "loss": 1.0082,
+      "step": 8202
+    },
+    {
+      "epoch": 0.22528906099805004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8065841124608177e-05,
+      "loss": 0.9658,
+      "step": 8203
+    },
+    {
+      "epoch": 0.22531652522589327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8065315282583963e-05,
+      "loss": 0.9991,
+      "step": 8204
+    },
+    {
+      "epoch": 0.22534398945373652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806478937674385e-05,
+      "loss": 1.0021,
+      "step": 8205
+    },
+    {
+      "epoch": 0.22537145368157974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8064263407092e-05,
+      "loss": 0.9771,
+      "step": 8206
+    },
+    {
+      "epoch": 0.22539891790942299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806373737363257e-05,
+      "loss": 0.9618,
+      "step": 8207
+    },
+    {
+      "epoch": 0.2254263821372662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806321127636973e-05,
+      "loss": 0.9863,
+      "step": 8208
+    },
+    {
+      "epoch": 0.22545384636510946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8062685115307643e-05,
+      "loss": 1.0222,
+      "step": 8209
+    },
+    {
+      "epoch": 0.22548131059295268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8062158890450467e-05,
+      "loss": 0.9872,
+      "step": 8210
+    },
+    {
+      "epoch": 0.22550877482079593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806163260180236e-05,
+      "loss": 1.0255,
+      "step": 8211
+    },
+    {
+      "epoch": 0.22553623904863915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8061106249367498e-05,
+      "loss": 0.9611,
+      "step": 8212
+    },
+    {
+      "epoch": 0.22556370327648237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806057983315004e-05,
+      "loss": 0.9989,
+      "step": 8213
+    },
+    {
+      "epoch": 0.22559116750432562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8060053353154153e-05,
+      "loss": 1.1001,
+      "step": 8214
+    },
+    {
+      "epoch": 0.22561863173216884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8059526809384e-05,
+      "loss": 0.9973,
+      "step": 8215
+    },
+    {
+      "epoch": 0.2256460959600121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805900020184375e-05,
+      "loss": 1.0432,
+      "step": 8216
+    },
+    {
+      "epoch": 0.2256735601878553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8058473530537568e-05,
+      "loss": 0.9962,
+      "step": 8217
+    },
+    {
+      "epoch": 0.22570102441569856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8057946795469625e-05,
+      "loss": 1.0408,
+      "step": 8218
+    },
+    {
+      "epoch": 0.22572848864354178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805741999664408e-05,
+      "loss": 1.0483,
+      "step": 8219
+    },
+    {
+      "epoch": 0.22575595287138503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805689313406511e-05,
+      "loss": 0.9534,
+      "step": 8220
+    },
+    {
+      "epoch": 0.22578341709922825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805636620773688e-05,
+      "loss": 1.0249,
+      "step": 8221
+    },
+    {
+      "epoch": 0.2258108813270715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805583921766356e-05,
+      "loss": 1.0648,
+      "step": 8222
+    },
+    {
+      "epoch": 0.22583834555491472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805531216384932e-05,
+      "loss": 0.9529,
+      "step": 8223
+    },
+    {
+      "epoch": 0.22586580978275797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805478504629833e-05,
+      "loss": 0.9986,
+      "step": 8224
+    },
+    {
+      "epoch": 0.2258932740106012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805425786501476e-05,
+      "loss": 0.9786,
+      "step": 8225
+    },
+    {
+      "epoch": 0.2259207382384444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8053730620002783e-05,
+      "loss": 0.9953,
+      "step": 8226
+    },
+    {
+      "epoch": 0.22594820246628766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8053203311266566e-05,
+      "loss": 0.9867,
+      "step": 8227
+    },
+    {
+      "epoch": 0.22597566669413088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8052675938810288e-05,
+      "loss": 1.0469,
+      "step": 8228
+    },
+    {
+      "epoch": 0.22600313092197413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8052148502638117e-05,
+      "loss": 0.9731,
+      "step": 8229
+    },
+    {
+      "epoch": 0.22603059514981735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805162100275423e-05,
+      "loss": 0.9395,
+      "step": 8230
+    },
+    {
+      "epoch": 0.2260580593776606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80510934391628e-05,
+      "loss": 0.9383,
+      "step": 8231
+    },
+    {
+      "epoch": 0.22608552360550382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8050565811867996e-05,
+      "loss": 0.9649,
+      "step": 8232
+    },
+    {
+      "epoch": 0.22611298783334707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8050038120874e-05,
+      "loss": 1.0427,
+      "step": 8233
+    },
+    {
+      "epoch": 0.2261404520611903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8049510366184982e-05,
+      "loss": 1.0944,
+      "step": 8234
+    },
+    {
+      "epoch": 0.22616791628903354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804898254780512e-05,
+      "loss": 1.0644,
+      "step": 8235
+    },
+    {
+      "epoch": 0.22619538051687677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8048454665738592e-05,
+      "loss": 1.0387,
+      "step": 8236
+    },
+    {
+      "epoch": 0.22622284474472001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804792671998957e-05,
+      "loss": 0.993,
+      "step": 8237
+    },
+    {
+      "epoch": 0.22625030897256324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8047398710562235e-05,
+      "loss": 1.0175,
+      "step": 8238
+    },
+    {
+      "epoch": 0.22627777320040646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8046870637460767e-05,
+      "loss": 0.9684,
+      "step": 8239
+    },
+    {
+      "epoch": 0.2263052374282497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804634250068934e-05,
+      "loss": 1.0354,
+      "step": 8240
+    },
+    {
+      "epoch": 0.22633270165609293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804581430025213e-05,
+      "loss": 0.9131,
+      "step": 8241
+    },
+    {
+      "epoch": 0.22636016588393618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8045286036153328e-05,
+      "loss": 1.024,
+      "step": 8242
+    },
+    {
+      "epoch": 0.2263876301117794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8044757708397103e-05,
+      "loss": 1.1326,
+      "step": 8243
+    },
+    {
+      "epoch": 0.22641509433962265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804422931698764e-05,
+      "loss": 0.8382,
+      "step": 8244
+    },
+    {
+      "epoch": 0.22644255856746587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8043700861929117e-05,
+      "loss": 0.9967,
+      "step": 8245
+    },
+    {
+      "epoch": 0.22647002279530912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8043172343225716e-05,
+      "loss": 0.9287,
+      "step": 8246
+    },
+    {
+      "epoch": 0.22649748702315234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804264376088162e-05,
+      "loss": 1.0602,
+      "step": 8247
+    },
+    {
+      "epoch": 0.2265249512509956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042115114901014e-05,
+      "loss": 1.0128,
+      "step": 8248
+    },
+    {
+      "epoch": 0.2265524154788388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8041586405288076e-05,
+      "loss": 0.9818,
+      "step": 8249
+    },
+    {
+      "epoch": 0.22657987970668206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804105763204699e-05,
+      "loss": 0.9898,
+      "step": 8250
+    },
+    {
+      "epoch": 0.22660734393452528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8040528795181945e-05,
+      "loss": 0.9581,
+      "step": 8251
+    },
+    {
+      "epoch": 0.22663480816236853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8039999894697118e-05,
+      "loss": 1.0178,
+      "step": 8252
+    },
+    {
+      "epoch": 0.22666227239021175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80394709305967e-05,
+      "loss": 0.9343,
+      "step": 8253
+    },
+    {
+      "epoch": 0.22668973661805497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8038941902884873e-05,
+      "loss": 1.0348,
+      "step": 8254
+    },
+    {
+      "epoch": 0.22671720084589822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803841281156582e-05,
+      "loss": 1.0111,
+      "step": 8255
+    },
+    {
+      "epoch": 0.22674466507374144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8037883656643735e-05,
+      "loss": 0.9612,
+      "step": 8256
+    },
+    {
+      "epoch": 0.2267721293015847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80373544381228e-05,
+      "loss": 0.9398,
+      "step": 8257
+    },
+    {
+      "epoch": 0.2267995935294279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8036825156007207e-05,
+      "loss": 0.9998,
+      "step": 8258
+    },
+    {
+      "epoch": 0.22682705775727116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8036295810301135e-05,
+      "loss": 0.9108,
+      "step": 8259
+    },
+    {
+      "epoch": 0.22685452198511438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803576640100878e-05,
+      "loss": 1.0665,
+      "step": 8260
+    },
+    {
+      "epoch": 0.22688198621295763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8035236928134327e-05,
+      "loss": 0.9957,
+      "step": 8261
+    },
+    {
+      "epoch": 0.22690945044080085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8034707391681968e-05,
+      "loss": 0.9272,
+      "step": 8262
+    },
+    {
+      "epoch": 0.2269369146686441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8034177791655893e-05,
+      "loss": 1.0901,
+      "step": 8263
+    },
+    {
+      "epoch": 0.22696437889648732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803364812806029e-05,
+      "loss": 1.0455,
+      "step": 8264
+    },
+    {
+      "epoch": 0.22699184312433057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803311840089935e-05,
+      "loss": 0.9363,
+      "step": 8265
+    },
+    {
+      "epoch": 0.2270193073521738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8032588610177268e-05,
+      "loss": 1.0478,
+      "step": 8266
+    },
+    {
+      "epoch": 0.22704677158001701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8032058755898228e-05,
+      "loss": 0.9797,
+      "step": 8267
+    },
+    {
+      "epoch": 0.22707423580786026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803152883806643e-05,
+      "loss": 1.0758,
+      "step": 8268
+    },
+    {
+      "epoch": 0.22710170003570349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803099885668607e-05,
+      "loss": 0.9486,
+      "step": 8269
+    },
+    {
+      "epoch": 0.22712916426354673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8030468811761332e-05,
+      "loss": 1.1194,
+      "step": 8270
+    },
+    {
+      "epoch": 0.22715662849138996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8029938703296412e-05,
+      "loss": 1.0446,
+      "step": 8271
+    },
+    {
+      "epoch": 0.2271840927192332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802940853129551e-05,
+      "loss": 0.924,
+      "step": 8272
+    },
+    {
+      "epoch": 0.22721155694707643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8028878295762817e-05,
+      "loss": 0.9115,
+      "step": 8273
+    },
+    {
+      "epoch": 0.22723902117491968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8028347996702527e-05,
+      "loss": 0.9722,
+      "step": 8274
+    },
+    {
+      "epoch": 0.2272664854027629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802781763411884e-05,
+      "loss": 1.0521,
+      "step": 8275
+    },
+    {
+      "epoch": 0.22729394963060615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8027287208015947e-05,
+      "loss": 0.9935,
+      "step": 8276
+    },
+    {
+      "epoch": 0.22732141385844937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802675671839805e-05,
+      "loss": 1.0196,
+      "step": 8277
+    },
+    {
+      "epoch": 0.22734887808629262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8026226165269347e-05,
+      "loss": 0.9356,
+      "step": 8278
+    },
+    {
+      "epoch": 0.22737634231413584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802569554863403e-05,
+      "loss": 0.9515,
+      "step": 8279
+    },
+    {
+      "epoch": 0.22740380654197906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8025164868496304e-05,
+      "loss": 0.9965,
+      "step": 8280
+    },
+    {
+      "epoch": 0.2274312707698223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802463412486036e-05,
+      "loss": 1.0807,
+      "step": 8281
+    },
+    {
+      "epoch": 0.22745873499766553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8024103317730408e-05,
+      "loss": 0.9749,
+      "step": 8282
+    },
+    {
+      "epoch": 0.22748619922550878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802357244711064e-05,
+      "loss": 1.069,
+      "step": 8283
+    },
+    {
+      "epoch": 0.227513663453352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8023041513005256e-05,
+      "loss": 1.0232,
+      "step": 8284
+    },
+    {
+      "epoch": 0.22754112768119525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8022510515418464e-05,
+      "loss": 0.9813,
+      "step": 8285
+    },
+    {
+      "epoch": 0.22756859190903847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802197945435446e-05,
+      "loss": 1.0995,
+      "step": 8286
+    },
+    {
+      "epoch": 0.22759605613688172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8021448329817443e-05,
+      "loss": 0.9711,
+      "step": 8287
+    },
+    {
+      "epoch": 0.22762352036472494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8020917141811627e-05,
+      "loss": 1.0538,
+      "step": 8288
+    },
+    {
+      "epoch": 0.2276509845925682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8020385890341203e-05,
+      "loss": 0.9402,
+      "step": 8289
+    },
+    {
+      "epoch": 0.2276784488204114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8019854575410377e-05,
+      "loss": 0.9465,
+      "step": 8290
+    },
+    {
+      "epoch": 0.22770591304825466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801932319702336e-05,
+      "loss": 1.0077,
+      "step": 8291
+    },
+    {
+      "epoch": 0.22773337727609788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801879175518435e-05,
+      "loss": 1.0787,
+      "step": 8292
+    },
+    {
+      "epoch": 0.22776084150394113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801826024989755e-05,
+      "loss": 1.0033,
+      "step": 8293
+    },
+    {
+      "epoch": 0.22778830573178435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801772868116717e-05,
+      "loss": 1.0184,
+      "step": 8294
+    },
+    {
+      "epoch": 0.22781576995962757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801719704899742e-05,
+      "loss": 1.0282,
+      "step": 8295
+    },
+    {
+      "epoch": 0.22784323418747082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80166653533925e-05,
+      "loss": 1.0219,
+      "step": 8296
+    },
+    {
+      "epoch": 0.22787069841531404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8016133594356614e-05,
+      "loss": 1.059,
+      "step": 8297
+    },
+    {
+      "epoch": 0.2278981626431573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015601771893976e-05,
+      "loss": 1.0948,
+      "step": 8298
+    },
+    {
+      "epoch": 0.22792562687100051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015069886008794e-05,
+      "loss": 0.9734,
+      "step": 8299
+    },
+    {
+      "epoch": 0.22795309109884376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801453793670527e-05,
+      "loss": 1.0096,
+      "step": 8300
+    },
+    {
+      "epoch": 0.22798055532668698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801400592398762e-05,
+      "loss": 1.0124,
+      "step": 8301
+    },
+    {
+      "epoch": 0.22800801955453023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801347384786005e-05,
+      "loss": 1.1072,
+      "step": 8302
+    },
+    {
+      "epoch": 0.22803548378237345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801294170832677e-05,
+      "loss": 1.0497,
+      "step": 8303
+    },
+    {
+      "epoch": 0.2280629480102167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8012409505391995e-05,
+      "loss": 0.9855,
+      "step": 8304
+    },
+    {
+      "epoch": 0.22809041223805993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8011877239059928e-05,
+      "loss": 0.9576,
+      "step": 8305
+    },
+    {
+      "epoch": 0.22811787646590317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8011344909334787e-05,
+      "loss": 1.0766,
+      "step": 8306
+    },
+    {
+      "epoch": 0.2281453406937464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801081251622078e-05,
+      "loss": 1.0194,
+      "step": 8307
+    },
+    {
+      "epoch": 0.22817280492158962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8010280059722123e-05,
+      "loss": 0.9744,
+      "step": 8308
+    },
+    {
+      "epoch": 0.22820026914943287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800974753984303e-05,
+      "loss": 0.9276,
+      "step": 8309
+    },
+    {
+      "epoch": 0.2282277333772761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8009214956587706e-05,
+      "loss": 1.0131,
+      "step": 8310
+    },
+    {
+      "epoch": 0.22825519760511934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8008682309960374e-05,
+      "loss": 0.9442,
+      "step": 8311
+    },
+    {
+      "epoch": 0.22828266183296256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8008149599965246e-05,
+      "loss": 1.0442,
+      "step": 8312
+    },
+    {
+      "epoch": 0.2283101260608058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8007616826606535e-05,
+      "loss": 1.1183,
+      "step": 8313
+    },
+    {
+      "epoch": 0.22833759028864903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8007083989888458e-05,
+      "loss": 0.9367,
+      "step": 8314
+    },
+    {
+      "epoch": 0.22836505451649228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800655108981523e-05,
+      "loss": 1.0039,
+      "step": 8315
+    },
+    {
+      "epoch": 0.2283925187443355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8006018126391067e-05,
+      "loss": 0.9589,
+      "step": 8316
+    },
+    {
+      "epoch": 0.22841998297217875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800548509962019e-05,
+      "loss": 1.0424,
+      "step": 8317
+    },
+    {
+      "epoch": 0.22844744720002197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8004952009506813e-05,
+      "loss": 1.0256,
+      "step": 8318
+    },
+    {
+      "epoch": 0.22847491142786522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8004418856055155e-05,
+      "loss": 0.8459,
+      "step": 8319
+    },
+    {
+      "epoch": 0.22850237565570844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8003885639269438e-05,
+      "loss": 1.0271,
+      "step": 8320
+    },
+    {
+      "epoch": 0.22852983988355166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8003352359153872e-05,
+      "loss": 0.9655,
+      "step": 8321
+    },
+    {
+      "epoch": 0.2285573041113949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8002819015712683e-05,
+      "loss": 0.9658,
+      "step": 8322
+    },
+    {
+      "epoch": 0.22858476833923813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800228560895009e-05,
+      "loss": 1.0464,
+      "step": 8323
+    },
+    {
+      "epoch": 0.22861223256708138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8001752138870316e-05,
+      "loss": 0.9977,
+      "step": 8324
+    },
+    {
+      "epoch": 0.2286396967949246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8001218605477578e-05,
+      "loss": 0.9195,
+      "step": 8325
+    },
+    {
+      "epoch": 0.22866716102276785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80006850087761e-05,
+      "loss": 1.0444,
+      "step": 8326
+    },
+    {
+      "epoch": 0.22869462525061107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80001513487701e-05,
+      "loss": 0.9845,
+      "step": 8327
+    },
+    {
+      "epoch": 0.22872208947845432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7999617625463805e-05,
+      "loss": 0.9044,
+      "step": 8328
+    },
+    {
+      "epoch": 0.22874955370629754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7999083838861436e-05,
+      "loss": 1.0262,
+      "step": 8329
+    },
+    {
+      "epoch": 0.2287770179341408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799854998896722e-05,
+      "loss": 1.1327,
+      "step": 8330
+    },
+    {
+      "epoch": 0.228804482161984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7998016075785373e-05,
+      "loss": 0.9666,
+      "step": 8331
+    },
+    {
+      "epoch": 0.22883194638982726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7997482099320128e-05,
+      "loss": 0.9715,
+      "step": 8332
+    },
+    {
+      "epoch": 0.22885941061767048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799694805957571e-05,
+      "loss": 0.9799,
+      "step": 8333
+    },
+    {
+      "epoch": 0.2288868748455137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7996413956556335e-05,
+      "loss": 1.0803,
+      "step": 8334
+    },
+    {
+      "epoch": 0.22891433907335695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799587979026624e-05,
+      "loss": 0.9554,
+      "step": 8335
+    },
+    {
+      "epoch": 0.22894180330120018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7995345560709645e-05,
+      "loss": 0.9241,
+      "step": 8336
+    },
+    {
+      "epoch": 0.22896926752904342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7994811267890778e-05,
+      "loss": 1.0627,
+      "step": 8337
+    },
+    {
+      "epoch": 0.22899673175688665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799427691181387e-05,
+      "loss": 0.9791,
+      "step": 8338
+    },
+    {
+      "epoch": 0.2290241959847299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7993742492483145e-05,
+      "loss": 0.9383,
+      "step": 8339
+    },
+    {
+      "epoch": 0.22905166021257312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799320800990283e-05,
+      "loss": 1.0258,
+      "step": 8340
+    },
+    {
+      "epoch": 0.22907912444041637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799267346407716e-05,
+      "loss": 1.053,
+      "step": 8341
+    },
+    {
+      "epoch": 0.2291065886682596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7992138855010362e-05,
+      "loss": 0.9527,
+      "step": 8342
+    },
+    {
+      "epoch": 0.22913405289610284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7991604182706663e-05,
+      "loss": 1.0305,
+      "step": 8343
+    },
+    {
+      "epoch": 0.22916151712394606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79910694471703e-05,
+      "loss": 0.9981,
+      "step": 8344
+    },
+    {
+      "epoch": 0.2291889813517893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7990534648405496e-05,
+      "loss": 1.0922,
+      "step": 8345
+    },
+    {
+      "epoch": 0.22921644557963253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7989999786416487e-05,
+      "loss": 1.0142,
+      "step": 8346
+    },
+    {
+      "epoch": 0.22924390980747578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7989464861207508e-05,
+      "loss": 1.0777,
+      "step": 8347
+    },
+    {
+      "epoch": 0.229271374035319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7988929872782782e-05,
+      "loss": 1.0154,
+      "step": 8348
+    },
+    {
+      "epoch": 0.22929883826316222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7988394821146556e-05,
+      "loss": 1.0589,
+      "step": 8349
+    },
+    {
+      "epoch": 0.22932630249100547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798785970630305e-05,
+      "loss": 0.9586,
+      "step": 8350
+    },
+    {
+      "epoch": 0.2293537667188487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7987324528256508e-05,
+      "loss": 0.9463,
+      "step": 8351
+    },
+    {
+      "epoch": 0.22938123094669194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7986789287011156e-05,
+      "loss": 1.027,
+      "step": 8352
+    },
+    {
+      "epoch": 0.22940869517453516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7986253982571235e-05,
+      "loss": 1.0186,
+      "step": 8353
+    },
+    {
+      "epoch": 0.2294361594023784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798571861494098e-05,
+      "loss": 1.0361,
+      "step": 8354
+    },
+    {
+      "epoch": 0.22946362363022163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7985183184124626e-05,
+      "loss": 0.9069,
+      "step": 8355
+    },
+    {
+      "epoch": 0.22949108785806488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798464769012641e-05,
+      "loss": 1.046,
+      "step": 8356
+    },
+    {
+      "epoch": 0.2295185520859081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7984112132950565e-05,
+      "loss": 1.0105,
+      "step": 8357
+    },
+    {
+      "epoch": 0.22954601631375135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7983576512601337e-05,
+      "loss": 0.9343,
+      "step": 8358
+    },
+    {
+      "epoch": 0.22957348054159457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7983040829082954e-05,
+      "loss": 1.017,
+      "step": 8359
+    },
+    {
+      "epoch": 0.22960094476943782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7982505082399663e-05,
+      "loss": 0.9783,
+      "step": 8360
+    },
+    {
+      "epoch": 0.22962840899728104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79819692725557e-05,
+      "loss": 0.9837,
+      "step": 8361
+    },
+    {
+      "epoch": 0.22965587322512426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79814333995553e-05,
+      "loss": 1.0015,
+      "step": 8362
+    },
+    {
+      "epoch": 0.2296833374529675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798089746340271e-05,
+      "loss": 0.9969,
+      "step": 8363
+    },
+    {
+      "epoch": 0.22971080168081073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798036146410217e-05,
+      "loss": 1.0033,
+      "step": 8364
+    },
+    {
+      "epoch": 0.22973826590865398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7979825401657916e-05,
+      "loss": 0.9677,
+      "step": 8365
+    },
+    {
+      "epoch": 0.2297657301364972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7979289276074193e-05,
+      "loss": 1.0129,
+      "step": 8366
+    },
+    {
+      "epoch": 0.22979319436434045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7978753087355243e-05,
+      "loss": 1.0514,
+      "step": 8367
+    },
+    {
+      "epoch": 0.22982065859218367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797821683550531e-05,
+      "loss": 0.9504,
+      "step": 8368
+    },
+    {
+      "epoch": 0.22984812282002692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7977680520528632e-05,
+      "loss": 0.9758,
+      "step": 8369
+    },
+    {
+      "epoch": 0.22987558704787014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7977144142429455e-05,
+      "loss": 0.9531,
+      "step": 8370
+    },
+    {
+      "epoch": 0.2299030512757134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7976607701212023e-05,
+      "loss": 0.9539,
+      "step": 8371
+    },
+    {
+      "epoch": 0.22993051550355662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7976071196880584e-05,
+      "loss": 1.0356,
+      "step": 8372
+    },
+    {
+      "epoch": 0.22995797973139986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7975534629439376e-05,
+      "loss": 1.0717,
+      "step": 8373
+    },
+    {
+      "epoch": 0.22998544395924309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797499799889265e-05,
+      "loss": 0.9563,
+      "step": 8374
+    },
+    {
+      "epoch": 0.2300129081870863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7974461305244652e-05,
+      "loss": 0.9777,
+      "step": 8375
+    },
+    {
+      "epoch": 0.23004037241492956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7973924548499627e-05,
+      "loss": 0.9899,
+      "step": 8376
+    },
+    {
+      "epoch": 0.23006783664277278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797338772866182e-05,
+      "loss": 0.9518,
+      "step": 8377
+    },
+    {
+      "epoch": 0.23009530087061603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7972850845735484e-05,
+      "loss": 0.9646,
+      "step": 8378
+    },
+    {
+      "epoch": 0.23012276509845925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7972313899724864e-05,
+      "loss": 0.9317,
+      "step": 8379
+    },
+    {
+      "epoch": 0.2301502293263025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7971776890634206e-05,
+      "loss": 1.005,
+      "step": 8380
+    },
+    {
+      "epoch": 0.23017769355414572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797123981846776e-05,
+      "loss": 1.0045,
+      "step": 8381
+    },
+    {
+      "epoch": 0.23020515778198897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7970702683229778e-05,
+      "loss": 0.9875,
+      "step": 8382
+    },
+    {
+      "epoch": 0.2302326220098322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7970165484924513e-05,
+      "loss": 1.1575,
+      "step": 8383
+    },
+    {
+      "epoch": 0.23026008623767544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7969628223556208e-05,
+      "loss": 1.0184,
+      "step": 8384
+    },
+    {
+      "epoch": 0.23028755046551866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7969090899129116e-05,
+      "loss": 0.9845,
+      "step": 8385
+    },
+    {
+      "epoch": 0.2303150146933619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796855351164749e-05,
+      "loss": 0.9668,
+      "step": 8386
+    },
+    {
+      "epoch": 0.23034247892120513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7968016061115584e-05,
+      "loss": 1.0729,
+      "step": 8387
+    },
+    {
+      "epoch": 0.23036994314904835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7967478547537645e-05,
+      "loss": 1.033,
+      "step": 8388
+    },
+    {
+      "epoch": 0.2303974073768916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966940970917935e-05,
+      "loss": 1.045,
+      "step": 8389
+    },
+    {
+      "epoch": 0.23042487160473482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966403331260697e-05,
+      "loss": 0.881,
+      "step": 8390
+    },
+    {
+      "epoch": 0.23045233583257807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7965865628570192e-05,
+      "loss": 1.0457,
+      "step": 8391
+    },
+    {
+      "epoch": 0.2304798000604213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7965327862850674e-05,
+      "loss": 0.9466,
+      "step": 8392
+    },
+    {
+      "epoch": 0.23050726428826454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7964790034106392e-05,
+      "loss": 0.9364,
+      "step": 8393
+    },
+    {
+      "epoch": 0.23053472851610776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796425214234161e-05,
+      "loss": 1.0441,
+      "step": 8394
+    },
+    {
+      "epoch": 0.230562192743951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7963714187560576e-05,
+      "loss": 1.0439,
+      "step": 8395
+    },
+    {
+      "epoch": 0.23058965697179423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7963176169767553e-05,
+      "loss": 1.013,
+      "step": 8396
+    },
+    {
+      "epoch": 0.23061712119963748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7962638088966798e-05,
+      "loss": 0.9608,
+      "step": 8397
+    },
+    {
+      "epoch": 0.2306445854274807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7962099945162564e-05,
+      "loss": 0.9561,
+      "step": 8398
+    },
+    {
+      "epoch": 0.23067204965532395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796156173835911e-05,
+      "loss": 1.0182,
+      "step": 8399
+    },
+    {
+      "epoch": 0.23069951388316717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7961023468560696e-05,
+      "loss": 1.0063,
+      "step": 8400
+    },
+    {
+      "epoch": 0.23072697811101042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7960485135771583e-05,
+      "loss": 0.9337,
+      "step": 8401
+    },
+    {
+      "epoch": 0.23075444233885364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7959946739996023e-05,
+      "loss": 1.0175,
+      "step": 8402
+    },
+    {
+      "epoch": 0.23078190656669686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795940828123828e-05,
+      "loss": 1.0358,
+      "step": 8403
+    },
+    {
+      "epoch": 0.23080937079454011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7958869759502623e-05,
+      "loss": 0.9748,
+      "step": 8404
+    },
+    {
+      "epoch": 0.23083683502238334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79583311747933e-05,
+      "loss": 1.0117,
+      "step": 8405
+    },
+    {
+      "epoch": 0.23086429925022658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795779252711458e-05,
+      "loss": 0.9682,
+      "step": 8406
+    },
+    {
+      "epoch": 0.2308917634780698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7957253816470724e-05,
+      "loss": 0.9416,
+      "step": 8407
+    },
+    {
+      "epoch": 0.23091922770591305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795671504286599e-05,
+      "loss": 0.9561,
+      "step": 8408
+    },
+    {
+      "epoch": 0.23094669193375628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7956176206304646e-05,
+      "loss": 0.9934,
+      "step": 8409
+    },
+    {
+      "epoch": 0.23097415616159953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7955637306790955e-05,
+      "loss": 1.0059,
+      "step": 8410
+    },
+    {
+      "epoch": 0.23100162038944275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795509834432918e-05,
+      "loss": 0.8973,
+      "step": 8411
+    },
+    {
+      "epoch": 0.231029084617286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7954559318923585e-05,
+      "loss": 1.0426,
+      "step": 8412
+    },
+    {
+      "epoch": 0.23105654884512922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7954020230578434e-05,
+      "loss": 0.9976,
+      "step": 8413
+    },
+    {
+      "epoch": 0.23108401307297247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7953481079297996e-05,
+      "loss": 0.9433,
+      "step": 8414
+    },
+    {
+      "epoch": 0.2311114773008157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7952941865086534e-05,
+      "loss": 0.9669,
+      "step": 8415
+    },
+    {
+      "epoch": 0.2311389415286589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7952402587948313e-05,
+      "loss": 1.0301,
+      "step": 8416
+    },
+    {
+      "epoch": 0.23116640575650216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7951863247887606e-05,
+      "loss": 1.001,
+      "step": 8417
+    },
+    {
+      "epoch": 0.23119386998434538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7951323844908675e-05,
+      "loss": 0.9629,
+      "step": 8418
+    },
+    {
+      "epoch": 0.23122133421218863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795078437901579e-05,
+      "loss": 0.9553,
+      "step": 8419
+    },
+    {
+      "epoch": 0.23124879844003185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7950244850213217e-05,
+      "loss": 0.9401,
+      "step": 8420
+    },
+    {
+      "epoch": 0.2312762626678751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794970525850523e-05,
+      "loss": 0.9753,
+      "step": 8421
+    },
+    {
+      "epoch": 0.23130372689571832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7949165603896096e-05,
+      "loss": 0.8795,
+      "step": 8422
+    },
+    {
+      "epoch": 0.23133119112356157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794862588639008e-05,
+      "loss": 0.9877,
+      "step": 8423
+    },
+    {
+      "epoch": 0.2313586553514048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7948086105991462e-05,
+      "loss": 1.1434,
+      "step": 8424
+    },
+    {
+      "epoch": 0.23138611957924804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7947546262704504e-05,
+      "loss": 1.0506,
+      "step": 8425
+    },
+    {
+      "epoch": 0.23141358380709126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7947006356533483e-05,
+      "loss": 1.2233,
+      "step": 8426
+    },
+    {
+      "epoch": 0.2314410480349345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794646638748267e-05,
+      "loss": 1.0675,
+      "step": 8427
+    },
+    {
+      "epoch": 0.23146851226277773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794592635555634e-05,
+      "loss": 1.0137,
+      "step": 8428
+    },
+    {
+      "epoch": 0.23149597649062095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794538626075876e-05,
+      "loss": 1.0386,
+      "step": 8429
+    },
+    {
+      "epoch": 0.2315234407184642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7944846103094203e-05,
+      "loss": 1.0444,
+      "step": 8430
+    },
+    {
+      "epoch": 0.23155090494630742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794430588256695e-05,
+      "loss": 0.9001,
+      "step": 8431
+    },
+    {
+      "epoch": 0.23157836917415067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7943765599181268e-05,
+      "loss": 0.9637,
+      "step": 8432
+    },
+    {
+      "epoch": 0.2316058334019939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7943225252941436e-05,
+      "loss": 0.9884,
+      "step": 8433
+    },
+    {
+      "epoch": 0.23163329762983714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7942684843851732e-05,
+      "loss": 1.068,
+      "step": 8434
+    },
+    {
+      "epoch": 0.23166076185768036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7942144371916425e-05,
+      "loss": 0.9525,
+      "step": 8435
+    },
+    {
+      "epoch": 0.2316882260855236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7941603837139798e-05,
+      "loss": 0.9877,
+      "step": 8436
+    },
+    {
+      "epoch": 0.23171569031336683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7941063239526124e-05,
+      "loss": 1.038,
+      "step": 8437
+    },
+    {
+      "epoch": 0.23174315454121008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794052257907968e-05,
+      "loss": 1.0801,
+      "step": 8438
+    },
+    {
+      "epoch": 0.2317706187690533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7939981855804752e-05,
+      "loss": 0.9763,
+      "step": 8439
+    },
+    {
+      "epoch": 0.23179808299689655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793944106970561e-05,
+      "loss": 1.0795,
+      "step": 8440
+    },
+    {
+      "epoch": 0.23182554722473978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793890022078653e-05,
+      "loss": 0.9996,
+      "step": 8441
+    },
+    {
+      "epoch": 0.23185301145258302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7938359309051797e-05,
+      "loss": 0.9501,
+      "step": 8442
+    },
+    {
+      "epoch": 0.23188047568042625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7937818334505693e-05,
+      "loss": 0.947,
+      "step": 8443
+    },
+    {
+      "epoch": 0.23190793990826947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7937277297152493e-05,
+      "loss": 1.075,
+      "step": 8444
+    },
+    {
+      "epoch": 0.23193540413611272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793673619699648e-05,
+      "loss": 0.9276,
+      "step": 8445
+    },
+    {
+      "epoch": 0.23196286836395594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7936195034041938e-05,
+      "loss": 0.9154,
+      "step": 8446
+    },
+    {
+      "epoch": 0.2319903325917992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7935653808293146e-05,
+      "loss": 0.9981,
+      "step": 8447
+    },
+    {
+      "epoch": 0.2320177968196424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793511251975439e-05,
+      "loss": 1.032,
+      "step": 8448
+    },
+    {
+      "epoch": 0.23204526104748566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7934571168429944e-05,
+      "loss": 1.1029,
+      "step": 8449
+    },
+    {
+      "epoch": 0.23207272527532888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79340297543241e-05,
+      "loss": 0.9378,
+      "step": 8450
+    },
+    {
+      "epoch": 0.23210018950317213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7933488277441143e-05,
+      "loss": 1.0742,
+      "step": 8451
+    },
+    {
+      "epoch": 0.23212765373101535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793294673778535e-05,
+      "loss": 0.8881,
+      "step": 8452
+    },
+    {
+      "epoch": 0.2321551179588586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793240513536101e-05,
+      "loss": 0.9963,
+      "step": 8453
+    },
+    {
+      "epoch": 0.23218258218670182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7931863470172403e-05,
+      "loss": 0.9203,
+      "step": 8454
+    },
+    {
+      "epoch": 0.23221004641454507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7931321742223828e-05,
+      "loss": 1.0532,
+      "step": 8455
+    },
+    {
+      "epoch": 0.2322375106423883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7930779951519557e-05,
+      "loss": 0.8542,
+      "step": 8456
+    },
+    {
+      "epoch": 0.2322649748702315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7930238098063885e-05,
+      "loss": 1.0101,
+      "step": 8457
+    },
+    {
+      "epoch": 0.23229243909807476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79296961818611e-05,
+      "loss": 0.8125,
+      "step": 8458
+    },
+    {
+      "epoch": 0.23231990332591798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7929154202915486e-05,
+      "loss": 1.0359,
+      "step": 8459
+    },
+    {
+      "epoch": 0.23234736755376123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7928612161231328e-05,
+      "loss": 1.0441,
+      "step": 8460
+    },
+    {
+      "epoch": 0.23237483178160445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7928070056812922e-05,
+      "loss": 0.9544,
+      "step": 8461
+    },
+    {
+      "epoch": 0.2324022960094477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7927527889664557e-05,
+      "loss": 1.084,
+      "step": 8462
+    },
+    {
+      "epoch": 0.23242976023729092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792698565979052e-05,
+      "loss": 1.0342,
+      "step": 8463
+    },
+    {
+      "epoch": 0.23245722446513417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7926443367195097e-05,
+      "loss": 1.0095,
+      "step": 8464
+    },
+    {
+      "epoch": 0.2324846886929774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7925901011882588e-05,
+      "loss": 1.0092,
+      "step": 8465
+    },
+    {
+      "epoch": 0.23251215292082064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792535859385728e-05,
+      "loss": 0.9224,
+      "step": 8466
+    },
+    {
+      "epoch": 0.23253961714866386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7924816113123468e-05,
+      "loss": 0.922,
+      "step": 8467
+    },
+    {
+      "epoch": 0.2325670813765071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7924273569685432e-05,
+      "loss": 1.0833,
+      "step": 8468
+    },
+    {
+      "epoch": 0.23259454560435033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7923730963547484e-05,
+      "loss": 1.0169,
+      "step": 8469
+    },
+    {
+      "epoch": 0.23262200983219355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79231882947139e-05,
+      "loss": 0.9973,
+      "step": 8470
+    },
+    {
+      "epoch": 0.2326494740600368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7922645563188984e-05,
+      "loss": 1.0266,
+      "step": 8471
+    },
+    {
+      "epoch": 0.23267693828788003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7922102768977026e-05,
+      "loss": 0.997,
+      "step": 8472
+    },
+    {
+      "epoch": 0.23270440251572327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7921559912082322e-05,
+      "loss": 1.0064,
+      "step": 8473
+    },
+    {
+      "epoch": 0.2327318667435665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792101699250917e-05,
+      "loss": 1.0124,
+      "step": 8474
+    },
+    {
+      "epoch": 0.23275933097140974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7920474010261863e-05,
+      "loss": 0.8871,
+      "step": 8475
+    },
+    {
+      "epoch": 0.23278679519925297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7919930965344696e-05,
+      "loss": 0.9944,
+      "step": 8476
+    },
+    {
+      "epoch": 0.23281425942709622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7919387857761966e-05,
+      "loss": 0.9199,
+      "step": 8477
+    },
+    {
+      "epoch": 0.23284172365493944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7918844687517977e-05,
+      "loss": 0.9738,
+      "step": 8478
+    },
+    {
+      "epoch": 0.23286918788278269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7918301454617017e-05,
+      "loss": 1.1089,
+      "step": 8479
+    },
+    {
+      "epoch": 0.2328966521106259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7917758159063387e-05,
+      "loss": 0.936,
+      "step": 8480
+    },
+    {
+      "epoch": 0.23292411633846916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7917214800861388e-05,
+      "loss": 1.0243,
+      "step": 8481
+    },
+    {
+      "epoch": 0.23295158056631238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791667138001532e-05,
+      "loss": 0.9718,
+      "step": 8482
+    },
+    {
+      "epoch": 0.2329790447941556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7916127896529484e-05,
+      "loss": 1.0098,
+      "step": 8483
+    },
+    {
+      "epoch": 0.23300650902199885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7915584350408174e-05,
+      "loss": 0.9706,
+      "step": 8484
+    },
+    {
+      "epoch": 0.23303397324984207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7915040741655696e-05,
+      "loss": 0.9789,
+      "step": 8485
+    },
+    {
+      "epoch": 0.23306143747768532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791449707027635e-05,
+      "loss": 1.0224,
+      "step": 8486
+    },
+    {
+      "epoch": 0.23308890170552854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7913953336274438e-05,
+      "loss": 1.0491,
+      "step": 8487
+    },
+    {
+      "epoch": 0.2331163659333718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7913409539654257e-05,
+      "loss": 0.9732,
+      "step": 8488
+    },
+    {
+      "epoch": 0.233143830161215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7912865680420117e-05,
+      "loss": 0.9797,
+      "step": 8489
+    },
+    {
+      "epoch": 0.23317129438905826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791232175857632e-05,
+      "loss": 0.9433,
+      "step": 8490
+    },
+    {
+      "epoch": 0.23319875861690148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791177777412717e-05,
+      "loss": 1.1743,
+      "step": 8491
+    },
+    {
+      "epoch": 0.23322622284474473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7911233727076967e-05,
+      "loss": 0.8922,
+      "step": 8492
+    },
+    {
+      "epoch": 0.23325368707258795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7910689617430018e-05,
+      "loss": 0.9492,
+      "step": 8493
+    },
+    {
+      "epoch": 0.2332811513004312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791014544519063e-05,
+      "loss": 0.9755,
+      "step": 8494
+    },
+    {
+      "epoch": 0.23330861552827442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7909601210363104e-05,
+      "loss": 1.0847,
+      "step": 8495
+    },
+    {
+      "epoch": 0.23333607975611767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7909056912951752e-05,
+      "loss": 1.0344,
+      "step": 8496
+    },
+    {
+      "epoch": 0.2333635439839609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7908512552960878e-05,
+      "loss": 0.979,
+      "step": 8497
+    },
+    {
+      "epoch": 0.2333910082118041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7907968130394787e-05,
+      "loss": 0.9648,
+      "step": 8498
+    },
+    {
+      "epoch": 0.23341847243964736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790742364525779e-05,
+      "loss": 1.0285,
+      "step": 8499
+    },
+    {
+      "epoch": 0.23344593666749058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7906879097554196e-05,
+      "loss": 0.9523,
+      "step": 8500
+    },
+    {
+      "epoch": 0.23347340089533383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790633448728831e-05,
+      "loss": 1.0447,
+      "step": 8501
+    },
+    {
+      "epoch": 0.23350086512317705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7905789814464443e-05,
+      "loss": 0.9454,
+      "step": 8502
+    },
+    {
+      "epoch": 0.2335283293510203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7905245079086906e-05,
+      "loss": 1.0277,
+      "step": 8503
+    },
+    {
+      "epoch": 0.23355579357886352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7904700281160005e-05,
+      "loss": 0.9835,
+      "step": 8504
+    },
+    {
+      "epoch": 0.23358325780670677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7904155420688055e-05,
+      "loss": 0.9117,
+      "step": 8505
+    },
+    {
+      "epoch": 0.23361072203455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7903610497675367e-05,
+      "loss": 0.9714,
+      "step": 8506
+    },
+    {
+      "epoch": 0.23363818626239324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7903065512126246e-05,
+      "loss": 0.9262,
+      "step": 8507
+    },
+    {
+      "epoch": 0.23366565049023647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7902520464045016e-05,
+      "loss": 1.0209,
+      "step": 8508
+    },
+    {
+      "epoch": 0.23369311471807971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790197535343598e-05,
+      "loss": 0.9907,
+      "step": 8509
+    },
+    {
+      "epoch": 0.23372057894592294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7901430180303455e-05,
+      "loss": 1.0921,
+      "step": 8510
+    },
+    {
+      "epoch": 0.23374804317376616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790088494465175e-05,
+      "loss": 0.9605,
+      "step": 8511
+    },
+    {
+      "epoch": 0.2337755074016094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7900339646485187e-05,
+      "loss": 0.984,
+      "step": 8512
+    },
+    {
+      "epoch": 0.23380297162945263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7899794285808078e-05,
+      "loss": 0.8849,
+      "step": 8513
+    },
+    {
+      "epoch": 0.23383043585729588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789924886262473e-05,
+      "loss": 1.0049,
+      "step": 8514
+    },
+    {
+      "epoch": 0.2338579000851391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7898703376939467e-05,
+      "loss": 1.0385,
+      "step": 8515
+    },
+    {
+      "epoch": 0.23388536431298235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7898157828756605e-05,
+      "loss": 1.0468,
+      "step": 8516
+    },
+    {
+      "epoch": 0.23391282854082557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789761221808046e-05,
+      "loss": 0.9691,
+      "step": 8517
+    },
+    {
+      "epoch": 0.23394029276866882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7897066544915345e-05,
+      "loss": 1.2063,
+      "step": 8518
+    },
+    {
+      "epoch": 0.23396775699651204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7896520809265583e-05,
+      "loss": 1.007,
+      "step": 8519
+    },
+    {
+      "epoch": 0.2339952212243553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789597501113549e-05,
+      "loss": 1.0469,
+      "step": 8520
+    },
+    {
+      "epoch": 0.2340226854521985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789542915052938e-05,
+      "loss": 1.0163,
+      "step": 8521
+    },
+    {
+      "epoch": 0.23405014968004176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789488322745158e-05,
+      "loss": 0.9978,
+      "step": 8522
+    },
+    {
+      "epoch": 0.23407761390788498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7894337241906403e-05,
+      "loss": 0.9329,
+      "step": 8523
+    },
+    {
+      "epoch": 0.2341050781357282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7893791193898174e-05,
+      "loss": 0.9097,
+      "step": 8524
+    },
+    {
+      "epoch": 0.23413254236357145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7893245083431214e-05,
+      "loss": 1.0284,
+      "step": 8525
+    },
+    {
+      "epoch": 0.23416000659141467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7892698910509834e-05,
+      "loss": 1.0167,
+      "step": 8526
+    },
+    {
+      "epoch": 0.23418747081925792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7892152675138368e-05,
+      "loss": 1.0507,
+      "step": 8527
+    },
+    {
+      "epoch": 0.23421493504710114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7891606377321133e-05,
+      "loss": 1.0155,
+      "step": 8528
+    },
+    {
+      "epoch": 0.2342423992749444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7891060017062448e-05,
+      "loss": 1.0585,
+      "step": 8529
+    },
+    {
+      "epoch": 0.2342698635027876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7890513594366644e-05,
+      "loss": 0.8789,
+      "step": 8530
+    },
+    {
+      "epoch": 0.23429732773063086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7889967109238037e-05,
+      "loss": 1.1033,
+      "step": 8531
+    },
+    {
+      "epoch": 0.23432479195847408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7889420561680954e-05,
+      "loss": 0.8341,
+      "step": 8532
+    },
+    {
+      "epoch": 0.23435225618631733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7888873951699718e-05,
+      "loss": 0.9851,
+      "step": 8533
+    },
+    {
+      "epoch": 0.23437972041416055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7888327279298655e-05,
+      "loss": 0.9982,
+      "step": 8534
+    },
+    {
+      "epoch": 0.2344071846420038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7887780544482096e-05,
+      "loss": 1.1058,
+      "step": 8535
+    },
+    {
+      "epoch": 0.23443464886984702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7887233747254357e-05,
+      "loss": 0.929,
+      "step": 8536
+    },
+    {
+      "epoch": 0.23446211309769027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7886686887619772e-05,
+      "loss": 0.9132,
+      "step": 8537
+    },
+    {
+      "epoch": 0.2344895773255335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7886139965582664e-05,
+      "loss": 0.9521,
+      "step": 8538
+    },
+    {
+      "epoch": 0.23451704155337671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7885592981147365e-05,
+      "loss": 1.0526,
+      "step": 8539
+    },
+    {
+      "epoch": 0.23454450578121996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78850459343182e-05,
+      "loss": 0.8994,
+      "step": 8540
+    },
+    {
+      "epoch": 0.23457197000906319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7884498825099492e-05,
+      "loss": 1.0186,
+      "step": 8541
+    },
+    {
+      "epoch": 0.23459943423690643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7883951653495578e-05,
+      "loss": 1.0113,
+      "step": 8542
+    },
+    {
+      "epoch": 0.23462689846474966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7883404419510784e-05,
+      "loss": 1.0372,
+      "step": 8543
+    },
+    {
+      "epoch": 0.2346543626925929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7882857123149443e-05,
+      "loss": 1.0156,
+      "step": 8544
+    },
+    {
+      "epoch": 0.23468182692043613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7882309764415876e-05,
+      "loss": 0.9589,
+      "step": 8545
+    },
+    {
+      "epoch": 0.23470929114827938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788176234331443e-05,
+      "loss": 1.001,
+      "step": 8546
+    },
+    {
+      "epoch": 0.2347367553761226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7881214859849422e-05,
+      "loss": 0.9642,
+      "step": 8547
+    },
+    {
+      "epoch": 0.23476421960396585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788066731402519e-05,
+      "loss": 1.0275,
+      "step": 8548
+    },
+    {
+      "epoch": 0.23479168383180907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7880119705846064e-05,
+      "loss": 0.9547,
+      "step": 8549
+    },
+    {
+      "epoch": 0.23481914805965232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787957203531638e-05,
+      "loss": 1.0712,
+      "step": 8550
+    },
+    {
+      "epoch": 0.23484661228749554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7879024302440468e-05,
+      "loss": 0.9304,
+      "step": 8551
+    },
+    {
+      "epoch": 0.23487407651533876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7878476507222668e-05,
+      "loss": 0.9532,
+      "step": 8552
+    },
+    {
+      "epoch": 0.234901540743182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7877928649667305e-05,
+      "loss": 1.0426,
+      "step": 8553
+    },
+    {
+      "epoch": 0.23492900497102523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7877380729778723e-05,
+      "loss": 1.0132,
+      "step": 8554
+    },
+    {
+      "epoch": 0.23495646919886848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787683274756125e-05,
+      "loss": 1.0754,
+      "step": 8555
+    },
+    {
+      "epoch": 0.2349839334267117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7876284703019226e-05,
+      "loss": 0.7054,
+      "step": 8556
+    },
+    {
+      "epoch": 0.23501139765455495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7875736596156988e-05,
+      "loss": 1.0163,
+      "step": 8557
+    },
+    {
+      "epoch": 0.23503886188239817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787518842697887e-05,
+      "loss": 0.9276,
+      "step": 8558
+    },
+    {
+      "epoch": 0.23506632611024142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7874640195489213e-05,
+      "loss": 0.922,
+      "step": 8559
+    },
+    {
+      "epoch": 0.23509379033808464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787409190169235e-05,
+      "loss": 0.9531,
+      "step": 8560
+    },
+    {
+      "epoch": 0.2351212545659279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7873543545592622e-05,
+      "loss": 1.0602,
+      "step": 8561
+    },
+    {
+      "epoch": 0.2351487187937711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7872995127194367e-05,
+      "loss": 1.001,
+      "step": 8562
+    },
+    {
+      "epoch": 0.23517618302161436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7872446646501927e-05,
+      "loss": 1.0319,
+      "step": 8563
+    },
+    {
+      "epoch": 0.23520364724945758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7871898103519636e-05,
+      "loss": 1.1069,
+      "step": 8564
+    },
+    {
+      "epoch": 0.2352311114773008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787134949825184e-05,
+      "loss": 0.976,
+      "step": 8565
+    },
+    {
+      "epoch": 0.23525857570514405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7870800830702875e-05,
+      "loss": 1.0702,
+      "step": 8566
+    },
+    {
+      "epoch": 0.23528603993298727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787025210087709e-05,
+      "loss": 0.9277,
+      "step": 8567
+    },
+    {
+      "epoch": 0.23531350416083052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7869703308778818e-05,
+      "loss": 1.007,
+      "step": 8568
+    },
+    {
+      "epoch": 0.23534096838867374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786915445441241e-05,
+      "loss": 1.0852,
+      "step": 8569
+    },
+    {
+      "epoch": 0.235368432616517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7868605537782198e-05,
+      "loss": 0.9536,
+      "step": 8570
+    },
+    {
+      "epoch": 0.23539589684436021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7868056558892534e-05,
+      "loss": 0.9633,
+      "step": 8571
+    },
+    {
+      "epoch": 0.23542336107220346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7867507517747758e-05,
+      "loss": 0.9909,
+      "step": 8572
+    },
+    {
+      "epoch": 0.23545082530004668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7866958414352216e-05,
+      "loss": 1.0197,
+      "step": 8573
+    },
+    {
+      "epoch": 0.23547828952788993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786640924871025e-05,
+      "loss": 1.0256,
+      "step": 8574
+    },
+    {
+      "epoch": 0.23550575375573315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7865860020826208e-05,
+      "loss": 0.9521,
+      "step": 8575
+    },
+    {
+      "epoch": 0.2355332179835764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786531073070443e-05,
+      "loss": 0.9194,
+      "step": 8576
+    },
+    {
+      "epoch": 0.23556068221141963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7864761378349274e-05,
+      "loss": 0.999,
+      "step": 8577
+    },
+    {
+      "epoch": 0.23558814643926285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7864211963765076e-05,
+      "loss": 1.0707,
+      "step": 8578
+    },
+    {
+      "epoch": 0.2356156106671061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7863662486956184e-05,
+      "loss": 0.984,
+      "step": 8579
+    },
+    {
+      "epoch": 0.23564307489494932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7863112947926953e-05,
+      "loss": 0.9918,
+      "step": 8580
+    },
+    {
+      "epoch": 0.23567053912279257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7862563346681725e-05,
+      "loss": 0.9355,
+      "step": 8581
+    },
+    {
+      "epoch": 0.2356980033506358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786201368322485e-05,
+      "loss": 0.9823,
+      "step": 8582
+    },
+    {
+      "epoch": 0.23572546757847904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7861463957560678e-05,
+      "loss": 1.014,
+      "step": 8583
+    },
+    {
+      "epoch": 0.23575293180632226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7860914169693556e-05,
+      "loss": 1.1064,
+      "step": 8584
+    },
+    {
+      "epoch": 0.2357803960341655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7860364319627836e-05,
+      "loss": 0.9466,
+      "step": 8585
+    },
+    {
+      "epoch": 0.23580786026200873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785981440736787e-05,
+      "loss": 1.0002,
+      "step": 8586
+    },
+    {
+      "epoch": 0.23583532448985198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7859264432918006e-05,
+      "loss": 0.9371,
+      "step": 8587
+    },
+    {
+      "epoch": 0.2358627887176952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78587143962826e-05,
+      "loss": 1.0749,
+      "step": 8588
+    },
+    {
+      "epoch": 0.23589025294553845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7858164297466e-05,
+      "loss": 1.0061,
+      "step": 8589
+    },
+    {
+      "epoch": 0.23591771717338167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785761413647256e-05,
+      "loss": 1.0325,
+      "step": 8590
+    },
+    {
+      "epoch": 0.23594518140122492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7857063913306636e-05,
+      "loss": 0.8347,
+      "step": 8591
+    },
+    {
+      "epoch": 0.23597264562906814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7856513627972575e-05,
+      "loss": 0.9144,
+      "step": 8592
+    },
+    {
+      "epoch": 0.23600010985691136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785596328047474e-05,
+      "loss": 0.9762,
+      "step": 8593
+    },
+    {
+      "epoch": 0.2360275740847546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7855412870817475e-05,
+      "loss": 1.0095,
+      "step": 8594
+    },
+    {
+      "epoch": 0.23605503831259783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7854862399005143e-05,
+      "loss": 0.8916,
+      "step": 8595
+    },
+    {
+      "epoch": 0.23608250254044108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7854311865042094e-05,
+      "loss": 1.0596,
+      "step": 8596
+    },
+    {
+      "epoch": 0.2361099667682843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785376126893269e-05,
+      "loss": 1.034,
+      "step": 8597
+    },
+    {
+      "epoch": 0.23613743099612755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7853210610681283e-05,
+      "loss": 0.9767,
+      "step": 8598
+    },
+    {
+      "epoch": 0.23616489522397077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7852659890292235e-05,
+      "loss": 1.0865,
+      "step": 8599
+    },
+    {
+      "epoch": 0.23619235945181402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78521091077699e-05,
+      "loss": 1.011,
+      "step": 8600
+    },
+    {
+      "epoch": 0.23621982367965724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785155826311863e-05,
+      "loss": 1.0204,
+      "step": 8601
+    },
+    {
+      "epoch": 0.2362472879075005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7851007356342795e-05,
+      "loss": 0.9405,
+      "step": 8602
+    },
+    {
+      "epoch": 0.2362747521353437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7850456387446745e-05,
+      "loss": 1.0029,
+      "step": 8603
+    },
+    {
+      "epoch": 0.23630221636318696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7849905356434846e-05,
+      "loss": 0.9678,
+      "step": 8604
+    },
+    {
+      "epoch": 0.23632968059103018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7849354263311454e-05,
+      "loss": 0.9876,
+      "step": 8605
+    },
+    {
+      "epoch": 0.2363571448188734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784880310808093e-05,
+      "loss": 0.8123,
+      "step": 8606
+    },
+    {
+      "epoch": 0.23638460904671665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7848251890747636e-05,
+      "loss": 0.9773,
+      "step": 8607
+    },
+    {
+      "epoch": 0.23641207327455988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784770061131593e-05,
+      "loss": 1.0972,
+      "step": 8608
+    },
+    {
+      "epoch": 0.23643953750240312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7847149269790183e-05,
+      "loss": 1.0465,
+      "step": 8609
+    },
+    {
+      "epoch": 0.23646700173024635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7846597866174748e-05,
+      "loss": 0.9573,
+      "step": 8610
+    },
+    {
+      "epoch": 0.2364944659580896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784604640047399e-05,
+      "loss": 0.9924,
+      "step": 8611
+    },
+    {
+      "epoch": 0.23652193018593282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7845494872692276e-05,
+      "loss": 0.9314,
+      "step": 8612
+    },
+    {
+      "epoch": 0.23654939441377607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7844943282833968e-05,
+      "loss": 1.0406,
+      "step": 8613
+    },
+    {
+      "epoch": 0.2365768586416193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7844391630903428e-05,
+      "loss": 0.8812,
+      "step": 8614
+    },
+    {
+      "epoch": 0.23660432286946254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7843839916905024e-05,
+      "loss": 0.9929,
+      "step": 8615
+    },
+    {
+      "epoch": 0.23663178709730576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784328814084312e-05,
+      "loss": 1.0416,
+      "step": 8616
+    },
+    {
+      "epoch": 0.236659251325149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784273630272208e-05,
+      "loss": 1.0661,
+      "step": 8617
+    },
+    {
+      "epoch": 0.23668671555299223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7842184402546275e-05,
+      "loss": 0.9891,
+      "step": 8618
+    },
+    {
+      "epoch": 0.23671417978083545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7841632440320065e-05,
+      "loss": 1.1659,
+      "step": 8619
+    },
+    {
+      "epoch": 0.2367416440086787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7841080416047828e-05,
+      "loss": 0.9951,
+      "step": 8620
+    },
+    {
+      "epoch": 0.23676910823652192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784052832973392e-05,
+      "loss": 1.0066,
+      "step": 8621
+    },
+    {
+      "epoch": 0.23679657246436517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7839976181382717e-05,
+      "loss": 1.0258,
+      "step": 8622
+    },
+    {
+      "epoch": 0.2368240366922084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7839423970998588e-05,
+      "loss": 1.022,
+      "step": 8623
+    },
+    {
+      "epoch": 0.23685150092005164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7838871698585897e-05,
+      "loss": 0.8335,
+      "step": 8624
+    },
+    {
+      "epoch": 0.23687896514789486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7838319364149015e-05,
+      "loss": 1.0894,
+      "step": 8625
+    },
+    {
+      "epoch": 0.2369064293757381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7837766967692317e-05,
+      "loss": 1.0352,
+      "step": 8626
+    },
+    {
+      "epoch": 0.23693389360358133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7837214509220166e-05,
+      "loss": 1.026,
+      "step": 8627
+    },
+    {
+      "epoch": 0.23696135783142458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7836661988736943e-05,
+      "loss": 1.0619,
+      "step": 8628
+    },
+    {
+      "epoch": 0.2369888220592678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783610940624701e-05,
+      "loss": 0.9977,
+      "step": 8629
+    },
+    {
+      "epoch": 0.23701628628711105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783555676175475e-05,
+      "loss": 1.089,
+      "step": 8630
+    },
+    {
+      "epoch": 0.23704375051495427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7835004055264528e-05,
+      "loss": 0.9422,
+      "step": 8631
+    },
+    {
+      "epoch": 0.2370712147427975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7834451286780717e-05,
+      "loss": 0.9757,
+      "step": 8632
+    },
+    {
+      "epoch": 0.23709867897064074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783389845630769e-05,
+      "loss": 0.9875,
+      "step": 8633
+    },
+    {
+      "epoch": 0.23712614319848396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783334556384983e-05,
+      "loss": 0.8358,
+      "step": 8634
+    },
+    {
+      "epoch": 0.2371536074263272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7832792609411503e-05,
+      "loss": 1.0576,
+      "step": 8635
+    },
+    {
+      "epoch": 0.23718107165417043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7832239592997085e-05,
+      "loss": 0.9841,
+      "step": 8636
+    },
+    {
+      "epoch": 0.23720853588201368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7831686514610956e-05,
+      "loss": 0.8501,
+      "step": 8637
+    },
+    {
+      "epoch": 0.2372360001098569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783113337425749e-05,
+      "loss": 1.0252,
+      "step": 8638
+    },
+    {
+      "epoch": 0.23726346433770015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7830580171941058e-05,
+      "loss": 0.9336,
+      "step": 8639
+    },
+    {
+      "epoch": 0.23729092856554337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7830026907666047e-05,
+      "loss": 1.1738,
+      "step": 8640
+    },
+    {
+      "epoch": 0.23731839279338662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7829473581436828e-05,
+      "loss": 0.9375,
+      "step": 8641
+    },
+    {
+      "epoch": 0.23734585702122984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7828920193257783e-05,
+      "loss": 1.0288,
+      "step": 8642
+    },
+    {
+      "epoch": 0.2373733212490731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782836674313329e-05,
+      "loss": 0.9614,
+      "step": 8643
+    },
+    {
+      "epoch": 0.23740078547691632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7827813231067726e-05,
+      "loss": 0.9951,
+      "step": 8644
+    },
+    {
+      "epoch": 0.23742824970475956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782725965706547e-05,
+      "loss": 1.0262,
+      "step": 8645
+    },
+    {
+      "epoch": 0.23745571393260279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78267060211309e-05,
+      "loss": 1.0421,
+      "step": 8646
+    },
+    {
+      "epoch": 0.237483178160446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7826152323268407e-05,
+      "loss": 1.0554,
+      "step": 8647
+    },
+    {
+      "epoch": 0.23751064238828926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782559856348236e-05,
+      "loss": 1.0514,
+      "step": 8648
+    },
+    {
+      "epoch": 0.23753810661613248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782504474177715e-05,
+      "loss": 0.9576,
+      "step": 8649
+    },
+    {
+      "epoch": 0.23756557084397573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7824490858157154e-05,
+      "loss": 1.0058,
+      "step": 8650
+    },
+    {
+      "epoch": 0.23759303507181895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7823936912626754e-05,
+      "loss": 1.0262,
+      "step": 8651
+    },
+    {
+      "epoch": 0.2376204992996622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7823382905190335e-05,
+      "loss": 0.9122,
+      "step": 8652
+    },
+    {
+      "epoch": 0.23764796352750542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7822828835852283e-05,
+      "loss": 0.9144,
+      "step": 8653
+    },
+    {
+      "epoch": 0.23767542775534867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7822274704616974e-05,
+      "loss": 1.077,
+      "step": 8654
+    },
+    {
+      "epoch": 0.2377028919831919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78217205114888e-05,
+      "loss": 0.9369,
+      "step": 8655
+    },
+    {
+      "epoch": 0.23773035621103514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7821166256472142e-05,
+      "loss": 1.0257,
+      "step": 8656
+    },
+    {
+      "epoch": 0.23775782043887836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782061193957139e-05,
+      "loss": 0.9906,
+      "step": 8657
+    },
+    {
+      "epoch": 0.2377852846667216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782005756079092e-05,
+      "loss": 1.0779,
+      "step": 8658
+    },
+    {
+      "epoch": 0.23781274889456483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7819503120135136e-05,
+      "loss": 0.9926,
+      "step": 8659
+    },
+    {
+      "epoch": 0.23784021312240805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7818948617608406e-05,
+      "loss": 1.0328,
+      "step": 8660
+    },
+    {
+      "epoch": 0.2378676773502513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781839405321513e-05,
+      "loss": 1.0336,
+      "step": 8661
+    },
+    {
+      "epoch": 0.23789514157809452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781783942695969e-05,
+      "loss": 1.057,
+      "step": 8662
+    },
+    {
+      "epoch": 0.23792260580593777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7817284738846473e-05,
+      "loss": 1.088,
+      "step": 8663
+    },
+    {
+      "epoch": 0.237950070033781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7816729988879874e-05,
+      "loss": 1.0477,
+      "step": 8664
+    },
+    {
+      "epoch": 0.23797753426162424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781617517706428e-05,
+      "loss": 1.0718,
+      "step": 8665
+    },
+    {
+      "epoch": 0.23800499848946746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7815620303404082e-05,
+      "loss": 0.9214,
+      "step": 8666
+    },
+    {
+      "epoch": 0.2380324627173107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7815065367903664e-05,
+      "loss": 0.9781,
+      "step": 8667
+    },
+    {
+      "epoch": 0.23805992694515393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7814510370567425e-05,
+      "loss": 0.9847,
+      "step": 8668
+    },
+    {
+      "epoch": 0.23808739117299718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781395531139975e-05,
+      "loss": 0.9893,
+      "step": 8669
+    },
+    {
+      "epoch": 0.2381148554008404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7813400190405036e-05,
+      "loss": 1.0567,
+      "step": 8670
+    },
+    {
+      "epoch": 0.23814231962868365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7812845007587677e-05,
+      "loss": 1.1051,
+      "step": 8671
+    },
+    {
+      "epoch": 0.23816978385652687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7812289762952058e-05,
+      "loss": 0.9898,
+      "step": 8672
+    },
+    {
+      "epoch": 0.2381972480843701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7811734456502573e-05,
+      "loss": 1.0385,
+      "step": 8673
+    },
+    {
+      "epoch": 0.23822471231221334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781117908824362e-05,
+      "loss": 0.9751,
+      "step": 8674
+    },
+    {
+      "epoch": 0.23825217654005656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7810623658179598e-05,
+      "loss": 1.0024,
+      "step": 8675
+    },
+    {
+      "epoch": 0.23827964076789981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7810068166314892e-05,
+      "loss": 0.9804,
+      "step": 8676
+    },
+    {
+      "epoch": 0.23830710499574304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7809512612653902e-05,
+      "loss": 0.9945,
+      "step": 8677
+    },
+    {
+      "epoch": 0.23833456922358628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780895699720102e-05,
+      "loss": 1.0009,
+      "step": 8678
+    },
+    {
+      "epoch": 0.2383620334514295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7808401319960648e-05,
+      "loss": 0.9272,
+      "step": 8679
+    },
+    {
+      "epoch": 0.23838949767927275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7807845580937183e-05,
+      "loss": 1.0058,
+      "step": 8680
+    },
+    {
+      "epoch": 0.23841696190711598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7807289780135015e-05,
+      "loss": 0.9658,
+      "step": 8681
+    },
+    {
+      "epoch": 0.23844442613495923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7806733917558547e-05,
+      "loss": 1.1161,
+      "step": 8682
+    },
+    {
+      "epoch": 0.23847189036280245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780617799321218e-05,
+      "loss": 0.975,
+      "step": 8683
+    },
+    {
+      "epoch": 0.2384993545906457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7805622007100302e-05,
+      "loss": 0.9108,
+      "step": 8684
+    },
+    {
+      "epoch": 0.23852681881848892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780506595922732e-05,
+      "loss": 0.9966,
+      "step": 8685
+    },
+    {
+      "epoch": 0.23855428304633217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7804509849597638e-05,
+      "loss": 0.9886,
+      "step": 8686
+    },
+    {
+      "epoch": 0.2385817472741754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7803953678215648e-05,
+      "loss": 0.9742,
+      "step": 8687
+    },
+    {
+      "epoch": 0.2386092115020186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7803397445085756e-05,
+      "loss": 1.0217,
+      "step": 8688
+    },
+    {
+      "epoch": 0.23863667572986186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780284115021236e-05,
+      "loss": 1.0051,
+      "step": 8689
+    },
+    {
+      "epoch": 0.23866413995770508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780228479359986e-05,
+      "loss": 0.9674,
+      "step": 8690
+    },
+    {
+      "epoch": 0.23869160418554833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7801728375252658e-05,
+      "loss": 0.9198,
+      "step": 8691
+    },
+    {
+      "epoch": 0.23871906841339155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780117189517516e-05,
+      "loss": 0.9552,
+      "step": 8692
+    },
+    {
+      "epoch": 0.2387465326412348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7800615353371772e-05,
+      "loss": 0.9614,
+      "step": 8693
+    },
+    {
+      "epoch": 0.23877399686907802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7800058749846892e-05,
+      "loss": 0.9634,
+      "step": 8694
+    },
+    {
+      "epoch": 0.23880146109692127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7799502084604924e-05,
+      "loss": 0.9522,
+      "step": 8695
+    },
+    {
+      "epoch": 0.2388289253247645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7798945357650277e-05,
+      "loss": 0.9189,
+      "step": 8696
+    },
+    {
+      "epoch": 0.23885638955260774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779838856898735e-05,
+      "loss": 0.9742,
+      "step": 8697
+    },
+    {
+      "epoch": 0.23888385378045096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7797831718620556e-05,
+      "loss": 0.9106,
+      "step": 8698
+    },
+    {
+      "epoch": 0.2389113180082942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7797274806554295e-05,
+      "loss": 1.0126,
+      "step": 8699
+    },
+    {
+      "epoch": 0.23893878223613743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7796717832792977e-05,
+      "loss": 0.9315,
+      "step": 8700
+    },
+    {
+      "epoch": 0.23896624646398065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7796160797341005e-05,
+      "loss": 0.9888,
+      "step": 8701
+    },
+    {
+      "epoch": 0.2389937106918239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779560370020279e-05,
+      "loss": 1.0111,
+      "step": 8702
+    },
+    {
+      "epoch": 0.23902117491966712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7795046541382735e-05,
+      "loss": 1.0343,
+      "step": 8703
+    },
+    {
+      "epoch": 0.23904863914751037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7794489320885256e-05,
+      "loss": 1.0445,
+      "step": 8704
+    },
+    {
+      "epoch": 0.2390761033753536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779393203871476e-05,
+      "loss": 1.0782,
+      "step": 8705
+    },
+    {
+      "epoch": 0.23910356760319684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779337469487565e-05,
+      "loss": 1.1307,
+      "step": 8706
+    },
+    {
+      "epoch": 0.23913103183104006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7792817289372343e-05,
+      "loss": 0.9869,
+      "step": 8707
+    },
+    {
+      "epoch": 0.2391584960588833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7792259822209244e-05,
+      "loss": 1.0144,
+      "step": 8708
+    },
+    {
+      "epoch": 0.23918596028672653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7791702293390773e-05,
+      "loss": 0.9736,
+      "step": 8709
+    },
+    {
+      "epoch": 0.23921342451456978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7791144702921333e-05,
+      "loss": 0.9773,
+      "step": 8710
+    },
+    {
+      "epoch": 0.239240888742413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7790587050805338e-05,
+      "loss": 0.9336,
+      "step": 8711
+    },
+    {
+      "epoch": 0.23926835297025625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7790029337047197e-05,
+      "loss": 0.8473,
+      "step": 8712
+    },
+    {
+      "epoch": 0.23929581719809948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778947156165133e-05,
+      "loss": 1.0273,
+      "step": 8713
+    },
+    {
+      "epoch": 0.2393232814259427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7788913724622142e-05,
+      "loss": 0.9561,
+      "step": 8714
+    },
+    {
+      "epoch": 0.23935074565378595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7788355825964057e-05,
+      "loss": 0.9825,
+      "step": 8715
+    },
+    {
+      "epoch": 0.23937820988162917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7787797865681483e-05,
+      "loss": 0.9729,
+      "step": 8716
+    },
+    {
+      "epoch": 0.23940567410947242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778723984377883e-05,
+      "loss": 0.865,
+      "step": 8717
+    },
+    {
+      "epoch": 0.23943313833731564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7786681760260526e-05,
+      "loss": 0.9401,
+      "step": 8718
+    },
+    {
+      "epoch": 0.2394606025651589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7786123615130978e-05,
+      "loss": 1.0188,
+      "step": 8719
+    },
+    {
+      "epoch": 0.2394880667930021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77855654083946e-05,
+      "loss": 1.0984,
+      "step": 8720
+    },
+    {
+      "epoch": 0.23951553102084536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7785007140055816e-05,
+      "loss": 1.0735,
+      "step": 8721
+    },
+    {
+      "epoch": 0.23954299524868858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7784448810119038e-05,
+      "loss": 0.9171,
+      "step": 8722
+    },
+    {
+      "epoch": 0.23957045947653183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7783890418588687e-05,
+      "loss": 0.925,
+      "step": 8723
+    },
+    {
+      "epoch": 0.23959792370437505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7783331965469178e-05,
+      "loss": 1.0668,
+      "step": 8724
+    },
+    {
+      "epoch": 0.2396253879322183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7782773450764935e-05,
+      "loss": 0.9692,
+      "step": 8725
+    },
+    {
+      "epoch": 0.23965285216006152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778221487448037e-05,
+      "loss": 0.9947,
+      "step": 8726
+    },
+    {
+      "epoch": 0.23968031638790474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778165623661991e-05,
+      "loss": 1.009,
+      "step": 8727
+    },
+    {
+      "epoch": 0.239707780615748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778109753718797e-05,
+      "loss": 0.9956,
+      "step": 8728
+    },
+    {
+      "epoch": 0.2397352448435912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7780538776188973e-05,
+      "loss": 1.0128,
+      "step": 8729
+    },
+    {
+      "epoch": 0.23976270907143446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777997995362734e-05,
+      "loss": 1.0213,
+      "step": 8730
+    },
+    {
+      "epoch": 0.23979017329927768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777942106950749e-05,
+      "loss": 1.0995,
+      "step": 8731
+    },
+    {
+      "epoch": 0.23981763752712093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7778862123833845e-05,
+      "loss": 0.911,
+      "step": 8732
+    },
+    {
+      "epoch": 0.23984510175496415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777830311661083e-05,
+      "loss": 1.042,
+      "step": 8733
+    },
+    {
+      "epoch": 0.2398725659828074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7777744047842872e-05,
+      "loss": 1.0237,
+      "step": 8734
+    },
+    {
+      "epoch": 0.23990003021065062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7777184917534386e-05,
+      "loss": 1.0229,
+      "step": 8735
+    },
+    {
+      "epoch": 0.23992749443849387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7776625725689803e-05,
+      "loss": 0.9187,
+      "step": 8736
+    },
+    {
+      "epoch": 0.2399549586663371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7776066472313545e-05,
+      "loss": 0.8504,
+      "step": 8737
+    },
+    {
+      "epoch": 0.23998242289418034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7775507157410033e-05,
+      "loss": 0.9912,
+      "step": 8738
+    },
+    {
+      "epoch": 0.24000988712202356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77749477809837e-05,
+      "loss": 0.9771,
+      "step": 8739
+    },
+    {
+      "epoch": 0.2400373513498668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777438834303897e-05,
+      "loss": 0.94,
+      "step": 8740
+    },
+    {
+      "epoch": 0.24006481557771003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7773828843580264e-05,
+      "loss": 0.8921,
+      "step": 8741
+    },
+    {
+      "epoch": 0.24009227980555325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7773269282612014e-05,
+      "loss": 0.9956,
+      "step": 8742
+    },
+    {
+      "epoch": 0.2401197440333965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7772709660138646e-05,
+      "loss": 1.0453,
+      "step": 8743
+    },
+    {
+      "epoch": 0.24014720826123973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777214997616459e-05,
+      "loss": 0.9605,
+      "step": 8744
+    },
+    {
+      "epoch": 0.24017467248908297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7771590230694272e-05,
+      "loss": 0.9119,
+      "step": 8745
+    },
+    {
+      "epoch": 0.2402021367169262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777103042373212e-05,
+      "loss": 1.0768,
+      "step": 8746
+    },
+    {
+      "epoch": 0.24022960094476944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777047055528257e-05,
+      "loss": 1.0126,
+      "step": 8747
+    },
+    {
+      "epoch": 0.24025706517261267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776991062535004e-05,
+      "loss": 0.9886,
+      "step": 8748
+    },
+    {
+      "epoch": 0.24028452940045592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7769350633938973e-05,
+      "loss": 0.907,
+      "step": 8749
+    },
+    {
+      "epoch": 0.24031199362829914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776879058105379e-05,
+      "loss": 1.0803,
+      "step": 8750
+    },
+    {
+      "epoch": 0.24033945785614239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7768230466698934e-05,
+      "loss": 1.0642,
+      "step": 8751
+    },
+    {
+      "epoch": 0.2403669220839856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776767029087882e-05,
+      "loss": 0.957,
+      "step": 8752
+    },
+    {
+      "epoch": 0.24039438631182886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7767110053597896e-05,
+      "loss": 1.0833,
+      "step": 8753
+    },
+    {
+      "epoch": 0.24042185053967208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7766549754860585e-05,
+      "loss": 0.9449,
+      "step": 8754
+    },
+    {
+      "epoch": 0.2404493147675153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7765989394671325e-05,
+      "loss": 1.0845,
+      "step": 8755
+    },
+    {
+      "epoch": 0.24047677899535855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776542897303455e-05,
+      "loss": 0.9761,
+      "step": 8756
+    },
+    {
+      "epoch": 0.24050424322320177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7764868489954695e-05,
+      "loss": 1.0174,
+      "step": 8757
+    },
+    {
+      "epoch": 0.24053170745104502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7764307945436188e-05,
+      "loss": 1.0345,
+      "step": 8758
+    },
+    {
+      "epoch": 0.24055917167888824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776374733948347e-05,
+      "loss": 0.9748,
+      "step": 8759
+    },
+    {
+      "epoch": 0.2405866359067315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7763186672100977e-05,
+      "loss": 0.9339,
+      "step": 8760
+    },
+    {
+      "epoch": 0.2406141001345747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7762625943293143e-05,
+      "loss": 1.0654,
+      "step": 8761
+    },
+    {
+      "epoch": 0.24064156436241796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7762065153064405e-05,
+      "loss": 1.0334,
+      "step": 8762
+    },
+    {
+      "epoch": 0.24066902859026118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7761504301419202e-05,
+      "loss": 1.0588,
+      "step": 8763
+    },
+    {
+      "epoch": 0.24069649281810443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776094338836197e-05,
+      "loss": 1.0748,
+      "step": 8764
+    },
+    {
+      "epoch": 0.24072395704594765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776038241389715e-05,
+      "loss": 0.8842,
+      "step": 8765
+    },
+    {
+      "epoch": 0.2407514212737909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775982137802917e-05,
+      "loss": 1.0551,
+      "step": 8766
+    },
+    {
+      "epoch": 0.24077888550163412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7759260280762486e-05,
+      "loss": 0.9737,
+      "step": 8767
+    },
+    {
+      "epoch": 0.24080634972947734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7758699122101524e-05,
+      "loss": 0.9801,
+      "step": 8768
+    },
+    {
+      "epoch": 0.2408338139573206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7758137902050735e-05,
+      "loss": 1.0158,
+      "step": 8769
+    },
+    {
+      "epoch": 0.2408612781851638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775757662061455e-05,
+      "loss": 0.9604,
+      "step": 8770
+    },
+    {
+      "epoch": 0.24088874241300706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7757015277797415e-05,
+      "loss": 1.0909,
+      "step": 8771
+    },
+    {
+      "epoch": 0.24091620664085028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7756453873603768e-05,
+      "loss": 1.0093,
+      "step": 8772
+    },
+    {
+      "epoch": 0.24094367086869353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7755892408038056e-05,
+      "loss": 1.0156,
+      "step": 8773
+    },
+    {
+      "epoch": 0.24097113509653675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7755330881104717e-05,
+      "loss": 0.9606,
+      "step": 8774
+    },
+    {
+      "epoch": 0.24099859932438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7754769292808194e-05,
+      "loss": 1.0302,
+      "step": 8775
+    },
+    {
+      "epoch": 0.24102606355222322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7754207643152933e-05,
+      "loss": 1.0283,
+      "step": 8776
+    },
+    {
+      "epoch": 0.24105352778006647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7753645932143386e-05,
+      "loss": 1.0311,
+      "step": 8777
+    },
+    {
+      "epoch": 0.2410809920079097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7753084159783977e-05,
+      "loss": 1.0684,
+      "step": 8778
+    },
+    {
+      "epoch": 0.24110845623575294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775252232607917e-05,
+      "loss": 1.0631,
+      "step": 8779
+    },
+    {
+      "epoch": 0.24113592046359617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7751960431033402e-05,
+      "loss": 0.9286,
+      "step": 8780
+    },
+    {
+      "epoch": 0.2411633846914394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775139847465112e-05,
+      "loss": 0.9396,
+      "step": 8781
+    },
+    {
+      "epoch": 0.24119084891928264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775083645693677e-05,
+      "loss": 1.0753,
+      "step": 8782
+    },
+    {
+      "epoch": 0.24121831314712586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7750274377894802e-05,
+      "loss": 1.0066,
+      "step": 8783
+    },
+    {
+      "epoch": 0.2412457773749691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7749712237529655e-05,
+      "loss": 1.116,
+      "step": 8784
+    },
+    {
+      "epoch": 0.24127324160281233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7749150035845786e-05,
+      "loss": 1.0261,
+      "step": 8785
+    },
+    {
+      "epoch": 0.24130070583065558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7748587772847643e-05,
+      "loss": 0.9525,
+      "step": 8786
+    },
+    {
+      "epoch": 0.2413281700584988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774802544853967e-05,
+      "loss": 0.969,
+      "step": 8787
+    },
+    {
+      "epoch": 0.24135563428634205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774746306292632e-05,
+      "loss": 0.9935,
+      "step": 8788
+    },
+    {
+      "epoch": 0.24138309851418527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7746900616012037e-05,
+      "loss": 1.015,
+      "step": 8789
+    },
+    {
+      "epoch": 0.24141056274202852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7746338107801277e-05,
+      "loss": 1.0506,
+      "step": 8790
+    },
+    {
+      "epoch": 0.24143802696987174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774577553829849e-05,
+      "loss": 1.0107,
+      "step": 8791
+    },
+    {
+      "epoch": 0.241465491197715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774521290750813e-05,
+      "loss": 1.0044,
+      "step": 8792
+    },
+    {
+      "epoch": 0.2414929554255582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774465021543464e-05,
+      "loss": 1.0318,
+      "step": 8793
+    },
+    {
+      "epoch": 0.24152041965340146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774408746208248e-05,
+      "loss": 0.9793,
+      "step": 8794
+    },
+    {
+      "epoch": 0.24154788388124468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7743524647456097e-05,
+      "loss": 1.1086,
+      "step": 8795
+    },
+    {
+      "epoch": 0.2415753481090879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774296177155995e-05,
+      "loss": 1.0023,
+      "step": 8796
+    },
+    {
+      "epoch": 0.24160281233693115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774239883439849e-05,
+      "loss": 1.0103,
+      "step": 8797
+    },
+    {
+      "epoch": 0.24163027656477437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7741835835976175e-05,
+      "loss": 1.0578,
+      "step": 8798
+    },
+    {
+      "epoch": 0.24165774079261762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7741272776297452e-05,
+      "loss": 0.9046,
+      "step": 8799
+    },
+    {
+      "epoch": 0.24168520502046084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7740709655366778e-05,
+      "loss": 1.0586,
+      "step": 8800
+    },
+    {
+      "epoch": 0.2417126692483041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7740146473188616e-05,
+      "loss": 0.9399,
+      "step": 8801
+    },
+    {
+      "epoch": 0.2417401334761473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7739583229767413e-05,
+      "loss": 0.9965,
+      "step": 8802
+    },
+    {
+      "epoch": 0.24176759770399056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773901992510763e-05,
+      "loss": 1.0261,
+      "step": 8803
+    },
+    {
+      "epoch": 0.24179506193183378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7738456559213726e-05,
+      "loss": 1.123,
+      "step": 8804
+    },
+    {
+      "epoch": 0.24182252615967703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7737893132090152e-05,
+      "loss": 1.0082,
+      "step": 8805
+    },
+    {
+      "epoch": 0.24184999038752025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7737329643741374e-05,
+      "loss": 0.9742,
+      "step": 8806
+    },
+    {
+      "epoch": 0.2418774546153635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7736766094171845e-05,
+      "loss": 1.0035,
+      "step": 8807
+    },
+    {
+      "epoch": 0.24190491884320672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7736202483386026e-05,
+      "loss": 0.9294,
+      "step": 8808
+    },
+    {
+      "epoch": 0.24193238307104994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7735638811388375e-05,
+      "loss": 1.0938,
+      "step": 8809
+    },
+    {
+      "epoch": 0.2419598472988932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7735075078183354e-05,
+      "loss": 1.07,
+      "step": 8810
+    },
+    {
+      "epoch": 0.24198731152673641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7734511283775425e-05,
+      "loss": 0.9496,
+      "step": 8811
+    },
+    {
+      "epoch": 0.24201477575457966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7733947428169044e-05,
+      "loss": 0.95,
+      "step": 8812
+    },
+    {
+      "epoch": 0.24204223998242289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7733383511368677e-05,
+      "loss": 0.916,
+      "step": 8813
+    },
+    {
+      "epoch": 0.24206970421026613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773281953337878e-05,
+      "loss": 1.1158,
+      "step": 8814
+    },
+    {
+      "epoch": 0.24209716843810936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7732255494203823e-05,
+      "loss": 0.9322,
+      "step": 8815
+    },
+    {
+      "epoch": 0.2421246326659526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7731691393848264e-05,
+      "loss": 0.9622,
+      "step": 8816
+    },
+    {
+      "epoch": 0.24215209689379583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773112723231657e-05,
+      "loss": 0.9323,
+      "step": 8817
+    },
+    {
+      "epoch": 0.24217956112163908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77305630096132e-05,
+      "loss": 0.9626,
+      "step": 8818
+    },
+    {
+      "epoch": 0.2422070253494823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7729998725742623e-05,
+      "loss": 1.0608,
+      "step": 8819
+    },
+    {
+      "epoch": 0.24223448957732555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7729434380709302e-05,
+      "loss": 0.8296,
+      "step": 8820
+    },
+    {
+      "epoch": 0.24226195380516877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77288699745177e-05,
+      "loss": 0.9565,
+      "step": 8821
+    },
+    {
+      "epoch": 0.242289418033012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7728305507172285e-05,
+      "loss": 1.0206,
+      "step": 8822
+    },
+    {
+      "epoch": 0.24231688226085524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7727740978677523e-05,
+      "loss": 0.8804,
+      "step": 8823
+    },
+    {
+      "epoch": 0.24234434648869846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7727176389037882e-05,
+      "loss": 1.0417,
+      "step": 8824
+    },
+    {
+      "epoch": 0.2423718107165417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7726611738257827e-05,
+      "loss": 0.9333,
+      "step": 8825
+    },
+    {
+      "epoch": 0.24239927494438493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7726047026341828e-05,
+      "loss": 0.9731,
+      "step": 8826
+    },
+    {
+      "epoch": 0.24242673917222818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7725482253294353e-05,
+      "loss": 0.9396,
+      "step": 8827
+    },
+    {
+      "epoch": 0.2424542034000714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7724917419119868e-05,
+      "loss": 0.8579,
+      "step": 8828
+    },
+    {
+      "epoch": 0.24248166762791465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7724352523822845e-05,
+      "loss": 0.9538,
+      "step": 8829
+    },
+    {
+      "epoch": 0.24250913185575787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7723787567407753e-05,
+      "loss": 0.9637,
+      "step": 8830
+    },
+    {
+      "epoch": 0.24253659608360112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772322254987906e-05,
+      "loss": 1.0284,
+      "step": 8831
+    },
+    {
+      "epoch": 0.24256406031144434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7722657471241242e-05,
+      "loss": 1.0039,
+      "step": 8832
+    },
+    {
+      "epoch": 0.2425915245392876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7722092331498764e-05,
+      "loss": 1.0681,
+      "step": 8833
+    },
+    {
+      "epoch": 0.2426189887671308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77215271306561e-05,
+      "loss": 0.9812,
+      "step": 8834
+    },
+    {
+      "epoch": 0.24264645299497406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7720961868717724e-05,
+      "loss": 1.0684,
+      "step": 8835
+    },
+    {
+      "epoch": 0.24267391722281728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7720396545688104e-05,
+      "loss": 0.9981,
+      "step": 8836
+    },
+    {
+      "epoch": 0.2427013814506605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771983116157172e-05,
+      "loss": 1.1085,
+      "step": 8837
+    },
+    {
+      "epoch": 0.24272884567850375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771926571637304e-05,
+      "loss": 0.9916,
+      "step": 8838
+    },
+    {
+      "epoch": 0.24275630990634697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7718700210096538e-05,
+      "loss": 1.038,
+      "step": 8839
+    },
+    {
+      "epoch": 0.24278377413419022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7718134642746688e-05,
+      "loss": 1.011,
+      "step": 8840
+    },
+    {
+      "epoch": 0.24281123836203344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7717569014327972e-05,
+      "loss": 0.9326,
+      "step": 8841
+    },
+    {
+      "epoch": 0.2428387025898767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771700332484486e-05,
+      "loss": 1.0038,
+      "step": 8842
+    },
+    {
+      "epoch": 0.24286616681771991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7716437574301823e-05,
+      "loss": 0.998,
+      "step": 8843
+    },
+    {
+      "epoch": 0.24289363104556316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7715871762703348e-05,
+      "loss": 0.903,
+      "step": 8844
+    },
+    {
+      "epoch": 0.24292109527340638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7715305890053906e-05,
+      "loss": 0.9896,
+      "step": 8845
+    },
+    {
+      "epoch": 0.24294855950124963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7714739956357974e-05,
+      "loss": 1.055,
+      "step": 8846
+    },
+    {
+      "epoch": 0.24297602372909285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7714173961620032e-05,
+      "loss": 0.9769,
+      "step": 8847
+    },
+    {
+      "epoch": 0.2430034879569361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7713607905844557e-05,
+      "loss": 0.9945,
+      "step": 8848
+    },
+    {
+      "epoch": 0.24303095218477933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771304178903603e-05,
+      "loss": 0.9409,
+      "step": 8849
+    },
+    {
+      "epoch": 0.24305841641262255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7712475611198922e-05,
+      "loss": 1.0204,
+      "step": 8850
+    },
+    {
+      "epoch": 0.2430858806404658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7711909372337727e-05,
+      "loss": 1.0919,
+      "step": 8851
+    },
+    {
+      "epoch": 0.24311334486830902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7711343072456915e-05,
+      "loss": 0.9795,
+      "step": 8852
+    },
+    {
+      "epoch": 0.24314080909615227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7710776711560972e-05,
+      "loss": 1.021,
+      "step": 8853
+    },
+    {
+      "epoch": 0.2431682733239955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771021028965437e-05,
+      "loss": 0.9882,
+      "step": 8854
+    },
+    {
+      "epoch": 0.24319573755183874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7709643806741604e-05,
+      "loss": 1.0267,
+      "step": 8855
+    },
+    {
+      "epoch": 0.24322320177968196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7709077262827147e-05,
+      "loss": 0.9978,
+      "step": 8856
+    },
+    {
+      "epoch": 0.2432506660075252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7708510657915487e-05,
+      "loss": 1.0526,
+      "step": 8857
+    },
+    {
+      "epoch": 0.24327813023536843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77079439920111e-05,
+      "loss": 0.9814,
+      "step": 8858
+    },
+    {
+      "epoch": 0.24330559446321168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7707377265118476e-05,
+      "loss": 1.0269,
+      "step": 8859
+    },
+    {
+      "epoch": 0.2433330586910549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7706810477242102e-05,
+      "loss": 1.004,
+      "step": 8860
+    },
+    {
+      "epoch": 0.24336052291889815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7706243628386453e-05,
+      "loss": 0.9526,
+      "step": 8861
+    },
+    {
+      "epoch": 0.24338798714674137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7705676718556023e-05,
+      "loss": 1.0345,
+      "step": 8862
+    },
+    {
+      "epoch": 0.2434154513745846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770510974775529e-05,
+      "loss": 1.0312,
+      "step": 8863
+    },
+    {
+      "epoch": 0.24344291560242784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7704542715988745e-05,
+      "loss": 0.9102,
+      "step": 8864
+    },
+    {
+      "epoch": 0.24347037983027106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7703975623260876e-05,
+      "loss": 1.002,
+      "step": 8865
+    },
+    {
+      "epoch": 0.2434978440581143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7703408469576166e-05,
+      "loss": 0.8996,
+      "step": 8866
+    },
+    {
+      "epoch": 0.24352530828595753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7702841254939107e-05,
+      "loss": 0.9962,
+      "step": 8867
+    },
+    {
+      "epoch": 0.24355277251380078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770227397935418e-05,
+      "loss": 1.01,
+      "step": 8868
+    },
+    {
+      "epoch": 0.243580236741644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770170664282588e-05,
+      "loss": 0.959,
+      "step": 8869
+    },
+    {
+      "epoch": 0.24360770096948725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770113924535869e-05,
+      "loss": 1.0823,
+      "step": 8870
+    },
+    {
+      "epoch": 0.24363516519733047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7700571786957107e-05,
+      "loss": 0.9825,
+      "step": 8871
+    },
+    {
+      "epoch": 0.24366262942517372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770000426762562e-05,
+      "loss": 0.9873,
+      "step": 8872
+    },
+    {
+      "epoch": 0.24369009365301694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769943668736871e-05,
+      "loss": 1.0021,
+      "step": 8873
+    },
+    {
+      "epoch": 0.2437175578808602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7698869046190878e-05,
+      "loss": 0.9705,
+      "step": 8874
+    },
+    {
+      "epoch": 0.2437450221087034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769830134409661e-05,
+      "loss": 0.9842,
+      "step": 8875
+    },
+    {
+      "epoch": 0.24377248633654663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76977335810904e-05,
+      "loss": 1.0037,
+      "step": 8876
+    },
+    {
+      "epoch": 0.24379995056438988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7697165757176743e-05,
+      "loss": 1.0831,
+      "step": 8877
+    },
+    {
+      "epoch": 0.2438274147922331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7696597872360128e-05,
+      "loss": 1.0252,
+      "step": 8878
+    },
+    {
+      "epoch": 0.24385487902007635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769602992664505e-05,
+      "loss": 0.9451,
+      "step": 8879
+    },
+    {
+      "epoch": 0.24388234324791958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7695461920036e-05,
+      "loss": 0.9747,
+      "step": 8880
+    },
+    {
+      "epoch": 0.24390980747576282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7694893852537472e-05,
+      "loss": 1.0779,
+      "step": 8881
+    },
+    {
+      "epoch": 0.24393727170360605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769432572415397e-05,
+      "loss": 0.9785,
+      "step": 8882
+    },
+    {
+      "epoch": 0.2439647359314493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7693757534889975e-05,
+      "loss": 0.9832,
+      "step": 8883
+    },
+    {
+      "epoch": 0.24399220015929252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7693189284749992e-05,
+      "loss": 1.0312,
+      "step": 8884
+    },
+    {
+      "epoch": 0.24401966438713577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7692620973738517e-05,
+      "loss": 1.057,
+      "step": 8885
+    },
+    {
+      "epoch": 0.244047128614979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7692052601860045e-05,
+      "loss": 0.9918,
+      "step": 8886
+    },
+    {
+      "epoch": 0.24407459284282224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7691484169119074e-05,
+      "loss": 1.068,
+      "step": 8887
+    },
+    {
+      "epoch": 0.24410205707066546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7690915675520097e-05,
+      "loss": 0.8824,
+      "step": 8888
+    },
+    {
+      "epoch": 0.2441295212985087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7690347121067623e-05,
+      "loss": 1.0272,
+      "step": 8889
+    },
+    {
+      "epoch": 0.24415698552635193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7689778505766135e-05,
+      "loss": 1.0143,
+      "step": 8890
+    },
+    {
+      "epoch": 0.24418444975419515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7689209829620147e-05,
+      "loss": 1.1416,
+      "step": 8891
+    },
+    {
+      "epoch": 0.2442119139820384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768864109263415e-05,
+      "loss": 1.0309,
+      "step": 8892
+    },
+    {
+      "epoch": 0.24423937820988162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7688072294812647e-05,
+      "loss": 1.1445,
+      "step": 8893
+    },
+    {
+      "epoch": 0.24426684243772487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7687503436160136e-05,
+      "loss": 1.0364,
+      "step": 8894
+    },
+    {
+      "epoch": 0.2442943066655681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7686934516681124e-05,
+      "loss": 0.9346,
+      "step": 8895
+    },
+    {
+      "epoch": 0.24432177089341134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7686365536380102e-05,
+      "loss": 0.8908,
+      "step": 8896
+    },
+    {
+      "epoch": 0.24434923512125456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7685796495261586e-05,
+      "loss": 1.0469,
+      "step": 8897
+    },
+    {
+      "epoch": 0.2443766993490978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7685227393330064e-05,
+      "loss": 1.0656,
+      "step": 8898
+    },
+    {
+      "epoch": 0.24440416357694103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768465823059005e-05,
+      "loss": 0.9594,
+      "step": 8899
+    },
+    {
+      "epoch": 0.24443162780478428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7684089007046042e-05,
+      "loss": 1.0194,
+      "step": 8900
+    },
+    {
+      "epoch": 0.2444590920326275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7683519722702547e-05,
+      "loss": 0.9874,
+      "step": 8901
+    },
+    {
+      "epoch": 0.24448655626047075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768295037756407e-05,
+      "loss": 0.9727,
+      "step": 8902
+    },
+    {
+      "epoch": 0.24451402048831397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7682380971635106e-05,
+      "loss": 0.9827,
+      "step": 8903
+    },
+    {
+      "epoch": 0.2445414847161572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7681811504920173e-05,
+      "loss": 0.8846,
+      "step": 8904
+    },
+    {
+      "epoch": 0.24456894894400044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768124197742377e-05,
+      "loss": 1.0702,
+      "step": 8905
+    },
+    {
+      "epoch": 0.24459641317184366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7680672389150407e-05,
+      "loss": 0.9551,
+      "step": 8906
+    },
+    {
+      "epoch": 0.2446238773996869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7680102740104588e-05,
+      "loss": 1.0681,
+      "step": 8907
+    },
+    {
+      "epoch": 0.24465134162753013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767953303029082e-05,
+      "loss": 0.8943,
+      "step": 8908
+    },
+    {
+      "epoch": 0.24467880585537338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767896325971361e-05,
+      "loss": 1.0032,
+      "step": 8909
+    },
+    {
+      "epoch": 0.2447062700832166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767839342837747e-05,
+      "loss": 0.9486,
+      "step": 8910
+    },
+    {
+      "epoch": 0.24473373431105985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7677823536286908e-05,
+      "loss": 0.9684,
+      "step": 8911
+    },
+    {
+      "epoch": 0.24476119853890307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7677253583446434e-05,
+      "loss": 0.9509,
+      "step": 8912
+    },
+    {
+      "epoch": 0.24478866276674632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7676683569860555e-05,
+      "loss": 0.9789,
+      "step": 8913
+    },
+    {
+      "epoch": 0.24481612699458954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767611349553378e-05,
+      "loss": 0.9962,
+      "step": 8914
+    },
+    {
+      "epoch": 0.2448435912224328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767554336047062e-05,
+      "loss": 1.1203,
+      "step": 8915
+    },
+    {
+      "epoch": 0.24487105545027602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767497316467559e-05,
+      "loss": 1.0012,
+      "step": 8916
+    },
+    {
+      "epoch": 0.24489851967811924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7674402908153198e-05,
+      "loss": 1.0446,
+      "step": 8917
+    },
+    {
+      "epoch": 0.24492598390596249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767383259090796e-05,
+      "loss": 1.011,
+      "step": 8918
+    },
+    {
+      "epoch": 0.2449534481338057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7673262212944383e-05,
+      "loss": 1.0319,
+      "step": 8919
+    },
+    {
+      "epoch": 0.24498091236164896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7672691774266986e-05,
+      "loss": 0.9844,
+      "step": 8920
+    },
+    {
+      "epoch": 0.24500837658949218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7672121274880278e-05,
+      "loss": 1.1003,
+      "step": 8921
+    },
+    {
+      "epoch": 0.24503584081733543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7671550714788775e-05,
+      "loss": 1.0753,
+      "step": 8922
+    },
+    {
+      "epoch": 0.24506330504517865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7670980093996987e-05,
+      "loss": 1.0459,
+      "step": 8923
+    },
+    {
+      "epoch": 0.2450907692730219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7670409412509437e-05,
+      "loss": 0.9029,
+      "step": 8924
+    },
+    {
+      "epoch": 0.24511823350086512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7669838670330638e-05,
+      "loss": 0.9553,
+      "step": 8925
+    },
+    {
+      "epoch": 0.24514569772870837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76692678674651e-05,
+      "loss": 0.9736,
+      "step": 8926
+    },
+    {
+      "epoch": 0.2451731619565516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766869700391735e-05,
+      "loss": 0.9579,
+      "step": 8927
+    },
+    {
+      "epoch": 0.24520062618439484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766812607969189e-05,
+      "loss": 1.0236,
+      "step": 8928
+    },
+    {
+      "epoch": 0.24522809041223806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7667555094793253e-05,
+      "loss": 0.9704,
+      "step": 8929
+    },
+    {
+      "epoch": 0.24525555464008128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766698404922595e-05,
+      "loss": 0.949,
+      "step": 8930
+    },
+    {
+      "epoch": 0.24528301886792453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7666412942994496e-05,
+      "loss": 1.0208,
+      "step": 8931
+    },
+    {
+      "epoch": 0.24531048309576775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7665841776103416e-05,
+      "loss": 0.9846,
+      "step": 8932
+    },
+    {
+      "epoch": 0.245337947323611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7665270548557225e-05,
+      "loss": 1.0591,
+      "step": 8933
+    },
+    {
+      "epoch": 0.24536541155145422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7664699260360443e-05,
+      "loss": 1.0518,
+      "step": 8934
+    },
+    {
+      "epoch": 0.24539287577929747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7664127911517596e-05,
+      "loss": 0.9595,
+      "step": 8935
+    },
+    {
+      "epoch": 0.2454203400071407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7663556502033193e-05,
+      "loss": 1.0302,
+      "step": 8936
+    },
+    {
+      "epoch": 0.24544780423498394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766298503191177e-05,
+      "loss": 1.0395,
+      "step": 8937
+    },
+    {
+      "epoch": 0.24547526846282716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7662413501157836e-05,
+      "loss": 1.0003,
+      "step": 8938
+    },
+    {
+      "epoch": 0.2455027326906704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766184190977592e-05,
+      "loss": 1.0274,
+      "step": 8939
+    },
+    {
+      "epoch": 0.24553019691851363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7661270257770538e-05,
+      "loss": 1.0077,
+      "step": 8940
+    },
+    {
+      "epoch": 0.24555766114635688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7660698545146227e-05,
+      "loss": 0.9526,
+      "step": 8941
+    },
+    {
+      "epoch": 0.2455851253742001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7660126771907496e-05,
+      "loss": 0.9557,
+      "step": 8942
+    },
+    {
+      "epoch": 0.24561258960204335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7659554938058872e-05,
+      "loss": 0.983,
+      "step": 8943
+    },
+    {
+      "epoch": 0.24564005382988657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765898304360489e-05,
+      "loss": 1.0977,
+      "step": 8944
+    },
+    {
+      "epoch": 0.2456675180577298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7658411088550065e-05,
+      "loss": 0.9415,
+      "step": 8945
+    },
+    {
+      "epoch": 0.24569498228557304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765783907289892e-05,
+      "loss": 0.9454,
+      "step": 8946
+    },
+    {
+      "epoch": 0.24572244651341626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765726699665599e-05,
+      "loss": 0.981,
+      "step": 8947
+    },
+    {
+      "epoch": 0.24574991074125951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7656694859825794e-05,
+      "loss": 0.9895,
+      "step": 8948
+    },
+    {
+      "epoch": 0.24577737496910274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765612266241287e-05,
+      "loss": 1.141,
+      "step": 8949
+    },
+    {
+      "epoch": 0.24580483919694598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765555040442173e-05,
+      "loss": 1.0436,
+      "step": 8950
+    },
+    {
+      "epoch": 0.2458323034247892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7654978085856912e-05,
+      "loss": 0.8987,
+      "step": 8951
+    },
+    {
+      "epoch": 0.24585976765263245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7654405706722944e-05,
+      "loss": 1.1206,
+      "step": 8952
+    },
+    {
+      "epoch": 0.24588723188047568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7653833267024352e-05,
+      "loss": 1.0597,
+      "step": 8953
+    },
+    {
+      "epoch": 0.24591469610831893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7653260766765665e-05,
+      "loss": 0.9232,
+      "step": 8954
+    },
+    {
+      "epoch": 0.24594216033616215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7652688205951414e-05,
+      "loss": 0.912,
+      "step": 8955
+    },
+    {
+      "epoch": 0.2459696245640054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7652115584586132e-05,
+      "loss": 0.9756,
+      "step": 8956
+    },
+    {
+      "epoch": 0.24599708879184862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7651542902674345e-05,
+      "loss": 1.1075,
+      "step": 8957
+    },
+    {
+      "epoch": 0.24602455301969184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7650970160220587e-05,
+      "loss": 1.0361,
+      "step": 8958
+    },
+    {
+      "epoch": 0.2460520172475351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765039735722939e-05,
+      "loss": 0.9025,
+      "step": 8959
+    },
+    {
+      "epoch": 0.2460794814753783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7649824493705287e-05,
+      "loss": 1.0431,
+      "step": 8960
+    },
+    {
+      "epoch": 0.24610694570322156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7649251569652808e-05,
+      "loss": 0.9931,
+      "step": 8961
+    },
+    {
+      "epoch": 0.24613440993106478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7648678585076488e-05,
+      "loss": 1.0304,
+      "step": 8962
+    },
+    {
+      "epoch": 0.24616187415890803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764810553998086e-05,
+      "loss": 0.9775,
+      "step": 8963
+    },
+    {
+      "epoch": 0.24618933838675125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7647532434370456e-05,
+      "loss": 0.9735,
+      "step": 8964
+    },
+    {
+      "epoch": 0.2462168026145945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764695926824982e-05,
+      "loss": 0.9103,
+      "step": 8965
+    },
+    {
+      "epoch": 0.24624426684243772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7646386041623473e-05,
+      "loss": 0.9098,
+      "step": 8966
+    },
+    {
+      "epoch": 0.24627173107028097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764581275449596e-05,
+      "loss": 1.0939,
+      "step": 8967
+    },
+    {
+      "epoch": 0.2462991952981242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7645239406871814e-05,
+      "loss": 0.9776,
+      "step": 8968
+    },
+    {
+      "epoch": 0.24632665952596744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7644665998755574e-05,
+      "loss": 0.8545,
+      "step": 8969
+    },
+    {
+      "epoch": 0.24635412375381066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7644092530151776e-05,
+      "loss": 0.963,
+      "step": 8970
+    },
+    {
+      "epoch": 0.24638158798165388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7643519001064954e-05,
+      "loss": 1.1061,
+      "step": 8971
+    },
+    {
+      "epoch": 0.24640905220949713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764294541149965e-05,
+      "loss": 0.9281,
+      "step": 8972
+    },
+    {
+      "epoch": 0.24643651643734035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7642371761460403e-05,
+      "loss": 1.0346,
+      "step": 8973
+    },
+    {
+      "epoch": 0.2464639806651836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7641798050951745e-05,
+      "loss": 0.8797,
+      "step": 8974
+    },
+    {
+      "epoch": 0.24649144489302682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7641224279978225e-05,
+      "loss": 1.09,
+      "step": 8975
+    },
+    {
+      "epoch": 0.24651890912087007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764065044854438e-05,
+      "loss": 0.9924,
+      "step": 8976
+    },
+    {
+      "epoch": 0.2465463733487133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7640076556654747e-05,
+      "loss": 0.9247,
+      "step": 8977
+    },
+    {
+      "epoch": 0.24657383757655654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763950260431387e-05,
+      "loss": 0.8662,
+      "step": 8978
+    },
+    {
+      "epoch": 0.24660130180439976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7638928591526287e-05,
+      "loss": 1.0415,
+      "step": 8979
+    },
+    {
+      "epoch": 0.246628766032243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7638354518296544e-05,
+      "loss": 1.0248,
+      "step": 8980
+    },
+    {
+      "epoch": 0.24665623026008623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763778038462918e-05,
+      "loss": 1.0161,
+      "step": 8981
+    },
+    {
+      "epoch": 0.24668369448792948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7637206190528736e-05,
+      "loss": 1.0753,
+      "step": 8982
+    },
+    {
+      "epoch": 0.2467111587157727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7636631935999763e-05,
+      "loss": 0.9314,
+      "step": 8983
+    },
+    {
+      "epoch": 0.24673862294361595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76360576210468e-05,
+      "loss": 0.9137,
+      "step": 8984
+    },
+    {
+      "epoch": 0.24676608717145918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763548324567439e-05,
+      "loss": 0.9535,
+      "step": 8985
+    },
+    {
+      "epoch": 0.2467935513993024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7634908809887078e-05,
+      "loss": 0.9402,
+      "step": 8986
+    },
+    {
+      "epoch": 0.24682101562714565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763433431368941e-05,
+      "loss": 1.054,
+      "step": 8987
+    },
+    {
+      "epoch": 0.24684847985498887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763375975708593e-05,
+      "loss": 0.9754,
+      "step": 8988
+    },
+    {
+      "epoch": 0.24687594408283212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763318514008119e-05,
+      "loss": 1.0251,
+      "step": 8989
+    },
+    {
+      "epoch": 0.24690340831067534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7632610462679728e-05,
+      "loss": 0.8767,
+      "step": 8990
+    },
+    {
+      "epoch": 0.2469308725385186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76320357248861e-05,
+      "loss": 1.0787,
+      "step": 8991
+    },
+    {
+      "epoch": 0.2469583367663618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7631460926704845e-05,
+      "loss": 1.0021,
+      "step": 8992
+    },
+    {
+      "epoch": 0.24698580099420506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7630886068140517e-05,
+      "loss": 0.9109,
+      "step": 8993
+    },
+    {
+      "epoch": 0.24701326522204828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763031114919766e-05,
+      "loss": 1.0161,
+      "step": 8994
+    },
+    {
+      "epoch": 0.24704072944989153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762973616988083e-05,
+      "loss": 1.0464,
+      "step": 8995
+    },
+    {
+      "epoch": 0.24706819367773475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762916113019457e-05,
+      "loss": 0.9653,
+      "step": 8996
+    },
+    {
+      "epoch": 0.247095657905578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762858603014343e-05,
+      "loss": 0.9702,
+      "step": 8997
+    },
+    {
+      "epoch": 0.24712312213342122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7628010869731965e-05,
+      "loss": 1.0671,
+      "step": 8998
+    },
+    {
+      "epoch": 0.24715058636126444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762743564896472e-05,
+      "loss": 0.9972,
+      "step": 8999
+    },
+    {
+      "epoch": 0.2471780505891077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7626860367846252e-05,
+      "loss": 0.9056,
+      "step": 9000
+    },
+    {
+      "epoch": 0.2472055148169509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7626285026381112e-05,
+      "loss": 0.9169,
+      "step": 9001
+    },
+    {
+      "epoch": 0.24723297904479416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762570962457385e-05,
+      "loss": 0.9723,
+      "step": 9002
+    },
+    {
+      "epoch": 0.24726044327263738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7625134162429015e-05,
+      "loss": 0.8863,
+      "step": 9003
+    },
+    {
+      "epoch": 0.24728790750048063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762455863995117e-05,
+      "loss": 0.9519,
+      "step": 9004
+    },
+    {
+      "epoch": 0.24731537172832385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762398305714486e-05,
+      "loss": 0.9845,
+      "step": 9005
+    },
+    {
+      "epoch": 0.2473428359561671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7623407414014645e-05,
+      "loss": 0.9154,
+      "step": 9006
+    },
+    {
+      "epoch": 0.24737030018401032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762283171056508e-05,
+      "loss": 0.972,
+      "step": 9007
+    },
+    {
+      "epoch": 0.24739776441185357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7622255946800712e-05,
+      "loss": 1.0126,
+      "step": 9008
+    },
+    {
+      "epoch": 0.2474252286396968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7621680122726102e-05,
+      "loss": 0.945,
+      "step": 9009
+    },
+    {
+      "epoch": 0.24745269286754004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7621104238345812e-05,
+      "loss": 1.0393,
+      "step": 9010
+    },
+    {
+      "epoch": 0.24748015709538326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762052829366439e-05,
+      "loss": 0.9321,
+      "step": 9011
+    },
+    {
+      "epoch": 0.24750762132322648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7619952288686395e-05,
+      "loss": 1.0662,
+      "step": 9012
+    },
+    {
+      "epoch": 0.24753508555106973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761937622341639e-05,
+      "loss": 0.9744,
+      "step": 9013
+    },
+    {
+      "epoch": 0.24756254977891295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7618800097858924e-05,
+      "loss": 1.0115,
+      "step": 9014
+    },
+    {
+      "epoch": 0.2475900140067562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7618223912018565e-05,
+      "loss": 0.8673,
+      "step": 9015
+    },
+    {
+      "epoch": 0.24761747823459943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7617647665899865e-05,
+      "loss": 0.8315,
+      "step": 9016
+    },
+    {
+      "epoch": 0.24764494246244267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7617071359507385e-05,
+      "loss": 1.0251,
+      "step": 9017
+    },
+    {
+      "epoch": 0.2476724066902859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7616494992845688e-05,
+      "loss": 1.0019,
+      "step": 9018
+    },
+    {
+      "epoch": 0.24769987091812914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7615918565919332e-05,
+      "loss": 1.0597,
+      "step": 9019
+    },
+    {
+      "epoch": 0.24772733514597237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7615342078732877e-05,
+      "loss": 0.9083,
+      "step": 9020
+    },
+    {
+      "epoch": 0.24775479937381562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7614765531290886e-05,
+      "loss": 0.9423,
+      "step": 9021
+    },
+    {
+      "epoch": 0.24778226360165884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7614188923597918e-05,
+      "loss": 0.9289,
+      "step": 9022
+    },
+    {
+      "epoch": 0.24780972782950209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761361225565854e-05,
+      "loss": 1.0707,
+      "step": 9023
+    },
+    {
+      "epoch": 0.2478371920573453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7613035527477314e-05,
+      "loss": 0.9343,
+      "step": 9024
+    },
+    {
+      "epoch": 0.24786465628518853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76124587390588e-05,
+      "loss": 0.9613,
+      "step": 9025
+    },
+    {
+      "epoch": 0.24789212051303178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7611881890407564e-05,
+      "loss": 0.9471,
+      "step": 9026
+    },
+    {
+      "epoch": 0.247919584740875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761130498152817e-05,
+      "loss": 1.0163,
+      "step": 9027
+    },
+    {
+      "epoch": 0.24794704896871825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7610728012425184e-05,
+      "loss": 0.9955,
+      "step": 9028
+    },
+    {
+      "epoch": 0.24797451319656147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761015098310317e-05,
+      "loss": 0.9825,
+      "step": 9029
+    },
+    {
+      "epoch": 0.24800197742440472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7609573893566692e-05,
+      "loss": 0.9373,
+      "step": 9030
+    },
+    {
+      "epoch": 0.24802944165224794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760899674382032e-05,
+      "loss": 1.0484,
+      "step": 9031
+    },
+    {
+      "epoch": 0.2480569058800912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760841953386861e-05,
+      "loss": 1.0718,
+      "step": 9032
+    },
+    {
+      "epoch": 0.2480843701079344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7607842263716145e-05,
+      "loss": 0.9869,
+      "step": 9033
+    },
+    {
+      "epoch": 0.24811183433577766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7607264933367487e-05,
+      "loss": 0.9119,
+      "step": 9034
+    },
+    {
+      "epoch": 0.24813929856362088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76066875428272e-05,
+      "loss": 0.9304,
+      "step": 9035
+    },
+    {
+      "epoch": 0.24816676279146413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7606110092099852e-05,
+      "loss": 1.0078,
+      "step": 9036
+    },
+    {
+      "epoch": 0.24819422701930735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7605532581190017e-05,
+      "loss": 0.9786,
+      "step": 9037
+    },
+    {
+      "epoch": 0.2482216912471506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760495501010226e-05,
+      "loss": 1.0547,
+      "step": 9038
+    },
+    {
+      "epoch": 0.24824915547499382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7604377378841155e-05,
+      "loss": 0.9918,
+      "step": 9039
+    },
+    {
+      "epoch": 0.24827661970283704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760379968741127e-05,
+      "loss": 1.0415,
+      "step": 9040
+    },
+    {
+      "epoch": 0.2483040839306803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7603221935817174e-05,
+      "loss": 1.1073,
+      "step": 9041
+    },
+    {
+      "epoch": 0.2483315481585235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7602644124063442e-05,
+      "loss": 1.0115,
+      "step": 9042
+    },
+    {
+      "epoch": 0.24835901238636676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7602066252154647e-05,
+      "loss": 0.9804,
+      "step": 9043
+    },
+    {
+      "epoch": 0.24838647661420998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7601488320095358e-05,
+      "loss": 0.9026,
+      "step": 9044
+    },
+    {
+      "epoch": 0.24841394084205323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7600910327890147e-05,
+      "loss": 0.9951,
+      "step": 9045
+    },
+    {
+      "epoch": 0.24844140506989645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760033227554359e-05,
+      "loss": 1.0627,
+      "step": 9046
+    },
+    {
+      "epoch": 0.2484688692977397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7599754163060264e-05,
+      "loss": 0.9471,
+      "step": 9047
+    },
+    {
+      "epoch": 0.24849633352558292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7599175990444733e-05,
+      "loss": 1.0828,
+      "step": 9048
+    },
+    {
+      "epoch": 0.24852379775342617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7598597757701583e-05,
+      "loss": 1.098,
+      "step": 9049
+    },
+    {
+      "epoch": 0.2485512619812694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7598019464835383e-05,
+      "loss": 0.9499,
+      "step": 9050
+    },
+    {
+      "epoch": 0.24857872620911264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7597441111850708e-05,
+      "loss": 1.0022,
+      "step": 9051
+    },
+    {
+      "epoch": 0.24860619043695587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759686269875214e-05,
+      "loss": 1.0642,
+      "step": 9052
+    },
+    {
+      "epoch": 0.2486336546647991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7596284225544246e-05,
+      "loss": 0.9587,
+      "step": 9053
+    },
+    {
+      "epoch": 0.24866111889264234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759570569223161e-05,
+      "loss": 0.9575,
+      "step": 9054
+    },
+    {
+      "epoch": 0.24868858312048556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7595127098818808e-05,
+      "loss": 0.9638,
+      "step": 9055
+    },
+    {
+      "epoch": 0.2487160473483288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759454844531042e-05,
+      "loss": 0.9707,
+      "step": 9056
+    },
+    {
+      "epoch": 0.24874351157617203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7593969731711024e-05,
+      "loss": 1.0034,
+      "step": 9057
+    },
+    {
+      "epoch": 0.24877097580401528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7593390958025196e-05,
+      "loss": 0.9508,
+      "step": 9058
+    },
+    {
+      "epoch": 0.2487984400318585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759281212425752e-05,
+      "loss": 1.0765,
+      "step": 9059
+    },
+    {
+      "epoch": 0.24882590425970175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759223323041257e-05,
+      "loss": 0.9751,
+      "step": 9060
+    },
+    {
+      "epoch": 0.24885336848754497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7591654276494933e-05,
+      "loss": 1.0199,
+      "step": 9061
+    },
+    {
+      "epoch": 0.24888083271538822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759107526250918e-05,
+      "loss": 1.1543,
+      "step": 9062
+    },
+    {
+      "epoch": 0.24890829694323144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7590496188459904e-05,
+      "loss": 1.0695,
+      "step": 9063
+    },
+    {
+      "epoch": 0.2489357611710747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758991705435168e-05,
+      "loss": 0.9169,
+      "step": 9064
+    },
+    {
+      "epoch": 0.2489632253989179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7589337860189096e-05,
+      "loss": 1.0609,
+      "step": 9065
+    },
+    {
+      "epoch": 0.24899068962676113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758875860597673e-05,
+      "loss": 0.9496,
+      "step": 9066
+    },
+    {
+      "epoch": 0.24901815385460438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758817929171916e-05,
+      "loss": 1.0618,
+      "step": 9067
+    },
+    {
+      "epoch": 0.2490456180824476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7587599917420983e-05,
+      "loss": 1.081,
+      "step": 9068
+    },
+    {
+      "epoch": 0.24907308231029085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7587020483086777e-05,
+      "loss": 0.959,
+      "step": 9069
+    },
+    {
+      "epoch": 0.24910054653813407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7586440988721123e-05,
+      "loss": 0.9786,
+      "step": 9070
+    },
+    {
+      "epoch": 0.24912801076597732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758586143432861e-05,
+      "loss": 0.9901,
+      "step": 9071
+    },
+    {
+      "epoch": 0.24915547499382054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7585281819913823e-05,
+      "loss": 0.9422,
+      "step": 9072
+    },
+    {
+      "epoch": 0.2491829392216638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7584702145481344e-05,
+      "loss": 1.021,
+      "step": 9073
+    },
+    {
+      "epoch": 0.249210403449507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7584122411035765e-05,
+      "loss": 1.0169,
+      "step": 9074
+    },
+    {
+      "epoch": 0.24923786767735026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7583542616581676e-05,
+      "loss": 0.9923,
+      "step": 9075
+    },
+    {
+      "epoch": 0.24926533190519348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7582962762123658e-05,
+      "loss": 0.9295,
+      "step": 9076
+    },
+    {
+      "epoch": 0.24929279613303673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7582382847666298e-05,
+      "loss": 0.9978,
+      "step": 9077
+    },
+    {
+      "epoch": 0.24932026036087995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7581802873214194e-05,
+      "loss": 0.9978,
+      "step": 9078
+    },
+    {
+      "epoch": 0.2493477245887232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7581222838771927e-05,
+      "loss": 1.0496,
+      "step": 9079
+    },
+    {
+      "epoch": 0.24937518881656642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7580642744344084e-05,
+      "loss": 0.9951,
+      "step": 9080
+    },
+    {
+      "epoch": 0.24940265304440964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7580062589935263e-05,
+      "loss": 0.992,
+      "step": 9081
+    },
+    {
+      "epoch": 0.2494301172722529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757948237555005e-05,
+      "loss": 1.0684,
+      "step": 9082
+    },
+    {
+      "epoch": 0.24945758150009611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7578902101193035e-05,
+      "loss": 0.9329,
+      "step": 9083
+    },
+    {
+      "epoch": 0.24948504572793936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7578321766868812e-05,
+      "loss": 0.8813,
+      "step": 9084
+    },
+    {
+      "epoch": 0.24951250995578259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7577741372581972e-05,
+      "loss": 0.9235,
+      "step": 9085
+    },
+    {
+      "epoch": 0.24953997418362583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7577160918337105e-05,
+      "loss": 1.0109,
+      "step": 9086
+    },
+    {
+      "epoch": 0.24956743841146906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757658040413881e-05,
+      "loss": 0.8408,
+      "step": 9087
+    },
+    {
+      "epoch": 0.2495949026393123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757599982999167e-05,
+      "loss": 0.9949,
+      "step": 9088
+    },
+    {
+      "epoch": 0.24962236686715553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7575419195900286e-05,
+      "loss": 1.0204,
+      "step": 9089
+    },
+    {
+      "epoch": 0.24964983109499878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7574838501869255e-05,
+      "loss": 0.9313,
+      "step": 9090
+    },
+    {
+      "epoch": 0.249677295322842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7574257747903165e-05,
+      "loss": 1.0605,
+      "step": 9091
+    },
+    {
+      "epoch": 0.24970475955068525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7573676934006614e-05,
+      "loss": 0.9572,
+      "step": 9092
+    },
+    {
+      "epoch": 0.24973222377852847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7573096060184197e-05,
+      "loss": 0.9845,
+      "step": 9093
+    },
+    {
+      "epoch": 0.2497596880063717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757251512644051e-05,
+      "loss": 0.9057,
+      "step": 9094
+    },
+    {
+      "epoch": 0.24978715223421494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757193413278015e-05,
+      "loss": 1.0195,
+      "step": 9095
+    },
+    {
+      "epoch": 0.24981461646205816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7571353079207715e-05,
+      "loss": 0.8328,
+      "step": 9096
+    },
+    {
+      "epoch": 0.2498420806899014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75707719657278e-05,
+      "loss": 1.0363,
+      "step": 9097
+    },
+    {
+      "epoch": 0.24986954491774463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7570190792345006e-05,
+      "loss": 1.0146,
+      "step": 9098
+    },
+    {
+      "epoch": 0.24989700914558788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756960955906393e-05,
+      "loss": 0.9569,
+      "step": 9099
+    },
+    {
+      "epoch": 0.2499244733734311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7569028265889172e-05,
+      "loss": 1.0212,
+      "step": 9100
+    },
+    {
+      "epoch": 0.24995193760127435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756844691282533e-05,
+      "loss": 1.0746,
+      "step": 9101
+    },
+    {
+      "epoch": 0.24997940182911757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7567865499877e-05,
+      "loss": 0.9673,
+      "step": 9102
+    },
+    {
+      "epoch": 0.2500068660569608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756728402704879e-05,
+      "loss": 0.9734,
+      "step": 9103
+    },
+    {
+      "epoch": 0.25003433028480404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75667024943453e-05,
+      "loss": 1.0795,
+      "step": 9104
+    },
+    {
+      "epoch": 0.25006179451264726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7566120901771126e-05,
+      "loss": 0.9459,
+      "step": 9105
+    },
+    {
+      "epoch": 0.25008925874049054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7565539249330873e-05,
+      "loss": 0.9621,
+      "step": 9106
+    },
+    {
+      "epoch": 0.25011672296833376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7564957537029142e-05,
+      "loss": 0.9744,
+      "step": 9107
+    },
+    {
+      "epoch": 0.250144187196177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7564375764870537e-05,
+      "loss": 1.0393,
+      "step": 9108
+    },
+    {
+      "epoch": 0.2501716514240202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7563793932859662e-05,
+      "loss": 0.9689,
+      "step": 9109
+    },
+    {
+      "epoch": 0.2501991156518634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756321204100112e-05,
+      "loss": 0.9875,
+      "step": 9110
+    },
+    {
+      "epoch": 0.2502265798797067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756263008929951e-05,
+      "loss": 0.8959,
+      "step": 9111
+    },
+    {
+      "epoch": 0.2502540441075499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7562048077759448e-05,
+      "loss": 1.1292,
+      "step": 9112
+    },
+    {
+      "epoch": 0.25028150833539314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7561466006385525e-05,
+      "loss": 1.0729,
+      "step": 9113
+    },
+    {
+      "epoch": 0.25030897256323636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756088387518236e-05,
+      "loss": 0.9508,
+      "step": 9114
+    },
+    {
+      "epoch": 0.25033643679107964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756030168415455e-05,
+      "loss": 1.0662,
+      "step": 9115
+    },
+    {
+      "epoch": 0.25036390101892286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7559719433306707e-05,
+      "loss": 0.8984,
+      "step": 9116
+    },
+    {
+      "epoch": 0.2503913652467661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7559137122643432e-05,
+      "loss": 0.9899,
+      "step": 9117
+    },
+    {
+      "epoch": 0.2504188294746093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755855475216934e-05,
+      "loss": 0.8289,
+      "step": 9118
+    },
+    {
+      "epoch": 0.2504462937024526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755797232188903e-05,
+      "loss": 0.9773,
+      "step": 9119
+    },
+    {
+      "epoch": 0.2504737579302958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7557389831807116e-05,
+      "loss": 1.0222,
+      "step": 9120
+    },
+    {
+      "epoch": 0.250501222158139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755680728192821e-05,
+      "loss": 0.9674,
+      "step": 9121
+    },
+    {
+      "epoch": 0.25052868638598225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7556224672256916e-05,
+      "loss": 1.0225,
+      "step": 9122
+    },
+    {
+      "epoch": 0.25055615061382547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7555642002797845e-05,
+      "loss": 0.9645,
+      "step": 9123
+    },
+    {
+      "epoch": 0.25058361484166874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7555059273555608e-05,
+      "loss": 1.0179,
+      "step": 9124
+    },
+    {
+      "epoch": 0.25061107906951197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7554476484534813e-05,
+      "loss": 1.0963,
+      "step": 9125
+    },
+    {
+      "epoch": 0.2506385432973552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755389363574008e-05,
+      "loss": 0.9997,
+      "step": 9126
+    },
+    {
+      "epoch": 0.2506660075251984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755331072717601e-05,
+      "loss": 0.9324,
+      "step": 9127
+    },
+    {
+      "epoch": 0.2506934717530417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7552727758847218e-05,
+      "loss": 0.9502,
+      "step": 9128
+    },
+    {
+      "epoch": 0.2507209359808849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755214473075832e-05,
+      "loss": 0.9079,
+      "step": 9129
+    },
+    {
+      "epoch": 0.25074840020872813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755156164291393e-05,
+      "loss": 1.0412,
+      "step": 9130
+    },
+    {
+      "epoch": 0.25077586443657135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7550978495318655e-05,
+      "loss": 0.9564,
+      "step": 9131
+    },
+    {
+      "epoch": 0.2508033286644146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7550395287977114e-05,
+      "loss": 1.0244,
+      "step": 9132
+    },
+    {
+      "epoch": 0.25083079289225785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7549812020893923e-05,
+      "loss": 1.085,
+      "step": 9133
+    },
+    {
+      "epoch": 0.25085825712010107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7549228694073696e-05,
+      "loss": 1.0016,
+      "step": 9134
+    },
+    {
+      "epoch": 0.2508857213479443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7548645307521044e-05,
+      "loss": 0.93,
+      "step": 9135
+    },
+    {
+      "epoch": 0.2509131855757875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754806186124059e-05,
+      "loss": 1.0728,
+      "step": 9136
+    },
+    {
+      "epoch": 0.2509406498036308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754747835523694e-05,
+      "loss": 0.9489,
+      "step": 9137
+    },
+    {
+      "epoch": 0.250968114031474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7546894789514724e-05,
+      "loss": 0.9144,
+      "step": 9138
+    },
+    {
+      "epoch": 0.25099557825931723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7546311164078547e-05,
+      "loss": 1.0905,
+      "step": 9139
+    },
+    {
+      "epoch": 0.25102304248716045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754572747893304e-05,
+      "loss": 1.0261,
+      "step": 9140
+    },
+    {
+      "epoch": 0.25105050671500373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7545143734082808e-05,
+      "loss": 1.0402,
+      "step": 9141
+    },
+    {
+      "epoch": 0.25107797094284695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7544559929532478e-05,
+      "loss": 1.056,
+      "step": 9142
+    },
+    {
+      "epoch": 0.25110543517069017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7543976065286668e-05,
+      "loss": 0.9244,
+      "step": 9143
+    },
+    {
+      "epoch": 0.2511328993985334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754339214135e-05,
+      "loss": 0.9513,
+      "step": 9144
+    },
+    {
+      "epoch": 0.25116036362637667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754280815772709e-05,
+      "loss": 0.9643,
+      "step": 9145
+    },
+    {
+      "epoch": 0.2511878278542199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7542224114422555e-05,
+      "loss": 1.004,
+      "step": 9146
+    },
+    {
+      "epoch": 0.2512152920820631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7541640011441024e-05,
+      "loss": 1.0733,
+      "step": 9147
+    },
+    {
+      "epoch": 0.25124275630990633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754105584878712e-05,
+      "loss": 0.9907,
+      "step": 9148
+    },
+    {
+      "epoch": 0.25127022053774956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7540471626465456e-05,
+      "loss": 1.045,
+      "step": 9149
+    },
+    {
+      "epoch": 0.25129768476559283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7539887344480662e-05,
+      "loss": 0.9343,
+      "step": 9150
+    },
+    {
+      "epoch": 0.25132514899343605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753930300283736e-05,
+      "loss": 1.1654,
+      "step": 9151
+    },
+    {
+      "epoch": 0.2513526132212793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7538718601540168e-05,
+      "loss": 0.9447,
+      "step": 9152
+    },
+    {
+      "epoch": 0.2513800774491225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7538134140593717e-05,
+      "loss": 1.0835,
+      "step": 9153
+    },
+    {
+      "epoch": 0.2514075416769658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753754962000263e-05,
+      "loss": 1.0681,
+      "step": 9154
+    },
+    {
+      "epoch": 0.251435005904809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7536965039771527e-05,
+      "loss": 0.9973,
+      "step": 9155
+    },
+    {
+      "epoch": 0.2514624701326522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753638039990504e-05,
+      "loss": 1.0185,
+      "step": 9156
+    },
+    {
+      "epoch": 0.25148993436049544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753579570040779e-05,
+      "loss": 0.9977,
+      "step": 9157
+    },
+    {
+      "epoch": 0.2515173985883387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7535210941284406e-05,
+      "loss": 0.986,
+      "step": 9158
+    },
+    {
+      "epoch": 0.25154486281618194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7534626122539515e-05,
+      "loss": 0.972,
+      "step": 9159
+    },
+    {
+      "epoch": 0.25157232704402516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7534041244177743e-05,
+      "loss": 1.0717,
+      "step": 9160
+    },
+    {
+      "epoch": 0.2515997912718684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7533456306203715e-05,
+      "loss": 0.9433,
+      "step": 9161
+    },
+    {
+      "epoch": 0.2516272554997116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753287130862207e-05,
+      "loss": 0.9591,
+      "step": 9162
+    },
+    {
+      "epoch": 0.2516547197275549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7532286251437424e-05,
+      "loss": 0.986,
+      "step": 9163
+    },
+    {
+      "epoch": 0.2516821839553981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753170113465441e-05,
+      "loss": 1.019,
+      "step": 9164
+    },
+    {
+      "epoch": 0.2517096481832413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7531115958277664e-05,
+      "loss": 1.0763,
+      "step": 9165
+    },
+    {
+      "epoch": 0.25173711241108454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7530530722311808e-05,
+      "loss": 0.9323,
+      "step": 9166
+    },
+    {
+      "epoch": 0.2517645766389278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7529945426761473e-05,
+      "loss": 0.9113,
+      "step": 9167
+    },
+    {
+      "epoch": 0.25179204086677104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75293600716313e-05,
+      "loss": 1.1279,
+      "step": 9168
+    },
+    {
+      "epoch": 0.25181950509461426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752877465692591e-05,
+      "loss": 0.9624,
+      "step": 9169
+    },
+    {
+      "epoch": 0.2518469693224575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752818918264994e-05,
+      "loss": 0.9265,
+      "step": 9170
+    },
+    {
+      "epoch": 0.25187443355030076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7527603648808016e-05,
+      "loss": 1.0293,
+      "step": 9171
+    },
+    {
+      "epoch": 0.251901897778144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752701805540478e-05,
+      "loss": 1.0148,
+      "step": 9172
+    },
+    {
+      "epoch": 0.2519293620059872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7526432402444862e-05,
+      "loss": 1.0009,
+      "step": 9173
+    },
+    {
+      "epoch": 0.2519568262338304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7525846689932896e-05,
+      "loss": 1.0533,
+      "step": 9174
+    },
+    {
+      "epoch": 0.25198429046167364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752526091787351e-05,
+      "loss": 0.8778,
+      "step": 9175
+    },
+    {
+      "epoch": 0.2520117546895169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752467508627135e-05,
+      "loss": 0.9398,
+      "step": 9176
+    },
+    {
+      "epoch": 0.25203921891736014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7524089195131045e-05,
+      "loss": 0.988,
+      "step": 9177
+    },
+    {
+      "epoch": 0.25206668314520336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7523503244457233e-05,
+      "loss": 0.9078,
+      "step": 9178
+    },
+    {
+      "epoch": 0.2520941473730466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7522917234254545e-05,
+      "loss": 0.9951,
+      "step": 9179
+    },
+    {
+      "epoch": 0.25212161160088986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7522331164527627e-05,
+      "loss": 0.9075,
+      "step": 9180
+    },
+    {
+      "epoch": 0.2521490758287331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7521745035281106e-05,
+      "loss": 1.1018,
+      "step": 9181
+    },
+    {
+      "epoch": 0.2521765400565763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7521158846519628e-05,
+      "loss": 0.9691,
+      "step": 9182
+    },
+    {
+      "epoch": 0.2522040042844195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7520572598247827e-05,
+      "loss": 1.03,
+      "step": 9183
+    },
+    {
+      "epoch": 0.2522314685122628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7519986290470342e-05,
+      "loss": 1.1274,
+      "step": 9184
+    },
+    {
+      "epoch": 0.252258932740106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751939992319181e-05,
+      "loss": 0.9111,
+      "step": 9185
+    },
+    {
+      "epoch": 0.25228639696794924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7518813496416877e-05,
+      "loss": 0.9903,
+      "step": 9186
+    },
+    {
+      "epoch": 0.25231386119579247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751822701015018e-05,
+      "loss": 0.9486,
+      "step": 9187
+    },
+    {
+      "epoch": 0.2523413254236357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7517640464396356e-05,
+      "loss": 0.9784,
+      "step": 9188
+    },
+    {
+      "epoch": 0.25236878965147896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7517053859160047e-05,
+      "loss": 1.0192,
+      "step": 9189
+    },
+    {
+      "epoch": 0.2523962538793222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7516467194445897e-05,
+      "loss": 1.0954,
+      "step": 9190
+    },
+    {
+      "epoch": 0.2524237181071654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7515880470258548e-05,
+      "loss": 0.8456,
+      "step": 9191
+    },
+    {
+      "epoch": 0.25245118233500863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7515293686602643e-05,
+      "loss": 1.0137,
+      "step": 9192
+    },
+    {
+      "epoch": 0.2524786465628519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7514706843482822e-05,
+      "loss": 1.0377,
+      "step": 9193
+    },
+    {
+      "epoch": 0.2525061107906951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7514119940903728e-05,
+      "loss": 0.9169,
+      "step": 9194
+    },
+    {
+      "epoch": 0.25253357501853835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7513532978870005e-05,
+      "loss": 1.0474,
+      "step": 9195
+    },
+    {
+      "epoch": 0.25256103924638157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7512945957386303e-05,
+      "loss": 0.8694,
+      "step": 9196
+    },
+    {
+      "epoch": 0.25258850347422485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751235887645726e-05,
+      "loss": 1.01,
+      "step": 9197
+    },
+    {
+      "epoch": 0.25261596770206807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7511771736087523e-05,
+      "loss": 0.9917,
+      "step": 9198
+    },
+    {
+      "epoch": 0.2526434319299113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7511184536281742e-05,
+      "loss": 0.9616,
+      "step": 9199
+    },
+    {
+      "epoch": 0.2526708961577545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7510597277044555e-05,
+      "loss": 0.9209,
+      "step": 9200
+    },
+    {
+      "epoch": 0.25269836038559773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7510009958380613e-05,
+      "loss": 0.859,
+      "step": 9201
+    },
+    {
+      "epoch": 0.252725824613441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7509422580294563e-05,
+      "loss": 1.0516,
+      "step": 9202
+    },
+    {
+      "epoch": 0.25275328884128423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7508835142791056e-05,
+      "loss": 1.0139,
+      "step": 9203
+    },
+    {
+      "epoch": 0.25278075306912745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7508247645874736e-05,
+      "loss": 1.0877,
+      "step": 9204
+    },
+    {
+      "epoch": 0.25280821729697067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750766008955025e-05,
+      "loss": 1.0676,
+      "step": 9205
+    },
+    {
+      "epoch": 0.25283568152481395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7507072473822252e-05,
+      "loss": 1.0239,
+      "step": 9206
+    },
+    {
+      "epoch": 0.25286314575265717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7506484798695388e-05,
+      "loss": 0.8934,
+      "step": 9207
+    },
+    {
+      "epoch": 0.2528906099805004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7505897064174306e-05,
+      "loss": 0.9539,
+      "step": 9208
+    },
+    {
+      "epoch": 0.2529180742083436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750530927026366e-05,
+      "loss": 0.9395,
+      "step": 9209
+    },
+    {
+      "epoch": 0.2529455384361869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75047214169681e-05,
+      "loss": 1.0377,
+      "step": 9210
+    },
+    {
+      "epoch": 0.2529730026640301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7504133504292277e-05,
+      "loss": 0.9594,
+      "step": 9211
+    },
+    {
+      "epoch": 0.25300046689187333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7503545532240843e-05,
+      "loss": 0.9905,
+      "step": 9212
+    },
+    {
+      "epoch": 0.25302793111971655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7502957500818453e-05,
+      "loss": 1.0764,
+      "step": 9213
+    },
+    {
+      "epoch": 0.25305539534755983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7502369410029753e-05,
+      "loss": 1.0745,
+      "step": 9214
+    },
+    {
+      "epoch": 0.25308285957540305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7501781259879402e-05,
+      "loss": 1.0046,
+      "step": 9215
+    },
+    {
+      "epoch": 0.2531103238032463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750119305037205e-05,
+      "loss": 1.0019,
+      "step": 9216
+    },
+    {
+      "epoch": 0.2531377880310895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7500604781512354e-05,
+      "loss": 1.0189,
+      "step": 9217
+    },
+    {
+      "epoch": 0.2531652522589327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7500016453304968e-05,
+      "loss": 1.0312,
+      "step": 9218
+    },
+    {
+      "epoch": 0.253192716486776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7499428065754545e-05,
+      "loss": 0.9571,
+      "step": 9219
+    },
+    {
+      "epoch": 0.2532201807146192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7498839618865743e-05,
+      "loss": 0.9555,
+      "step": 9220
+    },
+    {
+      "epoch": 0.25324764494246244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749825111264322e-05,
+      "loss": 1.0059,
+      "step": 9221
+    },
+    {
+      "epoch": 0.25327510917030566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7497662547091624e-05,
+      "loss": 1.0167,
+      "step": 9222
+    },
+    {
+      "epoch": 0.25330257339814893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749707392221562e-05,
+      "loss": 1.0493,
+      "step": 9223
+    },
+    {
+      "epoch": 0.25333003762599215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7496485238019863e-05,
+      "loss": 1.111,
+      "step": 9224
+    },
+    {
+      "epoch": 0.2533575018538354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7495896494509014e-05,
+      "loss": 0.9958,
+      "step": 9225
+    },
+    {
+      "epoch": 0.2533849660816786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7495307691687726e-05,
+      "loss": 0.9967,
+      "step": 9226
+    },
+    {
+      "epoch": 0.2534124303095219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749471882956066e-05,
+      "loss": 1.0192,
+      "step": 9227
+    },
+    {
+      "epoch": 0.2534398945373651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7494129908132475e-05,
+      "loss": 1.0111,
+      "step": 9228
+    },
+    {
+      "epoch": 0.2534673587652083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7493540927407833e-05,
+      "loss": 0.9741,
+      "step": 9229
+    },
+    {
+      "epoch": 0.25349482299305154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7492951887391393e-05,
+      "loss": 0.9908,
+      "step": 9230
+    },
+    {
+      "epoch": 0.25352228722089476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749236278808781e-05,
+      "loss": 0.9272,
+      "step": 9231
+    },
+    {
+      "epoch": 0.25354975144873804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7491773629501754e-05,
+      "loss": 1.0515,
+      "step": 9232
+    },
+    {
+      "epoch": 0.25357721567658126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7491184411637886e-05,
+      "loss": 0.9816,
+      "step": 9233
+    },
+    {
+      "epoch": 0.2536046799044245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7490595134500858e-05,
+      "loss": 1.0943,
+      "step": 9234
+    },
+    {
+      "epoch": 0.2536321441322677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7490005798095345e-05,
+      "loss": 0.9208,
+      "step": 9235
+    },
+    {
+      "epoch": 0.253659608360111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7489416402426003e-05,
+      "loss": 1.0231,
+      "step": 9236
+    },
+    {
+      "epoch": 0.2536870725879542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7488826947497498e-05,
+      "loss": 0.9881,
+      "step": 9237
+    },
+    {
+      "epoch": 0.2537145368157974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748823743331449e-05,
+      "loss": 1.018,
+      "step": 9238
+    },
+    {
+      "epoch": 0.25374200104364064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7487647859881655e-05,
+      "loss": 1.0216,
+      "step": 9239
+    },
+    {
+      "epoch": 0.2537694652714839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748705822720364e-05,
+      "loss": 0.9324,
+      "step": 9240
+    },
+    {
+      "epoch": 0.25379692949932714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748646853528513e-05,
+      "loss": 1.0229,
+      "step": 9241
+    },
+    {
+      "epoch": 0.25382439372717036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748587878413077e-05,
+      "loss": 1.0581,
+      "step": 9242
+    },
+    {
+      "epoch": 0.2538518579550136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7485288973745245e-05,
+      "loss": 1.0125,
+      "step": 9243
+    },
+    {
+      "epoch": 0.2538793221828568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748469910413321e-05,
+      "loss": 0.9919,
+      "step": 9244
+    },
+    {
+      "epoch": 0.2539067864107001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7484109175299342e-05,
+      "loss": 0.9704,
+      "step": 9245
+    },
+    {
+      "epoch": 0.2539342506385433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7483519187248297e-05,
+      "loss": 0.9893,
+      "step": 9246
+    },
+    {
+      "epoch": 0.2539617148663865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7482929139984752e-05,
+      "loss": 1.092,
+      "step": 9247
+    },
+    {
+      "epoch": 0.25398917909422974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7482339033513374e-05,
+      "loss": 1.0223,
+      "step": 9248
+    },
+    {
+      "epoch": 0.254016643322073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748174886783883e-05,
+      "loss": 1.0786,
+      "step": 9249
+    },
+    {
+      "epoch": 0.25404410754991624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748115864296579e-05,
+      "loss": 0.904,
+      "step": 9250
+    },
+    {
+      "epoch": 0.25407157177775946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7480568358898924e-05,
+      "loss": 0.9619,
+      "step": 9251
+    },
+    {
+      "epoch": 0.2540990360056027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7479978015642908e-05,
+      "loss": 0.9992,
+      "step": 9252
+    },
+    {
+      "epoch": 0.25412650023344596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7479387613202402e-05,
+      "loss": 0.9917,
+      "step": 9253
+    },
+    {
+      "epoch": 0.2541539644612892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747879715158209e-05,
+      "loss": 0.9478,
+      "step": 9254
+    },
+    {
+      "epoch": 0.2541814286891324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7478206630786636e-05,
+      "loss": 0.9868,
+      "step": 9255
+    },
+    {
+      "epoch": 0.2542088929169756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7477616050820714e-05,
+      "loss": 1.0514,
+      "step": 9256
+    },
+    {
+      "epoch": 0.25423635714481885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7477025411688997e-05,
+      "loss": 1.023,
+      "step": 9257
+    },
+    {
+      "epoch": 0.2542638213726621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747643471339616e-05,
+      "loss": 0.9937,
+      "step": 9258
+    },
+    {
+      "epoch": 0.25429128560050535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747584395594687e-05,
+      "loss": 0.9938,
+      "step": 9259
+    },
+    {
+      "epoch": 0.25431874982834857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7475253139345814e-05,
+      "loss": 0.9405,
+      "step": 9260
+    },
+    {
+      "epoch": 0.2543462140561918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7474662263597656e-05,
+      "loss": 0.9438,
+      "step": 9261
+    },
+    {
+      "epoch": 0.25437367828403507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7474071328707073e-05,
+      "loss": 1.0808,
+      "step": 9262
+    },
+    {
+      "epoch": 0.2544011425118783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7473480334678747e-05,
+      "loss": 1.0377,
+      "step": 9263
+    },
+    {
+      "epoch": 0.2544286067397215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747288928151735e-05,
+      "loss": 0.9723,
+      "step": 9264
+    },
+    {
+      "epoch": 0.25445607096756473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7472298169227554e-05,
+      "loss": 0.9158,
+      "step": 9265
+    },
+    {
+      "epoch": 0.254483535195408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747170699781404e-05,
+      "loss": 0.9129,
+      "step": 9266
+    },
+    {
+      "epoch": 0.2545109994232512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747111576728149e-05,
+      "loss": 0.998,
+      "step": 9267
+    },
+    {
+      "epoch": 0.25453846365109445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7470524477634575e-05,
+      "loss": 1.1207,
+      "step": 9268
+    },
+    {
+      "epoch": 0.25456592787893767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7469933128877978e-05,
+      "loss": 1.056,
+      "step": 9269
+    },
+    {
+      "epoch": 0.2545933921067809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7469341721016374e-05,
+      "loss": 1.0962,
+      "step": 9270
+    },
+    {
+      "epoch": 0.25462085633462417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746875025405445e-05,
+      "loss": 0.9314,
+      "step": 9271
+    },
+    {
+      "epoch": 0.2546483205624674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7468158727996875e-05,
+      "loss": 1.0198,
+      "step": 9272
+    },
+    {
+      "epoch": 0.2546757847903106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746756714284834e-05,
+      "loss": 0.9639,
+      "step": 9273
+    },
+    {
+      "epoch": 0.25470324901815383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746697549861352e-05,
+      "loss": 1.0652,
+      "step": 9274
+    },
+    {
+      "epoch": 0.2547307132459971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7466383795297096e-05,
+      "loss": 1.1061,
+      "step": 9275
+    },
+    {
+      "epoch": 0.25475817747384033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746579203290375e-05,
+      "loss": 0.9978,
+      "step": 9276
+    },
+    {
+      "epoch": 0.25478564170168355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746520021143817e-05,
+      "loss": 1.1213,
+      "step": 9277
+    },
+    {
+      "epoch": 0.2548131059295268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746460833090503e-05,
+      "loss": 0.9927,
+      "step": 9278
+    },
+    {
+      "epoch": 0.25484057015737005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746401639130902e-05,
+      "loss": 0.968,
+      "step": 9279
+    },
+    {
+      "epoch": 0.25486803438521327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746342439265482e-05,
+      "loss": 1.0353,
+      "step": 9280
+    },
+    {
+      "epoch": 0.2548954986130565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7462832334947113e-05,
+      "loss": 0.9613,
+      "step": 9281
+    },
+    {
+      "epoch": 0.2549229628408997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7462240218190588e-05,
+      "loss": 1.0542,
+      "step": 9282
+    },
+    {
+      "epoch": 0.25495042706874294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746164804238993e-05,
+      "loss": 0.9533,
+      "step": 9283
+    },
+    {
+      "epoch": 0.2549778912965862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746105580754982e-05,
+      "loss": 0.9128,
+      "step": 9284
+    },
+    {
+      "epoch": 0.25500535552442943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7460463513674947e-05,
+      "loss": 0.9325,
+      "step": 9285
+    },
+    {
+      "epoch": 0.25503281975227265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7459871160769993e-05,
+      "loss": 0.8774,
+      "step": 9286
+    },
+    {
+      "epoch": 0.2550602839801159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7459278748839654e-05,
+      "loss": 1.069,
+      "step": 9287
+    },
+    {
+      "epoch": 0.25508774820795915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7458686277888607e-05,
+      "loss": 1.0929,
+      "step": 9288
+    },
+    {
+      "epoch": 0.2551152124358024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7458093747921553e-05,
+      "loss": 0.9817,
+      "step": 9289
+    },
+    {
+      "epoch": 0.2551426766636456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.745750115894316e-05,
+      "loss": 1.0335,
+      "step": 9290
+    },
+    {
+      "epoch": 0.2551701408914888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7456908510958142e-05,
+      "loss": 1.0217,
+      "step": 9291
+    },
+    {
+      "epoch": 0.2551976051193321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7456315803971166e-05,
+      "loss": 0.9716,
+      "step": 9292
+    },
+    {
+      "epoch": 0.2552250693471753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7455723037986937e-05,
+      "loss": 0.8291,
+      "step": 9293
+    },
+    {
+      "epoch": 0.25525253357501854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7455130213010135e-05,
+      "loss": 0.9686,
+      "step": 9294
+    },
+    {
+      "epoch": 0.25527999780286176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7454537329045457e-05,
+      "loss": 0.9582,
+      "step": 9295
+    },
+    {
+      "epoch": 0.255307462030705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7453944386097593e-05,
+      "loss": 1.0271,
+      "step": 9296
+    },
+    {
+      "epoch": 0.25533492625854826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.745335138417123e-05,
+      "loss": 0.9905,
+      "step": 9297
+    },
+    {
+      "epoch": 0.2553623904863915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7452758323271068e-05,
+      "loss": 1.0048,
+      "step": 9298
+    },
+    {
+      "epoch": 0.2553898547142347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7452165203401792e-05,
+      "loss": 1.0173,
+      "step": 9299
+    },
+    {
+      "epoch": 0.2554173189420779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74515720245681e-05,
+      "loss": 1.0538,
+      "step": 9300
+    },
+    {
+      "epoch": 0.2554447831699212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7450978786774684e-05,
+      "loss": 1.0376,
+      "step": 9301
+    },
+    {
+      "epoch": 0.2554722473977644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7450385490026236e-05,
+      "loss": 0.9997,
+      "step": 9302
+    },
+    {
+      "epoch": 0.25549971162560764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7449792134327453e-05,
+      "loss": 1.0001,
+      "step": 9303
+    },
+    {
+      "epoch": 0.25552717585345086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7449198719683028e-05,
+      "loss": 0.9421,
+      "step": 9304
+    },
+    {
+      "epoch": 0.25555464008129414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744860524609766e-05,
+      "loss": 0.9433,
+      "step": 9305
+    },
+    {
+      "epoch": 0.25558210430913736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744801171357604e-05,
+      "loss": 1.0101,
+      "step": 9306
+    },
+    {
+      "epoch": 0.2556095685369806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7447418122122863e-05,
+      "loss": 1.0265,
+      "step": 9307
+    },
+    {
+      "epoch": 0.2556370327648238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7446824471742833e-05,
+      "loss": 1.0455,
+      "step": 9308
+    },
+    {
+      "epoch": 0.2556644969926671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7446230762440642e-05,
+      "loss": 0.955,
+      "step": 9309
+    },
+    {
+      "epoch": 0.2556919612205103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445636994220988e-05,
+      "loss": 0.8837,
+      "step": 9310
+    },
+    {
+      "epoch": 0.2557194254483535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445043167088573e-05,
+      "loss": 1.0234,
+      "step": 9311
+    },
+    {
+      "epoch": 0.25574688967619674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744444928104809e-05,
+      "loss": 1.0025,
+      "step": 9312
+    },
+    {
+      "epoch": 0.25577435390403996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7443855336104243e-05,
+      "loss": 0.9669,
+      "step": 9313
+    },
+    {
+      "epoch": 0.25580181813188324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7443261332261724e-05,
+      "loss": 0.9027,
+      "step": 9314
+    },
+    {
+      "epoch": 0.25582928235972646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744266726952524e-05,
+      "loss": 1.0264,
+      "step": 9315
+    },
+    {
+      "epoch": 0.2558567465875697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744207314789949e-05,
+      "loss": 1.195,
+      "step": 9316
+    },
+    {
+      "epoch": 0.2558842108154129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744147896738918e-05,
+      "loss": 0.9353,
+      "step": 9317
+    },
+    {
+      "epoch": 0.2559116750432562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7440884727998998e-05,
+      "loss": 0.9476,
+      "step": 9318
+    },
+    {
+      "epoch": 0.2559391392710994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7440290429733657e-05,
+      "loss": 1.0071,
+      "step": 9319
+    },
+    {
+      "epoch": 0.2559666034989426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7439696072597856e-05,
+      "loss": 1.0314,
+      "step": 9320
+    },
+    {
+      "epoch": 0.25599406772678585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7439101656596298e-05,
+      "loss": 0.9302,
+      "step": 9321
+    },
+    {
+      "epoch": 0.2560215319546291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7438507181733685e-05,
+      "loss": 1.0102,
+      "step": 9322
+    },
+    {
+      "epoch": 0.25604899618247234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7437912648014724e-05,
+      "loss": 1.1216,
+      "step": 9323
+    },
+    {
+      "epoch": 0.25607646041031557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7437318055444116e-05,
+      "loss": 0.9276,
+      "step": 9324
+    },
+    {
+      "epoch": 0.2561039246381588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7436723404026566e-05,
+      "loss": 0.9278,
+      "step": 9325
+    },
+    {
+      "epoch": 0.256131388866002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743612869376678e-05,
+      "loss": 0.9455,
+      "step": 9326
+    },
+    {
+      "epoch": 0.2561588530938453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7435533924669463e-05,
+      "loss": 1.0551,
+      "step": 9327
+    },
+    {
+      "epoch": 0.2561863173216885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743493909673932e-05,
+      "loss": 1.0467,
+      "step": 9328
+    },
+    {
+      "epoch": 0.2562137815495317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743434420998106e-05,
+      "loss": 1.0544,
+      "step": 9329
+    },
+    {
+      "epoch": 0.25624124577737495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743374926439939e-05,
+      "loss": 0.9752,
+      "step": 9330
+    },
+    {
+      "epoch": 0.2562687100052182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7433154259999018e-05,
+      "loss": 1.0468,
+      "step": 9331
+    },
+    {
+      "epoch": 0.25629617423306145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7432559196784647e-05,
+      "loss": 1.0746,
+      "step": 9332
+    },
+    {
+      "epoch": 0.25632363846090467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7431964074760988e-05,
+      "loss": 0.9769,
+      "step": 9333
+    },
+    {
+      "epoch": 0.2563511026887479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743136889393275e-05,
+      "loss": 1.0709,
+      "step": 9334
+    },
+    {
+      "epoch": 0.25637856691659117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7430773654304646e-05,
+      "loss": 0.8792,
+      "step": 9335
+    },
+    {
+      "epoch": 0.2564060311444344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743017835588138e-05,
+      "loss": 0.9438,
+      "step": 9336
+    },
+    {
+      "epoch": 0.2564334953722776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7429582998667665e-05,
+      "loss": 0.9437,
+      "step": 9337
+    },
+    {
+      "epoch": 0.25646095960012083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742898758266821e-05,
+      "loss": 1.0102,
+      "step": 9338
+    },
+    {
+      "epoch": 0.25648842382796405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7428392107887727e-05,
+      "loss": 0.9189,
+      "step": 9339
+    },
+    {
+      "epoch": 0.25651588805580733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7427796574330932e-05,
+      "loss": 0.9232,
+      "step": 9340
+    },
+    {
+      "epoch": 0.25654335228365055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7427200982002533e-05,
+      "loss": 0.9052,
+      "step": 9341
+    },
+    {
+      "epoch": 0.25657081651149377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742660533090724e-05,
+      "loss": 0.9862,
+      "step": 9342
+    },
+    {
+      "epoch": 0.256598280739337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742600962104977e-05,
+      "loss": 0.9435,
+      "step": 9343
+    },
+    {
+      "epoch": 0.25662574496718027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7425413852434835e-05,
+      "loss": 1.0598,
+      "step": 9344
+    },
+    {
+      "epoch": 0.2566532091950235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7424818025067153e-05,
+      "loss": 0.973,
+      "step": 9345
+    },
+    {
+      "epoch": 0.2566806734228667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742422213895143e-05,
+      "loss": 0.924,
+      "step": 9346
+    },
+    {
+      "epoch": 0.25670813765070993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7423626194092387e-05,
+      "loss": 0.9587,
+      "step": 9347
+    },
+    {
+      "epoch": 0.2567356018785532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742303019049474e-05,
+      "loss": 1.0037,
+      "step": 9348
+    },
+    {
+      "epoch": 0.25676306610639643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7422434128163197e-05,
+      "loss": 1.0882,
+      "step": 9349
+    },
+    {
+      "epoch": 0.25679053033423965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7421838007102483e-05,
+      "loss": 0.9282,
+      "step": 9350
+    },
+    {
+      "epoch": 0.2568179945620829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7421241827317314e-05,
+      "loss": 1.1208,
+      "step": 9351
+    },
+    {
+      "epoch": 0.2568454587899261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7420645588812402e-05,
+      "loss": 1.0341,
+      "step": 9352
+    },
+    {
+      "epoch": 0.2568729230177694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742004929159247e-05,
+      "loss": 1.0991,
+      "step": 9353
+    },
+    {
+      "epoch": 0.2569003872456126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7419452935662233e-05,
+      "loss": 1.0674,
+      "step": 9354
+    },
+    {
+      "epoch": 0.2569278514734558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741885652102641e-05,
+      "loss": 0.8755,
+      "step": 9355
+    },
+    {
+      "epoch": 0.25695531570129904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741826004768972e-05,
+      "loss": 1.0079,
+      "step": 9356
+    },
+    {
+      "epoch": 0.2569827799291423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7417663515656883e-05,
+      "loss": 0.9666,
+      "step": 9357
+    },
+    {
+      "epoch": 0.25701024415698553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7417066924932623e-05,
+      "loss": 1.0767,
+      "step": 9358
+    },
+    {
+      "epoch": 0.25703770838482876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741647027552165e-05,
+      "loss": 0.9816,
+      "step": 9359
+    },
+    {
+      "epoch": 0.257065172612672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7415873567428696e-05,
+      "loss": 0.9737,
+      "step": 9360
+    },
+    {
+      "epoch": 0.25709263684051525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7415276800658473e-05,
+      "loss": 0.9277,
+      "step": 9361
+    },
+    {
+      "epoch": 0.2571201010683585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741467997521571e-05,
+      "loss": 0.9644,
+      "step": 9362
+    },
+    {
+      "epoch": 0.2571475652962017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741408309110513e-05,
+      "loss": 0.9752,
+      "step": 9363
+    },
+    {
+      "epoch": 0.2571750295240449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7413486148331448e-05,
+      "loss": 1.0093,
+      "step": 9364
+    },
+    {
+      "epoch": 0.25720249375188814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7412889146899393e-05,
+      "loss": 1.0605,
+      "step": 9365
+    },
+    {
+      "epoch": 0.2572299579797314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7412292086813688e-05,
+      "loss": 1.0691,
+      "step": 9366
+    },
+    {
+      "epoch": 0.25725742220757464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741169496807906e-05,
+      "loss": 0.9119,
+      "step": 9367
+    },
+    {
+      "epoch": 0.25728488643541786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7411097790700227e-05,
+      "loss": 0.9874,
+      "step": 9368
+    },
+    {
+      "epoch": 0.2573123506632611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7410500554681916e-05,
+      "loss": 0.9327,
+      "step": 9369
+    },
+    {
+      "epoch": 0.25733981489110436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7409903260028858e-05,
+      "loss": 1.009,
+      "step": 9370
+    },
+    {
+      "epoch": 0.2573672791189476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740930590674577e-05,
+      "loss": 1.0531,
+      "step": 9371
+    },
+    {
+      "epoch": 0.2573947433467908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7408708494837386e-05,
+      "loss": 0.9812,
+      "step": 9372
+    },
+    {
+      "epoch": 0.257422207574634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7408111024308433e-05,
+      "loss": 0.9792,
+      "step": 9373
+    },
+    {
+      "epoch": 0.2574496718024773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7407513495163634e-05,
+      "loss": 1.0847,
+      "step": 9374
+    },
+    {
+      "epoch": 0.2574771360303205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740691590740772e-05,
+      "loss": 0.9927,
+      "step": 9375
+    },
+    {
+      "epoch": 0.25750460025816374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740631826104542e-05,
+      "loss": 1.0348,
+      "step": 9376
+    },
+    {
+      "epoch": 0.25753206448600696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7405720556081456e-05,
+      "loss": 0.9101,
+      "step": 9377
+    },
+    {
+      "epoch": 0.2575595287138502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7405122792520564e-05,
+      "loss": 0.8997,
+      "step": 9378
+    },
+    {
+      "epoch": 0.25758699294169346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740452497036747e-05,
+      "loss": 1.0355,
+      "step": 9379
+    },
+    {
+      "epoch": 0.2576144571695367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740392708962691e-05,
+      "loss": 0.9988,
+      "step": 9380
+    },
+    {
+      "epoch": 0.2576419213973799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740332915030361e-05,
+      "loss": 1.0657,
+      "step": 9381
+    },
+    {
+      "epoch": 0.2576693856252231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74027311524023e-05,
+      "loss": 0.9452,
+      "step": 9382
+    },
+    {
+      "epoch": 0.2576968498530664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7402133095927715e-05,
+      "loss": 0.9746,
+      "step": 9383
+    },
+    {
+      "epoch": 0.2577243140809096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7401534980884586e-05,
+      "loss": 1.0067,
+      "step": 9384
+    },
+    {
+      "epoch": 0.25775177830875284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7400936807277645e-05,
+      "loss": 0.9454,
+      "step": 9385
+    },
+    {
+      "epoch": 0.25777924253659606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7400338575111624e-05,
+      "loss": 0.8957,
+      "step": 9386
+    },
+    {
+      "epoch": 0.25780670676443934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739974028439126e-05,
+      "loss": 1.0439,
+      "step": 9387
+    },
+    {
+      "epoch": 0.25783417099228256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7399141935121283e-05,
+      "loss": 1.0347,
+      "step": 9388
+    },
+    {
+      "epoch": 0.2578616352201258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739854352730643e-05,
+      "loss": 1.0247,
+      "step": 9389
+    },
+    {
+      "epoch": 0.257889099447969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7397945060951432e-05,
+      "loss": 0.9665,
+      "step": 9390
+    },
+    {
+      "epoch": 0.2579165636758122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739734653606103e-05,
+      "loss": 1.0244,
+      "step": 9391
+    },
+    {
+      "epoch": 0.2579440279036555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7396747952639956e-05,
+      "loss": 0.9892,
+      "step": 9392
+    },
+    {
+      "epoch": 0.2579714921314987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7396149310692944e-05,
+      "loss": 1.021,
+      "step": 9393
+    },
+    {
+      "epoch": 0.25799895635934195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7395550610224737e-05,
+      "loss": 0.9986,
+      "step": 9394
+    },
+    {
+      "epoch": 0.25802642058718517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7394951851240066e-05,
+      "loss": 0.9971,
+      "step": 9395
+    },
+    {
+      "epoch": 0.25805388481502844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7394353033743674e-05,
+      "loss": 0.875,
+      "step": 9396
+    },
+    {
+      "epoch": 0.25808134904287167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7393754157740293e-05,
+      "loss": 0.9514,
+      "step": 9397
+    },
+    {
+      "epoch": 0.2581088132707149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739315522323467e-05,
+      "loss": 0.8088,
+      "step": 9398
+    },
+    {
+      "epoch": 0.2581362774985581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7392556230231537e-05,
+      "loss": 0.9926,
+      "step": 9399
+    },
+    {
+      "epoch": 0.2581637417264014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7391957178735635e-05,
+      "loss": 1.0181,
+      "step": 9400
+    },
+    {
+      "epoch": 0.2581912059542446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7391358068751705e-05,
+      "loss": 1.0317,
+      "step": 9401
+    },
+    {
+      "epoch": 0.25821867018208783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7390758900284487e-05,
+      "loss": 0.9909,
+      "step": 9402
+    },
+    {
+      "epoch": 0.25824613440993105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739015967333872e-05,
+      "loss": 1.0505,
+      "step": 9403
+    },
+    {
+      "epoch": 0.2582735986377743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738956038791915e-05,
+      "loss": 0.9879,
+      "step": 9404
+    },
+    {
+      "epoch": 0.25830106286561755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7388961044030513e-05,
+      "loss": 0.9774,
+      "step": 9405
+    },
+    {
+      "epoch": 0.25832852709346077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7388361641677556e-05,
+      "loss": 0.8643,
+      "step": 9406
+    },
+    {
+      "epoch": 0.258355991321304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7387762180865017e-05,
+      "loss": 1.0591,
+      "step": 9407
+    },
+    {
+      "epoch": 0.2583834555491472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7387162661597643e-05,
+      "loss": 0.9471,
+      "step": 9408
+    },
+    {
+      "epoch": 0.2584109197769905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7386563083880178e-05,
+      "loss": 1.0529,
+      "step": 9409
+    },
+    {
+      "epoch": 0.2584383840048337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7385963447717365e-05,
+      "loss": 0.8959,
+      "step": 9410
+    },
+    {
+      "epoch": 0.25846584823267693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7385363753113944e-05,
+      "loss": 0.9752,
+      "step": 9411
+    },
+    {
+      "epoch": 0.25849331246052015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738476400007467e-05,
+      "loss": 1.052,
+      "step": 9412
+    },
+    {
+      "epoch": 0.25852077668836343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7384164188604276e-05,
+      "loss": 1.0055,
+      "step": 9413
+    },
+    {
+      "epoch": 0.25854824091620665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738356431870752e-05,
+      "loss": 1.061,
+      "step": 9414
+    },
+    {
+      "epoch": 0.25857570514404987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738296439038914e-05,
+      "loss": 1.0648,
+      "step": 9415
+    },
+    {
+      "epoch": 0.2586031693718931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738236440365389e-05,
+      "loss": 0.8619,
+      "step": 9416
+    },
+    {
+      "epoch": 0.25863063359973637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7381764358506507e-05,
+      "loss": 0.9264,
+      "step": 9417
+    },
+    {
+      "epoch": 0.2586580978275796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738116425495175e-05,
+      "loss": 1.0306,
+      "step": 9418
+    },
+    {
+      "epoch": 0.2586855620554228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7380564092994357e-05,
+      "loss": 0.9037,
+      "step": 9419
+    },
+    {
+      "epoch": 0.25871302628326603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7379963872639087e-05,
+      "loss": 1.0789,
+      "step": 9420
+    },
+    {
+      "epoch": 0.25874049051110926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7379363593890683e-05,
+      "loss": 0.9544,
+      "step": 9421
+    },
+    {
+      "epoch": 0.25876795473895253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7378763256753893e-05,
+      "loss": 0.9555,
+      "step": 9422
+    },
+    {
+      "epoch": 0.25879541896679575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7378162861233473e-05,
+      "loss": 1.0753,
+      "step": 9423
+    },
+    {
+      "epoch": 0.258822883194639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737756240733417e-05,
+      "loss": 1.0153,
+      "step": 9424
+    },
+    {
+      "epoch": 0.2588503474224822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7376961895060735e-05,
+      "loss": 0.9599,
+      "step": 9425
+    },
+    {
+      "epoch": 0.2588778116503255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737636132441792e-05,
+      "loss": 1.1006,
+      "step": 9426
+    },
+    {
+      "epoch": 0.2589052758781687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7375760695410476e-05,
+      "loss": 0.9144,
+      "step": 9427
+    },
+    {
+      "epoch": 0.2589327401060119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7375160008043157e-05,
+      "loss": 0.9471,
+      "step": 9428
+    },
+    {
+      "epoch": 0.25896020433385514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7374559262320716e-05,
+      "loss": 1.0554,
+      "step": 9429
+    },
+    {
+      "epoch": 0.2589876685616984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7373958458247906e-05,
+      "loss": 1.0323,
+      "step": 9430
+    },
+    {
+      "epoch": 0.25901513278954164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7373357595829475e-05,
+      "loss": 0.936,
+      "step": 9431
+    },
+    {
+      "epoch": 0.25904259701738486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737275667507019e-05,
+      "loss": 0.957,
+      "step": 9432
+    },
+    {
+      "epoch": 0.2590700612452281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7372155695974795e-05,
+      "loss": 1.0031,
+      "step": 9433
+    },
+    {
+      "epoch": 0.2590975254730713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7371554658548048e-05,
+      "loss": 0.9314,
+      "step": 9434
+    },
+    {
+      "epoch": 0.2591249897009146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7370953562794703e-05,
+      "loss": 0.955,
+      "step": 9435
+    },
+    {
+      "epoch": 0.2591524539287578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737035240871952e-05,
+      "loss": 0.9965,
+      "step": 9436
+    },
+    {
+      "epoch": 0.259179918156601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7369751196327255e-05,
+      "loss": 0.9285,
+      "step": 9437
+    },
+    {
+      "epoch": 0.25920738238444424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7369149925622663e-05,
+      "loss": 0.9189,
+      "step": 9438
+    },
+    {
+      "epoch": 0.2592348466122875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7368548596610502e-05,
+      "loss": 0.9142,
+      "step": 9439
+    },
+    {
+      "epoch": 0.25926231084013074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736794720929553e-05,
+      "loss": 1.0676,
+      "step": 9440
+    },
+    {
+      "epoch": 0.25928977506797396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7367345763682505e-05,
+      "loss": 1.11,
+      "step": 9441
+    },
+    {
+      "epoch": 0.2593172392958172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736674425977619e-05,
+      "loss": 0.93,
+      "step": 9442
+    },
+    {
+      "epoch": 0.25934470352366046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7366142697581337e-05,
+      "loss": 1.0121,
+      "step": 9443
+    },
+    {
+      "epoch": 0.2593721677515037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736554107710271e-05,
+      "loss": 0.8776,
+      "step": 9444
+    },
+    {
+      "epoch": 0.2593996319793469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736493939834507e-05,
+      "loss": 1.0657,
+      "step": 9445
+    },
+    {
+      "epoch": 0.2594270962071901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7364337661313178e-05,
+      "loss": 0.9291,
+      "step": 9446
+    },
+    {
+      "epoch": 0.25945456043503334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736373586601179e-05,
+      "loss": 0.8714,
+      "step": 9447
+    },
+    {
+      "epoch": 0.2594820246628766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7363134012445675e-05,
+      "loss": 1.0339,
+      "step": 9448
+    },
+    {
+      "epoch": 0.25950948889071984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7362532100619592e-05,
+      "loss": 0.894,
+      "step": 9449
+    },
+    {
+      "epoch": 0.25953695311856306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7361930130538297e-05,
+      "loss": 1.0393,
+      "step": 9450
+    },
+    {
+      "epoch": 0.2595644173464063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7361328102206566e-05,
+      "loss": 1.0447,
+      "step": 9451
+    },
+    {
+      "epoch": 0.25959188157424956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7360726015629153e-05,
+      "loss": 0.9889,
+      "step": 9452
+    },
+    {
+      "epoch": 0.2596193458020928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7360123870810822e-05,
+      "loss": 1.1187,
+      "step": 9453
+    },
+    {
+      "epoch": 0.259646810029936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735952166775634e-05,
+      "loss": 1.0359,
+      "step": 9454
+    },
+    {
+      "epoch": 0.2596742742577792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7358919406470477e-05,
+      "loss": 0.959,
+      "step": 9455
+    },
+    {
+      "epoch": 0.2597017384856225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7358317086957987e-05,
+      "loss": 0.9326,
+      "step": 9456
+    },
+    {
+      "epoch": 0.2597292027134657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7357714709223644e-05,
+      "loss": 0.9565,
+      "step": 9457
+    },
+    {
+      "epoch": 0.25975666694130894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7357112273272215e-05,
+      "loss": 0.9102,
+      "step": 9458
+    },
+    {
+      "epoch": 0.25978413116915217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735650977910846e-05,
+      "loss": 0.8408,
+      "step": 9459
+    },
+    {
+      "epoch": 0.2598115953969954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7355907226737154e-05,
+      "loss": 1.0586,
+      "step": 9460
+    },
+    {
+      "epoch": 0.25983905962483866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735530461616306e-05,
+      "loss": 0.85,
+      "step": 9461
+    },
+    {
+      "epoch": 0.2598665238526819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735470194739094e-05,
+      "loss": 1.0502,
+      "step": 9462
+    },
+    {
+      "epoch": 0.2598939880805251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7354099220425577e-05,
+      "loss": 1.0089,
+      "step": 9463
+    },
+    {
+      "epoch": 0.25992145230836833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7353496435271727e-05,
+      "loss": 0.9672,
+      "step": 9464
+    },
+    {
+      "epoch": 0.2599489165362116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7352893591934168e-05,
+      "loss": 1.0095,
+      "step": 9465
+    },
+    {
+      "epoch": 0.2599763807640548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7352290690417666e-05,
+      "loss": 0.9354,
+      "step": 9466
+    },
+    {
+      "epoch": 0.26000384499189805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7351687730726992e-05,
+      "loss": 1.0077,
+      "step": 9467
+    },
+    {
+      "epoch": 0.26003130921974127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735108471286692e-05,
+      "loss": 0.9805,
+      "step": 9468
+    },
+    {
+      "epoch": 0.26005877344758455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7350481636842213e-05,
+      "loss": 0.7898,
+      "step": 9469
+    },
+    {
+      "epoch": 0.26008623767542777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734987850265765e-05,
+      "loss": 1.0381,
+      "step": 9470
+    },
+    {
+      "epoch": 0.260113701903271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7349275310318e-05,
+      "loss": 0.9625,
+      "step": 9471
+    },
+    {
+      "epoch": 0.2601411661311142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7348672059828037e-05,
+      "loss": 0.9111,
+      "step": 9472
+    },
+    {
+      "epoch": 0.26016863035895743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7348068751192538e-05,
+      "loss": 1.0643,
+      "step": 9473
+    },
+    {
+      "epoch": 0.2601960945868007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7347465384416268e-05,
+      "loss": 1.0035,
+      "step": 9474
+    },
+    {
+      "epoch": 0.26022355881464393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734686195950401e-05,
+      "loss": 1.0308,
+      "step": 9475
+    },
+    {
+      "epoch": 0.26025102304248715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7346258476460528e-05,
+      "loss": 0.9299,
+      "step": 9476
+    },
+    {
+      "epoch": 0.26027848727033037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734565493529061e-05,
+      "loss": 0.9783,
+      "step": 9477
+    },
+    {
+      "epoch": 0.26030595149817365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734505133599902e-05,
+      "loss": 0.9111,
+      "step": 9478
+    },
+    {
+      "epoch": 0.26033341572601687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734444767859054e-05,
+      "loss": 1.019,
+      "step": 9479
+    },
+    {
+      "epoch": 0.2603608799538601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7343843963069948e-05,
+      "loss": 0.9879,
+      "step": 9480
+    },
+    {
+      "epoch": 0.2603883441817033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7343240189442015e-05,
+      "loss": 0.987,
+      "step": 9481
+    },
+    {
+      "epoch": 0.2604158084095466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7342636357711518e-05,
+      "loss": 1.0124,
+      "step": 9482
+    },
+    {
+      "epoch": 0.2604432726373898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7342032467883245e-05,
+      "loss": 0.9793,
+      "step": 9483
+    },
+    {
+      "epoch": 0.26047073686523303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7341428519961962e-05,
+      "loss": 0.9902,
+      "step": 9484
+    },
+    {
+      "epoch": 0.26049820109307625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7340824513952454e-05,
+      "loss": 1.0754,
+      "step": 9485
+    },
+    {
+      "epoch": 0.2605256653209195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7340220449859498e-05,
+      "loss": 0.9701,
+      "step": 9486
+    },
+    {
+      "epoch": 0.26055312954876275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7339616327687875e-05,
+      "loss": 0.9708,
+      "step": 9487
+    },
+    {
+      "epoch": 0.260580593776606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7339012147442365e-05,
+      "loss": 1.051,
+      "step": 9488
+    },
+    {
+      "epoch": 0.2606080580044492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733840790912775e-05,
+      "loss": 0.9733,
+      "step": 9489
+    },
+    {
+      "epoch": 0.2606355222322924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7337803612748808e-05,
+      "loss": 0.9993,
+      "step": 9490
+    },
+    {
+      "epoch": 0.2606629864601357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733719925831032e-05,
+      "loss": 0.9453,
+      "step": 9491
+    },
+    {
+      "epoch": 0.2606904506879789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7336594845817068e-05,
+      "loss": 0.9924,
+      "step": 9492
+    },
+    {
+      "epoch": 0.26071791491582214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7335990375273838e-05,
+      "loss": 1.0392,
+      "step": 9493
+    },
+    {
+      "epoch": 0.26074537914366536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7335385846685412e-05,
+      "loss": 1.0394,
+      "step": 9494
+    },
+    {
+      "epoch": 0.26077284337150863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7334781260056567e-05,
+      "loss": 1.0178,
+      "step": 9495
+    },
+    {
+      "epoch": 0.26080030759935185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7334176615392094e-05,
+      "loss": 0.9963,
+      "step": 9496
+    },
+    {
+      "epoch": 0.2608277718271951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7333571912696775e-05,
+      "loss": 0.9265,
+      "step": 9497
+    },
+    {
+      "epoch": 0.2608552360550383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733296715197539e-05,
+      "loss": 0.9405,
+      "step": 9498
+    },
+    {
+      "epoch": 0.2608827002828816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7332362333232736e-05,
+      "loss": 1.0267,
+      "step": 9499
+    },
+    {
+      "epoch": 0.2609101645107248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733175745647359e-05,
+      "loss": 0.9337,
+      "step": 9500
+    },
+    {
+      "epoch": 0.260937628738568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7331152521702733e-05,
+      "loss": 0.9496,
+      "step": 9501
+    },
+    {
+      "epoch": 0.26096509296641124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733054752892496e-05,
+      "loss": 0.8887,
+      "step": 9502
+    },
+    {
+      "epoch": 0.26099255719425446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7329942478145057e-05,
+      "loss": 1.0247,
+      "step": 9503
+    },
+    {
+      "epoch": 0.26102002142209774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732933736936781e-05,
+      "loss": 0.9816,
+      "step": 9504
+    },
+    {
+      "epoch": 0.26104748564994096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7328732202598006e-05,
+      "loss": 0.9908,
+      "step": 9505
+    },
+    {
+      "epoch": 0.2610749498777842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732812697784043e-05,
+      "loss": 0.8983,
+      "step": 9506
+    },
+    {
+      "epoch": 0.2611024141056274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732752169509988e-05,
+      "loss": 0.9548,
+      "step": 9507
+    },
+    {
+      "epoch": 0.2611298783334707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732691635438114e-05,
+      "loss": 1.1237,
+      "step": 9508
+    },
+    {
+      "epoch": 0.2611573425613139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7326310955688997e-05,
+      "loss": 0.9761,
+      "step": 9509
+    },
+    {
+      "epoch": 0.2611848067891571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7325705499028244e-05,
+      "loss": 1.0252,
+      "step": 9510
+    },
+    {
+      "epoch": 0.26121227101700034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732509998440367e-05,
+      "loss": 1.0239,
+      "step": 9511
+    },
+    {
+      "epoch": 0.2612397352448436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732449441182007e-05,
+      "loss": 1.0638,
+      "step": 9512
+    },
+    {
+      "epoch": 0.26126719947268684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732388878128223e-05,
+      "loss": 0.8746,
+      "step": 9513
+    },
+    {
+      "epoch": 0.26129466370053006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732328309279495e-05,
+      "loss": 0.9822,
+      "step": 9514
+    },
+    {
+      "epoch": 0.2613221279283733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7322677346363014e-05,
+      "loss": 0.9932,
+      "step": 9515
+    },
+    {
+      "epoch": 0.2613495921562165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7322071541991218e-05,
+      "loss": 0.9406,
+      "step": 9516
+    },
+    {
+      "epoch": 0.2613770563840598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7321465679684358e-05,
+      "loss": 0.892,
+      "step": 9517
+    },
+    {
+      "epoch": 0.261404520611903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7320859759447223e-05,
+      "loss": 1.0351,
+      "step": 9518
+    },
+    {
+      "epoch": 0.2614319848397462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7320253781284612e-05,
+      "loss": 0.9898,
+      "step": 9519
+    },
+    {
+      "epoch": 0.26145944906758944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7319647745201316e-05,
+      "loss": 0.9871,
+      "step": 9520
+    },
+    {
+      "epoch": 0.2614869132954327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7319041651202132e-05,
+      "loss": 1.0394,
+      "step": 9521
+    },
+    {
+      "epoch": 0.26151437752327594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7318435499291854e-05,
+      "loss": 1.0382,
+      "step": 9522
+    },
+    {
+      "epoch": 0.26154184175111916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731782928947528e-05,
+      "loss": 1.0189,
+      "step": 9523
+    },
+    {
+      "epoch": 0.2615693059789624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7317223021757207e-05,
+      "loss": 1.0685,
+      "step": 9524
+    },
+    {
+      "epoch": 0.26159677020680566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7316616696142434e-05,
+      "loss": 0.9528,
+      "step": 9525
+    },
+    {
+      "epoch": 0.2616242344346489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731601031263575e-05,
+      "loss": 0.9717,
+      "step": 9526
+    },
+    {
+      "epoch": 0.2616516986624921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731540387124196e-05,
+      "loss": 0.9299,
+      "step": 9527
+    },
+    {
+      "epoch": 0.2616791628903353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7314797371965863e-05,
+      "loss": 1.0851,
+      "step": 9528
+    },
+    {
+      "epoch": 0.26170662711817855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731419081481225e-05,
+      "loss": 1.0072,
+      "step": 9529
+    },
+    {
+      "epoch": 0.2617340913460218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7313584199785934e-05,
+      "loss": 1.093,
+      "step": 9530
+    },
+    {
+      "epoch": 0.26176155557386505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7312977526891704e-05,
+      "loss": 0.9862,
+      "step": 9531
+    },
+    {
+      "epoch": 0.26178901980170827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731237079613436e-05,
+      "loss": 0.9757,
+      "step": 9532
+    },
+    {
+      "epoch": 0.2618164840295515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731176400751871e-05,
+      "loss": 1.0339,
+      "step": 9533
+    },
+    {
+      "epoch": 0.26184394825739477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731115716104955e-05,
+      "loss": 0.9504,
+      "step": 9534
+    },
+    {
+      "epoch": 0.261871412485238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7310550256731678e-05,
+      "loss": 1.0416,
+      "step": 9535
+    },
+    {
+      "epoch": 0.2618988767130812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7309943294569903e-05,
+      "loss": 0.8925,
+      "step": 9536
+    },
+    {
+      "epoch": 0.26192634094092443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7309336274569025e-05,
+      "loss": 0.996,
+      "step": 9537
+    },
+    {
+      "epoch": 0.2619538051687677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730872919673385e-05,
+      "loss": 0.9591,
+      "step": 9538
+    },
+    {
+      "epoch": 0.2619812693966109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7308122061069176e-05,
+      "loss": 1.1246,
+      "step": 9539
+    },
+    {
+      "epoch": 0.26200873362445415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730751486757981e-05,
+      "loss": 1.025,
+      "step": 9540
+    },
+    {
+      "epoch": 0.26203619785229737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7306907616270555e-05,
+      "loss": 0.8658,
+      "step": 9541
+    },
+    {
+      "epoch": 0.2620636620801406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730630030714622e-05,
+      "loss": 1.0136,
+      "step": 9542
+    },
+    {
+      "epoch": 0.26209112630798387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.73056929402116e-05,
+      "loss": 1.0437,
+      "step": 9543
+    },
+    {
+      "epoch": 0.2621185905358271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7305085515471513e-05,
+      "loss": 1.0058,
+      "step": 9544
+    },
+    {
+      "epoch": 0.2621460547636703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7304478032930757e-05,
+      "loss": 0.9047,
+      "step": 9545
+    },
+    {
+      "epoch": 0.26217351899151353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730387049259414e-05,
+      "loss": 0.8869,
+      "step": 9546
+    },
+    {
+      "epoch": 0.2622009832193568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7303262894466474e-05,
+      "loss": 1.0448,
+      "step": 9547
+    },
+    {
+      "epoch": 0.26222844744720003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730265523855256e-05,
+      "loss": 0.9998,
+      "step": 9548
+    },
+    {
+      "epoch": 0.26225591167504325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730204752485721e-05,
+      "loss": 0.9707,
+      "step": 9549
+    },
+    {
+      "epoch": 0.2622833759028865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7301439753385235e-05,
+      "loss": 1.0286,
+      "step": 9550
+    },
+    {
+      "epoch": 0.26231084013072975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7300831924141433e-05,
+      "loss": 0.9117,
+      "step": 9551
+    },
+    {
+      "epoch": 0.26233830435857297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7300224037130628e-05,
+      "loss": 1.0094,
+      "step": 9552
+    },
+    {
+      "epoch": 0.2623657685864162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729961609235762e-05,
+      "loss": 0.9763,
+      "step": 9553
+    },
+    {
+      "epoch": 0.2623932328142594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729900808982722e-05,
+      "loss": 1.0357,
+      "step": 9554
+    },
+    {
+      "epoch": 0.26242069704210264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7298400029544242e-05,
+      "loss": 0.9509,
+      "step": 9555
+    },
+    {
+      "epoch": 0.2624481612699459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7297791911513497e-05,
+      "loss": 0.9375,
+      "step": 9556
+    },
+    {
+      "epoch": 0.26247562549778913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7297183735739795e-05,
+      "loss": 0.9911,
+      "step": 9557
+    },
+    {
+      "epoch": 0.26250308972563235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7296575502227947e-05,
+      "loss": 1.0081,
+      "step": 9558
+    },
+    {
+      "epoch": 0.2625305539534756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7295967210982766e-05,
+      "loss": 0.9738,
+      "step": 9559
+    },
+    {
+      "epoch": 0.26255801818131885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729535886200907e-05,
+      "loss": 1.0648,
+      "step": 9560
+    },
+    {
+      "epoch": 0.2625854824091621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7294750455311667e-05,
+      "loss": 0.962,
+      "step": 9561
+    },
+    {
+      "epoch": 0.2626129466370053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729414199089537e-05,
+      "loss": 0.9801,
+      "step": 9562
+    },
+    {
+      "epoch": 0.2626404108648485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7293533468765004e-05,
+      "loss": 0.9929,
+      "step": 9563
+    },
+    {
+      "epoch": 0.2626678750926918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7292924888925368e-05,
+      "loss": 0.9508,
+      "step": 9564
+    },
+    {
+      "epoch": 0.262695339320535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729231625138129e-05,
+      "loss": 1.063,
+      "step": 9565
+    },
+    {
+      "epoch": 0.26272280354837824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729170755613758e-05,
+      "loss": 1.0353,
+      "step": 9566
+    },
+    {
+      "epoch": 0.26275026777622146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7291098803199052e-05,
+      "loss": 0.9917,
+      "step": 9567
+    },
+    {
+      "epoch": 0.2627777320040647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729048999257053e-05,
+      "loss": 0.983,
+      "step": 9568
+    },
+    {
+      "epoch": 0.26280519623190796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7289881124256823e-05,
+      "loss": 0.9815,
+      "step": 9569
+    },
+    {
+      "epoch": 0.2628326604597512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7289272198262756e-05,
+      "loss": 0.9473,
+      "step": 9570
+    },
+    {
+      "epoch": 0.2628601246875944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7288663214593146e-05,
+      "loss": 0.9754,
+      "step": 9571
+    },
+    {
+      "epoch": 0.2628875889154376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7288054173252806e-05,
+      "loss": 0.9458,
+      "step": 9572
+    },
+    {
+      "epoch": 0.2629150531432809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7287445074246557e-05,
+      "loss": 1.0006,
+      "step": 9573
+    },
+    {
+      "epoch": 0.2629425173711241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7286835917579222e-05,
+      "loss": 1.0113,
+      "step": 9574
+    },
+    {
+      "epoch": 0.26296998159896734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7286226703255618e-05,
+      "loss": 1.0835,
+      "step": 9575
+    },
+    {
+      "epoch": 0.26299744582681056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7285617431280566e-05,
+      "loss": 0.9144,
+      "step": 9576
+    },
+    {
+      "epoch": 0.26302491005465384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7285008101658885e-05,
+      "loss": 1.0767,
+      "step": 9577
+    },
+    {
+      "epoch": 0.26305237428249706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7284398714395402e-05,
+      "loss": 0.9568,
+      "step": 9578
+    },
+    {
+      "epoch": 0.2630798385103403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7283789269494932e-05,
+      "loss": 1.0833,
+      "step": 9579
+    },
+    {
+      "epoch": 0.2631073027381835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72831797669623e-05,
+      "loss": 1.0241,
+      "step": 9580
+    },
+    {
+      "epoch": 0.2631347669660267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7282570206802326e-05,
+      "loss": 0.9317,
+      "step": 9581
+    },
+    {
+      "epoch": 0.26316223119387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728196058901984e-05,
+      "loss": 1.0903,
+      "step": 9582
+    },
+    {
+      "epoch": 0.2631896954217132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7281350913619658e-05,
+      "loss": 0.9266,
+      "step": 9583
+    },
+    {
+      "epoch": 0.26321715964955644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728074118060661e-05,
+      "loss": 1.0601,
+      "step": 9584
+    },
+    {
+      "epoch": 0.26324462387739966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728013138998551e-05,
+      "loss": 0.9621,
+      "step": 9585
+    },
+    {
+      "epoch": 0.26327208810524294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7279521541761203e-05,
+      "loss": 1.0384,
+      "step": 9586
+    },
+    {
+      "epoch": 0.26329955233308616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727891163593849e-05,
+      "loss": 0.9696,
+      "step": 9587
+    },
+    {
+      "epoch": 0.2633270165609294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7278301672522215e-05,
+      "loss": 0.9086,
+      "step": 9588
+    },
+    {
+      "epoch": 0.2633544807887726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7277691651517195e-05,
+      "loss": 0.9608,
+      "step": 9589
+    },
+    {
+      "epoch": 0.2633819450166159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7277081572928263e-05,
+      "loss": 0.9137,
+      "step": 9590
+    },
+    {
+      "epoch": 0.2634094092444591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7276471436760237e-05,
+      "loss": 1.0605,
+      "step": 9591
+    },
+    {
+      "epoch": 0.2634368734723023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7275861243017956e-05,
+      "loss": 1.0089,
+      "step": 9592
+    },
+    {
+      "epoch": 0.26346433770014555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727525099170624e-05,
+      "loss": 1.0019,
+      "step": 9593
+    },
+    {
+      "epoch": 0.26349180192798877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727464068282992e-05,
+      "loss": 1.0026,
+      "step": 9594
+    },
+    {
+      "epoch": 0.26351926615583204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7274030316393826e-05,
+      "loss": 0.9863,
+      "step": 9595
+    },
+    {
+      "epoch": 0.26354673038367527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7273419892402785e-05,
+      "loss": 0.9149,
+      "step": 9596
+    },
+    {
+      "epoch": 0.2635741946115185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7272809410861628e-05,
+      "loss": 1.0145,
+      "step": 9597
+    },
+    {
+      "epoch": 0.2636016588393617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7272198871775184e-05,
+      "loss": 0.9481,
+      "step": 9598
+    },
+    {
+      "epoch": 0.263629123067205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727158827514829e-05,
+      "loss": 0.9774,
+      "step": 9599
+    },
+    {
+      "epoch": 0.2636565872950482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727097762098577e-05,
+      "loss": 0.9812,
+      "step": 9600
+    },
+    {
+      "epoch": 0.2636840515228914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727036690929246e-05,
+      "loss": 0.954,
+      "step": 9601
+    },
+    {
+      "epoch": 0.26371151575073465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7269756140073193e-05,
+      "loss": 1.0116,
+      "step": 9602
+    },
+    {
+      "epoch": 0.2637389799785779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7269145313332795e-05,
+      "loss": 0.9169,
+      "step": 9603
+    },
+    {
+      "epoch": 0.26376644420642115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7268534429076105e-05,
+      "loss": 0.9841,
+      "step": 9604
+    },
+    {
+      "epoch": 0.26379390843426437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7267923487307955e-05,
+      "loss": 0.9788,
+      "step": 9605
+    },
+    {
+      "epoch": 0.2638213726621076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726731248803318e-05,
+      "loss": 0.9767,
+      "step": 9606
+    },
+    {
+      "epoch": 0.26384883688995087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726670143125661e-05,
+      "loss": 1.0427,
+      "step": 9607
+    },
+    {
+      "epoch": 0.2638763011177941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726609031698309e-05,
+      "loss": 0.8092,
+      "step": 9608
+    },
+    {
+      "epoch": 0.2639037653456373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7265479145217445e-05,
+      "loss": 0.9685,
+      "step": 9609
+    },
+    {
+      "epoch": 0.26393122957348053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264867915964513e-05,
+      "loss": 0.9769,
+      "step": 9610
+    },
+    {
+      "epoch": 0.26395869380132375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264256629229134e-05,
+      "loss": 1.0215,
+      "step": 9611
+    },
+    {
+      "epoch": 0.26398615802916703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7263645285016143e-05,
+      "loss": 1.0022,
+      "step": 9612
+    },
+    {
+      "epoch": 0.26401362225701025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7263033883330377e-05,
+      "loss": 0.9689,
+      "step": 9613
+    },
+    {
+      "epoch": 0.26404108648485347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726242242417667e-05,
+      "loss": 1.0005,
+      "step": 9614
+    },
+    {
+      "epoch": 0.2640685507126967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7261810907559867e-05,
+      "loss": 1.0163,
+      "step": 9615
+    },
+    {
+      "epoch": 0.26409601494053997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72611993334848e-05,
+      "loss": 1.0754,
+      "step": 9616
+    },
+    {
+      "epoch": 0.2641234791683832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7260587701956315e-05,
+      "loss": 1.0438,
+      "step": 9617
+    },
+    {
+      "epoch": 0.2641509433962264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7259976012979247e-05,
+      "loss": 0.9778,
+      "step": 9618
+    },
+    {
+      "epoch": 0.26417840762406963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7259364266558437e-05,
+      "loss": 1.0725,
+      "step": 9619
+    },
+    {
+      "epoch": 0.2642058718519129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725875246269872e-05,
+      "loss": 0.9774,
+      "step": 9620
+    },
+    {
+      "epoch": 0.26423333607975613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7258140601404945e-05,
+      "loss": 0.9393,
+      "step": 9621
+    },
+    {
+      "epoch": 0.26426080030759935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725752868268195e-05,
+      "loss": 0.9379,
+      "step": 9622
+    },
+    {
+      "epoch": 0.2642882645354426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725691670653458e-05,
+      "loss": 0.9652,
+      "step": 9623
+    },
+    {
+      "epoch": 0.2643157287632858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725630467296767e-05,
+      "loss": 0.9427,
+      "step": 9624
+    },
+    {
+      "epoch": 0.2643431929911291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7255692581986068e-05,
+      "loss": 0.9804,
+      "step": 9625
+    },
+    {
+      "epoch": 0.2643706572189723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7255080433594616e-05,
+      "loss": 0.9498,
+      "step": 9626
+    },
+    {
+      "epoch": 0.2643981214468155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7254468227798152e-05,
+      "loss": 1.016,
+      "step": 9627
+    },
+    {
+      "epoch": 0.26442558567465874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725385596460153e-05,
+      "loss": 1.0591,
+      "step": 9628
+    },
+    {
+      "epoch": 0.264453049902502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7253243644009592e-05,
+      "loss": 1.0813,
+      "step": 9629
+    },
+    {
+      "epoch": 0.26448051413034523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7252631266027177e-05,
+      "loss": 1.0465,
+      "step": 9630
+    },
+    {
+      "epoch": 0.26450797835818846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7252018830659136e-05,
+      "loss": 0.9404,
+      "step": 9631
+    },
+    {
+      "epoch": 0.2645354425860317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725140633791031e-05,
+      "loss": 0.9823,
+      "step": 9632
+    },
+    {
+      "epoch": 0.26456290681387495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7250793787785552e-05,
+      "loss": 0.9173,
+      "step": 9633
+    },
+    {
+      "epoch": 0.2645903710417182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7250181180289703e-05,
+      "loss": 0.9658,
+      "step": 9634
+    },
+    {
+      "epoch": 0.2646178352695614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7249568515427613e-05,
+      "loss": 1.0237,
+      "step": 9635
+    },
+    {
+      "epoch": 0.2646452994974046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7248955793204126e-05,
+      "loss": 0.9738,
+      "step": 9636
+    },
+    {
+      "epoch": 0.26467276372524784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7248343013624094e-05,
+      "loss": 1.056,
+      "step": 9637
+    },
+    {
+      "epoch": 0.2647002279530911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7247730176692367e-05,
+      "loss": 0.9495,
+      "step": 9638
+    },
+    {
+      "epoch": 0.26472769218093434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724711728241379e-05,
+      "loss": 1.145,
+      "step": 9639
+    },
+    {
+      "epoch": 0.26475515640877756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7246504330793213e-05,
+      "loss": 0.9491,
+      "step": 9640
+    },
+    {
+      "epoch": 0.2647826206366208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7245891321835487e-05,
+      "loss": 0.9017,
+      "step": 9641
+    },
+    {
+      "epoch": 0.26481008486446406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724527825554546e-05,
+      "loss": 1.0804,
+      "step": 9642
+    },
+    {
+      "epoch": 0.2648375490923073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7244665131927986e-05,
+      "loss": 0.9974,
+      "step": 9643
+    },
+    {
+      "epoch": 0.2648650133201505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7244051950987918e-05,
+      "loss": 0.9749,
+      "step": 9644
+    },
+    {
+      "epoch": 0.2648924775479937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72434387127301e-05,
+      "loss": 0.9553,
+      "step": 9645
+    },
+    {
+      "epoch": 0.264919941775837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7242825417159395e-05,
+      "loss": 1.0005,
+      "step": 9646
+    },
+    {
+      "epoch": 0.2649474060036802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7242212064280644e-05,
+      "loss": 0.9239,
+      "step": 9647
+    },
+    {
+      "epoch": 0.26497487023152344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724159865409871e-05,
+      "loss": 1.0104,
+      "step": 9648
+    },
+    {
+      "epoch": 0.26500233445936666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724098518661844e-05,
+      "loss": 0.8699,
+      "step": 9649
+    },
+    {
+      "epoch": 0.2650297986872099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724037166184469e-05,
+      "loss": 1.073,
+      "step": 9650
+    },
+    {
+      "epoch": 0.26505726291505316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7239758079782317e-05,
+      "loss": 0.962,
+      "step": 9651
+    },
+    {
+      "epoch": 0.2650847271428964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7239144440436173e-05,
+      "loss": 0.9399,
+      "step": 9652
+    },
+    {
+      "epoch": 0.2651121913707396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7238530743811113e-05,
+      "loss": 0.9132,
+      "step": 9653
+    },
+    {
+      "epoch": 0.2651396555985828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7237916989911993e-05,
+      "loss": 0.9785,
+      "step": 9654
+    },
+    {
+      "epoch": 0.2651671198264261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723730317874367e-05,
+      "loss": 1.063,
+      "step": 9655
+    },
+    {
+      "epoch": 0.2651945840542693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7236689310311002e-05,
+      "loss": 0.9346,
+      "step": 9656
+    },
+    {
+      "epoch": 0.26522204828211254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7236075384618848e-05,
+      "loss": 0.8517,
+      "step": 9657
+    },
+    {
+      "epoch": 0.26524951250995576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7235461401672058e-05,
+      "loss": 0.9581,
+      "step": 9658
+    },
+    {
+      "epoch": 0.26527697673779904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7234847361475496e-05,
+      "loss": 0.9401,
+      "step": 9659
+    },
+    {
+      "epoch": 0.26530444096564226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7234233264034022e-05,
+      "loss": 0.8969,
+      "step": 9660
+    },
+    {
+      "epoch": 0.2653319051934855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7233619109352488e-05,
+      "loss": 0.9443,
+      "step": 9661
+    },
+    {
+      "epoch": 0.2653593694213287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7233004897435758e-05,
+      "loss": 1.0201,
+      "step": 9662
+    },
+    {
+      "epoch": 0.2653868336491719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7232390628288695e-05,
+      "loss": 0.9451,
+      "step": 9663
+    },
+    {
+      "epoch": 0.2654142978770152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7231776301916154e-05,
+      "loss": 1.0723,
+      "step": 9664
+    },
+    {
+      "epoch": 0.2654417621048584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7231161918322993e-05,
+      "loss": 0.943,
+      "step": 9665
+    },
+    {
+      "epoch": 0.26546922633270165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7230547477514084e-05,
+      "loss": 1.0537,
+      "step": 9666
+    },
+    {
+      "epoch": 0.26549669056054487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722993297949428e-05,
+      "loss": 1.0605,
+      "step": 9667
+    },
+    {
+      "epoch": 0.26552415478838814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7229318424268443e-05,
+      "loss": 0.9823,
+      "step": 9668
+    },
+    {
+      "epoch": 0.26555161901623137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7228703811841444e-05,
+      "loss": 0.9965,
+      "step": 9669
+    },
+    {
+      "epoch": 0.2655790832440746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7228089142218135e-05,
+      "loss": 1.01,
+      "step": 9670
+    },
+    {
+      "epoch": 0.2656065474719178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7227474415403388e-05,
+      "loss": 1.0756,
+      "step": 9671
+    },
+    {
+      "epoch": 0.2656340116997611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7226859631402063e-05,
+      "loss": 0.9371,
+      "step": 9672
+    },
+    {
+      "epoch": 0.2656614759276043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7226244790219024e-05,
+      "loss": 0.9278,
+      "step": 9673
+    },
+    {
+      "epoch": 0.26568894015544753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7225629891859143e-05,
+      "loss": 1.0499,
+      "step": 9674
+    },
+    {
+      "epoch": 0.26571640438329075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7225014936327273e-05,
+      "loss": 0.9563,
+      "step": 9675
+    },
+    {
+      "epoch": 0.26574386861113397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7224399923628285e-05,
+      "loss": 0.9559,
+      "step": 9676
+    },
+    {
+      "epoch": 0.26577133283897725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722378485376705e-05,
+      "loss": 0.9377,
+      "step": 9677
+    },
+    {
+      "epoch": 0.26579879706682047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722316972674843e-05,
+      "loss": 0.9279,
+      "step": 9678
+    },
+    {
+      "epoch": 0.2658262612946637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7222554542577296e-05,
+      "loss": 0.986,
+      "step": 9679
+    },
+    {
+      "epoch": 0.2658537255225069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7221939301258508e-05,
+      "loss": 0.9264,
+      "step": 9680
+    },
+    {
+      "epoch": 0.2658811897503502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722132400279694e-05,
+      "loss": 0.9038,
+      "step": 9681
+    },
+    {
+      "epoch": 0.2659086539781934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7220708647197463e-05,
+      "loss": 1.1146,
+      "step": 9682
+    },
+    {
+      "epoch": 0.26593611820603663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7220093234464936e-05,
+      "loss": 0.9682,
+      "step": 9683
+    },
+    {
+      "epoch": 0.26596358243387985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7219477764604238e-05,
+      "loss": 0.943,
+      "step": 9684
+    },
+    {
+      "epoch": 0.26599104666172313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7218862237620235e-05,
+      "loss": 1.0167,
+      "step": 9685
+    },
+    {
+      "epoch": 0.26601851088956635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7218246653517797e-05,
+      "loss": 1.046,
+      "step": 9686
+    },
+    {
+      "epoch": 0.26604597511740957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7217631012301796e-05,
+      "loss": 0.9466,
+      "step": 9687
+    },
+    {
+      "epoch": 0.2660734393452528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7217015313977105e-05,
+      "loss": 0.9695,
+      "step": 9688
+    },
+    {
+      "epoch": 0.266100903573096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7216399558548586e-05,
+      "loss": 1.0623,
+      "step": 9689
+    },
+    {
+      "epoch": 0.2661283678009393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7215783746021126e-05,
+      "loss": 0.9764,
+      "step": 9690
+    },
+    {
+      "epoch": 0.2661558320287825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7215167876399587e-05,
+      "loss": 0.9169,
+      "step": 9691
+    },
+    {
+      "epoch": 0.26618329625662573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7214551949688843e-05,
+      "loss": 0.9,
+      "step": 9692
+    },
+    {
+      "epoch": 0.26621076048446896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721393596589377e-05,
+      "loss": 0.8607,
+      "step": 9693
+    },
+    {
+      "epoch": 0.26623822471231223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7213319925019246e-05,
+      "loss": 1.0046,
+      "step": 9694
+    },
+    {
+      "epoch": 0.26626568894015545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212703827070133e-05,
+      "loss": 0.941,
+      "step": 9695
+    },
+    {
+      "epoch": 0.2662931531679987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212087672051318e-05,
+      "loss": 1.004,
+      "step": 9696
+    },
+    {
+      "epoch": 0.2663206173958419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7211471459967673e-05,
+      "loss": 0.9296,
+      "step": 9697
+    },
+    {
+      "epoch": 0.2663480816236852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721085519082407e-05,
+      "loss": 0.9707,
+      "step": 9698
+    },
+    {
+      "epoch": 0.2663755458515284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721023886462539e-05,
+      "loss": 1.0534,
+      "step": 9699
+    },
+    {
+      "epoch": 0.2664030100793716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7209622481376504e-05,
+      "loss": 0.978,
+      "step": 9700
+    },
+    {
+      "epoch": 0.26643047430721484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720900604108229e-05,
+      "loss": 1.0092,
+      "step": 9701
+    },
+    {
+      "epoch": 0.2664579385350581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7208389543747633e-05,
+      "loss": 0.8906,
+      "step": 9702
+    },
+    {
+      "epoch": 0.26648540276290134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7207772989377405e-05,
+      "loss": 1.0434,
+      "step": 9703
+    },
+    {
+      "epoch": 0.26651286699074456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7207156377976482e-05,
+      "loss": 0.9912,
+      "step": 9704
+    },
+    {
+      "epoch": 0.2665403312185878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7206539709549748e-05,
+      "loss": 0.9883,
+      "step": 9705
+    },
+    {
+      "epoch": 0.266567795446431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720592298410208e-05,
+      "loss": 1.0096,
+      "step": 9706
+    },
+    {
+      "epoch": 0.2665952596742743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7205306201638357e-05,
+      "loss": 0.9881,
+      "step": 9707
+    },
+    {
+      "epoch": 0.2666227239021175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7204689362163463e-05,
+      "loss": 0.927,
+      "step": 9708
+    },
+    {
+      "epoch": 0.2666501881299607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7204072465682272e-05,
+      "loss": 1.0142,
+      "step": 9709
+    },
+    {
+      "epoch": 0.26667765235780394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720345551219967e-05,
+      "loss": 0.9724,
+      "step": 9710
+    },
+    {
+      "epoch": 0.2667051165856472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720283850172054e-05,
+      "loss": 0.8906,
+      "step": 9711
+    },
+    {
+      "epoch": 0.26673258081349044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720222143424976e-05,
+      "loss": 0.9485,
+      "step": 9712
+    },
+    {
+      "epoch": 0.26676004504133366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7201604309792216e-05,
+      "loss": 0.9405,
+      "step": 9713
+    },
+    {
+      "epoch": 0.2667875092691769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7200987128352788e-05,
+      "loss": 0.9118,
+      "step": 9714
+    },
+    {
+      "epoch": 0.26681497349702016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720036988993636e-05,
+      "loss": 1.0194,
+      "step": 9715
+    },
+    {
+      "epoch": 0.2668424377248634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7199752594547818e-05,
+      "loss": 0.8821,
+      "step": 9716
+    },
+    {
+      "epoch": 0.2668699019527066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719913524219204e-05,
+      "loss": 0.8595,
+      "step": 9717
+    },
+    {
+      "epoch": 0.2668973661805498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719851783287392e-05,
+      "loss": 1.0431,
+      "step": 9718
+    },
+    {
+      "epoch": 0.26692483040839304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7197900366598336e-05,
+      "loss": 1.0375,
+      "step": 9719
+    },
+    {
+      "epoch": 0.2669522946362363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7197282843370176e-05,
+      "loss": 1.0436,
+      "step": 9720
+    },
+    {
+      "epoch": 0.26697975886407954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7196665263194328e-05,
+      "loss": 1.0077,
+      "step": 9721
+    },
+    {
+      "epoch": 0.26700722309192276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7196047626075673e-05,
+      "loss": 0.9309,
+      "step": 9722
+    },
+    {
+      "epoch": 0.267034687319766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7195429932019108e-05,
+      "loss": 1.0385,
+      "step": 9723
+    },
+    {
+      "epoch": 0.26706215154760926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719481218102951e-05,
+      "loss": 0.9897,
+      "step": 9724
+    },
+    {
+      "epoch": 0.2670896157754525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719419437311177e-05,
+      "loss": 0.9896,
+      "step": 9725
+    },
+    {
+      "epoch": 0.2671170800032957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719357650827078e-05,
+      "loss": 0.9628,
+      "step": 9726
+    },
+    {
+      "epoch": 0.2671445442311389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7192958586511425e-05,
+      "loss": 0.9353,
+      "step": 9727
+    },
+    {
+      "epoch": 0.2671720084589822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7192340607838596e-05,
+      "loss": 0.9678,
+      "step": 9728
+    },
+    {
+      "epoch": 0.2671994726868254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719172257225718e-05,
+      "loss": 0.8847,
+      "step": 9729
+    },
+    {
+      "epoch": 0.26722693691466864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719110447977207e-05,
+      "loss": 1.0444,
+      "step": 9730
+    },
+    {
+      "epoch": 0.26725440114251187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7190486330388155e-05,
+      "loss": 1.0067,
+      "step": 9731
+    },
+    {
+      "epoch": 0.2672818653703551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7189868124110332e-05,
+      "loss": 0.9441,
+      "step": 9732
+    },
+    {
+      "epoch": 0.26730932959819836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7189249860943483e-05,
+      "loss": 1.0061,
+      "step": 9733
+    },
+    {
+      "epoch": 0.2673367938260416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7188631540892504e-05,
+      "loss": 1.0035,
+      "step": 9734
+    },
+    {
+      "epoch": 0.2673642580538848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718801316396229e-05,
+      "loss": 0.9869,
+      "step": 9735
+    },
+    {
+      "epoch": 0.26739172228172803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718739473015773e-05,
+      "loss": 1.0317,
+      "step": 9736
+    },
+    {
+      "epoch": 0.2674191865095713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7186776239483716e-05,
+      "loss": 0.9711,
+      "step": 9737
+    },
+    {
+      "epoch": 0.2674466507374145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718615769194515e-05,
+      "loss": 1.0122,
+      "step": 9738
+    },
+    {
+      "epoch": 0.26747411496525775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7185539087546917e-05,
+      "loss": 0.9377,
+      "step": 9739
+    },
+    {
+      "epoch": 0.26750157919310097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7184920426293915e-05,
+      "loss": 1.0122,
+      "step": 9740
+    },
+    {
+      "epoch": 0.26752904342094425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718430170819104e-05,
+      "loss": 0.9865,
+      "step": 9741
+    },
+    {
+      "epoch": 0.26755650764878747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7183682933243192e-05,
+      "loss": 1.0138,
+      "step": 9742
+    },
+    {
+      "epoch": 0.2675839718766307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7183064101455255e-05,
+      "loss": 1.0431,
+      "step": 9743
+    },
+    {
+      "epoch": 0.2676114361044739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7182445212832135e-05,
+      "loss": 0.9961,
+      "step": 9744
+    },
+    {
+      "epoch": 0.26763890033231713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7181826267378726e-05,
+      "loss": 0.9959,
+      "step": 9745
+    },
+    {
+      "epoch": 0.2676663645601604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7181207265099927e-05,
+      "loss": 1.0464,
+      "step": 9746
+    },
+    {
+      "epoch": 0.26769382878800363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7180588206000632e-05,
+      "loss": 0.9271,
+      "step": 9747
+    },
+    {
+      "epoch": 0.26772129301584685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7179969090085742e-05,
+      "loss": 0.9092,
+      "step": 9748
+    },
+    {
+      "epoch": 0.26774875724369007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717934991736016e-05,
+      "loss": 1.012,
+      "step": 9749
+    },
+    {
+      "epoch": 0.26777622147153335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7178730687828772e-05,
+      "loss": 1.0465,
+      "step": 9750
+    },
+    {
+      "epoch": 0.26780368569937657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717811140149649e-05,
+      "loss": 0.9331,
+      "step": 9751
+    },
+    {
+      "epoch": 0.2678311499272198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717749205836821e-05,
+      "loss": 0.9794,
+      "step": 9752
+    },
+    {
+      "epoch": 0.267858614155063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7176872658448836e-05,
+      "loss": 0.9677,
+      "step": 9753
+    },
+    {
+      "epoch": 0.2678860783829063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7176253201743263e-05,
+      "loss": 1.0151,
+      "step": 9754
+    },
+    {
+      "epoch": 0.2679135426107495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7175633688256393e-05,
+      "loss": 0.9142,
+      "step": 9755
+    },
+    {
+      "epoch": 0.26794100683859273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7175014117993132e-05,
+      "loss": 0.9915,
+      "step": 9756
+    },
+    {
+      "epoch": 0.26796847106643595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717439449095838e-05,
+      "loss": 0.9752,
+      "step": 9757
+    },
+    {
+      "epoch": 0.2679959352942792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717377480715704e-05,
+      "loss": 0.9624,
+      "step": 9758
+    },
+    {
+      "epoch": 0.26802339952212245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7173155066594012e-05,
+      "loss": 0.8973,
+      "step": 9759
+    },
+    {
+      "epoch": 0.2680508637499657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7172535269274204e-05,
+      "loss": 1.0885,
+      "step": 9760
+    },
+    {
+      "epoch": 0.2680783279778089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7171915415202522e-05,
+      "loss": 0.9658,
+      "step": 9761
+    },
+    {
+      "epoch": 0.2681057922056521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717129550438386e-05,
+      "loss": 0.9244,
+      "step": 9762
+    },
+    {
+      "epoch": 0.2681332564334954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7170675536823136e-05,
+      "loss": 1.0319,
+      "step": 9763
+    },
+    {
+      "epoch": 0.2681607206613386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7170055512525244e-05,
+      "loss": 0.9952,
+      "step": 9764
+    },
+    {
+      "epoch": 0.26818818488918184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71694354314951e-05,
+      "loss": 1.0557,
+      "step": 9765
+    },
+    {
+      "epoch": 0.26821564911702506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7168815293737604e-05,
+      "loss": 1.0592,
+      "step": 9766
+    },
+    {
+      "epoch": 0.26824311334486833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7168195099257668e-05,
+      "loss": 1.0067,
+      "step": 9767
+    },
+    {
+      "epoch": 0.26827057757271155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7167574848060192e-05,
+      "loss": 0.9938,
+      "step": 9768
+    },
+    {
+      "epoch": 0.2682980418005548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7166954540150086e-05,
+      "loss": 0.8975,
+      "step": 9769
+    },
+    {
+      "epoch": 0.268325506028398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7166334175532262e-05,
+      "loss": 1.1041,
+      "step": 9770
+    },
+    {
+      "epoch": 0.2683529702562412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7165713754211627e-05,
+      "loss": 0.9377,
+      "step": 9771
+    },
+    {
+      "epoch": 0.2683804344840845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7165093276193087e-05,
+      "loss": 1.0021,
+      "step": 9772
+    },
+    {
+      "epoch": 0.2684078987119277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7164472741481554e-05,
+      "loss": 0.9776,
+      "step": 9773
+    },
+    {
+      "epoch": 0.26843536293977094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7163852150081938e-05,
+      "loss": 0.8674,
+      "step": 9774
+    },
+    {
+      "epoch": 0.26846282716761416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7163231501999148e-05,
+      "loss": 0.9642,
+      "step": 9775
+    },
+    {
+      "epoch": 0.26849029139545744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7162610797238094e-05,
+      "loss": 0.996,
+      "step": 9776
+    },
+    {
+      "epoch": 0.26851775562330066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716199003580369e-05,
+      "loss": 0.9803,
+      "step": 9777
+    },
+    {
+      "epoch": 0.2685452198511439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716136921770085e-05,
+      "loss": 1.0457,
+      "step": 9778
+    },
+    {
+      "epoch": 0.2685726840789871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716074834293448e-05,
+      "loss": 0.9604,
+      "step": 9779
+    },
+    {
+      "epoch": 0.2686001483068304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7160127411509493e-05,
+      "loss": 0.9481,
+      "step": 9780
+    },
+    {
+      "epoch": 0.2686276125346736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715950642343081e-05,
+      "loss": 0.9462,
+      "step": 9781
+    },
+    {
+      "epoch": 0.2686550767625168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7158885378703333e-05,
+      "loss": 1.0573,
+      "step": 9782
+    },
+    {
+      "epoch": 0.26868254099036004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7158264277331983e-05,
+      "loss": 1.0352,
+      "step": 9783
+    },
+    {
+      "epoch": 0.26871000521820326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7157643119321678e-05,
+      "loss": 0.9171,
+      "step": 9784
+    },
+    {
+      "epoch": 0.26873746944604654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7157021904677323e-05,
+      "loss": 1.0089,
+      "step": 9785
+    },
+    {
+      "epoch": 0.26876493367388976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715640063340384e-05,
+      "loss": 1.0695,
+      "step": 9786
+    },
+    {
+      "epoch": 0.268792397901733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715577930550614e-05,
+      "loss": 0.9238,
+      "step": 9787
+    },
+    {
+      "epoch": 0.2688198621295762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7155157920989146e-05,
+      "loss": 0.8339,
+      "step": 9788
+    },
+    {
+      "epoch": 0.2688473263574195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7154536479857773e-05,
+      "loss": 1.0315,
+      "step": 9789
+    },
+    {
+      "epoch": 0.2688747905852627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7153914982116932e-05,
+      "loss": 1.0735,
+      "step": 9790
+    },
+    {
+      "epoch": 0.2689022548131059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7153293427771545e-05,
+      "loss": 1.0339,
+      "step": 9791
+    },
+    {
+      "epoch": 0.26892971904094914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715267181682653e-05,
+      "loss": 0.9233,
+      "step": 9792
+    },
+    {
+      "epoch": 0.2689571832687924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7152050149286804e-05,
+      "loss": 1.0861,
+      "step": 9793
+    },
+    {
+      "epoch": 0.26898464749663564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7151428425157288e-05,
+      "loss": 0.9868,
+      "step": 9794
+    },
+    {
+      "epoch": 0.26901211172447886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71508066444429e-05,
+      "loss": 1.0468,
+      "step": 9795
+    },
+    {
+      "epoch": 0.2690395759523221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7150184807148558e-05,
+      "loss": 0.964,
+      "step": 9796
+    },
+    {
+      "epoch": 0.26906704018016536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7149562913279186e-05,
+      "loss": 0.9994,
+      "step": 9797
+    },
+    {
+      "epoch": 0.2690945044080086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7148940962839706e-05,
+      "loss": 0.9421,
+      "step": 9798
+    },
+    {
+      "epoch": 0.2691219686358518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714831895583503e-05,
+      "loss": 0.9661,
+      "step": 9799
+    },
+    {
+      "epoch": 0.269149432863695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7147696892270088e-05,
+      "loss": 1.0319,
+      "step": 9800
+    },
+    {
+      "epoch": 0.26917689709153825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71470747721498e-05,
+      "loss": 0.8808,
+      "step": 9801
+    },
+    {
+      "epoch": 0.2692043613193815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7146452595479085e-05,
+      "loss": 0.8893,
+      "step": 9802
+    },
+    {
+      "epoch": 0.26923182554722475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7145830362262872e-05,
+      "loss": 1.0309,
+      "step": 9803
+    },
+    {
+      "epoch": 0.26925928977506797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714520807250608e-05,
+      "loss": 1.0533,
+      "step": 9804
+    },
+    {
+      "epoch": 0.2692867540029112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7144585726213636e-05,
+      "loss": 0.8748,
+      "step": 9805
+    },
+    {
+      "epoch": 0.26931421823075447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7143963323390456e-05,
+      "loss": 1.0505,
+      "step": 9806
+    },
+    {
+      "epoch": 0.2693416824585977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7143340864041476e-05,
+      "loss": 1.1069,
+      "step": 9807
+    },
+    {
+      "epoch": 0.2693691466864409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7142718348171613e-05,
+      "loss": 0.8934,
+      "step": 9808
+    },
+    {
+      "epoch": 0.26939661091428413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71420957757858e-05,
+      "loss": 1.016,
+      "step": 9809
+    },
+    {
+      "epoch": 0.2694240751421274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7141473146888955e-05,
+      "loss": 0.8983,
+      "step": 9810
+    },
+    {
+      "epoch": 0.2694515393699706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714085046148601e-05,
+      "loss": 0.9804,
+      "step": 9811
+    },
+    {
+      "epoch": 0.26947900359781385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714022771958189e-05,
+      "loss": 1.0568,
+      "step": 9812
+    },
+    {
+      "epoch": 0.26950646782565707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7139604921181522e-05,
+      "loss": 0.8945,
+      "step": 9813
+    },
+    {
+      "epoch": 0.2695339320535003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7138982066289832e-05,
+      "loss": 0.9714,
+      "step": 9814
+    },
+    {
+      "epoch": 0.26956139628134357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7138359154911755e-05,
+      "loss": 0.9256,
+      "step": 9815
+    },
+    {
+      "epoch": 0.2695888605091868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7137736187052212e-05,
+      "loss": 1.0471,
+      "step": 9816
+    },
+    {
+      "epoch": 0.26961632473703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7137113162716135e-05,
+      "loss": 0.9838,
+      "step": 9817
+    },
+    {
+      "epoch": 0.26964378896487323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7136490081908456e-05,
+      "loss": 1.1207,
+      "step": 9818
+    },
+    {
+      "epoch": 0.2696712531927165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7135866944634104e-05,
+      "loss": 1.0042,
+      "step": 9819
+    },
+    {
+      "epoch": 0.26969871742055973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7135243750898007e-05,
+      "loss": 0.9949,
+      "step": 9820
+    },
+    {
+      "epoch": 0.26972618164840295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7134620500705094e-05,
+      "loss": 0.9313,
+      "step": 9821
+    },
+    {
+      "epoch": 0.2697536458762462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7133997194060304e-05,
+      "loss": 1.1062,
+      "step": 9822
+    },
+    {
+      "epoch": 0.26978111010408945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7133373830968568e-05,
+      "loss": 0.9666,
+      "step": 9823
+    },
+    {
+      "epoch": 0.26980857433193267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713275041143481e-05,
+      "loss": 0.9861,
+      "step": 9824
+    },
+    {
+      "epoch": 0.2698360385597759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713212693546397e-05,
+      "loss": 1.0886,
+      "step": 9825
+    },
+    {
+      "epoch": 0.2698635027876191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7131503403060975e-05,
+      "loss": 0.8882,
+      "step": 9826
+    },
+    {
+      "epoch": 0.26989096701546234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7130879814230764e-05,
+      "loss": 0.9409,
+      "step": 9827
+    },
+    {
+      "epoch": 0.2699184312433056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7130256168978273e-05,
+      "loss": 0.9067,
+      "step": 9828
+    },
+    {
+      "epoch": 0.26994589547114883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7129632467308433e-05,
+      "loss": 1.0269,
+      "step": 9829
+    },
+    {
+      "epoch": 0.26997335969899205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7129008709226176e-05,
+      "loss": 1.0175,
+      "step": 9830
+    },
+    {
+      "epoch": 0.2700008239268353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712838489473644e-05,
+      "loss": 1.1125,
+      "step": 9831
+    },
+    {
+      "epoch": 0.27002828815467855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7127761023844166e-05,
+      "loss": 1.1042,
+      "step": 9832
+    },
+    {
+      "epoch": 0.2700557523825218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7127137096554284e-05,
+      "loss": 0.9191,
+      "step": 9833
+    },
+    {
+      "epoch": 0.270083216610365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712651311287173e-05,
+      "loss": 0.912,
+      "step": 9834
+    },
+    {
+      "epoch": 0.2701106808382082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7125889072801446e-05,
+      "loss": 0.9474,
+      "step": 9835
+    },
+    {
+      "epoch": 0.2701381450660515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7125264976348365e-05,
+      "loss": 0.9418,
+      "step": 9836
+    },
+    {
+      "epoch": 0.2701656092938947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712464082351743e-05,
+      "loss": 1.0938,
+      "step": 9837
+    },
+    {
+      "epoch": 0.27019307352173794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7124016614313577e-05,
+      "loss": 0.9657,
+      "step": 9838
+    },
+    {
+      "epoch": 0.27022053774958116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7123392348741744e-05,
+      "loss": 1.0403,
+      "step": 9839
+    },
+    {
+      "epoch": 0.2702480019774244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7122768026806872e-05,
+      "loss": 0.9866,
+      "step": 9840
+    },
+    {
+      "epoch": 0.27027546620526766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71221436485139e-05,
+      "loss": 0.9804,
+      "step": 9841
+    },
+    {
+      "epoch": 0.2703029304331109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7121519213867766e-05,
+      "loss": 1.0133,
+      "step": 9842
+    },
+    {
+      "epoch": 0.2703303946609541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712089472287342e-05,
+      "loss": 1.0043,
+      "step": 9843
+    },
+    {
+      "epoch": 0.2703578588887973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712027017553579e-05,
+      "loss": 0.9185,
+      "step": 9844
+    },
+    {
+      "epoch": 0.2703853231166406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7119645571859828e-05,
+      "loss": 0.88,
+      "step": 9845
+    },
+    {
+      "epoch": 0.2704127873444838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711902091185047e-05,
+      "loss": 1.0838,
+      "step": 9846
+    },
+    {
+      "epoch": 0.27044025157232704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711839619551266e-05,
+      "loss": 1.0721,
+      "step": 9847
+    },
+    {
+      "epoch": 0.27046771580017026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7117771422851347e-05,
+      "loss": 0.9804,
+      "step": 9848
+    },
+    {
+      "epoch": 0.27049518002801354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7117146593871464e-05,
+      "loss": 0.9903,
+      "step": 9849
+    },
+    {
+      "epoch": 0.27052264425585676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7116521708577964e-05,
+      "loss": 0.9757,
+      "step": 9850
+    },
+    {
+      "epoch": 0.2705501084837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7115896766975786e-05,
+      "loss": 1.0037,
+      "step": 9851
+    },
+    {
+      "epoch": 0.2705775727115432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7115271769069876e-05,
+      "loss": 0.9724,
+      "step": 9852
+    },
+    {
+      "epoch": 0.2706050369393864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711464671486518e-05,
+      "loss": 0.8611,
+      "step": 9853
+    },
+    {
+      "epoch": 0.2706325011672297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711402160436664e-05,
+      "loss": 0.9914,
+      "step": 9854
+    },
+    {
+      "epoch": 0.2706599653950729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711339643757921e-05,
+      "loss": 0.9658,
+      "step": 9855
+    },
+    {
+      "epoch": 0.27068742962291614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711277121450783e-05,
+      "loss": 0.8702,
+      "step": 9856
+    },
+    {
+      "epoch": 0.27071489385075936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711214593515745e-05,
+      "loss": 1.0436,
+      "step": 9857
+    },
+    {
+      "epoch": 0.27074235807860264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7111520599533015e-05,
+      "loss": 0.9432,
+      "step": 9858
+    },
+    {
+      "epoch": 0.27076982230644586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7110895207639477e-05,
+      "loss": 0.9042,
+      "step": 9859
+    },
+    {
+      "epoch": 0.2707972865342891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711026975948178e-05,
+      "loss": 0.8954,
+      "step": 9860
+    },
+    {
+      "epoch": 0.2708247507621323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7109644255064873e-05,
+      "loss": 0.9305,
+      "step": 9861
+    },
+    {
+      "epoch": 0.2708522149899756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7109018694393708e-05,
+      "loss": 0.9464,
+      "step": 9862
+    },
+    {
+      "epoch": 0.2708796792178188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7108393077473233e-05,
+      "loss": 0.9659,
+      "step": 9863
+    },
+    {
+      "epoch": 0.270907143445662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7107767404308394e-05,
+      "loss": 0.9737,
+      "step": 9864
+    },
+    {
+      "epoch": 0.27093460767350525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7107141674904153e-05,
+      "loss": 1.0418,
+      "step": 9865
+    },
+    {
+      "epoch": 0.27096207190134847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7106515889265455e-05,
+      "loss": 0.8983,
+      "step": 9866
+    },
+    {
+      "epoch": 0.27098953612919174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7105890047397244e-05,
+      "loss": 1.0255,
+      "step": 9867
+    },
+    {
+      "epoch": 0.27101700035703497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710526414930448e-05,
+      "loss": 1.1725,
+      "step": 9868
+    },
+    {
+      "epoch": 0.2710444645848782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7104638194992117e-05,
+      "loss": 0.8698,
+      "step": 9869
+    },
+    {
+      "epoch": 0.2710719288127214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71040121844651e-05,
+      "loss": 0.96,
+      "step": 9870
+    },
+    {
+      "epoch": 0.2710993930405647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710338611772839e-05,
+      "loss": 1.0112,
+      "step": 9871
+    },
+    {
+      "epoch": 0.2711268572684079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7102759994786938e-05,
+      "loss": 1.0281,
+      "step": 9872
+    },
+    {
+      "epoch": 0.2711543214962511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7102133815645695e-05,
+      "loss": 0.9432,
+      "step": 9873
+    },
+    {
+      "epoch": 0.27118178572409435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710150758030962e-05,
+      "loss": 0.9906,
+      "step": 9874
+    },
+    {
+      "epoch": 0.2712092499519376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7100881288783665e-05,
+      "loss": 0.9772,
+      "step": 9875
+    },
+    {
+      "epoch": 0.27123671417978085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710025494107279e-05,
+      "loss": 0.9374,
+      "step": 9876
+    },
+    {
+      "epoch": 0.27126417840762407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7099628537181944e-05,
+      "loss": 1.0432,
+      "step": 9877
+    },
+    {
+      "epoch": 0.2712916426354673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7099002077116087e-05,
+      "loss": 0.922,
+      "step": 9878
+    },
+    {
+      "epoch": 0.2713191068633105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7098375560880176e-05,
+      "loss": 1.0337,
+      "step": 9879
+    },
+    {
+      "epoch": 0.2713465710911538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7097748988479167e-05,
+      "loss": 0.9567,
+      "step": 9880
+    },
+    {
+      "epoch": 0.271374035318997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709712235991802e-05,
+      "loss": 0.9952,
+      "step": 9881
+    },
+    {
+      "epoch": 0.27140149954684023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7096495675201693e-05,
+      "loss": 0.9654,
+      "step": 9882
+    },
+    {
+      "epoch": 0.27142896377468345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709586893433514e-05,
+      "loss": 1.0023,
+      "step": 9883
+    },
+    {
+      "epoch": 0.27145642800252673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7095242137323326e-05,
+      "loss": 0.9661,
+      "step": 9884
+    },
+    {
+      "epoch": 0.27148389223036995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7094615284171207e-05,
+      "loss": 1.004,
+      "step": 9885
+    },
+    {
+      "epoch": 0.27151135645821317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7093988374883745e-05,
+      "loss": 1.0146,
+      "step": 9886
+    },
+    {
+      "epoch": 0.2715388206860564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7093361409465896e-05,
+      "loss": 0.9653,
+      "step": 9887
+    },
+    {
+      "epoch": 0.27156628491389967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7092734387922627e-05,
+      "loss": 0.8846,
+      "step": 9888
+    },
+    {
+      "epoch": 0.2715937491417429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7092107310258893e-05,
+      "loss": 1.1412,
+      "step": 9889
+    },
+    {
+      "epoch": 0.2716212133695861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7091480176479663e-05,
+      "loss": 0.9281,
+      "step": 9890
+    },
+    {
+      "epoch": 0.27164867759742933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709085298658989e-05,
+      "loss": 0.8715,
+      "step": 9891
+    },
+    {
+      "epoch": 0.2716761418252726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7090225740594543e-05,
+      "loss": 1.0297,
+      "step": 9892
+    },
+    {
+      "epoch": 0.27170360605311583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7089598438498583e-05,
+      "loss": 0.8914,
+      "step": 9893
+    },
+    {
+      "epoch": 0.27173107028095905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7088971080306976e-05,
+      "loss": 1.0854,
+      "step": 9894
+    },
+    {
+      "epoch": 0.2717585345088023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7088343666024684e-05,
+      "loss": 0.936,
+      "step": 9895
+    },
+    {
+      "epoch": 0.2717859987366455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.708771619565667e-05,
+      "loss": 0.9269,
+      "step": 9896
+    },
+    {
+      "epoch": 0.2718134629644888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7087088669207902e-05,
+      "loss": 1.0658,
+      "step": 9897
+    },
+    {
+      "epoch": 0.271840927192332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.708646108668334e-05,
+      "loss": 0.9965,
+      "step": 9898
+    },
+    {
+      "epoch": 0.2718683914201752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7085833448087953e-05,
+      "loss": 1.0385,
+      "step": 9899
+    },
+    {
+      "epoch": 0.27189585564801844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7085205753426707e-05,
+      "loss": 1.0182,
+      "step": 9900
+    },
+    {
+      "epoch": 0.2719233198758617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7084578002704568e-05,
+      "loss": 1.0573,
+      "step": 9901
+    },
+    {
+      "epoch": 0.27195078410370493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7083950195926508e-05,
+      "loss": 0.9534,
+      "step": 9902
+    },
+    {
+      "epoch": 0.27197824833154816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7083322333097485e-05,
+      "loss": 0.9162,
+      "step": 9903
+    },
+    {
+      "epoch": 0.2720057125593914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7082694414222474e-05,
+      "loss": 1.0114,
+      "step": 9904
+    },
+    {
+      "epoch": 0.27203317678723465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7082066439306442e-05,
+      "loss": 1.1289,
+      "step": 9905
+    },
+    {
+      "epoch": 0.2720606410150779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7081438408354356e-05,
+      "loss": 0.9888,
+      "step": 9906
+    },
+    {
+      "epoch": 0.2720881052429211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7080810321371183e-05,
+      "loss": 1.0163,
+      "step": 9907
+    },
+    {
+      "epoch": 0.2721155694707643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7080182178361903e-05,
+      "loss": 1.0173,
+      "step": 9908
+    },
+    {
+      "epoch": 0.27214303369860754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7079553979331474e-05,
+      "loss": 1.0604,
+      "step": 9909
+    },
+    {
+      "epoch": 0.2721704979264508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7078925724284875e-05,
+      "loss": 1.0017,
+      "step": 9910
+    },
+    {
+      "epoch": 0.27219796215429404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7078297413227067e-05,
+      "loss": 1.0156,
+      "step": 9911
+    },
+    {
+      "epoch": 0.27222542638213726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7077669046163034e-05,
+      "loss": 1.0608,
+      "step": 9912
+    },
+    {
+      "epoch": 0.2722528906099805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707704062309774e-05,
+      "loss": 0.9895,
+      "step": 9913
+    },
+    {
+      "epoch": 0.27228035483782376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707641214403616e-05,
+      "loss": 0.9526,
+      "step": 9914
+    },
+    {
+      "epoch": 0.272307819065667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7075783608983265e-05,
+      "loss": 0.9667,
+      "step": 9915
+    },
+    {
+      "epoch": 0.2723352832935102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707515501794403e-05,
+      "loss": 0.9705,
+      "step": 9916
+    },
+    {
+      "epoch": 0.2723627475213534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7074526370923424e-05,
+      "loss": 1.0393,
+      "step": 9917
+    },
+    {
+      "epoch": 0.2723902117491967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7073897667926432e-05,
+      "loss": 0.9736,
+      "step": 9918
+    },
+    {
+      "epoch": 0.2724176759770399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707326890895802e-05,
+      "loss": 0.8721,
+      "step": 9919
+    },
+    {
+      "epoch": 0.27244514020488314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7072640094023157e-05,
+      "loss": 1.0496,
+      "step": 9920
+    },
+    {
+      "epoch": 0.27247260443272636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707201122312683e-05,
+      "loss": 0.8774,
+      "step": 9921
+    },
+    {
+      "epoch": 0.2725000686605696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7071382296274017e-05,
+      "loss": 1.0448,
+      "step": 9922
+    },
+    {
+      "epoch": 0.27252753288841286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7070753313469683e-05,
+      "loss": 0.875,
+      "step": 9923
+    },
+    {
+      "epoch": 0.2725549971162561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707012427471881e-05,
+      "loss": 1.0549,
+      "step": 9924
+    },
+    {
+      "epoch": 0.2725824613440993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7069495180026373e-05,
+      "loss": 1.0099,
+      "step": 9925
+    },
+    {
+      "epoch": 0.2726099255719425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7068866029397355e-05,
+      "loss": 0.9676,
+      "step": 9926
+    },
+    {
+      "epoch": 0.2726373897997858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7068236822836728e-05,
+      "loss": 0.9439,
+      "step": 9927
+    },
+    {
+      "epoch": 0.272664854027629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7067607560349474e-05,
+      "loss": 0.8618,
+      "step": 9928
+    },
+    {
+      "epoch": 0.27269231825547224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7066978241940573e-05,
+      "loss": 0.8693,
+      "step": 9929
+    },
+    {
+      "epoch": 0.27271978248331546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7066348867615002e-05,
+      "loss": 1.0477,
+      "step": 9930
+    },
+    {
+      "epoch": 0.27274724671115874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706571943737774e-05,
+      "loss": 0.9426,
+      "step": 9931
+    },
+    {
+      "epoch": 0.27277471093900196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7065089951233772e-05,
+      "loss": 0.964,
+      "step": 9932
+    },
+    {
+      "epoch": 0.2728021751668452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7064460409188074e-05,
+      "loss": 0.9613,
+      "step": 9933
+    },
+    {
+      "epoch": 0.2728296393946884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7063830811245625e-05,
+      "loss": 0.8785,
+      "step": 9934
+    },
+    {
+      "epoch": 0.2728571036225316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7063201157411413e-05,
+      "loss": 0.9306,
+      "step": 9935
+    },
+    {
+      "epoch": 0.2728845678503749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706257144769042e-05,
+      "loss": 0.9572,
+      "step": 9936
+    },
+    {
+      "epoch": 0.2729120320782181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706194168208762e-05,
+      "loss": 0.9672,
+      "step": 9937
+    },
+    {
+      "epoch": 0.27293949630606135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7061311860608008e-05,
+      "loss": 0.9457,
+      "step": 9938
+    },
+    {
+      "epoch": 0.27296696053390457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706068198325656e-05,
+      "loss": 0.9019,
+      "step": 9939
+    },
+    {
+      "epoch": 0.27299442476174784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706005205003826e-05,
+      "loss": 0.9709,
+      "step": 9940
+    },
+    {
+      "epoch": 0.27302188898959107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7059422060958093e-05,
+      "loss": 0.963,
+      "step": 9941
+    },
+    {
+      "epoch": 0.2730493532174343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7058792016021044e-05,
+      "loss": 0.9357,
+      "step": 9942
+    },
+    {
+      "epoch": 0.2730768174452775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70581619152321e-05,
+      "loss": 0.9799,
+      "step": 9943
+    },
+    {
+      "epoch": 0.2731042816731208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7057531758596245e-05,
+      "loss": 0.9307,
+      "step": 9944
+    },
+    {
+      "epoch": 0.273131745900964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7056901546118464e-05,
+      "loss": 0.8518,
+      "step": 9945
+    },
+    {
+      "epoch": 0.27315921012880723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7056271277803742e-05,
+      "loss": 0.8561,
+      "step": 9946
+    },
+    {
+      "epoch": 0.27318667435665045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7055640953657068e-05,
+      "loss": 0.8988,
+      "step": 9947
+    },
+    {
+      "epoch": 0.27321413858449367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7055010573683434e-05,
+      "loss": 0.9509,
+      "step": 9948
+    },
+    {
+      "epoch": 0.27324160281233695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7054380137887822e-05,
+      "loss": 0.946,
+      "step": 9949
+    },
+    {
+      "epoch": 0.27326906704018017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705374964627522e-05,
+      "loss": 0.9628,
+      "step": 9950
+    },
+    {
+      "epoch": 0.2732965312680234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705311909885062e-05,
+      "loss": 1.0103,
+      "step": 9951
+    },
+    {
+      "epoch": 0.2733239954958666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705248849561901e-05,
+      "loss": 0.9841,
+      "step": 9952
+    },
+    {
+      "epoch": 0.2733514597237099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7051857836585376e-05,
+      "loss": 0.9815,
+      "step": 9953
+    },
+    {
+      "epoch": 0.2733789239515531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7051227121754716e-05,
+      "loss": 1.0142,
+      "step": 9954
+    },
+    {
+      "epoch": 0.27340638817939633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705059635113201e-05,
+      "loss": 0.9955,
+      "step": 9955
+    },
+    {
+      "epoch": 0.27343385240723955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704996552472226e-05,
+      "loss": 0.9349,
+      "step": 9956
+    },
+    {
+      "epoch": 0.27346131663508283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704933464253045e-05,
+      "loss": 1.0444,
+      "step": 9957
+    },
+    {
+      "epoch": 0.27348878086292605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704870370456157e-05,
+      "loss": 1.0054,
+      "step": 9958
+    },
+    {
+      "epoch": 0.27351624509076927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704807271082062e-05,
+      "loss": 0.9241,
+      "step": 9959
+    },
+    {
+      "epoch": 0.2735437093186125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704744166131259e-05,
+      "loss": 0.9296,
+      "step": 9960
+    },
+    {
+      "epoch": 0.2735711735464557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7046810556042467e-05,
+      "loss": 0.9763,
+      "step": 9961
+    },
+    {
+      "epoch": 0.273598637774299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7046179395015254e-05,
+      "loss": 1.0874,
+      "step": 9962
+    },
+    {
+      "epoch": 0.2736261020021422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7045548178235938e-05,
+      "loss": 0.9749,
+      "step": 9963
+    },
+    {
+      "epoch": 0.27365356622998543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7044916905709512e-05,
+      "loss": 1.0073,
+      "step": 9964
+    },
+    {
+      "epoch": 0.27368103045782866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704428557744098e-05,
+      "loss": 1.0217,
+      "step": 9965
+    },
+    {
+      "epoch": 0.27370849468567193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704365419343533e-05,
+      "loss": 0.978,
+      "step": 9966
+    },
+    {
+      "epoch": 0.27373595891351515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704302275369756e-05,
+      "loss": 1.063,
+      "step": 9967
+    },
+    {
+      "epoch": 0.2737634231413584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704239125823266e-05,
+      "loss": 1.0256,
+      "step": 9968
+    },
+    {
+      "epoch": 0.2737908873692016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7041759707045643e-05,
+      "loss": 0.9932,
+      "step": 9969
+    },
+    {
+      "epoch": 0.2738183515970449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7041128100141485e-05,
+      "loss": 1.0381,
+      "step": 9970
+    },
+    {
+      "epoch": 0.2738458158248881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70404964375252e-05,
+      "loss": 0.9773,
+      "step": 9971
+    },
+    {
+      "epoch": 0.2738732800527313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703986471920178e-05,
+      "loss": 0.9534,
+      "step": 9972
+    },
+    {
+      "epoch": 0.27390074428057454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703923294517622e-05,
+      "loss": 0.9722,
+      "step": 9973
+    },
+    {
+      "epoch": 0.27392820850841776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7038601115453527e-05,
+      "loss": 0.883,
+      "step": 9974
+    },
+    {
+      "epoch": 0.27395567273626104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7037969230038692e-05,
+      "loss": 0.9667,
+      "step": 9975
+    },
+    {
+      "epoch": 0.27398313696410426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703733728893672e-05,
+      "loss": 1.1035,
+      "step": 9976
+    },
+    {
+      "epoch": 0.2740106011919475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7036705292152607e-05,
+      "loss": 0.9102,
+      "step": 9977
+    },
+    {
+      "epoch": 0.2740380654197907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703607323969136e-05,
+      "loss": 0.9987,
+      "step": 9978
+    },
+    {
+      "epoch": 0.274065529647634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7035441131557974e-05,
+      "loss": 0.9993,
+      "step": 9979
+    },
+    {
+      "epoch": 0.2740929938754772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7034808967757446e-05,
+      "loss": 0.9154,
+      "step": 9980
+    },
+    {
+      "epoch": 0.2741204581033204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7034176748294796e-05,
+      "loss": 1.0247,
+      "step": 9981
+    },
+    {
+      "epoch": 0.27414792233116364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7033544473175006e-05,
+      "loss": 0.9418,
+      "step": 9982
+    },
+    {
+      "epoch": 0.2741753865590069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703291214240309e-05,
+      "loss": 1.0486,
+      "step": 9983
+    },
+    {
+      "epoch": 0.27420285078685014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703227975598405e-05,
+      "loss": 0.9172,
+      "step": 9984
+    },
+    {
+      "epoch": 0.27423031501469336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703164731392289e-05,
+      "loss": 0.9424,
+      "step": 9985
+    },
+    {
+      "epoch": 0.2742577792425366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7031014816224612e-05,
+      "loss": 0.9189,
+      "step": 9986
+    },
+    {
+      "epoch": 0.2742852434703798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703038226289422e-05,
+      "loss": 0.8781,
+      "step": 9987
+    },
+    {
+      "epoch": 0.2743127076982231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7029749653936718e-05,
+      "loss": 0.8461,
+      "step": 9988
+    },
+    {
+      "epoch": 0.2743401719260663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7029116989357115e-05,
+      "loss": 1.0524,
+      "step": 9989
+    },
+    {
+      "epoch": 0.2743676361539095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7028484269160417e-05,
+      "loss": 0.9402,
+      "step": 9990
+    },
+    {
+      "epoch": 0.27439510038175274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702785149335163e-05,
+      "loss": 0.9848,
+      "step": 9991
+    },
+    {
+      "epoch": 0.274422564609596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7027218661935755e-05,
+      "loss": 0.9206,
+      "step": 9992
+    },
+    {
+      "epoch": 0.27445002883743924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702658577491781e-05,
+      "loss": 0.9122,
+      "step": 9993
+    },
+    {
+      "epoch": 0.27447749306528246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702595283230279e-05,
+      "loss": 0.8708,
+      "step": 9994
+    },
+    {
+      "epoch": 0.2745049572931257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7025319834095715e-05,
+      "loss": 0.986,
+      "step": 9995
+    },
+    {
+      "epoch": 0.27453242152096896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7024686780301586e-05,
+      "loss": 1.0011,
+      "step": 9996
+    },
+    {
+      "epoch": 0.2745598857488122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7024053670925412e-05,
+      "loss": 0.9367,
+      "step": 9997
+    },
+    {
+      "epoch": 0.2745873499766554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702342050597221e-05,
+      "loss": 0.8979,
+      "step": 9998
+    },
+    {
+      "epoch": 0.2746148142044986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7022787285446976e-05,
+      "loss": 1.0076,
+      "step": 9999
+    },
+    {
+      "epoch": 0.2746422784323419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7022154009354735e-05,
+      "loss": 0.976,
+      "step": 10000
+    },
+    {
+      "epoch": 0.2746697426601851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702152067770049e-05,
+      "loss": 0.9685,
+      "step": 10001
+    },
+    {
+      "epoch": 0.27469720688802834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7020887290489257e-05,
+      "loss": 1.0676,
+      "step": 10002
+    },
+    {
+      "epoch": 0.27472467111587157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702025384772604e-05,
+      "loss": 0.9836,
+      "step": 10003
+    },
+    {
+      "epoch": 0.2747521353437148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7019620349415856e-05,
+      "loss": 1.033,
+      "step": 10004
+    },
+    {
+      "epoch": 0.27477959957155806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7018986795563718e-05,
+      "loss": 0.9744,
+      "step": 10005
+    },
+    {
+      "epoch": 0.2748070637994013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701835318617463e-05,
+      "loss": 1.0669,
+      "step": 10006
+    },
+    {
+      "epoch": 0.2748345280272445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7017719521253623e-05,
+      "loss": 0.9346,
+      "step": 10007
+    },
+    {
+      "epoch": 0.27486199225508773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7017085800805696e-05,
+      "loss": 0.9238,
+      "step": 10008
+    },
+    {
+      "epoch": 0.274889456482931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7016452024835867e-05,
+      "loss": 0.9283,
+      "step": 10009
+    },
+    {
+      "epoch": 0.2749169207107742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7015818193349155e-05,
+      "loss": 0.9437,
+      "step": 10010
+    },
+    {
+      "epoch": 0.27494438493861745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701518430635057e-05,
+      "loss": 0.9878,
+      "step": 10011
+    },
+    {
+      "epoch": 0.27497184916646067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701455036384513e-05,
+      "loss": 1.0306,
+      "step": 10012
+    },
+    {
+      "epoch": 0.27499931339430395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701391636583785e-05,
+      "loss": 0.9185,
+      "step": 10013
+    },
+    {
+      "epoch": 0.27502677762214717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7013282312333746e-05,
+      "loss": 0.9564,
+      "step": 10014
+    },
+    {
+      "epoch": 0.2750542418499904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7012648203337833e-05,
+      "loss": 1.078,
+      "step": 10015
+    },
+    {
+      "epoch": 0.2750817060778336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7012014038855137e-05,
+      "loss": 0.9633,
+      "step": 10016
+    },
+    {
+      "epoch": 0.27510917030567683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7011379818890668e-05,
+      "loss": 0.94,
+      "step": 10017
+    },
+    {
+      "epoch": 0.2751366345335201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7010745543449443e-05,
+      "loss": 1.024,
+      "step": 10018
+    },
+    {
+      "epoch": 0.27516409876136333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7010111212536486e-05,
+      "loss": 1.0411,
+      "step": 10019
+    },
+    {
+      "epoch": 0.27519156298920655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7009476826156815e-05,
+      "loss": 0.972,
+      "step": 10020
+    },
+    {
+      "epoch": 0.27521902721704977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7008842384315444e-05,
+      "loss": 1.0122,
+      "step": 10021
+    },
+    {
+      "epoch": 0.27524649144489305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7008207887017402e-05,
+      "loss": 0.8376,
+      "step": 10022
+    },
+    {
+      "epoch": 0.27527395567273627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70075733342677e-05,
+      "loss": 0.7652,
+      "step": 10023
+    },
+    {
+      "epoch": 0.2753014199005795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7006938726071366e-05,
+      "loss": 1.0106,
+      "step": 10024
+    },
+    {
+      "epoch": 0.2753288841284227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7006304062433417e-05,
+      "loss": 0.9549,
+      "step": 10025
+    },
+    {
+      "epoch": 0.275356348356266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7005669343358877e-05,
+      "loss": 1.0392,
+      "step": 10026
+    },
+    {
+      "epoch": 0.2753838125841092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700503456885277e-05,
+      "loss": 1.0212,
+      "step": 10027
+    },
+    {
+      "epoch": 0.27541127681195243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700439973892011e-05,
+      "loss": 0.9855,
+      "step": 10028
+    },
+    {
+      "epoch": 0.27543874103979565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700376485356593e-05,
+      "loss": 1.0056,
+      "step": 10029
+    },
+    {
+      "epoch": 0.2754662052676389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7003129912795246e-05,
+      "loss": 1.0256,
+      "step": 10030
+    },
+    {
+      "epoch": 0.27549366949548215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700249491661309e-05,
+      "loss": 0.986,
+      "step": 10031
+    },
+    {
+      "epoch": 0.2755211337233254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700185986502448e-05,
+      "loss": 0.9224,
+      "step": 10032
+    },
+    {
+      "epoch": 0.2755485979511686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700122475803444e-05,
+      "loss": 0.9819,
+      "step": 10033
+    },
+    {
+      "epoch": 0.2755760621790118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7000589595648006e-05,
+      "loss": 0.931,
+      "step": 10034
+    },
+    {
+      "epoch": 0.2756035264068551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6999954377870187e-05,
+      "loss": 0.9969,
+      "step": 10035
+    },
+    {
+      "epoch": 0.2756309906346983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699931910470602e-05,
+      "loss": 1.0212,
+      "step": 10036
+    },
+    {
+      "epoch": 0.27565845486254154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699868377616053e-05,
+      "loss": 1.0214,
+      "step": 10037
+    },
+    {
+      "epoch": 0.27568591909038476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699804839223874e-05,
+      "loss": 1.0521,
+      "step": 10038
+    },
+    {
+      "epoch": 0.27571338331822803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699741295294568e-05,
+      "loss": 0.9858,
+      "step": 10039
+    },
+    {
+      "epoch": 0.27574084754607125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6996777458286383e-05,
+      "loss": 1.0958,
+      "step": 10040
+    },
+    {
+      "epoch": 0.2757683117739145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699614190826587e-05,
+      "loss": 1.0133,
+      "step": 10041
+    },
+    {
+      "epoch": 0.2757957760017577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699550630288917e-05,
+      "loss": 1.0089,
+      "step": 10042
+    },
+    {
+      "epoch": 0.2758232402296009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6994870642161315e-05,
+      "loss": 1.0131,
+      "step": 10043
+    },
+    {
+      "epoch": 0.2758507044574442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699423492608734e-05,
+      "loss": 1.0213,
+      "step": 10044
+    },
+    {
+      "epoch": 0.2758781686852874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6993599154672265e-05,
+      "loss": 1.0452,
+      "step": 10045
+    },
+    {
+      "epoch": 0.27590563291313064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.699296332792112e-05,
+      "loss": 0.8595,
+      "step": 10046
+    },
+    {
+      "epoch": 0.27593309714097386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6992327445838947e-05,
+      "loss": 0.8752,
+      "step": 10047
+    },
+    {
+      "epoch": 0.27596056136881714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6991691508430765e-05,
+      "loss": 1.0168,
+      "step": 10048
+    },
+    {
+      "epoch": 0.27598802559666036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6991055515701615e-05,
+      "loss": 1.0063,
+      "step": 10049
+    },
+    {
+      "epoch": 0.2760154898245036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6990419467656524e-05,
+      "loss": 0.9471,
+      "step": 10050
+    },
+    {
+      "epoch": 0.2760429540523468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6989783364300527e-05,
+      "loss": 1.0802,
+      "step": 10051
+    },
+    {
+      "epoch": 0.2760704182801901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698914720563866e-05,
+      "loss": 0.9927,
+      "step": 10052
+    },
+    {
+      "epoch": 0.2760978825080333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6988510991675945e-05,
+      "loss": 1.0664,
+      "step": 10053
+    },
+    {
+      "epoch": 0.2761253467358765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698787472241743e-05,
+      "loss": 0.906,
+      "step": 10054
+    },
+    {
+      "epoch": 0.27615281096371974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6987238397868143e-05,
+      "loss": 0.9826,
+      "step": 10055
+    },
+    {
+      "epoch": 0.27618027519156296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6986602018033118e-05,
+      "loss": 1.0216,
+      "step": 10056
+    },
+    {
+      "epoch": 0.27620773941940624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698596558291739e-05,
+      "loss": 1.0124,
+      "step": 10057
+    },
+    {
+      "epoch": 0.27623520364724946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6985329092525997e-05,
+      "loss": 0.9382,
+      "step": 10058
+    },
+    {
+      "epoch": 0.2762626678750927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6984692546863978e-05,
+      "loss": 1.0444,
+      "step": 10059
+    },
+    {
+      "epoch": 0.2762901321029359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6984055945936362e-05,
+      "loss": 0.9854,
+      "step": 10060
+    },
+    {
+      "epoch": 0.2763175963307792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6983419289748192e-05,
+      "loss": 1.0772,
+      "step": 10061
+    },
+    {
+      "epoch": 0.2763450605586224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69827825783045e-05,
+      "loss": 0.968,
+      "step": 10062
+    },
+    {
+      "epoch": 0.2763725247864656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6982145811610332e-05,
+      "loss": 1.0623,
+      "step": 10063
+    },
+    {
+      "epoch": 0.27639998901430884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698150898967072e-05,
+      "loss": 0.9496,
+      "step": 10064
+    },
+    {
+      "epoch": 0.2764274532421521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69808721124907e-05,
+      "loss": 1.0257,
+      "step": 10065
+    },
+    {
+      "epoch": 0.27645491746999534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698023518007532e-05,
+      "loss": 0.9414,
+      "step": 10066
+    },
+    {
+      "epoch": 0.27648238169783856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6979598192429616e-05,
+      "loss": 0.8808,
+      "step": 10067
+    },
+    {
+      "epoch": 0.2765098459256818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6978961149558627e-05,
+      "loss": 0.9335,
+      "step": 10068
+    },
+    {
+      "epoch": 0.276537310153525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6978324051467394e-05,
+      "loss": 0.9841,
+      "step": 10069
+    },
+    {
+      "epoch": 0.2765647743813683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6977686898160957e-05,
+      "loss": 0.9536,
+      "step": 10070
+    },
+    {
+      "epoch": 0.2765922386092115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6977049689644357e-05,
+      "loss": 1.0615,
+      "step": 10071
+    },
+    {
+      "epoch": 0.2766197028370547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6976412425922642e-05,
+      "loss": 1.04,
+      "step": 10072
+    },
+    {
+      "epoch": 0.27664716706489795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6975775107000843e-05,
+      "loss": 1.0422,
+      "step": 10073
+    },
+    {
+      "epoch": 0.2766746312927412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6975137732884017e-05,
+      "loss": 1.0027,
+      "step": 10074
+    },
+    {
+      "epoch": 0.27670209552058445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6974500303577196e-05,
+      "loss": 1.0205,
+      "step": 10075
+    },
+    {
+      "epoch": 0.27672955974842767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6973862819085423e-05,
+      "loss": 0.9787,
+      "step": 10076
+    },
+    {
+      "epoch": 0.2767570239762709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697322527941375e-05,
+      "loss": 1.0469,
+      "step": 10077
+    },
+    {
+      "epoch": 0.27678448820411417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6972587684567215e-05,
+      "loss": 0.927,
+      "step": 10078
+    },
+    {
+      "epoch": 0.2768119524319574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697195003455087e-05,
+      "loss": 0.9669,
+      "step": 10079
+    },
+    {
+      "epoch": 0.2768394166598006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6971312329369748e-05,
+      "loss": 0.9683,
+      "step": 10080
+    },
+    {
+      "epoch": 0.27686688088764383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697067456902891e-05,
+      "loss": 0.9605,
+      "step": 10081
+    },
+    {
+      "epoch": 0.27689434511548705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6970036753533392e-05,
+      "loss": 0.9254,
+      "step": 10082
+    },
+    {
+      "epoch": 0.2769218093433303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6969398882888243e-05,
+      "loss": 0.9641,
+      "step": 10083
+    },
+    {
+      "epoch": 0.27694927357117355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696876095709851e-05,
+      "loss": 0.9518,
+      "step": 10084
+    },
+    {
+      "epoch": 0.27697673779901677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696812297616924e-05,
+      "loss": 0.975,
+      "step": 10085
+    },
+    {
+      "epoch": 0.27700420202686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6967484940105482e-05,
+      "loss": 0.8054,
+      "step": 10086
+    },
+    {
+      "epoch": 0.27703166625470327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6966846848912287e-05,
+      "loss": 0.9296,
+      "step": 10087
+    },
+    {
+      "epoch": 0.2770591304825465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69662087025947e-05,
+      "loss": 0.9599,
+      "step": 10088
+    },
+    {
+      "epoch": 0.2770865947103897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696557050115777e-05,
+      "loss": 0.9716,
+      "step": 10089
+    },
+    {
+      "epoch": 0.27711405893823293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6964932244606546e-05,
+      "loss": 1.0301,
+      "step": 10090
+    },
+    {
+      "epoch": 0.2771415231660762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6964293932946083e-05,
+      "loss": 1.0361,
+      "step": 10091
+    },
+    {
+      "epoch": 0.27716898739391943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696365556618143e-05,
+      "loss": 0.9492,
+      "step": 10092
+    },
+    {
+      "epoch": 0.27719645162176265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6963017144317633e-05,
+      "loss": 0.9966,
+      "step": 10093
+    },
+    {
+      "epoch": 0.2772239158496059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6962378667359748e-05,
+      "loss": 1.0491,
+      "step": 10094
+    },
+    {
+      "epoch": 0.27725138007744915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696174013531283e-05,
+      "loss": 0.9403,
+      "step": 10095
+    },
+    {
+      "epoch": 0.27727884430529237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696110154818192e-05,
+      "loss": 0.9924,
+      "step": 10096
+    },
+    {
+      "epoch": 0.2773063085331356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6960462905972082e-05,
+      "loss": 0.922,
+      "step": 10097
+    },
+    {
+      "epoch": 0.2773337727609788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6959824208688368e-05,
+      "loss": 0.9717,
+      "step": 10098
+    },
+    {
+      "epoch": 0.27736123698882204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6959185456335826e-05,
+      "loss": 0.9921,
+      "step": 10099
+    },
+    {
+      "epoch": 0.2773887012166653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6958546648919513e-05,
+      "loss": 1.0128,
+      "step": 10100
+    },
+    {
+      "epoch": 0.27741616544450853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6957907786444482e-05,
+      "loss": 1.0665,
+      "step": 10101
+    },
+    {
+      "epoch": 0.27744362967235175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6957268868915796e-05,
+      "loss": 0.9587,
+      "step": 10102
+    },
+    {
+      "epoch": 0.277471093900195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6956629896338497e-05,
+      "loss": 0.9582,
+      "step": 10103
+    },
+    {
+      "epoch": 0.27749855812803825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6955990868717652e-05,
+      "loss": 0.9254,
+      "step": 10104
+    },
+    {
+      "epoch": 0.2775260223558815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6955351786058308e-05,
+      "loss": 0.8873,
+      "step": 10105
+    },
+    {
+      "epoch": 0.2775534865837247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695471264836553e-05,
+      "loss": 1.006,
+      "step": 10106
+    },
+    {
+      "epoch": 0.2775809508115679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695407345564437e-05,
+      "loss": 0.9649,
+      "step": 10107
+    },
+    {
+      "epoch": 0.2776084150394112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6953434207899886e-05,
+      "loss": 0.9988,
+      "step": 10108
+    },
+    {
+      "epoch": 0.2776358792672544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695279490513714e-05,
+      "loss": 1.0469,
+      "step": 10109
+    },
+    {
+      "epoch": 0.27766334349509764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6952155547361185e-05,
+      "loss": 0.898,
+      "step": 10110
+    },
+    {
+      "epoch": 0.27769080772294086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6951516134577085e-05,
+      "loss": 1.019,
+      "step": 10111
+    },
+    {
+      "epoch": 0.2777182719507841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695087666678989e-05,
+      "loss": 0.8549,
+      "step": 10112
+    },
+    {
+      "epoch": 0.27774573617862736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6950237144004675e-05,
+      "loss": 1.0271,
+      "step": 10113
+    },
+    {
+      "epoch": 0.2777732004064706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6949597566226485e-05,
+      "loss": 1.0049,
+      "step": 10114
+    },
+    {
+      "epoch": 0.2778006646343138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694895793346039e-05,
+      "loss": 0.9026,
+      "step": 10115
+    },
+    {
+      "epoch": 0.277828128862157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694831824571145e-05,
+      "loss": 1.0105,
+      "step": 10116
+    },
+    {
+      "epoch": 0.2778555930900003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694767850298472e-05,
+      "loss": 0.9875,
+      "step": 10117
+    },
+    {
+      "epoch": 0.2778830573178435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694703870528527e-05,
+      "loss": 0.9411,
+      "step": 10118
+    },
+    {
+      "epoch": 0.27791052154568674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6946398852618156e-05,
+      "loss": 0.9433,
+      "step": 10119
+    },
+    {
+      "epoch": 0.27793798577352996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6945758944988443e-05,
+      "loss": 1.049,
+      "step": 10120
+    },
+    {
+      "epoch": 0.27796545000137324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6945118982401197e-05,
+      "loss": 1.0125,
+      "step": 10121
+    },
+    {
+      "epoch": 0.27799291422921646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6944478964861478e-05,
+      "loss": 0.944,
+      "step": 10122
+    },
+    {
+      "epoch": 0.2780203784570597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6943838892374352e-05,
+      "loss": 0.8781,
+      "step": 10123
+    },
+    {
+      "epoch": 0.2780478426849029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6943198764944882e-05,
+      "loss": 1.0516,
+      "step": 10124
+    },
+    {
+      "epoch": 0.2780753069127461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694255858257814e-05,
+      "loss": 0.9839,
+      "step": 10125
+    },
+    {
+      "epoch": 0.2781027711405894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6941918345279175e-05,
+      "loss": 0.8643,
+      "step": 10126
+    },
+    {
+      "epoch": 0.2781302353684326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6941278053053068e-05,
+      "loss": 0.8875,
+      "step": 10127
+    },
+    {
+      "epoch": 0.27815769959627584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6940637705904884e-05,
+      "loss": 1.053,
+      "step": 10128
+    },
+    {
+      "epoch": 0.27818516382411906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693999730383968e-05,
+      "loss": 0.907,
+      "step": 10129
+    },
+    {
+      "epoch": 0.27821262805196234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693935684686253e-05,
+      "loss": 0.8124,
+      "step": 10130
+    },
+    {
+      "epoch": 0.27824009227980556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6938716334978503e-05,
+      "loss": 0.9218,
+      "step": 10131
+    },
+    {
+      "epoch": 0.2782675565076488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6938075768192663e-05,
+      "loss": 0.9421,
+      "step": 10132
+    },
+    {
+      "epoch": 0.278295020735492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6937435146510083e-05,
+      "loss": 0.8507,
+      "step": 10133
+    },
+    {
+      "epoch": 0.2783224849633353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6936794469935824e-05,
+      "loss": 1.0245,
+      "step": 10134
+    },
+    {
+      "epoch": 0.2783499491911785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6936153738474962e-05,
+      "loss": 1.0065,
+      "step": 10135
+    },
+    {
+      "epoch": 0.2783774134190217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6935512952132565e-05,
+      "loss": 0.8943,
+      "step": 10136
+    },
+    {
+      "epoch": 0.27840487764686495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6934872110913706e-05,
+      "loss": 0.9392,
+      "step": 10137
+    },
+    {
+      "epoch": 0.27843234187470817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693423121482345e-05,
+      "loss": 1.0703,
+      "step": 10138
+    },
+    {
+      "epoch": 0.27845980610255144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693359026386687e-05,
+      "loss": 0.9697,
+      "step": 10139
+    },
+    {
+      "epoch": 0.27848727033039467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6932949258049037e-05,
+      "loss": 1.0571,
+      "step": 10140
+    },
+    {
+      "epoch": 0.2785147345582379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6932308197375023e-05,
+      "loss": 0.908,
+      "step": 10141
+    },
+    {
+      "epoch": 0.2785421987860811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6931667081849904e-05,
+      "loss": 0.9792,
+      "step": 10142
+    },
+    {
+      "epoch": 0.2785696630139244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693102591147875e-05,
+      "loss": 0.9403,
+      "step": 10143
+    },
+    {
+      "epoch": 0.2785971272417676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6930384686266635e-05,
+      "loss": 0.9922,
+      "step": 10144
+    },
+    {
+      "epoch": 0.2786245914696108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6929743406218627e-05,
+      "loss": 0.9823,
+      "step": 10145
+    },
+    {
+      "epoch": 0.27865205569745405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692910207133981e-05,
+      "loss": 1.0725,
+      "step": 10146
+    },
+    {
+      "epoch": 0.2786795199252973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692846068163525e-05,
+      "loss": 1.0685,
+      "step": 10147
+    },
+    {
+      "epoch": 0.27870698415314055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6927819237110027e-05,
+      "loss": 0.9774,
+      "step": 10148
+    },
+    {
+      "epoch": 0.27873444838098377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6927177737769214e-05,
+      "loss": 0.9213,
+      "step": 10149
+    },
+    {
+      "epoch": 0.278761912608827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692653618361789e-05,
+      "loss": 0.9962,
+      "step": 10150
+    },
+    {
+      "epoch": 0.2787893768366702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6925894574661124e-05,
+      "loss": 0.9061,
+      "step": 10151
+    },
+    {
+      "epoch": 0.2788168410645135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6925252910904003e-05,
+      "loss": 1.1014,
+      "step": 10152
+    },
+    {
+      "epoch": 0.2788443052923567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692461119235159e-05,
+      "loss": 0.9289,
+      "step": 10153
+    },
+    {
+      "epoch": 0.27887176952019993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692396941900898e-05,
+      "loss": 0.9892,
+      "step": 10154
+    },
+    {
+      "epoch": 0.27889923374804315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6923327590881234e-05,
+      "loss": 0.8981,
+      "step": 10155
+    },
+    {
+      "epoch": 0.27892669797588643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6922685707973446e-05,
+      "loss": 1.0206,
+      "step": 10156
+    },
+    {
+      "epoch": 0.27895416220372965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692204377029068e-05,
+      "loss": 1.0334,
+      "step": 10157
+    },
+    {
+      "epoch": 0.27898162643157287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6921401777838026e-05,
+      "loss": 1.0151,
+      "step": 10158
+    },
+    {
+      "epoch": 0.2790090906594161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692075973062056e-05,
+      "loss": 0.9621,
+      "step": 10159
+    },
+    {
+      "epoch": 0.27903655488725937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692011762864336e-05,
+      "loss": 1.0628,
+      "step": 10160
+    },
+    {
+      "epoch": 0.2790640191151026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6919475471911513e-05,
+      "loss": 0.9838,
+      "step": 10161
+    },
+    {
+      "epoch": 0.2790914833429458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6918833260430093e-05,
+      "loss": 1.0308,
+      "step": 10162
+    },
+    {
+      "epoch": 0.27911894757078903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6918190994204186e-05,
+      "loss": 1.014,
+      "step": 10163
+    },
+    {
+      "epoch": 0.27914641179863225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.691754867323887e-05,
+      "loss": 0.9308,
+      "step": 10164
+    },
+    {
+      "epoch": 0.27917387602647553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.691690629753923e-05,
+      "loss": 0.9887,
+      "step": 10165
+    },
+    {
+      "epoch": 0.27920134025431875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6916263867110347e-05,
+      "loss": 0.9506,
+      "step": 10166
+    },
+    {
+      "epoch": 0.279228804482162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6915621381957302e-05,
+      "loss": 0.9992,
+      "step": 10167
+    },
+    {
+      "epoch": 0.2792562687100052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6914978842085187e-05,
+      "loss": 1.0444,
+      "step": 10168
+    },
+    {
+      "epoch": 0.2792837329378485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.691433624749908e-05,
+      "loss": 1.012,
+      "step": 10169
+    },
+    {
+      "epoch": 0.2793111971656917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6913693598204067e-05,
+      "loss": 0.9918,
+      "step": 10170
+    },
+    {
+      "epoch": 0.2793386613935349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.691305089420523e-05,
+      "loss": 0.9366,
+      "step": 10171
+    },
+    {
+      "epoch": 0.27936612562137814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6912408135507656e-05,
+      "loss": 1.0357,
+      "step": 10172
+    },
+    {
+      "epoch": 0.2793935898492214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911765322116432e-05,
+      "loss": 0.9566,
+      "step": 10173
+    },
+    {
+      "epoch": 0.27942105407706463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911122454036644e-05,
+      "loss": 0.9651,
+      "step": 10174
+    },
+    {
+      "epoch": 0.27944851830490786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.691047953127338e-05,
+      "loss": 0.942,
+      "step": 10175
+    },
+    {
+      "epoch": 0.2794759825327511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690983655383172e-05,
+      "loss": 0.9727,
+      "step": 10176
+    },
+    {
+      "epoch": 0.2795034467605943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690919352171676e-05,
+      "loss": 0.935,
+      "step": 10177
+    },
+    {
+      "epoch": 0.2795309109884376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6908550434933583e-05,
+      "loss": 1.0308,
+      "step": 10178
+    },
+    {
+      "epoch": 0.2795583752162808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690790729348728e-05,
+      "loss": 0.9285,
+      "step": 10179
+    },
+    {
+      "epoch": 0.279585839444124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690726409738294e-05,
+      "loss": 0.9634,
+      "step": 10180
+    },
+    {
+      "epoch": 0.27961330367196724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690662084662565e-05,
+      "loss": 0.9664,
+      "step": 10181
+    },
+    {
+      "epoch": 0.2796407678998105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6905977541220498e-05,
+      "loss": 1.0855,
+      "step": 10182
+    },
+    {
+      "epoch": 0.27966823212765374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690533418117258e-05,
+      "loss": 0.9407,
+      "step": 10183
+    },
+    {
+      "epoch": 0.27969569635549696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690469076648698e-05,
+      "loss": 1.0621,
+      "step": 10184
+    },
+    {
+      "epoch": 0.2797231605833402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6904047297168795e-05,
+      "loss": 0.9605,
+      "step": 10185
+    },
+    {
+      "epoch": 0.27975062481118346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6903403773223113e-05,
+      "loss": 0.9741,
+      "step": 10186
+    },
+    {
+      "epoch": 0.2797780890390267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6902760194655023e-05,
+      "loss": 1.1025,
+      "step": 10187
+    },
+    {
+      "epoch": 0.2798055532668699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6902116561469625e-05,
+      "loss": 1.0085,
+      "step": 10188
+    },
+    {
+      "epoch": 0.2798330174947131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6901472873672002e-05,
+      "loss": 0.9889,
+      "step": 10189
+    },
+    {
+      "epoch": 0.2798604817225564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6900829131267256e-05,
+      "loss": 0.9374,
+      "step": 10190
+    },
+    {
+      "epoch": 0.2798879459503996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690018533426048e-05,
+      "loss": 1.0209,
+      "step": 10191
+    },
+    {
+      "epoch": 0.27991541017824284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6899541482656756e-05,
+      "loss": 1.0021,
+      "step": 10192
+    },
+    {
+      "epoch": 0.27994287440608606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689889757646119e-05,
+      "loss": 0.9408,
+      "step": 10193
+    },
+    {
+      "epoch": 0.2799703386339293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6898253615678876e-05,
+      "loss": 0.9174,
+      "step": 10194
+    },
+    {
+      "epoch": 0.27999780286177256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689760960031491e-05,
+      "loss": 0.9069,
+      "step": 10195
+    },
+    {
+      "epoch": 0.2800252670896158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689696553037438e-05,
+      "loss": 1.045,
+      "step": 10196
+    },
+    {
+      "epoch": 0.280052731317459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689632140586239e-05,
+      "loss": 0.9882,
+      "step": 10197
+    },
+    {
+      "epoch": 0.2800801955453022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6895677226784034e-05,
+      "loss": 0.8927,
+      "step": 10198
+    },
+    {
+      "epoch": 0.2801076597731455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6895032993144407e-05,
+      "loss": 0.9401,
+      "step": 10199
+    },
+    {
+      "epoch": 0.2801351240009887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6894388704948608e-05,
+      "loss": 1.0127,
+      "step": 10200
+    },
+    {
+      "epoch": 0.28016258822883194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6893744362201735e-05,
+      "loss": 0.9695,
+      "step": 10201
+    },
+    {
+      "epoch": 0.28019005245667516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6893099964908885e-05,
+      "loss": 0.9916,
+      "step": 10202
+    },
+    {
+      "epoch": 0.28021751668451844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689245551307516e-05,
+      "loss": 0.9716,
+      "step": 10203
+    },
+    {
+      "epoch": 0.28024498091236166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6891811006705655e-05,
+      "loss": 0.8874,
+      "step": 10204
+    },
+    {
+      "epoch": 0.2802724451402049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6891166445805477e-05,
+      "loss": 0.9898,
+      "step": 10205
+    },
+    {
+      "epoch": 0.2802999093680481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6890521830379715e-05,
+      "loss": 0.8989,
+      "step": 10206
+    },
+    {
+      "epoch": 0.2803273735958913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6889877160433478e-05,
+      "loss": 0.9711,
+      "step": 10207
+    },
+    {
+      "epoch": 0.2803548378237346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688923243597186e-05,
+      "loss": 0.9909,
+      "step": 10208
+    },
+    {
+      "epoch": 0.2803823020515778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688858765699997e-05,
+      "loss": 1.0135,
+      "step": 10209
+    },
+    {
+      "epoch": 0.28040976627942105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6887942823522907e-05,
+      "loss": 0.9565,
+      "step": 10210
+    },
+    {
+      "epoch": 0.28043723050726427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688729793554577e-05,
+      "loss": 1.0288,
+      "step": 10211
+    },
+    {
+      "epoch": 0.28046469473510754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6886652993073666e-05,
+      "loss": 0.9811,
+      "step": 10212
+    },
+    {
+      "epoch": 0.28049215896295077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6886007996111694e-05,
+      "loss": 1.0192,
+      "step": 10213
+    },
+    {
+      "epoch": 0.280519623190794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6885362944664962e-05,
+      "loss": 0.8824,
+      "step": 10214
+    },
+    {
+      "epoch": 0.2805470874186372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6884717838738568e-05,
+      "loss": 0.9586,
+      "step": 10215
+    },
+    {
+      "epoch": 0.2805745516464805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6884072678337618e-05,
+      "loss": 0.9589,
+      "step": 10216
+    },
+    {
+      "epoch": 0.2806020158743237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6883427463467222e-05,
+      "loss": 0.9186,
+      "step": 10217
+    },
+    {
+      "epoch": 0.28062948010216693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6882782194132482e-05,
+      "loss": 0.9149,
+      "step": 10218
+    },
+    {
+      "epoch": 0.28065694433001015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68821368703385e-05,
+      "loss": 1.0066,
+      "step": 10219
+    },
+    {
+      "epoch": 0.28068440855785337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688149149209039e-05,
+      "loss": 0.8741,
+      "step": 10220
+    },
+    {
+      "epoch": 0.28071187278569665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688084605939325e-05,
+      "loss": 1.0092,
+      "step": 10221
+    },
+    {
+      "epoch": 0.28073933701353987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688020057225219e-05,
+      "loss": 1.0276,
+      "step": 10222
+    },
+    {
+      "epoch": 0.2807668012413831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687955503067232e-05,
+      "loss": 0.9935,
+      "step": 10223
+    },
+    {
+      "epoch": 0.2807942654692263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6878909434658745e-05,
+      "loss": 1.0418,
+      "step": 10224
+    },
+    {
+      "epoch": 0.2808217296970696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6878263784216577e-05,
+      "loss": 0.9537,
+      "step": 10225
+    },
+    {
+      "epoch": 0.2808491939249128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6877618079350916e-05,
+      "loss": 0.9081,
+      "step": 10226
+    },
+    {
+      "epoch": 0.28087665815275603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6876972320066877e-05,
+      "loss": 1.0497,
+      "step": 10227
+    },
+    {
+      "epoch": 0.28090412238059925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687632650636957e-05,
+      "loss": 0.9734,
+      "step": 10228
+    },
+    {
+      "epoch": 0.28093158660844253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6875680638264107e-05,
+      "loss": 0.9797,
+      "step": 10229
+    },
+    {
+      "epoch": 0.28095905083628575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6875034715755593e-05,
+      "loss": 0.948,
+      "step": 10230
+    },
+    {
+      "epoch": 0.28098651506412897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687438873884914e-05,
+      "loss": 0.9878,
+      "step": 10231
+    },
+    {
+      "epoch": 0.2810139792919722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6873742707549862e-05,
+      "loss": 1.0241,
+      "step": 10232
+    },
+    {
+      "epoch": 0.2810414435198154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687309662186287e-05,
+      "loss": 0.9875,
+      "step": 10233
+    },
+    {
+      "epoch": 0.2810689077476587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6872450481793277e-05,
+      "loss": 1.0542,
+      "step": 10234
+    },
+    {
+      "epoch": 0.2810963719755019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687180428734619e-05,
+      "loss": 0.9072,
+      "step": 10235
+    },
+    {
+      "epoch": 0.28112383620334513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6871158038526725e-05,
+      "loss": 1.0397,
+      "step": 10236
+    },
+    {
+      "epoch": 0.28115130043118836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6870511735340002e-05,
+      "loss": 0.9812,
+      "step": 10237
+    },
+    {
+      "epoch": 0.28117876465903163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686986537779112e-05,
+      "loss": 1.0229,
+      "step": 10238
+    },
+    {
+      "epoch": 0.28120622888687485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686921896588521e-05,
+      "loss": 1.0634,
+      "step": 10239
+    },
+    {
+      "epoch": 0.2812336931147181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686857249962737e-05,
+      "loss": 1.0126,
+      "step": 10240
+    },
+    {
+      "epoch": 0.2812611573425613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6867925979022734e-05,
+      "loss": 0.9838,
+      "step": 10241
+    },
+    {
+      "epoch": 0.2812886215704046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6867279404076396e-05,
+      "loss": 0.9407,
+      "step": 10242
+    },
+    {
+      "epoch": 0.2813160857982478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686663277479349e-05,
+      "loss": 0.9316,
+      "step": 10243
+    },
+    {
+      "epoch": 0.281343550026091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686598609117913e-05,
+      "loss": 1.0016,
+      "step": 10244
+    },
+    {
+      "epoch": 0.28137101425393424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686533935323842e-05,
+      "loss": 0.9715,
+      "step": 10245
+    },
+    {
+      "epoch": 0.28139847848177746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6864692560976486e-05,
+      "loss": 1.0712,
+      "step": 10246
+    },
+    {
+      "epoch": 0.28142594270962074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6864045714398453e-05,
+      "loss": 0.9091,
+      "step": 10247
+    },
+    {
+      "epoch": 0.28145340693746396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6863398813509423e-05,
+      "loss": 0.9624,
+      "step": 10248
+    },
+    {
+      "epoch": 0.2814808711653072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6862751858314528e-05,
+      "loss": 1.0834,
+      "step": 10249
+    },
+    {
+      "epoch": 0.2815083353931504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6862104848818882e-05,
+      "loss": 1.009,
+      "step": 10250
+    },
+    {
+      "epoch": 0.2815357996209937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68614577850276e-05,
+      "loss": 0.9991,
+      "step": 10251
+    },
+    {
+      "epoch": 0.2815632638488369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686081066694581e-05,
+      "loss": 0.981,
+      "step": 10252
+    },
+    {
+      "epoch": 0.2815907280766801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686016349457863e-05,
+      "loss": 0.9974,
+      "step": 10253
+    },
+    {
+      "epoch": 0.28161819230452334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6859516267931172e-05,
+      "loss": 0.9599,
+      "step": 10254
+    },
+    {
+      "epoch": 0.2816456565323666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6858868987008572e-05,
+      "loss": 1.0054,
+      "step": 10255
+    },
+    {
+      "epoch": 0.28167312076020984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685822165181594e-05,
+      "loss": 0.9327,
+      "step": 10256
+    },
+    {
+      "epoch": 0.28170058498805306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68575742623584e-05,
+      "loss": 0.9363,
+      "step": 10257
+    },
+    {
+      "epoch": 0.2817280492158963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685692681864108e-05,
+      "loss": 1.0377,
+      "step": 10258
+    },
+    {
+      "epoch": 0.2817555134437395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68562793206691e-05,
+      "loss": 1.0001,
+      "step": 10259
+    },
+    {
+      "epoch": 0.2817829776715828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6855631768447575e-05,
+      "loss": 1.014,
+      "step": 10260
+    },
+    {
+      "epoch": 0.281810441899426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685498416198164e-05,
+      "loss": 1.0255,
+      "step": 10261
+    },
+    {
+      "epoch": 0.2818379061272692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6854336501276416e-05,
+      "loss": 1.0061,
+      "step": 10262
+    },
+    {
+      "epoch": 0.28186537035511244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6853688786337024e-05,
+      "loss": 1.0467,
+      "step": 10263
+    },
+    {
+      "epoch": 0.2818928345829557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6853041017168594e-05,
+      "loss": 0.9441,
+      "step": 10264
+    },
+    {
+      "epoch": 0.28192029881079894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6852393193776248e-05,
+      "loss": 1.015,
+      "step": 10265
+    },
+    {
+      "epoch": 0.28194776303864216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685174531616511e-05,
+      "loss": 1.0424,
+      "step": 10266
+    },
+    {
+      "epoch": 0.2819752272664854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6851097384340313e-05,
+      "loss": 1.0063,
+      "step": 10267
+    },
+    {
+      "epoch": 0.28200269149432866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685044939830698e-05,
+      "loss": 0.8445,
+      "step": 10268
+    },
+    {
+      "epoch": 0.2820301557221719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684980135807023e-05,
+      "loss": 0.9746,
+      "step": 10269
+    },
+    {
+      "epoch": 0.2820576199500151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6849153263635206e-05,
+      "loss": 0.9916,
+      "step": 10270
+    },
+    {
+      "epoch": 0.2820850841778583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6848505115007027e-05,
+      "loss": 0.9348,
+      "step": 10271
+    },
+    {
+      "epoch": 0.28211254840570155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6847856912190823e-05,
+      "loss": 0.9884,
+      "step": 10272
+    },
+    {
+      "epoch": 0.2821400126335448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684720865519172e-05,
+      "loss": 0.8595,
+      "step": 10273
+    },
+    {
+      "epoch": 0.28216747686138804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6846560344014852e-05,
+      "loss": 0.9847,
+      "step": 10274
+    },
+    {
+      "epoch": 0.28219494108923127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6845911978665344e-05,
+      "loss": 0.9585,
+      "step": 10275
+    },
+    {
+      "epoch": 0.2822224053170745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6845263559148334e-05,
+      "loss": 1.0324,
+      "step": 10276
+    },
+    {
+      "epoch": 0.28224986954491776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684461508546894e-05,
+      "loss": 1.025,
+      "step": 10277
+    },
+    {
+      "epoch": 0.282277333772761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6843966557632306e-05,
+      "loss": 0.9717,
+      "step": 10278
+    },
+    {
+      "epoch": 0.2823047980006042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6843317975643555e-05,
+      "loss": 0.9794,
+      "step": 10279
+    },
+    {
+      "epoch": 0.28233226222844743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684266933950782e-05,
+      "loss": 1.0227,
+      "step": 10280
+    },
+    {
+      "epoch": 0.2823597264562907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6842020649230236e-05,
+      "loss": 0.9751,
+      "step": 10281
+    },
+    {
+      "epoch": 0.2823871906841339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6841371904815934e-05,
+      "loss": 1.0843,
+      "step": 10282
+    },
+    {
+      "epoch": 0.28241465491197715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6840723106270047e-05,
+      "loss": 0.8893,
+      "step": 10283
+    },
+    {
+      "epoch": 0.28244211913982037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6840074253597708e-05,
+      "loss": 0.8968,
+      "step": 10284
+    },
+    {
+      "epoch": 0.28246958336766365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6839425346804056e-05,
+      "loss": 1.1024,
+      "step": 10285
+    },
+    {
+      "epoch": 0.28249704759550687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6838776385894218e-05,
+      "loss": 1.0311,
+      "step": 10286
+    },
+    {
+      "epoch": 0.2825245118233501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6838127370873333e-05,
+      "loss": 0.9753,
+      "step": 10287
+    },
+    {
+      "epoch": 0.2825519760511933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6837478301746535e-05,
+      "loss": 0.8589,
+      "step": 10288
+    },
+    {
+      "epoch": 0.28257944027903653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683682917851896e-05,
+      "loss": 0.9681,
+      "step": 10289
+    },
+    {
+      "epoch": 0.2826069045068798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6836180001195745e-05,
+      "loss": 0.9927,
+      "step": 10290
+    },
+    {
+      "epoch": 0.28263436873472303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6835530769782025e-05,
+      "loss": 1.0444,
+      "step": 10291
+    },
+    {
+      "epoch": 0.28266183296256625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834881484282938e-05,
+      "loss": 1.0236,
+      "step": 10292
+    },
+    {
+      "epoch": 0.28268929719040947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834232144703622e-05,
+      "loss": 0.9136,
+      "step": 10293
+    },
+    {
+      "epoch": 0.28271676141825275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6833582751049215e-05,
+      "loss": 0.9861,
+      "step": 10294
+    },
+    {
+      "epoch": 0.28274422564609597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683293330332485e-05,
+      "loss": 0.931,
+      "step": 10295
+    },
+    {
+      "epoch": 0.2827716898739392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6832283801535672e-05,
+      "loss": 0.9788,
+      "step": 10296
+    },
+    {
+      "epoch": 0.2827991541017824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6831634245686815e-05,
+      "loss": 0.9821,
+      "step": 10297
+    },
+    {
+      "epoch": 0.2828266183296257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6830984635783426e-05,
+      "loss": 0.9822,
+      "step": 10298
+    },
+    {
+      "epoch": 0.2828540825574689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683033497183064e-05,
+      "loss": 1.0033,
+      "step": 10299
+    },
+    {
+      "epoch": 0.28288154678531213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6829685253833597e-05,
+      "loss": 1.0128,
+      "step": 10300
+    },
+    {
+      "epoch": 0.28290901101315535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6829035481797436e-05,
+      "loss": 1.0361,
+      "step": 10301
+    },
+    {
+      "epoch": 0.2829364752409986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6828385655727306e-05,
+      "loss": 0.9303,
+      "step": 10302
+    },
+    {
+      "epoch": 0.28296393946884185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682773577562834e-05,
+      "loss": 1.0872,
+      "step": 10303
+    },
+    {
+      "epoch": 0.2829914036966851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6827085841505687e-05,
+      "loss": 1.1099,
+      "step": 10304
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6826435853364486e-05,
+      "loss": 0.8921,
+      "step": 10305
+    },
+    {
+      "epoch": 0.2830463321523715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6825785811209877e-05,
+      "loss": 1.0157,
+      "step": 10306
+    },
+    {
+      "epoch": 0.2830737963802148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6825135715047008e-05,
+      "loss": 0.9977,
+      "step": 10307
+    },
+    {
+      "epoch": 0.283101260608058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682448556488102e-05,
+      "loss": 0.9514,
+      "step": 10308
+    },
+    {
+      "epoch": 0.28312872483590124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682383536071706e-05,
+      "loss": 0.9666,
+      "step": 10309
+    },
+    {
+      "epoch": 0.28315618906374446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6823185102560275e-05,
+      "loss": 0.9132,
+      "step": 10310
+    },
+    {
+      "epoch": 0.28318365329158773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68225347904158e-05,
+      "loss": 1.0695,
+      "step": 10311
+    },
+    {
+      "epoch": 0.28321111751943095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682188442428879e-05,
+      "loss": 1.0122,
+      "step": 10312
+    },
+    {
+      "epoch": 0.2832385817472742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6821234004184387e-05,
+      "loss": 0.9618,
+      "step": 10313
+    },
+    {
+      "epoch": 0.2832660459751174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682058353010774e-05,
+      "loss": 1.0842,
+      "step": 10314
+    },
+    {
+      "epoch": 0.2832935102029606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6819933002063994e-05,
+      "loss": 1.0149,
+      "step": 10315
+    },
+    {
+      "epoch": 0.2833209744308039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6819282420058293e-05,
+      "loss": 1.0123,
+      "step": 10316
+    },
+    {
+      "epoch": 0.2833484386586471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681863178409579e-05,
+      "loss": 0.96,
+      "step": 10317
+    },
+    {
+      "epoch": 0.28337590288649034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6817981094181633e-05,
+      "loss": 0.9407,
+      "step": 10318
+    },
+    {
+      "epoch": 0.28340336711433356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681733035032097e-05,
+      "loss": 0.8715,
+      "step": 10319
+    },
+    {
+      "epoch": 0.28343083134217684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681667955251894e-05,
+      "loss": 0.9705,
+      "step": 10320
+    },
+    {
+      "epoch": 0.28345829557002006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6816028700780708e-05,
+      "loss": 1.0216,
+      "step": 10321
+    },
+    {
+      "epoch": 0.2834857597978633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6815377795111415e-05,
+      "loss": 0.9729,
+      "step": 10322
+    },
+    {
+      "epoch": 0.2835132240257065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6814726835516214e-05,
+      "loss": 0.9028,
+      "step": 10323
+    },
+    {
+      "epoch": 0.2835406882535498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6814075822000254e-05,
+      "loss": 1.0421,
+      "step": 10324
+    },
+    {
+      "epoch": 0.283568152481393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6813424754568686e-05,
+      "loss": 0.8208,
+      "step": 10325
+    },
+    {
+      "epoch": 0.2835956167092362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681277363322666e-05,
+      "loss": 0.9584,
+      "step": 10326
+    },
+    {
+      "epoch": 0.28362308093707944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6812122457979332e-05,
+      "loss": 0.9216,
+      "step": 10327
+    },
+    {
+      "epoch": 0.28365054516492266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6811471228831853e-05,
+      "loss": 0.8633,
+      "step": 10328
+    },
+    {
+      "epoch": 0.28367800939276594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6810819945789373e-05,
+      "loss": 0.981,
+      "step": 10329
+    },
+    {
+      "epoch": 0.28370547362060916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681016860885705e-05,
+      "loss": 0.9383,
+      "step": 10330
+    },
+    {
+      "epoch": 0.2837329378484524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6809517218040032e-05,
+      "loss": 0.9254,
+      "step": 10331
+    },
+    {
+      "epoch": 0.2837604020762956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680886577334348e-05,
+      "loss": 0.885,
+      "step": 10332
+    },
+    {
+      "epoch": 0.2837878663041389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680821427477254e-05,
+      "loss": 0.8731,
+      "step": 10333
+    },
+    {
+      "epoch": 0.2838153305319821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6807562722332375e-05,
+      "loss": 0.9398,
+      "step": 10334
+    },
+    {
+      "epoch": 0.2838427947598253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680691111602814e-05,
+      "loss": 0.9254,
+      "step": 10335
+    },
+    {
+      "epoch": 0.28387025898766854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680625945586498e-05,
+      "loss": 0.9713,
+      "step": 10336
+    },
+    {
+      "epoch": 0.2838977232155118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680560774184806e-05,
+      "loss": 0.9354,
+      "step": 10337
+    },
+    {
+      "epoch": 0.28392518744335504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6804955973982538e-05,
+      "loss": 0.9874,
+      "step": 10338
+    },
+    {
+      "epoch": 0.28395265167119826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6804304152273565e-05,
+      "loss": 0.9905,
+      "step": 10339
+    },
+    {
+      "epoch": 0.2839801158990415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6803652276726306e-05,
+      "loss": 0.9538,
+      "step": 10340
+    },
+    {
+      "epoch": 0.2840075801268847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6803000347345913e-05,
+      "loss": 0.9861,
+      "step": 10341
+    },
+    {
+      "epoch": 0.284035044354728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6802348364137544e-05,
+      "loss": 0.9939,
+      "step": 10342
+    },
+    {
+      "epoch": 0.2840625085825712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680169632710636e-05,
+      "loss": 0.936,
+      "step": 10343
+    },
+    {
+      "epoch": 0.2840899728104144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6801044236257524e-05,
+      "loss": 0.9253,
+      "step": 10344
+    },
+    {
+      "epoch": 0.28411743703825765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680039209159619e-05,
+      "loss": 0.9595,
+      "step": 10345
+    },
+    {
+      "epoch": 0.2841449012661009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679973989312752e-05,
+      "loss": 0.9639,
+      "step": 10346
+    },
+    {
+      "epoch": 0.28417236549394415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6799087640856672e-05,
+      "loss": 0.9464,
+      "step": 10347
+    },
+    {
+      "epoch": 0.28419982972178737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6798435334788808e-05,
+      "loss": 0.936,
+      "step": 10348
+    },
+    {
+      "epoch": 0.2842272939496306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6797782974929092e-05,
+      "loss": 0.9122,
+      "step": 10349
+    },
+    {
+      "epoch": 0.28425475817747387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679713056128268e-05,
+      "loss": 0.91,
+      "step": 10350
+    },
+    {
+      "epoch": 0.2842822224053171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6796478093854744e-05,
+      "loss": 0.9372,
+      "step": 10351
+    },
+    {
+      "epoch": 0.2843096866331603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679582557265044e-05,
+      "loss": 0.9741,
+      "step": 10352
+    },
+    {
+      "epoch": 0.28433715086100353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679517299767493e-05,
+      "loss": 0.9036,
+      "step": 10353
+    },
+    {
+      "epoch": 0.28436461508884675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6794520368933377e-05,
+      "loss": 0.9304,
+      "step": 10354
+    },
+    {
+      "epoch": 0.28439207931669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679386768643095e-05,
+      "loss": 0.9377,
+      "step": 10355
+    },
+    {
+      "epoch": 0.28441954354453325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6793214950172807e-05,
+      "loss": 1.0574,
+      "step": 10356
+    },
+    {
+      "epoch": 0.28444700777237647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6792562160164117e-05,
+      "loss": 0.9493,
+      "step": 10357
+    },
+    {
+      "epoch": 0.2844744720002197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6791909316410047e-05,
+      "loss": 1.0069,
+      "step": 10358
+    },
+    {
+      "epoch": 0.28450193622806297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6791256418915756e-05,
+      "loss": 0.9608,
+      "step": 10359
+    },
+    {
+      "epoch": 0.2845294004559062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6790603467686416e-05,
+      "loss": 1.0276,
+      "step": 10360
+    },
+    {
+      "epoch": 0.2845568646837494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678995046272719e-05,
+      "loss": 1.0688,
+      "step": 10361
+    },
+    {
+      "epoch": 0.28458432891159263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6789297404043244e-05,
+      "loss": 1.0155,
+      "step": 10362
+    },
+    {
+      "epoch": 0.2846117931394359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6788644291639747e-05,
+      "loss": 1.0101,
+      "step": 10363
+    },
+    {
+      "epoch": 0.28463925736727913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678799112552187e-05,
+      "loss": 0.9727,
+      "step": 10364
+    },
+    {
+      "epoch": 0.28466672159512235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6787337905694775e-05,
+      "loss": 1.0416,
+      "step": 10365
+    },
+    {
+      "epoch": 0.2846941858229656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6786684632163636e-05,
+      "loss": 0.9296,
+      "step": 10366
+    },
+    {
+      "epoch": 0.2847216500508088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6786031304933616e-05,
+      "loss": 0.9195,
+      "step": 10367
+    },
+    {
+      "epoch": 0.28474911427865207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678537792400989e-05,
+      "loss": 0.9981,
+      "step": 10368
+    },
+    {
+      "epoch": 0.2847765785064953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6784724489397623e-05,
+      "loss": 1.0163,
+      "step": 10369
+    },
+    {
+      "epoch": 0.2848040427343385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678407100110199e-05,
+      "loss": 1.071,
+      "step": 10370
+    },
+    {
+      "epoch": 0.28483150696218174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678341745912816e-05,
+      "loss": 0.9854,
+      "step": 10371
+    },
+    {
+      "epoch": 0.284858971190025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67827638634813e-05,
+      "loss": 1.0237,
+      "step": 10372
+    },
+    {
+      "epoch": 0.28488643541786823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678211021416659e-05,
+      "loss": 0.9753,
+      "step": 10373
+    },
+    {
+      "epoch": 0.28491389964571145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6781456511189193e-05,
+      "loss": 1.0166,
+      "step": 10374
+    },
+    {
+      "epoch": 0.2849413638735547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6780802754554286e-05,
+      "loss": 0.9722,
+      "step": 10375
+    },
+    {
+      "epoch": 0.28496882810139795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678014894426704e-05,
+      "loss": 0.966,
+      "step": 10376
+    },
+    {
+      "epoch": 0.2849962923292412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677949508033263e-05,
+      "loss": 1.0049,
+      "step": 10377
+    },
+    {
+      "epoch": 0.2850237565570844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677884116275623e-05,
+      "loss": 1.1042,
+      "step": 10378
+    },
+    {
+      "epoch": 0.2850512207849276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6778187191543015e-05,
+      "loss": 0.9339,
+      "step": 10379
+    },
+    {
+      "epoch": 0.28507868501277084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677753316669815e-05,
+      "loss": 1.0607,
+      "step": 10380
+    },
+    {
+      "epoch": 0.2851061492406141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6776879088226824e-05,
+      "loss": 0.9708,
+      "step": 10381
+    },
+    {
+      "epoch": 0.28513361346845734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6776224956134202e-05,
+      "loss": 1.0573,
+      "step": 10382
+    },
+    {
+      "epoch": 0.28516107769630056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6775570770425465e-05,
+      "loss": 0.9983,
+      "step": 10383
+    },
+    {
+      "epoch": 0.2851885419241438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6774916531105783e-05,
+      "loss": 0.9777,
+      "step": 10384
+    },
+    {
+      "epoch": 0.28521600615198706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6774262238180344e-05,
+      "loss": 0.9025,
+      "step": 10385
+    },
+    {
+      "epoch": 0.2852434703798303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6773607891654315e-05,
+      "loss": 0.9785,
+      "step": 10386
+    },
+    {
+      "epoch": 0.2852709346076735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677295349153288e-05,
+      "loss": 1.0968,
+      "step": 10387
+    },
+    {
+      "epoch": 0.2852983988355167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6772299037821206e-05,
+      "loss": 1.0647,
+      "step": 10388
+    },
+    {
+      "epoch": 0.28532586306336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6771644530524485e-05,
+      "loss": 1.0485,
+      "step": 10389
+    },
+    {
+      "epoch": 0.2853533272912032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6770989969647887e-05,
+      "loss": 0.9412,
+      "step": 10390
+    },
+    {
+      "epoch": 0.28538079151904644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6770335355196592e-05,
+      "loss": 1.0236,
+      "step": 10391
+    },
+    {
+      "epoch": 0.28540825574688966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6769680687175782e-05,
+      "loss": 1.0485,
+      "step": 10392
+    },
+    {
+      "epoch": 0.28543571997473294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6769025965590638e-05,
+      "loss": 0.8591,
+      "step": 10393
+    },
+    {
+      "epoch": 0.28546318420257616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6768371190446337e-05,
+      "loss": 0.9399,
+      "step": 10394
+    },
+    {
+      "epoch": 0.2854906484304194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676771636174806e-05,
+      "loss": 0.897,
+      "step": 10395
+    },
+    {
+      "epoch": 0.2855181126582626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676706147950099e-05,
+      "loss": 0.9471,
+      "step": 10396
+    },
+    {
+      "epoch": 0.2855455768861058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6766406543710312e-05,
+      "loss": 1.0105,
+      "step": 10397
+    },
+    {
+      "epoch": 0.2855730411139491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67657515543812e-05,
+      "loss": 0.9221,
+      "step": 10398
+    },
+    {
+      "epoch": 0.2856005053417923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6765096511518842e-05,
+      "loss": 0.8389,
+      "step": 10399
+    },
+    {
+      "epoch": 0.28562796956963554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6764441415128422e-05,
+      "loss": 0.9469,
+      "step": 10400
+    },
+    {
+      "epoch": 0.28565543379747876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676378626521512e-05,
+      "loss": 1.0084,
+      "step": 10401
+    },
+    {
+      "epoch": 0.28568289802532204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6763131061784118e-05,
+      "loss": 1.0092,
+      "step": 10402
+    },
+    {
+      "epoch": 0.28571036225316526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6762475804840606e-05,
+      "loss": 1.019,
+      "step": 10403
+    },
+    {
+      "epoch": 0.2857378264810085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6761820494389767e-05,
+      "loss": 1.0541,
+      "step": 10404
+    },
+    {
+      "epoch": 0.2857652907088517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6761165130436784e-05,
+      "loss": 1.0482,
+      "step": 10405
+    },
+    {
+      "epoch": 0.285792754936695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6760509712986844e-05,
+      "loss": 1.0009,
+      "step": 10406
+    },
+    {
+      "epoch": 0.2858202191645382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675985424204513e-05,
+      "loss": 0.9887,
+      "step": 10407
+    },
+    {
+      "epoch": 0.2858476833923814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6759198717616835e-05,
+      "loss": 1.0379,
+      "step": 10408
+    },
+    {
+      "epoch": 0.28587514762022465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6758543139707137e-05,
+      "loss": 1.0318,
+      "step": 10409
+    },
+    {
+      "epoch": 0.28590261184806787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6757887508321228e-05,
+      "loss": 0.9896,
+      "step": 10410
+    },
+    {
+      "epoch": 0.28593007607591114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6757231823464297e-05,
+      "loss": 1.0254,
+      "step": 10411
+    },
+    {
+      "epoch": 0.28595754030375437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6756576085141532e-05,
+      "loss": 0.8468,
+      "step": 10412
+    },
+    {
+      "epoch": 0.2859850045315976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6755920293358116e-05,
+      "loss": 0.9789,
+      "step": 10413
+    },
+    {
+      "epoch": 0.2860124687594408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6755264448119244e-05,
+      "loss": 0.947,
+      "step": 10414
+    },
+    {
+      "epoch": 0.2860399329872841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6754608549430102e-05,
+      "loss": 0.8251,
+      "step": 10415
+    },
+    {
+      "epoch": 0.2860673972151273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6753952597295882e-05,
+      "loss": 0.9246,
+      "step": 10416
+    },
+    {
+      "epoch": 0.2860948614429705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675329659172177e-05,
+      "loss": 1.0189,
+      "step": 10417
+    },
+    {
+      "epoch": 0.28612232567081375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675264053271296e-05,
+      "loss": 1.0342,
+      "step": 10418
+    },
+    {
+      "epoch": 0.286149789898657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6751984420274646e-05,
+      "loss": 0.88,
+      "step": 10419
+    },
+    {
+      "epoch": 0.28617725412650025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6751328254412014e-05,
+      "loss": 0.8665,
+      "step": 10420
+    },
+    {
+      "epoch": 0.28620471835434347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6750672035130257e-05,
+      "loss": 1.0534,
+      "step": 10421
+    },
+    {
+      "epoch": 0.2862321825821867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6750015762434568e-05,
+      "loss": 1.0972,
+      "step": 10422
+    },
+    {
+      "epoch": 0.2862596468100299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674935943633014e-05,
+      "loss": 0.9811,
+      "step": 10423
+    },
+    {
+      "epoch": 0.2862871110378732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6748703056822164e-05,
+      "loss": 0.9755,
+      "step": 10424
+    },
+    {
+      "epoch": 0.2863145752657164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6748046623915838e-05,
+      "loss": 0.979,
+      "step": 10425
+    },
+    {
+      "epoch": 0.28634203949355963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6747390137616353e-05,
+      "loss": 0.9059,
+      "step": 10426
+    },
+    {
+      "epoch": 0.28636950372140285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6746733597928903e-05,
+      "loss": 1.0153,
+      "step": 10427
+    },
+    {
+      "epoch": 0.28639696794924613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6746077004858682e-05,
+      "loss": 0.9353,
+      "step": 10428
+    },
+    {
+      "epoch": 0.28642443217708935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6745420358410887e-05,
+      "loss": 1.0226,
+      "step": 10429
+    },
+    {
+      "epoch": 0.28645189640493257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674476365859071e-05,
+      "loss": 0.9772,
+      "step": 10430
+    },
+    {
+      "epoch": 0.2864793606327758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6744106905403354e-05,
+      "loss": 0.9898,
+      "step": 10431
+    },
+    {
+      "epoch": 0.28650682486061907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6743450098854013e-05,
+      "loss": 1.0005,
+      "step": 10432
+    },
+    {
+      "epoch": 0.2865342890884623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674279323894788e-05,
+      "loss": 0.9774,
+      "step": 10433
+    },
+    {
+      "epoch": 0.2865617533163055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6742136325690154e-05,
+      "loss": 0.9952,
+      "step": 10434
+    },
+    {
+      "epoch": 0.28658921754414873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6741479359086036e-05,
+      "loss": 0.9613,
+      "step": 10435
+    },
+    {
+      "epoch": 0.28661668177199195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6740822339140724e-05,
+      "loss": 0.851,
+      "step": 10436
+    },
+    {
+      "epoch": 0.28664414599983523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674016526585941e-05,
+      "loss": 1.0123,
+      "step": 10437
+    },
+    {
+      "epoch": 0.28667161022767845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67395081392473e-05,
+      "loss": 0.9146,
+      "step": 10438
+    },
+    {
+      "epoch": 0.2866990744555217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673885095930959e-05,
+      "loss": 0.9714,
+      "step": 10439
+    },
+    {
+      "epoch": 0.2867265386833649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673819372605148e-05,
+      "loss": 1.0509,
+      "step": 10440
+    },
+    {
+      "epoch": 0.2867540029112082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6737536439478175e-05,
+      "loss": 0.8582,
+      "step": 10441
+    },
+    {
+      "epoch": 0.2867814671390514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673687909959487e-05,
+      "loss": 0.9298,
+      "step": 10442
+    },
+    {
+      "epoch": 0.2868089313668946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6736221706406767e-05,
+      "loss": 0.9276,
+      "step": 10443
+    },
+    {
+      "epoch": 0.28683639559473784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6735564259919068e-05,
+      "loss": 1.0092,
+      "step": 10444
+    },
+    {
+      "epoch": 0.2868638598225811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6734906760136977e-05,
+      "loss": 0.9037,
+      "step": 10445
+    },
+    {
+      "epoch": 0.28689132405042433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6734249207065695e-05,
+      "loss": 0.9381,
+      "step": 10446
+    },
+    {
+      "epoch": 0.28691878827826756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6733591600710424e-05,
+      "loss": 1.034,
+      "step": 10447
+    },
+    {
+      "epoch": 0.2869462525061108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6732933941076366e-05,
+      "loss": 0.8783,
+      "step": 10448
+    },
+    {
+      "epoch": 0.286973716733954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6732276228168728e-05,
+      "loss": 1.0219,
+      "step": 10449
+    },
+    {
+      "epoch": 0.2870011809617973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6731618461992715e-05,
+      "loss": 0.9511,
+      "step": 10450
+    },
+    {
+      "epoch": 0.2870286451896405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6730960642553526e-05,
+      "loss": 1.0004,
+      "step": 10451
+    },
+    {
+      "epoch": 0.2870561094174837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673030276985637e-05,
+      "loss": 1.0154,
+      "step": 10452
+    },
+    {
+      "epoch": 0.28708357364532694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672964484390645e-05,
+      "loss": 1.0557,
+      "step": 10453
+    },
+    {
+      "epoch": 0.2871110378731702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6728986864708977e-05,
+      "loss": 1.0539,
+      "step": 10454
+    },
+    {
+      "epoch": 0.28713850210101344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672832883226915e-05,
+      "loss": 1.0163,
+      "step": 10455
+    },
+    {
+      "epoch": 0.28716596632885666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672767074659218e-05,
+      "loss": 0.922,
+      "step": 10456
+    },
+    {
+      "epoch": 0.2871934305566999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6727012607683277e-05,
+      "loss": 0.8757,
+      "step": 10457
+    },
+    {
+      "epoch": 0.28722089478454316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6726354415547636e-05,
+      "loss": 1.047,
+      "step": 10458
+    },
+    {
+      "epoch": 0.2872483590123864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672569617019048e-05,
+      "loss": 1.0189,
+      "step": 10459
+    },
+    {
+      "epoch": 0.2872758232402296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6725037871617006e-05,
+      "loss": 1.0282,
+      "step": 10460
+    },
+    {
+      "epoch": 0.2873032874680728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672437951983243e-05,
+      "loss": 0.9525,
+      "step": 10461
+    },
+    {
+      "epoch": 0.28733075169591604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6723721114841958e-05,
+      "loss": 1.0505,
+      "step": 10462
+    },
+    {
+      "epoch": 0.2873582159237593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6723062656650802e-05,
+      "loss": 0.865,
+      "step": 10463
+    },
+    {
+      "epoch": 0.28738568015160254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6722404145264167e-05,
+      "loss": 0.9392,
+      "step": 10464
+    },
+    {
+      "epoch": 0.28741314437944576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6721745580687268e-05,
+      "loss": 0.951,
+      "step": 10465
+    },
+    {
+      "epoch": 0.287440608607289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6721086962925316e-05,
+      "loss": 0.9255,
+      "step": 10466
+    },
+    {
+      "epoch": 0.28746807283513226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6720428291983516e-05,
+      "loss": 0.8595,
+      "step": 10467
+    },
+    {
+      "epoch": 0.2874955370629755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6719769567867087e-05,
+      "loss": 0.9601,
+      "step": 10468
+    },
+    {
+      "epoch": 0.2875230012908187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671911079058124e-05,
+      "loss": 0.9701,
+      "step": 10469
+    },
+    {
+      "epoch": 0.2875504655186619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671845196013118e-05,
+      "loss": 0.9475,
+      "step": 10470
+    },
+    {
+      "epoch": 0.2875779297465052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671779307652213e-05,
+      "loss": 0.8881,
+      "step": 10471
+    },
+    {
+      "epoch": 0.2876053939743484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67171341397593e-05,
+      "loss": 0.9697,
+      "step": 10472
+    },
+    {
+      "epoch": 0.28763285820219164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67164751498479e-05,
+      "loss": 0.9825,
+      "step": 10473
+    },
+    {
+      "epoch": 0.28766032243003486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6715816106793145e-05,
+      "loss": 1.0193,
+      "step": 10474
+    },
+    {
+      "epoch": 0.2876877866578781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6715157010600256e-05,
+      "loss": 1.0998,
+      "step": 10475
+    },
+    {
+      "epoch": 0.28771525088572136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671449786127444e-05,
+      "loss": 0.9986,
+      "step": 10476
+    },
+    {
+      "epoch": 0.2877427151135646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671383865882092e-05,
+      "loss": 0.9795,
+      "step": 10477
+    },
+    {
+      "epoch": 0.2877701793414078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6713179403244905e-05,
+      "loss": 0.903,
+      "step": 10478
+    },
+    {
+      "epoch": 0.287797643569251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6712520094551616e-05,
+      "loss": 1.0253,
+      "step": 10479
+    },
+    {
+      "epoch": 0.2878251077970943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6711860732746267e-05,
+      "loss": 0.9923,
+      "step": 10480
+    },
+    {
+      "epoch": 0.2878525720249375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6711201317834073e-05,
+      "loss": 0.9717,
+      "step": 10481
+    },
+    {
+      "epoch": 0.28788003625278075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6710541849820256e-05,
+      "loss": 0.9437,
+      "step": 10482
+    },
+    {
+      "epoch": 0.28790750048062397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6709882328710036e-05,
+      "loss": 1.0228,
+      "step": 10483
+    },
+    {
+      "epoch": 0.28793496470846724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6709222754508624e-05,
+      "loss": 1.0182,
+      "step": 10484
+    },
+    {
+      "epoch": 0.28796242893631047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6708563127221245e-05,
+      "loss": 0.9029,
+      "step": 10485
+    },
+    {
+      "epoch": 0.2879898931641537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6707903446853114e-05,
+      "loss": 1.0281,
+      "step": 10486
+    },
+    {
+      "epoch": 0.2880173573919969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670724371340945e-05,
+      "loss": 1.0114,
+      "step": 10487
+    },
+    {
+      "epoch": 0.2880448216198402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670658392689548e-05,
+      "loss": 0.9231,
+      "step": 10488
+    },
+    {
+      "epoch": 0.2880722858476834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670592408731642e-05,
+      "loss": 0.938,
+      "step": 10489
+    },
+    {
+      "epoch": 0.28809975007552663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6705264194677488e-05,
+      "loss": 0.9726,
+      "step": 10490
+    },
+    {
+      "epoch": 0.28812721430336985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6704604248983912e-05,
+      "loss": 1.0442,
+      "step": 10491
+    },
+    {
+      "epoch": 0.28815467853121307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6703944250240907e-05,
+      "loss": 0.9201,
+      "step": 10492
+    },
+    {
+      "epoch": 0.28818214275905635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67032841984537e-05,
+      "loss": 1.0124,
+      "step": 10493
+    },
+    {
+      "epoch": 0.28820960698689957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670262409362751e-05,
+      "loss": 0.8772,
+      "step": 10494
+    },
+    {
+      "epoch": 0.2882370712147428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6701963935767566e-05,
+      "loss": 1.0293,
+      "step": 10495
+    },
+    {
+      "epoch": 0.288264535442586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670130372487908e-05,
+      "loss": 0.9153,
+      "step": 10496
+    },
+    {
+      "epoch": 0.2882919996704293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670064346096729e-05,
+      "loss": 0.9423,
+      "step": 10497
+    },
+    {
+      "epoch": 0.2883194638982725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669998314403741e-05,
+      "loss": 0.9778,
+      "step": 10498
+    },
+    {
+      "epoch": 0.28834692812611573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6699322774094667e-05,
+      "loss": 0.9599,
+      "step": 10499
+    },
+    {
+      "epoch": 0.28837439235395895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6698662351144294e-05,
+      "loss": 0.9101,
+      "step": 10500
+    },
+    {
+      "epoch": 0.28840185658180223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66980018751915e-05,
+      "loss": 1.0247,
+      "step": 10501
+    },
+    {
+      "epoch": 0.28842932080964545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6697341346241528e-05,
+      "loss": 0.9887,
+      "step": 10502
+    },
+    {
+      "epoch": 0.28845678503748867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6696680764299595e-05,
+      "loss": 0.9904,
+      "step": 10503
+    },
+    {
+      "epoch": 0.2884842492653319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6696020129370925e-05,
+      "loss": 1.0283,
+      "step": 10504
+    },
+    {
+      "epoch": 0.2885117134931751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6695359441460755e-05,
+      "loss": 1.0321,
+      "step": 10505
+    },
+    {
+      "epoch": 0.2885391777210184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669469870057431e-05,
+      "loss": 1.0453,
+      "step": 10506
+    },
+    {
+      "epoch": 0.2885666419488616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669403790671681e-05,
+      "loss": 1.0594,
+      "step": 10507
+    },
+    {
+      "epoch": 0.28859410617670483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6693377059893492e-05,
+      "loss": 1.091,
+      "step": 10508
+    },
+    {
+      "epoch": 0.28862157040454806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669271616010958e-05,
+      "loss": 0.9366,
+      "step": 10509
+    },
+    {
+      "epoch": 0.28864903463239133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6692055207370307e-05,
+      "loss": 0.9108,
+      "step": 10510
+    },
+    {
+      "epoch": 0.28867649886023455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66913942016809e-05,
+      "loss": 0.9762,
+      "step": 10511
+    },
+    {
+      "epoch": 0.2887039630880778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6690733143046588e-05,
+      "loss": 0.9852,
+      "step": 10512
+    },
+    {
+      "epoch": 0.288731427315921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669007203147261e-05,
+      "loss": 0.9134,
+      "step": 10513
+    },
+    {
+      "epoch": 0.2887588915437643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6689410866964185e-05,
+      "loss": 0.894,
+      "step": 10514
+    },
+    {
+      "epoch": 0.2887863557716075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6688749649526554e-05,
+      "loss": 0.9891,
+      "step": 10515
+    },
+    {
+      "epoch": 0.2888138199994507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668808837916494e-05,
+      "loss": 1.0047,
+      "step": 10516
+    },
+    {
+      "epoch": 0.28884128422729394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668742705588458e-05,
+      "loss": 0.9974,
+      "step": 10517
+    },
+    {
+      "epoch": 0.28886874845513716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668676567969071e-05,
+      "loss": 0.9605,
+      "step": 10518
+    },
+    {
+      "epoch": 0.28889621268298044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668610425058856e-05,
+      "loss": 0.9766,
+      "step": 10519
+    },
+    {
+      "epoch": 0.28892367691082366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668544276858336e-05,
+      "loss": 0.9473,
+      "step": 10520
+    },
+    {
+      "epoch": 0.2889511411386669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6684781233680352e-05,
+      "loss": 0.9362,
+      "step": 10521
+    },
+    {
+      "epoch": 0.2889786053665101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668411964588476e-05,
+      "loss": 0.9026,
+      "step": 10522
+    },
+    {
+      "epoch": 0.2890060695943534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6683458005201828e-05,
+      "loss": 0.9367,
+      "step": 10523
+    },
+    {
+      "epoch": 0.2890335338221966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668279631163679e-05,
+      "loss": 1.0109,
+      "step": 10524
+    },
+    {
+      "epoch": 0.2890609980500398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668213456519487e-05,
+      "loss": 1.0239,
+      "step": 10525
+    },
+    {
+      "epoch": 0.28908846227788304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668147276588132e-05,
+      "loss": 0.846,
+      "step": 10526
+    },
+    {
+      "epoch": 0.2891159265057263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6680810913701368e-05,
+      "loss": 0.9711,
+      "step": 10527
+    },
+    {
+      "epoch": 0.28914339073356954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668014900866025e-05,
+      "loss": 1.0952,
+      "step": 10528
+    },
+    {
+      "epoch": 0.28917085496141276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667948705076321e-05,
+      "loss": 1.0514,
+      "step": 10529
+    },
+    {
+      "epoch": 0.289198319189256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6678825040015475e-05,
+      "loss": 0.941,
+      "step": 10530
+    },
+    {
+      "epoch": 0.2892257834170992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6678162976422294e-05,
+      "loss": 0.9827,
+      "step": 10531
+    },
+    {
+      "epoch": 0.2892532476449425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6677500859988896e-05,
+      "loss": 1.0217,
+      "step": 10532
+    },
+    {
+      "epoch": 0.2892807118727857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6676838690720528e-05,
+      "loss": 1.0183,
+      "step": 10533
+    },
+    {
+      "epoch": 0.2893081761006289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6676176468622424e-05,
+      "loss": 1.0219,
+      "step": 10534
+    },
+    {
+      "epoch": 0.28933564032847214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6675514193699828e-05,
+      "loss": 1.0276,
+      "step": 10535
+    },
+    {
+      "epoch": 0.2893631045563154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667485186595798e-05,
+      "loss": 0.9244,
+      "step": 10536
+    },
+    {
+      "epoch": 0.28939056878415864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6674189485402114e-05,
+      "loss": 0.9561,
+      "step": 10537
+    },
+    {
+      "epoch": 0.28941803301200186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667352705203748e-05,
+      "loss": 0.9378,
+      "step": 10538
+    },
+    {
+      "epoch": 0.2894454972398451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6672864565869308e-05,
+      "loss": 1.0072,
+      "step": 10539
+    },
+    {
+      "epoch": 0.28947296146768836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6672202026902852e-05,
+      "loss": 1.0346,
+      "step": 10540
+    },
+    {
+      "epoch": 0.2895004256955316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6671539435143344e-05,
+      "loss": 0.9883,
+      "step": 10541
+    },
+    {
+      "epoch": 0.2895278899233748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6670876790596034e-05,
+      "loss": 1.0028,
+      "step": 10542
+    },
+    {
+      "epoch": 0.289555354151218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6670214093266164e-05,
+      "loss": 0.9638,
+      "step": 10543
+    },
+    {
+      "epoch": 0.28958281837906125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666955134315898e-05,
+      "loss": 0.9598,
+      "step": 10544
+    },
+    {
+      "epoch": 0.2896102826069045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6668888540279714e-05,
+      "loss": 0.9106,
+      "step": 10545
+    },
+    {
+      "epoch": 0.28963774683474774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6668225684633627e-05,
+      "loss": 0.9584,
+      "step": 10546
+    },
+    {
+      "epoch": 0.28966521106259097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666756277622595e-05,
+      "loss": 0.995,
+      "step": 10547
+    },
+    {
+      "epoch": 0.2896926752904342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666689981506193e-05,
+      "loss": 0.8808,
+      "step": 10548
+    },
+    {
+      "epoch": 0.28972013951827746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6666236801146826e-05,
+      "loss": 0.9027,
+      "step": 10549
+    },
+    {
+      "epoch": 0.2897476037461207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6665573734485868e-05,
+      "loss": 1.0808,
+      "step": 10550
+    },
+    {
+      "epoch": 0.2897750679739639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666491061508431e-05,
+      "loss": 0.9198,
+      "step": 10551
+    },
+    {
+      "epoch": 0.28980253220180713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6664247442947393e-05,
+      "loss": 1.0711,
+      "step": 10552
+    },
+    {
+      "epoch": 0.2898299964296504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6663584218080376e-05,
+      "loss": 0.9298,
+      "step": 10553
+    },
+    {
+      "epoch": 0.2898574606574936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6662920940488495e-05,
+      "loss": 0.9516,
+      "step": 10554
+    },
+    {
+      "epoch": 0.28988492488533685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6662257610177003e-05,
+      "loss": 1.0129,
+      "step": 10555
+    },
+    {
+      "epoch": 0.28991238911318007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666159422715115e-05,
+      "loss": 0.9034,
+      "step": 10556
+    },
+    {
+      "epoch": 0.2899398533410233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6660930791416185e-05,
+      "loss": 0.9618,
+      "step": 10557
+    },
+    {
+      "epoch": 0.28996731756886657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666026730297735e-05,
+      "loss": 0.8984,
+      "step": 10558
+    },
+    {
+      "epoch": 0.2899947817967098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6659603761839905e-05,
+      "loss": 1.0958,
+      "step": 10559
+    },
+    {
+      "epoch": 0.290022246024553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665894016800909e-05,
+      "loss": 0.9365,
+      "step": 10560
+    },
+    {
+      "epoch": 0.29004971025239623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6658276521490163e-05,
+      "loss": 0.904,
+      "step": 10561
+    },
+    {
+      "epoch": 0.2900771744802395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6657612822288375e-05,
+      "loss": 1.101,
+      "step": 10562
+    },
+    {
+      "epoch": 0.29010463870808273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6656949070408974e-05,
+      "loss": 0.975,
+      "step": 10563
+    },
+    {
+      "epoch": 0.29013210293592595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6656285265857212e-05,
+      "loss": 0.966,
+      "step": 10564
+    },
+    {
+      "epoch": 0.29015956716376917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6655621408638348e-05,
+      "loss": 1.0063,
+      "step": 10565
+    },
+    {
+      "epoch": 0.29018703139161245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6654957498757626e-05,
+      "loss": 0.9717,
+      "step": 10566
+    },
+    {
+      "epoch": 0.29021449561945567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66542935362203e-05,
+      "loss": 0.9094,
+      "step": 10567
+    },
+    {
+      "epoch": 0.2902419598472989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6653629521031628e-05,
+      "loss": 1.0449,
+      "step": 10568
+    },
+    {
+      "epoch": 0.2902694240751421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665296545319686e-05,
+      "loss": 1.0239,
+      "step": 10569
+    },
+    {
+      "epoch": 0.29029688830298533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6652301332721256e-05,
+      "loss": 0.9708,
+      "step": 10570
+    },
+    {
+      "epoch": 0.2903243525308286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6651637159610063e-05,
+      "loss": 0.886,
+      "step": 10571
+    },
+    {
+      "epoch": 0.29035181675867183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6650972933868543e-05,
+      "loss": 0.921,
+      "step": 10572
+    },
+    {
+      "epoch": 0.29037928098651505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6650308655501946e-05,
+      "loss": 0.8923,
+      "step": 10573
+    },
+    {
+      "epoch": 0.2904067452143583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664964432451553e-05,
+      "loss": 0.9423,
+      "step": 10574
+    },
+    {
+      "epoch": 0.29043420944220155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6648979940914557e-05,
+      "loss": 1.0771,
+      "step": 10575
+    },
+    {
+      "epoch": 0.2904616736700448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664831550470428e-05,
+      "loss": 1.0082,
+      "step": 10576
+    },
+    {
+      "epoch": 0.290489137897888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664765101588995e-05,
+      "loss": 0.9415,
+      "step": 10577
+    },
+    {
+      "epoch": 0.2905166021257312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6646986474476835e-05,
+      "loss": 1.0256,
+      "step": 10578
+    },
+    {
+      "epoch": 0.2905440663535745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6646321880470184e-05,
+      "loss": 1.02,
+      "step": 10579
+    },
+    {
+      "epoch": 0.2905715305814177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6645657233875264e-05,
+      "loss": 1.0405,
+      "step": 10580
+    },
+    {
+      "epoch": 0.29059899480926094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6644992534697326e-05,
+      "loss": 0.9678,
+      "step": 10581
+    },
+    {
+      "epoch": 0.29062645903710416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6644327782941633e-05,
+      "loss": 0.9773,
+      "step": 10582
+    },
+    {
+      "epoch": 0.29065392326494743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6643662978613448e-05,
+      "loss": 0.9796,
+      "step": 10583
+    },
+    {
+      "epoch": 0.29068138749279065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6642998121718026e-05,
+      "loss": 0.9461,
+      "step": 10584
+    },
+    {
+      "epoch": 0.2907088517206339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6642333212260627e-05,
+      "loss": 0.8537,
+      "step": 10585
+    },
+    {
+      "epoch": 0.2907363159484771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664166825024652e-05,
+      "loss": 0.856,
+      "step": 10586
+    },
+    {
+      "epoch": 0.2907637801763203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6641003235680958e-05,
+      "loss": 1.0261,
+      "step": 10587
+    },
+    {
+      "epoch": 0.2907912444041636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6640338168569207e-05,
+      "loss": 1.0618,
+      "step": 10588
+    },
+    {
+      "epoch": 0.2908187086320068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6639673048916528e-05,
+      "loss": 0.9526,
+      "step": 10589
+    },
+    {
+      "epoch": 0.29084617285985004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663900787672818e-05,
+      "loss": 1.0115,
+      "step": 10590
+    },
+    {
+      "epoch": 0.29087363708769326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6638342652009437e-05,
+      "loss": 0.9252,
+      "step": 10591
+    },
+    {
+      "epoch": 0.29090110131553654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6637677374765552e-05,
+      "loss": 0.9648,
+      "step": 10592
+    },
+    {
+      "epoch": 0.29092856554337976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663701204500179e-05,
+      "loss": 1.1329,
+      "step": 10593
+    },
+    {
+      "epoch": 0.290956029771223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663634666272342e-05,
+      "loss": 0.9018,
+      "step": 10594
+    },
+    {
+      "epoch": 0.2909834939990662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6635681227935704e-05,
+      "loss": 1.0453,
+      "step": 10595
+    },
+    {
+      "epoch": 0.2910109582269095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663501574064391e-05,
+      "loss": 0.963,
+      "step": 10596
+    },
+    {
+      "epoch": 0.2910384224547527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66343502008533e-05,
+      "loss": 1.0126,
+      "step": 10597
+    },
+    {
+      "epoch": 0.2910658866825959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663368460856914e-05,
+      "loss": 0.9387,
+      "step": 10598
+    },
+    {
+      "epoch": 0.29109335091043914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6633018963796697e-05,
+      "loss": 0.9833,
+      "step": 10599
+    },
+    {
+      "epoch": 0.29112081513828236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6632353266541243e-05,
+      "loss": 0.92,
+      "step": 10600
+    },
+    {
+      "epoch": 0.29114827936612564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663168751680804e-05,
+      "loss": 0.9353,
+      "step": 10601
+    },
+    {
+      "epoch": 0.29117574359396886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663102171460235e-05,
+      "loss": 1.0422,
+      "step": 10602
+    },
+    {
+      "epoch": 0.2912032078218121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6630355859929454e-05,
+      "loss": 0.9356,
+      "step": 10603
+    },
+    {
+      "epoch": 0.2912306720496553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662968995279461e-05,
+      "loss": 1.0568,
+      "step": 10604
+    },
+    {
+      "epoch": 0.2912581362774986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6629023993203094e-05,
+      "loss": 1.0162,
+      "step": 10605
+    },
+    {
+      "epoch": 0.2912856005053418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662835798116017e-05,
+      "loss": 0.7948,
+      "step": 10606
+    },
+    {
+      "epoch": 0.291313064733185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6627691916671113e-05,
+      "loss": 0.9359,
+      "step": 10607
+    },
+    {
+      "epoch": 0.29134052896102824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662702579974119e-05,
+      "loss": 1.0524,
+      "step": 10608
+    },
+    {
+      "epoch": 0.2913679931888715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662635963037567e-05,
+      "loss": 1.026,
+      "step": 10609
+    },
+    {
+      "epoch": 0.29139545741671474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6625693408579827e-05,
+      "loss": 0.9664,
+      "step": 10610
+    },
+    {
+      "epoch": 0.29142292164455796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6625027134358932e-05,
+      "loss": 1.0205,
+      "step": 10611
+    },
+    {
+      "epoch": 0.2914503858724012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6624360807718255e-05,
+      "loss": 1.0278,
+      "step": 10612
+    },
+    {
+      "epoch": 0.2914778501002444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6623694428663068e-05,
+      "loss": 1.016,
+      "step": 10613
+    },
+    {
+      "epoch": 0.2915053143280877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6623027997198645e-05,
+      "loss": 0.9684,
+      "step": 10614
+    },
+    {
+      "epoch": 0.2915327785559309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662236151333026e-05,
+      "loss": 0.9564,
+      "step": 10615
+    },
+    {
+      "epoch": 0.2915602427837741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6621694977063185e-05,
+      "loss": 1.0569,
+      "step": 10616
+    },
+    {
+      "epoch": 0.29158770701161735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6621028388402695e-05,
+      "loss": 0.8771,
+      "step": 10617
+    },
+    {
+      "epoch": 0.2916151712394606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6620361747354064e-05,
+      "loss": 0.8998,
+      "step": 10618
+    },
+    {
+      "epoch": 0.29164263546730385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6619695053922564e-05,
+      "loss": 0.8669,
+      "step": 10619
+    },
+    {
+      "epoch": 0.29167009969514707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6619028308113474e-05,
+      "loss": 0.8352,
+      "step": 10620
+    },
+    {
+      "epoch": 0.2916975639229903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661836150993207e-05,
+      "loss": 0.8817,
+      "step": 10621
+    },
+    {
+      "epoch": 0.29172502815083357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661769465938362e-05,
+      "loss": 1.0089,
+      "step": 10622
+    },
+    {
+      "epoch": 0.2917524923786768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6617027756473413e-05,
+      "loss": 0.9333,
+      "step": 10623
+    },
+    {
+      "epoch": 0.29177995660652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6616360801206716e-05,
+      "loss": 0.9533,
+      "step": 10624
+    },
+    {
+      "epoch": 0.29180742083436323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661569379358881e-05,
+      "loss": 0.9383,
+      "step": 10625
+    },
+    {
+      "epoch": 0.29183488506220645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6615026733624974e-05,
+      "loss": 1.0286,
+      "step": 10626
+    },
+    {
+      "epoch": 0.2918623492900497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6614359621320482e-05,
+      "loss": 1.0117,
+      "step": 10627
+    },
+    {
+      "epoch": 0.29188981351789295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6613692456680613e-05,
+      "loss": 1.0304,
+      "step": 10628
+    },
+    {
+      "epoch": 0.29191727774573617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661302523971065e-05,
+      "loss": 1.024,
+      "step": 10629
+    },
+    {
+      "epoch": 0.2919447419735794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6612357970415872e-05,
+      "loss": 0.9008,
+      "step": 10630
+    },
+    {
+      "epoch": 0.29197220620142267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661169064880155e-05,
+      "loss": 0.9401,
+      "step": 10631
+    },
+    {
+      "epoch": 0.2919996704292659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6611023274872976e-05,
+      "loss": 0.9296,
+      "step": 10632
+    },
+    {
+      "epoch": 0.2920271346571091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661035584863542e-05,
+      "loss": 1.0776,
+      "step": 10633
+    },
+    {
+      "epoch": 0.29205459888495233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6609688370094172e-05,
+      "loss": 0.8881,
+      "step": 10634
+    },
+    {
+      "epoch": 0.2920820631127956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660902083925451e-05,
+      "loss": 0.9018,
+      "step": 10635
+    },
+    {
+      "epoch": 0.29210952734063883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6608353256121713e-05,
+      "loss": 0.8815,
+      "step": 10636
+    },
+    {
+      "epoch": 0.29213699156848205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660768562070106e-05,
+      "loss": 0.9015,
+      "step": 10637
+    },
+    {
+      "epoch": 0.2921644557963253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6607017932997845e-05,
+      "loss": 1.0351,
+      "step": 10638
+    },
+    {
+      "epoch": 0.2921919200241685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6606350193017346e-05,
+      "loss": 0.9222,
+      "step": 10639
+    },
+    {
+      "epoch": 0.29221938425201177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6605682400764843e-05,
+      "loss": 1.0258,
+      "step": 10640
+    },
+    {
+      "epoch": 0.292246848479855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660501455624562e-05,
+      "loss": 0.9043,
+      "step": 10641
+    },
+    {
+      "epoch": 0.2922743127076982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6604346659464967e-05,
+      "loss": 1.0015,
+      "step": 10642
+    },
+    {
+      "epoch": 0.29230177693554144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660367871042816e-05,
+      "loss": 0.9017,
+      "step": 10643
+    },
+    {
+      "epoch": 0.2923292411633847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6603010709140494e-05,
+      "loss": 0.91,
+      "step": 10644
+    },
+    {
+      "epoch": 0.29235670539122793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660234265560725e-05,
+      "loss": 1.013,
+      "step": 10645
+    },
+    {
+      "epoch": 0.29238416961907115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6601674549833707e-05,
+      "loss": 1.0052,
+      "step": 10646
+    },
+    {
+      "epoch": 0.2924116338469144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6601006391825165e-05,
+      "loss": 1.076,
+      "step": 10647
+    },
+    {
+      "epoch": 0.29243909807475765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6600338181586898e-05,
+      "loss": 0.9845,
+      "step": 10648
+    },
+    {
+      "epoch": 0.2924665623026009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65996699191242e-05,
+      "loss": 1.0575,
+      "step": 10649
+    },
+    {
+      "epoch": 0.2924940265304441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6599001604442363e-05,
+      "loss": 0.9022,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2925214907582873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6598333237546663e-05,
+      "loss": 0.9462,
+      "step": 10651
+    },
+    {
+      "epoch": 0.29254895498613054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6597664818442394e-05,
+      "loss": 0.8412,
+      "step": 10652
+    },
+    {
+      "epoch": 0.2925764192139738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6596996347134847e-05,
+      "loss": 0.8523,
+      "step": 10653
+    },
+    {
+      "epoch": 0.29260388344181704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6596327823629308e-05,
+      "loss": 0.9527,
+      "step": 10654
+    },
+    {
+      "epoch": 0.29263134766966026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659565924793107e-05,
+      "loss": 0.9488,
+      "step": 10655
+    },
+    {
+      "epoch": 0.2926588118975035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6594990620045423e-05,
+      "loss": 0.9366,
+      "step": 10656
+    },
+    {
+      "epoch": 0.29268627612534676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659432193997765e-05,
+      "loss": 1.1029,
+      "step": 10657
+    },
+    {
+      "epoch": 0.29271374035319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6593653207733053e-05,
+      "loss": 0.8995,
+      "step": 10658
+    },
+    {
+      "epoch": 0.2927412045810332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6592984423316914e-05,
+      "loss": 0.9547,
+      "step": 10659
+    },
+    {
+      "epoch": 0.2927686688088764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659231558673453e-05,
+      "loss": 1.0251,
+      "step": 10660
+    },
+    {
+      "epoch": 0.2927961330367197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659164669799119e-05,
+      "loss": 1.0042,
+      "step": 10661
+    },
+    {
+      "epoch": 0.2928235972645629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6590977757092188e-05,
+      "loss": 0.9827,
+      "step": 10662
+    },
+    {
+      "epoch": 0.29285106149240614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659030876404282e-05,
+      "loss": 0.8899,
+      "step": 10663
+    },
+    {
+      "epoch": 0.29287852572024936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6589639718848373e-05,
+      "loss": 1.0129,
+      "step": 10664
+    },
+    {
+      "epoch": 0.2929059899480926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6588970621514146e-05,
+      "loss": 1.0704,
+      "step": 10665
+    },
+    {
+      "epoch": 0.29293345417593586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658830147204543e-05,
+      "loss": 0.9577,
+      "step": 10666
+    },
+    {
+      "epoch": 0.2929609184037791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658763227044752e-05,
+      "loss": 1.0362,
+      "step": 10667
+    },
+    {
+      "epoch": 0.2929883826316223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6586963016725714e-05,
+      "loss": 0.9316,
+      "step": 10668
+    },
+    {
+      "epoch": 0.2930158468594655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65862937108853e-05,
+      "loss": 0.8255,
+      "step": 10669
+    },
+    {
+      "epoch": 0.2930433110873088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6585624352931583e-05,
+      "loss": 1.0165,
+      "step": 10670
+    },
+    {
+      "epoch": 0.293070775315152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658495494286986e-05,
+      "loss": 1.032,
+      "step": 10671
+    },
+    {
+      "epoch": 0.29309823954299524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6584285480705413e-05,
+      "loss": 0.9515,
+      "step": 10672
+    },
+    {
+      "epoch": 0.29312570377083846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6583615966443555e-05,
+      "loss": 1.0242,
+      "step": 10673
+    },
+    {
+      "epoch": 0.29315316799868174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6582946400089574e-05,
+      "loss": 1.0205,
+      "step": 10674
+    },
+    {
+      "epoch": 0.29318063222652496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6582276781648773e-05,
+      "loss": 0.9184,
+      "step": 10675
+    },
+    {
+      "epoch": 0.2932080964543682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6581607111126447e-05,
+      "loss": 1.0231,
+      "step": 10676
+    },
+    {
+      "epoch": 0.2932355606822114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6580937388527898e-05,
+      "loss": 0.9706,
+      "step": 10677
+    },
+    {
+      "epoch": 0.2932630249100547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6580267613858422e-05,
+      "loss": 0.9097,
+      "step": 10678
+    },
+    {
+      "epoch": 0.2932904891378979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6579597787123318e-05,
+      "loss": 0.9601,
+      "step": 10679
+    },
+    {
+      "epoch": 0.2933179533657411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6578927908327892e-05,
+      "loss": 0.9611,
+      "step": 10680
+    },
+    {
+      "epoch": 0.29334541759358435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6578257977477433e-05,
+      "loss": 0.9862,
+      "step": 10681
+    },
+    {
+      "epoch": 0.29337288182142757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6577587994577254e-05,
+      "loss": 1.0152,
+      "step": 10682
+    },
+    {
+      "epoch": 0.29340034604927084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657691795963265e-05,
+      "loss": 0.9979,
+      "step": 10683
+    },
+    {
+      "epoch": 0.29342781027711407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6576247872648922e-05,
+      "loss": 0.9463,
+      "step": 10684
+    },
+    {
+      "epoch": 0.2934552745049573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6575577733631375e-05,
+      "loss": 1.0237,
+      "step": 10685
+    },
+    {
+      "epoch": 0.2934827387328005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657490754258531e-05,
+      "loss": 0.9526,
+      "step": 10686
+    },
+    {
+      "epoch": 0.2935102029606438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657423729951603e-05,
+      "loss": 1.0859,
+      "step": 10687
+    },
+    {
+      "epoch": 0.293537667188487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6573567004428837e-05,
+      "loss": 0.8809,
+      "step": 10688
+    },
+    {
+      "epoch": 0.2935651314163302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6572896657329037e-05,
+      "loss": 0.9338,
+      "step": 10689
+    },
+    {
+      "epoch": 0.29359259564417345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657222625822193e-05,
+      "loss": 0.9476,
+      "step": 10690
+    },
+    {
+      "epoch": 0.2936200598720167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6571555807112822e-05,
+      "loss": 0.9537,
+      "step": 10691
+    },
+    {
+      "epoch": 0.29364752409985995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6570885304007023e-05,
+      "loss": 1.0869,
+      "step": 10692
+    },
+    {
+      "epoch": 0.29367498832770317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6570214748909833e-05,
+      "loss": 0.8878,
+      "step": 10693
+    },
+    {
+      "epoch": 0.2937024525555464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6569544141826558e-05,
+      "loss": 0.9772,
+      "step": 10694
+    },
+    {
+      "epoch": 0.2937299167833896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6568873482762504e-05,
+      "loss": 1.002,
+      "step": 10695
+    },
+    {
+      "epoch": 0.2937573810112329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6568202771722977e-05,
+      "loss": 0.9359,
+      "step": 10696
+    },
+    {
+      "epoch": 0.2937848452390761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656753200871329e-05,
+      "loss": 0.9765,
+      "step": 10697
+    },
+    {
+      "epoch": 0.29381230946691933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6566861193738745e-05,
+      "loss": 1.0197,
+      "step": 10698
+    },
+    {
+      "epoch": 0.29383977369476255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6566190326804647e-05,
+      "loss": 0.9001,
+      "step": 10699
+    },
+    {
+      "epoch": 0.29386723792260583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656551940791631e-05,
+      "loss": 0.9552,
+      "step": 10700
+    },
+    {
+      "epoch": 0.29389470215044905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656484843707904e-05,
+      "loss": 0.9142,
+      "step": 10701
+    },
+    {
+      "epoch": 0.29392216637829227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6564177414298146e-05,
+      "loss": 1.0253,
+      "step": 10702
+    },
+    {
+      "epoch": 0.2939496306061355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6563506339578935e-05,
+      "loss": 1.026,
+      "step": 10703
+    },
+    {
+      "epoch": 0.29397709483397877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6562835212926725e-05,
+      "loss": 0.9636,
+      "step": 10704
+    },
+    {
+      "epoch": 0.294004559061822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6562164034346817e-05,
+      "loss": 0.9354,
+      "step": 10705
+    },
+    {
+      "epoch": 0.2940320232896652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6561492803844525e-05,
+      "loss": 0.9966,
+      "step": 10706
+    },
+    {
+      "epoch": 0.29405948751750843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656082152142516e-05,
+      "loss": 1.0468,
+      "step": 10707
+    },
+    {
+      "epoch": 0.29408695174535165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6560150187094034e-05,
+      "loss": 0.9717,
+      "step": 10708
+    },
+    {
+      "epoch": 0.29411441597319493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6559478800856464e-05,
+      "loss": 0.9739,
+      "step": 10709
+    },
+    {
+      "epoch": 0.29414188020103815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655880736271775e-05,
+      "loss": 1.1725,
+      "step": 10710
+    },
+    {
+      "epoch": 0.2941693444288814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6558135872683215e-05,
+      "loss": 1.0956,
+      "step": 10711
+    },
+    {
+      "epoch": 0.2941968086567246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6557464330758167e-05,
+      "loss": 0.9538,
+      "step": 10712
+    },
+    {
+      "epoch": 0.2942242728845679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655679273694792e-05,
+      "loss": 1.0363,
+      "step": 10713
+    },
+    {
+      "epoch": 0.2942517371124111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655612109125779e-05,
+      "loss": 0.8907,
+      "step": 10714
+    },
+    {
+      "epoch": 0.2942792013402543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6555449393693092e-05,
+      "loss": 1.0624,
+      "step": 10715
+    },
+    {
+      "epoch": 0.29430666556809754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6554777644259138e-05,
+      "loss": 0.9346,
+      "step": 10716
+    },
+    {
+      "epoch": 0.2943341297959408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6554105842961245e-05,
+      "loss": 0.9771,
+      "step": 10717
+    },
+    {
+      "epoch": 0.29436159402378403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6553433989804725e-05,
+      "loss": 0.8951,
+      "step": 10718
+    },
+    {
+      "epoch": 0.29438905825162726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65527620847949e-05,
+      "loss": 0.9041,
+      "step": 10719
+    },
+    {
+      "epoch": 0.2944165224794705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655209012793708e-05,
+      "loss": 0.915,
+      "step": 10720
+    },
+    {
+      "epoch": 0.2944439867073137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6551418119236585e-05,
+      "loss": 0.8205,
+      "step": 10721
+    },
+    {
+      "epoch": 0.294471450935157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6550746058698736e-05,
+      "loss": 1.0015,
+      "step": 10722
+    },
+    {
+      "epoch": 0.2944989151630002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6550073946328843e-05,
+      "loss": 0.9986,
+      "step": 10723
+    },
+    {
+      "epoch": 0.2945263793908434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654940178213223e-05,
+      "loss": 0.9312,
+      "step": 10724
+    },
+    {
+      "epoch": 0.29455384361868664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654872956611421e-05,
+      "loss": 0.9852,
+      "step": 10725
+    },
+    {
+      "epoch": 0.2945813078465299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6548057298280105e-05,
+      "loss": 0.9808,
+      "step": 10726
+    },
+    {
+      "epoch": 0.29460877207437314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654738497863524e-05,
+      "loss": 0.9854,
+      "step": 10727
+    },
+    {
+      "epoch": 0.29463623630221636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6546712607184924e-05,
+      "loss": 1.0128,
+      "step": 10728
+    },
+    {
+      "epoch": 0.2946637005300596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6546040183934482e-05,
+      "loss": 1.0734,
+      "step": 10729
+    },
+    {
+      "epoch": 0.29469116475790286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6545367708889235e-05,
+      "loss": 1.0901,
+      "step": 10730
+    },
+    {
+      "epoch": 0.2947186289857461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6544695182054503e-05,
+      "loss": 0.9659,
+      "step": 10731
+    },
+    {
+      "epoch": 0.2947460932135893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654402260343561e-05,
+      "loss": 0.9722,
+      "step": 10732
+    },
+    {
+      "epoch": 0.2947735574414325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6543349973037872e-05,
+      "loss": 0.954,
+      "step": 10733
+    },
+    {
+      "epoch": 0.29480102166927574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6542677290866612e-05,
+      "loss": 1.0073,
+      "step": 10734
+    },
+    {
+      "epoch": 0.294828485897119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654200455692716e-05,
+      "loss": 0.9641,
+      "step": 10735
+    },
+    {
+      "epoch": 0.29485595012496224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6541331771224832e-05,
+      "loss": 0.9013,
+      "step": 10736
+    },
+    {
+      "epoch": 0.29488341435280546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6540658933764954e-05,
+      "loss": 0.9894,
+      "step": 10737
+    },
+    {
+      "epoch": 0.2949108785806487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6539986044552848e-05,
+      "loss": 1.0618,
+      "step": 10738
+    },
+    {
+      "epoch": 0.29493834280849196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6539313103593842e-05,
+      "loss": 0.9312,
+      "step": 10739
+    },
+    {
+      "epoch": 0.2949658070363352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6538640110893253e-05,
+      "loss": 0.9716,
+      "step": 10740
+    },
+    {
+      "epoch": 0.2949932712641784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6537967066456412e-05,
+      "loss": 0.973,
+      "step": 10741
+    },
+    {
+      "epoch": 0.2950207354920216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6537293970288644e-05,
+      "loss": 1.0061,
+      "step": 10742
+    },
+    {
+      "epoch": 0.2950481997198649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6536620822395274e-05,
+      "loss": 0.9152,
+      "step": 10743
+    },
+    {
+      "epoch": 0.2950756639477081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653594762278163e-05,
+      "loss": 0.8965,
+      "step": 10744
+    },
+    {
+      "epoch": 0.29510312817555134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653527437145303e-05,
+      "loss": 0.9621,
+      "step": 10745
+    },
+    {
+      "epoch": 0.29513059240339456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6534601068414812e-05,
+      "loss": 0.973,
+      "step": 10746
+    },
+    {
+      "epoch": 0.2951580566312378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65339277136723e-05,
+      "loss": 1.0038,
+      "step": 10747
+    },
+    {
+      "epoch": 0.29518552085908106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653325430723082e-05,
+      "loss": 0.9583,
+      "step": 10748
+    },
+    {
+      "epoch": 0.2952129850869243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65325808490957e-05,
+      "loss": 0.9935,
+      "step": 10749
+    },
+    {
+      "epoch": 0.2952404493147675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653190733927227e-05,
+      "loss": 0.989,
+      "step": 10750
+    },
+    {
+      "epoch": 0.2952679135426107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653123377776586e-05,
+      "loss": 0.9985,
+      "step": 10751
+    },
+    {
+      "epoch": 0.295295377770454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65305601645818e-05,
+      "loss": 0.9039,
+      "step": 10752
+    },
+    {
+      "epoch": 0.2953228419982972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6529886499725414e-05,
+      "loss": 0.9255,
+      "step": 10753
+    },
+    {
+      "epoch": 0.29535030622614045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652921278320204e-05,
+      "loss": 1.0517,
+      "step": 10754
+    },
+    {
+      "epoch": 0.29537777045398367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6528539015017006e-05,
+      "loss": 0.9914,
+      "step": 10755
+    },
+    {
+      "epoch": 0.29540523468182694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6527865195175638e-05,
+      "loss": 0.9552,
+      "step": 10756
+    },
+    {
+      "epoch": 0.29543269890967017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6527191323683275e-05,
+      "loss": 0.9153,
+      "step": 10757
+    },
+    {
+      "epoch": 0.2954601631375134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6526517400545247e-05,
+      "loss": 1.0526,
+      "step": 10758
+    },
+    {
+      "epoch": 0.2954876273653566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6525843425766882e-05,
+      "loss": 1.0511,
+      "step": 10759
+    },
+    {
+      "epoch": 0.29551509159319983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6525169399353518e-05,
+      "loss": 1.0102,
+      "step": 10760
+    },
+    {
+      "epoch": 0.2955425558210431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6524495321310487e-05,
+      "loss": 0.983,
+      "step": 10761
+    },
+    {
+      "epoch": 0.29557002004888633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652382119164312e-05,
+      "loss": 0.939,
+      "step": 10762
+    },
+    {
+      "epoch": 0.29559748427672955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6523147010356753e-05,
+      "loss": 1.0211,
+      "step": 10763
+    },
+    {
+      "epoch": 0.29562494850457277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652247277745672e-05,
+      "loss": 0.8825,
+      "step": 10764
+    },
+    {
+      "epoch": 0.29565241273241605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6521798492948353e-05,
+      "loss": 0.9074,
+      "step": 10765
+    },
+    {
+      "epoch": 0.29567987696025927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6521124156836996e-05,
+      "loss": 1.0413,
+      "step": 10766
+    },
+    {
+      "epoch": 0.2957073411881025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6520449769127976e-05,
+      "loss": 0.9883,
+      "step": 10767
+    },
+    {
+      "epoch": 0.2957348054159457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651977532982663e-05,
+      "loss": 0.9401,
+      "step": 10768
+    },
+    {
+      "epoch": 0.295762269643789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6519100838938296e-05,
+      "loss": 0.9108,
+      "step": 10769
+    },
+    {
+      "epoch": 0.2957897338716322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651842629646831e-05,
+      "loss": 0.9877,
+      "step": 10770
+    },
+    {
+      "epoch": 0.29581719809947543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6517751702422016e-05,
+      "loss": 0.994,
+      "step": 10771
+    },
+    {
+      "epoch": 0.29584466232731865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651707705680474e-05,
+      "loss": 1.0054,
+      "step": 10772
+    },
+    {
+      "epoch": 0.2958721265551619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651640235962183e-05,
+      "loss": 1.0679,
+      "step": 10773
+    },
+    {
+      "epoch": 0.29589959078300515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6515727610878616e-05,
+      "loss": 1.1462,
+      "step": 10774
+    },
+    {
+      "epoch": 0.29592705501084837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6515052810580443e-05,
+      "loss": 0.9882,
+      "step": 10775
+    },
+    {
+      "epoch": 0.2959545192386916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651437795873265e-05,
+      "loss": 1.0009,
+      "step": 10776
+    },
+    {
+      "epoch": 0.2959819834665348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6513703055340575e-05,
+      "loss": 0.9131,
+      "step": 10777
+    },
+    {
+      "epoch": 0.2960094476943781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6513028100409555e-05,
+      "loss": 1.0153,
+      "step": 10778
+    },
+    {
+      "epoch": 0.2960369119222213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6512353093944935e-05,
+      "loss": 0.9649,
+      "step": 10779
+    },
+    {
+      "epoch": 0.29606437615006453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6511678035952056e-05,
+      "loss": 0.9432,
+      "step": 10780
+    },
+    {
+      "epoch": 0.29609184037790776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651100292643626e-05,
+      "loss": 0.9095,
+      "step": 10781
+    },
+    {
+      "epoch": 0.29611930460575103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6510327765402882e-05,
+      "loss": 0.9216,
+      "step": 10782
+    },
+    {
+      "epoch": 0.29614676883359425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6509652552857274e-05,
+      "loss": 0.9515,
+      "step": 10783
+    },
+    {
+      "epoch": 0.2961742330614375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650897728880477e-05,
+      "loss": 1.0201,
+      "step": 10784
+    },
+    {
+      "epoch": 0.2962016972892807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6508301973250714e-05,
+      "loss": 1.0416,
+      "step": 10785
+    },
+    {
+      "epoch": 0.296229161517124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650762660620046e-05,
+      "loss": 1.0059,
+      "step": 10786
+    },
+    {
+      "epoch": 0.2962566257449672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6506951187659335e-05,
+      "loss": 1.0009,
+      "step": 10787
+    },
+    {
+      "epoch": 0.2962840899728104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6506275717632693e-05,
+      "loss": 0.9636,
+      "step": 10788
+    },
+    {
+      "epoch": 0.29631155420065364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650560019612588e-05,
+      "loss": 0.9448,
+      "step": 10789
+    },
+    {
+      "epoch": 0.29633901842849686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6504924623144237e-05,
+      "loss": 1.0063,
+      "step": 10790
+    },
+    {
+      "epoch": 0.29636648265634014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6504248998693114e-05,
+      "loss": 1.0049,
+      "step": 10791
+    },
+    {
+      "epoch": 0.29639394688418336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650357332277785e-05,
+      "loss": 0.9603,
+      "step": 10792
+    },
+    {
+      "epoch": 0.2964214111120266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6502897595403795e-05,
+      "loss": 1.1059,
+      "step": 10793
+    },
+    {
+      "epoch": 0.2964488753398698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6502221816576292e-05,
+      "loss": 0.9366,
+      "step": 10794
+    },
+    {
+      "epoch": 0.2964763395677131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65015459863007e-05,
+      "loss": 0.8848,
+      "step": 10795
+    },
+    {
+      "epoch": 0.2965038037955563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650087010458235e-05,
+      "loss": 1.0372,
+      "step": 10796
+    },
+    {
+      "epoch": 0.2965312680233995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65001941714266e-05,
+      "loss": 1.0292,
+      "step": 10797
+    },
+    {
+      "epoch": 0.29655873225124274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6499518186838795e-05,
+      "loss": 0.9586,
+      "step": 10798
+    },
+    {
+      "epoch": 0.296586196479086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6498842150824283e-05,
+      "loss": 0.9227,
+      "step": 10799
+    },
+    {
+      "epoch": 0.29661366070692924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6498166063388413e-05,
+      "loss": 0.9221,
+      "step": 10800
+    },
+    {
+      "epoch": 0.29664112493477246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6497489924536542e-05,
+      "loss": 0.9494,
+      "step": 10801
+    },
+    {
+      "epoch": 0.2966685891626157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6496813734274014e-05,
+      "loss": 0.8265,
+      "step": 10802
+    },
+    {
+      "epoch": 0.2966960533904589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6496137492606174e-05,
+      "loss": 0.9913,
+      "step": 10803
+    },
+    {
+      "epoch": 0.2967235176183022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6495461199538382e-05,
+      "loss": 0.9313,
+      "step": 10804
+    },
+    {
+      "epoch": 0.2967509818461454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649478485507598e-05,
+      "loss": 0.9088,
+      "step": 10805
+    },
+    {
+      "epoch": 0.2967784460739886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6494108459224327e-05,
+      "loss": 0.918,
+      "step": 10806
+    },
+    {
+      "epoch": 0.29680591030183184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6493432011988777e-05,
+      "loss": 0.8,
+      "step": 10807
+    },
+    {
+      "epoch": 0.2968333745296751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649275551337467e-05,
+      "loss": 1.0187,
+      "step": 10808
+    },
+    {
+      "epoch": 0.29686083875751834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649207896338737e-05,
+      "loss": 1.013,
+      "step": 10809
+    },
+    {
+      "epoch": 0.29688830298536156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6491402362032227e-05,
+      "loss": 0.9642,
+      "step": 10810
+    },
+    {
+      "epoch": 0.2969157672132048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6490725709314595e-05,
+      "loss": 1.1117,
+      "step": 10811
+    },
+    {
+      "epoch": 0.29694323144104806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6490049005239824e-05,
+      "loss": 0.9457,
+      "step": 10812
+    },
+    {
+      "epoch": 0.2969706956688913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648937224981327e-05,
+      "loss": 0.8435,
+      "step": 10813
+    },
+    {
+      "epoch": 0.2969981598967345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648869544304029e-05,
+      "loss": 1.0504,
+      "step": 10814
+    },
+    {
+      "epoch": 0.2970256241245777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648801858492624e-05,
+      "loss": 0.9067,
+      "step": 10815
+    },
+    {
+      "epoch": 0.29705308835242095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6487341675476476e-05,
+      "loss": 0.9893,
+      "step": 10816
+    },
+    {
+      "epoch": 0.2970805525802642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648666471469635e-05,
+      "loss": 0.9901,
+      "step": 10817
+    },
+    {
+      "epoch": 0.29710801680810744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6485987702591216e-05,
+      "loss": 0.9982,
+      "step": 10818
+    },
+    {
+      "epoch": 0.29713548103595067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6485310639166437e-05,
+      "loss": 1.0034,
+      "step": 10819
+    },
+    {
+      "epoch": 0.2971629452637939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648463352442737e-05,
+      "loss": 1.0535,
+      "step": 10820
+    },
+    {
+      "epoch": 0.29719040949163716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648395635837937e-05,
+      "loss": 1.0262,
+      "step": 10821
+    },
+    {
+      "epoch": 0.2972178737194804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6483279141027795e-05,
+      "loss": 1.0181,
+      "step": 10822
+    },
+    {
+      "epoch": 0.2972453379473236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6482601872378006e-05,
+      "loss": 1.014,
+      "step": 10823
+    },
+    {
+      "epoch": 0.29727280217516683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6481924552435357e-05,
+      "loss": 0.8973,
+      "step": 10824
+    },
+    {
+      "epoch": 0.2973002664030101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6481247181205212e-05,
+      "loss": 1.0186,
+      "step": 10825
+    },
+    {
+      "epoch": 0.2973277306308533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6480569758692925e-05,
+      "loss": 0.8477,
+      "step": 10826
+    },
+    {
+      "epoch": 0.29735519485869655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6479892284903865e-05,
+      "loss": 0.9772,
+      "step": 10827
+    },
+    {
+      "epoch": 0.29738265908653977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6479214759843385e-05,
+      "loss": 0.9311,
+      "step": 10828
+    },
+    {
+      "epoch": 0.297410123314383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6478537183516847e-05,
+      "loss": 0.9148,
+      "step": 10829
+    },
+    {
+      "epoch": 0.29743758754222627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6477859555929615e-05,
+      "loss": 1.01,
+      "step": 10830
+    },
+    {
+      "epoch": 0.2974650517700695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647718187708705e-05,
+      "loss": 0.8995,
+      "step": 10831
+    },
+    {
+      "epoch": 0.2974925159979127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6476504146994506e-05,
+      "loss": 1.0124,
+      "step": 10832
+    },
+    {
+      "epoch": 0.29751998022575593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647582636565736e-05,
+      "loss": 1.0132,
+      "step": 10833
+    },
+    {
+      "epoch": 0.2975474444535992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6475148533080963e-05,
+      "loss": 0.9976,
+      "step": 10834
+    },
+    {
+      "epoch": 0.29757490868144243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6474470649270684e-05,
+      "loss": 0.9381,
+      "step": 10835
+    },
+    {
+      "epoch": 0.29760237290928565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6473792714231886e-05,
+      "loss": 1.0543,
+      "step": 10836
+    },
+    {
+      "epoch": 0.29762983713712887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6473114727969932e-05,
+      "loss": 0.9906,
+      "step": 10837
+    },
+    {
+      "epoch": 0.29765730136497215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6472436690490186e-05,
+      "loss": 0.9086,
+      "step": 10838
+    },
+    {
+      "epoch": 0.29768476559281537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6471758601798016e-05,
+      "loss": 0.9584,
+      "step": 10839
+    },
+    {
+      "epoch": 0.2977122298206586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6471080461898783e-05,
+      "loss": 0.9955,
+      "step": 10840
+    },
+    {
+      "epoch": 0.2977396940485018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6470402270797852e-05,
+      "loss": 1.0245,
+      "step": 10841
+    },
+    {
+      "epoch": 0.29776715827634503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6469724028500597e-05,
+      "loss": 1.0585,
+      "step": 10842
+    },
+    {
+      "epoch": 0.2977946225041883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646904573501238e-05,
+      "loss": 0.9664,
+      "step": 10843
+    },
+    {
+      "epoch": 0.29782208673203153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6468367390338562e-05,
+      "loss": 0.9919,
+      "step": 10844
+    },
+    {
+      "epoch": 0.29784955095987475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646768899448452e-05,
+      "loss": 0.9183,
+      "step": 10845
+    },
+    {
+      "epoch": 0.297877015187718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6467010547455613e-05,
+      "loss": 0.9928,
+      "step": 10846
+    },
+    {
+      "epoch": 0.29790447941556125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6466332049257215e-05,
+      "loss": 0.8747,
+      "step": 10847
+    },
+    {
+      "epoch": 0.2979319436434045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6465653499894697e-05,
+      "loss": 0.9895,
+      "step": 10848
+    },
+    {
+      "epoch": 0.2979594078712477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646497489937342e-05,
+      "loss": 0.9914,
+      "step": 10849
+    },
+    {
+      "epoch": 0.2979868720990909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646429624769876e-05,
+      "loss": 0.9436,
+      "step": 10850
+    },
+    {
+      "epoch": 0.2980143363269342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6463617544876084e-05,
+      "loss": 0.9337,
+      "step": 10851
+    },
+    {
+      "epoch": 0.2980418005547774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6462938790910755e-05,
+      "loss": 1.0621,
+      "step": 10852
+    },
+    {
+      "epoch": 0.29806926478262064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6462259985808157e-05,
+      "loss": 0.9487,
+      "step": 10853
+    },
+    {
+      "epoch": 0.29809672901046386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6461581129573655e-05,
+      "loss": 1.0459,
+      "step": 10854
+    },
+    {
+      "epoch": 0.2981241932383071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646090222221262e-05,
+      "loss": 1.0098,
+      "step": 10855
+    },
+    {
+      "epoch": 0.29815165746615035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6460223263730425e-05,
+      "loss": 0.9998,
+      "step": 10856
+    },
+    {
+      "epoch": 0.2981791216939936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6459544254132437e-05,
+      "loss": 0.929,
+      "step": 10857
+    },
+    {
+      "epoch": 0.2982065859218368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6458865193424033e-05,
+      "loss": 0.9267,
+      "step": 10858
+    },
+    {
+      "epoch": 0.29823405014968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6458186081610587e-05,
+      "loss": 0.9049,
+      "step": 10859
+    },
+    {
+      "epoch": 0.2982615143775233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6457506918697473e-05,
+      "loss": 1.0879,
+      "step": 10860
+    },
+    {
+      "epoch": 0.2982889786053665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6456827704690065e-05,
+      "loss": 0.9653,
+      "step": 10861
+    },
+    {
+      "epoch": 0.29831644283320974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645614843959373e-05,
+      "loss": 1.007,
+      "step": 10862
+    },
+    {
+      "epoch": 0.29834390706105296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6455469123413846e-05,
+      "loss": 0.8342,
+      "step": 10863
+    },
+    {
+      "epoch": 0.29837137128889624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6454789756155792e-05,
+      "loss": 0.9617,
+      "step": 10864
+    },
+    {
+      "epoch": 0.29839883551673946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6454110337824945e-05,
+      "loss": 1.0547,
+      "step": 10865
+    },
+    {
+      "epoch": 0.2984262997445827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6453430868426675e-05,
+      "loss": 0.9789,
+      "step": 10866
+    },
+    {
+      "epoch": 0.2984537639724259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6452751347966358e-05,
+      "loss": 1.0651,
+      "step": 10867
+    },
+    {
+      "epoch": 0.2984812282002691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6452071776449372e-05,
+      "loss": 0.9577,
+      "step": 10868
+    },
+    {
+      "epoch": 0.2985086924281124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64513921538811e-05,
+      "loss": 0.9129,
+      "step": 10869
+    },
+    {
+      "epoch": 0.2985361566559556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645071248026691e-05,
+      "loss": 0.9755,
+      "step": 10870
+    },
+    {
+      "epoch": 0.29856362088379884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6450032755612184e-05,
+      "loss": 0.9573,
+      "step": 10871
+    },
+    {
+      "epoch": 0.29859108511164206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64493529799223e-05,
+      "loss": 0.9115,
+      "step": 10872
+    },
+    {
+      "epoch": 0.29861854933948534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.644867315320264e-05,
+      "loss": 0.9966,
+      "step": 10873
+    },
+    {
+      "epoch": 0.29864601356732856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6447993275458577e-05,
+      "loss": 0.9364,
+      "step": 10874
+    },
+    {
+      "epoch": 0.2986734777951718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6447313346695498e-05,
+      "loss": 0.9965,
+      "step": 10875
+    },
+    {
+      "epoch": 0.298700942023015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6446633366918772e-05,
+      "loss": 1.024,
+      "step": 10876
+    },
+    {
+      "epoch": 0.2987284062508583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.644595333613379e-05,
+      "loss": 0.9016,
+      "step": 10877
+    },
+    {
+      "epoch": 0.2987558704787015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6445273254345928e-05,
+      "loss": 0.8581,
+      "step": 10878
+    },
+    {
+      "epoch": 0.2987833347065447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6444593121560567e-05,
+      "loss": 1.0177,
+      "step": 10879
+    },
+    {
+      "epoch": 0.29881079893438794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6443912937783088e-05,
+      "loss": 0.9929,
+      "step": 10880
+    },
+    {
+      "epoch": 0.2988382631622312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6443232703018874e-05,
+      "loss": 1.0395,
+      "step": 10881
+    },
+    {
+      "epoch": 0.29886572739007444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6442552417273305e-05,
+      "loss": 0.9089,
+      "step": 10882
+    },
+    {
+      "epoch": 0.29889319161791766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.644187208055177e-05,
+      "loss": 0.8933,
+      "step": 10883
+    },
+    {
+      "epoch": 0.2989206558457609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6441191692859645e-05,
+      "loss": 0.853,
+      "step": 10884
+    },
+    {
+      "epoch": 0.2989481200736041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6440511254202315e-05,
+      "loss": 0.9717,
+      "step": 10885
+    },
+    {
+      "epoch": 0.2989755843014474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6439830764585166e-05,
+      "loss": 0.9809,
+      "step": 10886
+    },
+    {
+      "epoch": 0.2990030485292906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6439150224013584e-05,
+      "loss": 0.9994,
+      "step": 10887
+    },
+    {
+      "epoch": 0.2990305127571338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643846963249295e-05,
+      "loss": 1.0792,
+      "step": 10888
+    },
+    {
+      "epoch": 0.29905797698497705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643778899002865e-05,
+      "loss": 0.8841,
+      "step": 10889
+    },
+    {
+      "epoch": 0.2990854412128203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6437108296626067e-05,
+      "loss": 0.9585,
+      "step": 10890
+    },
+    {
+      "epoch": 0.29911290544066355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6436427552290592e-05,
+      "loss": 0.9357,
+      "step": 10891
+    },
+    {
+      "epoch": 0.29914036966850677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643574675702761e-05,
+      "loss": 0.9157,
+      "step": 10892
+    },
+    {
+      "epoch": 0.29916783389635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6435065910842506e-05,
+      "loss": 0.9308,
+      "step": 10893
+    },
+    {
+      "epoch": 0.29919529812419327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6434385013740665e-05,
+      "loss": 0.9238,
+      "step": 10894
+    },
+    {
+      "epoch": 0.2992227623520365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6433704065727482e-05,
+      "loss": 0.9011,
+      "step": 10895
+    },
+    {
+      "epoch": 0.2992502265798797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6433023066808338e-05,
+      "loss": 0.8983,
+      "step": 10896
+    },
+    {
+      "epoch": 0.29927769080772293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643234201698862e-05,
+      "loss": 0.9238,
+      "step": 10897
+    },
+    {
+      "epoch": 0.29930515503556615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6431660916273728e-05,
+      "loss": 0.8811,
+      "step": 10898
+    },
+    {
+      "epoch": 0.2993326192634094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643097976466904e-05,
+      "loss": 1.0421,
+      "step": 10899
+    },
+    {
+      "epoch": 0.29936008349125265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6430298562179947e-05,
+      "loss": 0.9735,
+      "step": 10900
+    },
+    {
+      "epoch": 0.29938754771909587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6429617308811844e-05,
+      "loss": 0.9407,
+      "step": 10901
+    },
+    {
+      "epoch": 0.2994150119469391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6428936004570112e-05,
+      "loss": 0.9274,
+      "step": 10902
+    },
+    {
+      "epoch": 0.29944247617478237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6428254649460155e-05,
+      "loss": 0.9294,
+      "step": 10903
+    },
+    {
+      "epoch": 0.2994699404026256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6427573243487357e-05,
+      "loss": 0.9753,
+      "step": 10904
+    },
+    {
+      "epoch": 0.2994974046304688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6426891786657105e-05,
+      "loss": 0.9866,
+      "step": 10905
+    },
+    {
+      "epoch": 0.29952486885831203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6426210278974797e-05,
+      "loss": 0.9199,
+      "step": 10906
+    },
+    {
+      "epoch": 0.2995523330861553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6425528720445825e-05,
+      "loss": 1.0342,
+      "step": 10907
+    },
+    {
+      "epoch": 0.29957979731399853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6424847111075582e-05,
+      "loss": 1.0081,
+      "step": 10908
+    },
+    {
+      "epoch": 0.29960726154184175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6424165450869456e-05,
+      "loss": 0.904,
+      "step": 10909
+    },
+    {
+      "epoch": 0.299634725769685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642348373983285e-05,
+      "loss": 1.0364,
+      "step": 10910
+    },
+    {
+      "epoch": 0.2996621899975282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6422801977971147e-05,
+      "loss": 1.0055,
+      "step": 10911
+    },
+    {
+      "epoch": 0.29968965422537147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642212016528975e-05,
+      "loss": 0.9434,
+      "step": 10912
+    },
+    {
+      "epoch": 0.2997171184532147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6421438301794054e-05,
+      "loss": 1.0152,
+      "step": 10913
+    },
+    {
+      "epoch": 0.2997445826810579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6420756387489442e-05,
+      "loss": 1.0522,
+      "step": 10914
+    },
+    {
+      "epoch": 0.29977204690890114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6420074422381325e-05,
+      "loss": 1.0424,
+      "step": 10915
+    },
+    {
+      "epoch": 0.2997995111367444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641939240647509e-05,
+      "loss": 0.9247,
+      "step": 10916
+    },
+    {
+      "epoch": 0.29982697536458763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6418710339776137e-05,
+      "loss": 1.0825,
+      "step": 10917
+    },
+    {
+      "epoch": 0.29985443959243085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6418028222289856e-05,
+      "loss": 0.8814,
+      "step": 10918
+    },
+    {
+      "epoch": 0.2998819038202741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641734605402166e-05,
+      "loss": 0.9845,
+      "step": 10919
+    },
+    {
+      "epoch": 0.29990936804811735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6416663834976928e-05,
+      "loss": 1.0009,
+      "step": 10920
+    },
+    {
+      "epoch": 0.2999368322759606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6415981565161065e-05,
+      "loss": 0.941,
+      "step": 10921
+    },
+    {
+      "epoch": 0.2999642965038038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6415299244579475e-05,
+      "loss": 0.9813,
+      "step": 10922
+    },
+    {
+      "epoch": 0.299991760731647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6414616873237552e-05,
+      "loss": 0.9773,
+      "step": 10923
+    },
+    {
+      "epoch": 0.30001922495949024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6413934451140693e-05,
+      "loss": 0.8939,
+      "step": 10924
+    },
+    {
+      "epoch": 0.3000466891873335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6413251978294306e-05,
+      "loss": 0.9002,
+      "step": 10925
+    },
+    {
+      "epoch": 0.30007415341517674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641256945470378e-05,
+      "loss": 1.0892,
+      "step": 10926
+    },
+    {
+      "epoch": 0.30010161764301996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641188688037452e-05,
+      "loss": 0.9825,
+      "step": 10927
+    },
+    {
+      "epoch": 0.3001290818708632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641120425531193e-05,
+      "loss": 0.9752,
+      "step": 10928
+    },
+    {
+      "epoch": 0.30015654609870646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6410521579521405e-05,
+      "loss": 0.9444,
+      "step": 10929
+    },
+    {
+      "epoch": 0.3001840103265497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6409838853008352e-05,
+      "loss": 0.8924,
+      "step": 10930
+    },
+    {
+      "epoch": 0.3002114745543929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.640915607577817e-05,
+      "loss": 1.0098,
+      "step": 10931
+    },
+    {
+      "epoch": 0.3002389387822361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6408473247836263e-05,
+      "loss": 0.8178,
+      "step": 10932
+    },
+    {
+      "epoch": 0.3002664030100794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6407790369188036e-05,
+      "loss": 1.0277,
+      "step": 10933
+    },
+    {
+      "epoch": 0.3002938672379226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6407107439838887e-05,
+      "loss": 0.9172,
+      "step": 10934
+    },
+    {
+      "epoch": 0.30032133146576584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6406424459794222e-05,
+      "loss": 0.9295,
+      "step": 10935
+    },
+    {
+      "epoch": 0.30034879569360906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6405741429059444e-05,
+      "loss": 1.0232,
+      "step": 10936
+    },
+    {
+      "epoch": 0.3003762599214523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6405058347639965e-05,
+      "loss": 0.9774,
+      "step": 10937
+    },
+    {
+      "epoch": 0.30040372414929556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6404375215541174e-05,
+      "loss": 0.8051,
+      "step": 10938
+    },
+    {
+      "epoch": 0.3004311883771388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.640369203276849e-05,
+      "loss": 0.8919,
+      "step": 10939
+    },
+    {
+      "epoch": 0.300458652604982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6403008799327317e-05,
+      "loss": 1.0121,
+      "step": 10940
+    },
+    {
+      "epoch": 0.3004861168328252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6402325515223053e-05,
+      "loss": 0.9487,
+      "step": 10941
+    },
+    {
+      "epoch": 0.3005135810606685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6401642180461112e-05,
+      "loss": 0.9633,
+      "step": 10942
+    },
+    {
+      "epoch": 0.3005410452885117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6400958795046896e-05,
+      "loss": 0.8984,
+      "step": 10943
+    },
+    {
+      "epoch": 0.30056850951635494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6400275358985815e-05,
+      "loss": 0.9797,
+      "step": 10944
+    },
+    {
+      "epoch": 0.30059597374419816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6399591872283277e-05,
+      "loss": 0.9311,
+      "step": 10945
+    },
+    {
+      "epoch": 0.30062343797204144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639890833494469e-05,
+      "loss": 0.9541,
+      "step": 10946
+    },
+    {
+      "epoch": 0.30065090219988466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6398224746975463e-05,
+      "loss": 0.8507,
+      "step": 10947
+    },
+    {
+      "epoch": 0.3006783664277279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6397541108381002e-05,
+      "loss": 0.9778,
+      "step": 10948
+    },
+    {
+      "epoch": 0.3007058306555711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6396857419166717e-05,
+      "loss": 0.9805,
+      "step": 10949
+    },
+    {
+      "epoch": 0.3007332948834143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6396173679338015e-05,
+      "loss": 0.9087,
+      "step": 10950
+    },
+    {
+      "epoch": 0.3007607591112576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6395489888900314e-05,
+      "loss": 1.0737,
+      "step": 10951
+    },
+    {
+      "epoch": 0.3007882233391008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639480604785902e-05,
+      "loss": 1.0271,
+      "step": 10952
+    },
+    {
+      "epoch": 0.30081568756694405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639412215621954e-05,
+      "loss": 0.9775,
+      "step": 10953
+    },
+    {
+      "epoch": 0.30084315179478727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639343821398729e-05,
+      "loss": 0.9261,
+      "step": 10954
+    },
+    {
+      "epoch": 0.30087061602263054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639275422116768e-05,
+      "loss": 0.9053,
+      "step": 10955
+    },
+    {
+      "epoch": 0.30089808025047377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6392070177766123e-05,
+      "loss": 0.8962,
+      "step": 10956
+    },
+    {
+      "epoch": 0.300925544478317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639138608378803e-05,
+      "loss": 0.8903,
+      "step": 10957
+    },
+    {
+      "epoch": 0.3009530087061602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6390701939238814e-05,
+      "loss": 1.0474,
+      "step": 10958
+    },
+    {
+      "epoch": 0.3009804729340035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639001774412389e-05,
+      "loss": 0.9149,
+      "step": 10959
+    },
+    {
+      "epoch": 0.3010079371618467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6389333498448673e-05,
+      "loss": 0.8618,
+      "step": 10960
+    },
+    {
+      "epoch": 0.3010354013896899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638864920221857e-05,
+      "loss": 0.9796,
+      "step": 10961
+    },
+    {
+      "epoch": 0.30106286561753315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6387964855438998e-05,
+      "loss": 0.8723,
+      "step": 10962
+    },
+    {
+      "epoch": 0.30109032984537637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6387280458115378e-05,
+      "loss": 1.014,
+      "step": 10963
+    },
+    {
+      "epoch": 0.30111779407321965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638659601025312e-05,
+      "loss": 0.8571,
+      "step": 10964
+    },
+    {
+      "epoch": 0.30114525830106287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6385911511857638e-05,
+      "loss": 1.0165,
+      "step": 10965
+    },
+    {
+      "epoch": 0.3011727225289061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6385226962934354e-05,
+      "loss": 0.9927,
+      "step": 10966
+    },
+    {
+      "epoch": 0.3012001867567493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638454236348868e-05,
+      "loss": 0.9782,
+      "step": 10967
+    },
+    {
+      "epoch": 0.3012276509845926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6383857713526033e-05,
+      "loss": 0.9425,
+      "step": 10968
+    },
+    {
+      "epoch": 0.3012551152124358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638317301305183e-05,
+      "loss": 0.995,
+      "step": 10969
+    },
+    {
+      "epoch": 0.30128257944027903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638248826207149e-05,
+      "loss": 0.9571,
+      "step": 10970
+    },
+    {
+      "epoch": 0.30131004366812225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638180346059043e-05,
+      "loss": 0.8674,
+      "step": 10971
+    },
+    {
+      "epoch": 0.30133750789596553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638111860861407e-05,
+      "loss": 1.0439,
+      "step": 10972
+    },
+    {
+      "epoch": 0.30136497212380875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6380433706147828e-05,
+      "loss": 1.0237,
+      "step": 10973
+    },
+    {
+      "epoch": 0.30139243635165197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6379748753197123e-05,
+      "loss": 0.9533,
+      "step": 10974
+    },
+    {
+      "epoch": 0.3014199005794952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6379063749767374e-05,
+      "loss": 0.9791,
+      "step": 10975
+    },
+    {
+      "epoch": 0.30144736480733847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6378378695864002e-05,
+      "loss": 0.9889,
+      "step": 10976
+    },
+    {
+      "epoch": 0.3014748290351817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6377693591492426e-05,
+      "loss": 0.9334,
+      "step": 10977
+    },
+    {
+      "epoch": 0.3015022932630249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637700843665807e-05,
+      "loss": 0.9115,
+      "step": 10978
+    },
+    {
+      "epoch": 0.30152975749086813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6376323231366352e-05,
+      "loss": 0.9422,
+      "step": 10979
+    },
+    {
+      "epoch": 0.30155722171871135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6375637975622693e-05,
+      "loss": 1.0148,
+      "step": 10980
+    },
+    {
+      "epoch": 0.30158468594655463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637495266943252e-05,
+      "loss": 1.0389,
+      "step": 10981
+    },
+    {
+      "epoch": 0.30161215017439785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6374267312801248e-05,
+      "loss": 0.9037,
+      "step": 10982
+    },
+    {
+      "epoch": 0.3016396144022411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6373581905734302e-05,
+      "loss": 0.983,
+      "step": 10983
+    },
+    {
+      "epoch": 0.3016670786300843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6372896448237116e-05,
+      "loss": 1.0324,
+      "step": 10984
+    },
+    {
+      "epoch": 0.3016945428579276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63722109403151e-05,
+      "loss": 0.8038,
+      "step": 10985
+    },
+    {
+      "epoch": 0.3017220070857708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637152538197368e-05,
+      "loss": 0.9722,
+      "step": 10986
+    },
+    {
+      "epoch": 0.301749471313614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6370839773218284e-05,
+      "loss": 0.9711,
+      "step": 10987
+    },
+    {
+      "epoch": 0.30177693554145724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6370154114054337e-05,
+      "loss": 0.9562,
+      "step": 10988
+    },
+    {
+      "epoch": 0.3018043997693005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6369468404487264e-05,
+      "loss": 0.9746,
+      "step": 10989
+    },
+    {
+      "epoch": 0.30183186399714373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636878264452249e-05,
+      "loss": 0.8424,
+      "step": 10990
+    },
+    {
+      "epoch": 0.30185932822498696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636809683416544e-05,
+      "loss": 1.0276,
+      "step": 10991
+    },
+    {
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6367410973421538e-05,
+      "loss": 1.0562,
+      "step": 10992
+    },
+    {
+      "epoch": 0.3019142566806734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636672506229622e-05,
+      "loss": 0.9854,
+      "step": 10993
+    },
+    {
+      "epoch": 0.3019417209085167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63660391007949e-05,
+      "loss": 0.973,
+      "step": 10994
+    },
+    {
+      "epoch": 0.3019691851363599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636535308892302e-05,
+      "loss": 0.9667,
+      "step": 10995
+    },
+    {
+      "epoch": 0.3019966493642031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6364667026685996e-05,
+      "loss": 0.9615,
+      "step": 10996
+    },
+    {
+      "epoch": 0.30202411359204634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6363980914089262e-05,
+      "loss": 0.9446,
+      "step": 10997
+    },
+    {
+      "epoch": 0.3020515778198896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6363294751138247e-05,
+      "loss": 0.9267,
+      "step": 10998
+    },
+    {
+      "epoch": 0.30207904204773284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636260853783838e-05,
+      "loss": 0.955,
+      "step": 10999
+    },
+    {
+      "epoch": 0.30210650627557606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6361922274195085e-05,
+      "loss": 0.9904,
+      "step": 11000
+    },
+    {
+      "epoch": 0.3021339705034193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63612359602138e-05,
+      "loss": 0.8796,
+      "step": 11001
+    },
+    {
+      "epoch": 0.30216143473126256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636054959589995e-05,
+      "loss": 0.9252,
+      "step": 11002
+    },
+    {
+      "epoch": 0.3021888989591058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6359863181258967e-05,
+      "loss": 0.9462,
+      "step": 11003
+    },
+    {
+      "epoch": 0.302216363186949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6359176716296287e-05,
+      "loss": 0.9503,
+      "step": 11004
+    },
+    {
+      "epoch": 0.3022438274147922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6358490201017336e-05,
+      "loss": 1.0483,
+      "step": 11005
+    },
+    {
+      "epoch": 0.30227129164263544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635780363542755e-05,
+      "loss": 1.1022,
+      "step": 11006
+    },
+    {
+      "epoch": 0.3022987558704787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6357117019532352e-05,
+      "loss": 0.9018,
+      "step": 11007
+    },
+    {
+      "epoch": 0.30232622009832194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6356430353337185e-05,
+      "loss": 0.9788,
+      "step": 11008
+    },
+    {
+      "epoch": 0.30235368432616516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635574363684748e-05,
+      "loss": 0.9751,
+      "step": 11009
+    },
+    {
+      "epoch": 0.3023811485540084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6355056870068667e-05,
+      "loss": 1.047,
+      "step": 11010
+    },
+    {
+      "epoch": 0.30240861278185166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6354370053006185e-05,
+      "loss": 1.004,
+      "step": 11011
+    },
+    {
+      "epoch": 0.3024360770096949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6353683185665464e-05,
+      "loss": 0.9346,
+      "step": 11012
+    },
+    {
+      "epoch": 0.3024635412375381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635299626805194e-05,
+      "loss": 0.9211,
+      "step": 11013
+    },
+    {
+      "epoch": 0.3024910054653813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6352309300171052e-05,
+      "loss": 0.9714,
+      "step": 11014
+    },
+    {
+      "epoch": 0.3025184696932246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6351622282028227e-05,
+      "loss": 1.0414,
+      "step": 11015
+    },
+    {
+      "epoch": 0.3025459339210678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635093521362891e-05,
+      "loss": 1.0319,
+      "step": 11016
+    },
+    {
+      "epoch": 0.30257339814891104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6350248094978532e-05,
+      "loss": 0.9825,
+      "step": 11017
+    },
+    {
+      "epoch": 0.30260086237675426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634956092608253e-05,
+      "loss": 0.9509,
+      "step": 11018
+    },
+    {
+      "epoch": 0.3026283266045975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6348873706946347e-05,
+      "loss": 0.9916,
+      "step": 11019
+    },
+    {
+      "epoch": 0.30265579083244076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6348186437575413e-05,
+      "loss": 0.9504,
+      "step": 11020
+    },
+    {
+      "epoch": 0.302683255060284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634749911797517e-05,
+      "loss": 1.0095,
+      "step": 11021
+    },
+    {
+      "epoch": 0.3027107192881272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6346811748151054e-05,
+      "loss": 0.9957,
+      "step": 11022
+    },
+    {
+      "epoch": 0.3027381835159704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63461243281085e-05,
+      "loss": 0.955,
+      "step": 11023
+    },
+    {
+      "epoch": 0.3027656477438137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634543685785296e-05,
+      "loss": 0.992,
+      "step": 11024
+    },
+    {
+      "epoch": 0.3027931119716569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6344749337389862e-05,
+      "loss": 0.9157,
+      "step": 11025
+    },
+    {
+      "epoch": 0.30282057619950015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634406176672465e-05,
+      "loss": 1.0494,
+      "step": 11026
+    },
+    {
+      "epoch": 0.30284804042734337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6343374145862762e-05,
+      "loss": 1.0045,
+      "step": 11027
+    },
+    {
+      "epoch": 0.30287550465518664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6342686474809645e-05,
+      "loss": 0.9935,
+      "step": 11028
+    },
+    {
+      "epoch": 0.30290296888302987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6341998753570734e-05,
+      "loss": 0.8915,
+      "step": 11029
+    },
+    {
+      "epoch": 0.3029304331108731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634131098215147e-05,
+      "loss": 0.9967,
+      "step": 11030
+    },
+    {
+      "epoch": 0.3029578973387163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6340623160557298e-05,
+      "loss": 0.9271,
+      "step": 11031
+    },
+    {
+      "epoch": 0.30298536156655953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633993528879366e-05,
+      "loss": 0.9577,
+      "step": 11032
+    },
+    {
+      "epoch": 0.3030128257944028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6339247366865998e-05,
+      "loss": 0.9401,
+      "step": 11033
+    },
+    {
+      "epoch": 0.30304029002224603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6338559394779757e-05,
+      "loss": 1.0768,
+      "step": 11034
+    },
+    {
+      "epoch": 0.30306775425008925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6337871372540376e-05,
+      "loss": 0.9714,
+      "step": 11035
+    },
+    {
+      "epoch": 0.30309521847793247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63371833001533e-05,
+      "loss": 0.9955,
+      "step": 11036
+    },
+    {
+      "epoch": 0.30312268270577575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633649517762398e-05,
+      "loss": 0.9876,
+      "step": 11037
+    },
+    {
+      "epoch": 0.30315014693361897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6335807004957853e-05,
+      "loss": 1.0369,
+      "step": 11038
+    },
+    {
+      "epoch": 0.3031776111614622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6335118782160366e-05,
+      "loss": 0.9645,
+      "step": 11039
+    },
+    {
+      "epoch": 0.3032050753893054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6334430509236966e-05,
+      "loss": 1.0768,
+      "step": 11040
+    },
+    {
+      "epoch": 0.3032325396171487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63337421861931e-05,
+      "loss": 0.9998,
+      "step": 11041
+    },
+    {
+      "epoch": 0.3032600038449919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6333053813034206e-05,
+      "loss": 0.8524,
+      "step": 11042
+    },
+    {
+      "epoch": 0.30328746807283513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6332365389765745e-05,
+      "loss": 0.9295,
+      "step": 11043
+    },
+    {
+      "epoch": 0.30331493230067835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633167691639315e-05,
+      "loss": 1.046,
+      "step": 11044
+    },
+    {
+      "epoch": 0.3033423965285216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6330988392921878e-05,
+      "loss": 1.0331,
+      "step": 11045
+    },
+    {
+      "epoch": 0.30336986075636485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6330299819357375e-05,
+      "loss": 0.9429,
+      "step": 11046
+    },
+    {
+      "epoch": 0.30339732498420807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6329611195705086e-05,
+      "loss": 1.0176,
+      "step": 11047
+    },
+    {
+      "epoch": 0.3034247892120513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632892252197046e-05,
+      "loss": 1.0557,
+      "step": 11048
+    },
+    {
+      "epoch": 0.3034522534398945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632823379815895e-05,
+      "loss": 0.8815,
+      "step": 11049
+    },
+    {
+      "epoch": 0.3034797176677378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6327545024276e-05,
+      "loss": 1.0266,
+      "step": 11050
+    },
+    {
+      "epoch": 0.303507181895581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632685620032706e-05,
+      "loss": 0.9301,
+      "step": 11051
+    },
+    {
+      "epoch": 0.30353464612342423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632616732631759e-05,
+      "loss": 1.0384,
+      "step": 11052
+    },
+    {
+      "epoch": 0.30356211035126746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632547840225303e-05,
+      "loss": 0.9701,
+      "step": 11053
+    },
+    {
+      "epoch": 0.30358957457911073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6324789428138834e-05,
+      "loss": 0.8998,
+      "step": 11054
+    },
+    {
+      "epoch": 0.30361703880695395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6324100403980457e-05,
+      "loss": 0.8589,
+      "step": 11055
+    },
+    {
+      "epoch": 0.3036445030347972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6323411329783344e-05,
+      "loss": 0.9321,
+      "step": 11056
+    },
+    {
+      "epoch": 0.3036719672626404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322722205552954e-05,
+      "loss": 1.0972,
+      "step": 11057
+    },
+    {
+      "epoch": 0.3036994314904836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322033031294738e-05,
+      "loss": 1.0071,
+      "step": 11058
+    },
+    {
+      "epoch": 0.3037268957183269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6321343807014144e-05,
+      "loss": 0.9985,
+      "step": 11059
+    },
+    {
+      "epoch": 0.3037543599461701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6320654532716627e-05,
+      "loss": 0.9878,
+      "step": 11060
+    },
+    {
+      "epoch": 0.30378182417401334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631996520840765e-05,
+      "loss": 0.9498,
+      "step": 11061
+    },
+    {
+      "epoch": 0.30380928840185656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6319275834092655e-05,
+      "loss": 1.0889,
+      "step": 11062
+    },
+    {
+      "epoch": 0.30383675262969984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6318586409777103e-05,
+      "loss": 1.0892,
+      "step": 11063
+    },
+    {
+      "epoch": 0.30386421685754306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631789693546645e-05,
+      "loss": 0.9649,
+      "step": 11064
+    },
+    {
+      "epoch": 0.3038916810853863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6317207411166146e-05,
+      "loss": 1.0257,
+      "step": 11065
+    },
+    {
+      "epoch": 0.3039191453132295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6316517836881652e-05,
+      "loss": 1.0841,
+      "step": 11066
+    },
+    {
+      "epoch": 0.3039466095410728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631582821261842e-05,
+      "loss": 0.978,
+      "step": 11067
+    },
+    {
+      "epoch": 0.303974073768916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631513853838191e-05,
+      "loss": 0.9026,
+      "step": 11068
+    },
+    {
+      "epoch": 0.3040015379967592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6314448814177578e-05,
+      "loss": 0.9816,
+      "step": 11069
+    },
+    {
+      "epoch": 0.30402900222460244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6313759040010883e-05,
+      "loss": 1.0033,
+      "step": 11070
+    },
+    {
+      "epoch": 0.3040564664524457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6313069215887275e-05,
+      "loss": 1.0027,
+      "step": 11071
+    },
+    {
+      "epoch": 0.30408393068028894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6312379341812223e-05,
+      "loss": 1.0124,
+      "step": 11072
+    },
+    {
+      "epoch": 0.30411139490813216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631168941779118e-05,
+      "loss": 0.8985,
+      "step": 11073
+    },
+    {
+      "epoch": 0.3041388591359754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6310999443829602e-05,
+      "loss": 0.8772,
+      "step": 11074
+    },
+    {
+      "epoch": 0.3041663233638186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6310309419932955e-05,
+      "loss": 0.9947,
+      "step": 11075
+    },
+    {
+      "epoch": 0.3041937875916619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6309619346106694e-05,
+      "loss": 0.9586,
+      "step": 11076
+    },
+    {
+      "epoch": 0.3042212518195051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630892922235628e-05,
+      "loss": 0.9603,
+      "step": 11077
+    },
+    {
+      "epoch": 0.3042487160473483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6308239048687177e-05,
+      "loss": 0.9812,
+      "step": 11078
+    },
+    {
+      "epoch": 0.30427618027519154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630754882510484e-05,
+      "loss": 1.011,
+      "step": 11079
+    },
+    {
+      "epoch": 0.3043036445030348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6306858551614736e-05,
+      "loss": 1.064,
+      "step": 11080
+    },
+    {
+      "epoch": 0.30433110873087804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6306168228222323e-05,
+      "loss": 1.0045,
+      "step": 11081
+    },
+    {
+      "epoch": 0.30435857295872126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6305477854933066e-05,
+      "loss": 0.9388,
+      "step": 11082
+    },
+    {
+      "epoch": 0.3043860371865645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630478743175242e-05,
+      "loss": 0.8736,
+      "step": 11083
+    },
+    {
+      "epoch": 0.30441350141440776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6304096958685857e-05,
+      "loss": 0.9432,
+      "step": 11084
+    },
+    {
+      "epoch": 0.304440965642251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630340643573884e-05,
+      "loss": 0.9937,
+      "step": 11085
+    },
+    {
+      "epoch": 0.3044684298700942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6302715862916824e-05,
+      "loss": 1.0284,
+      "step": 11086
+    },
+    {
+      "epoch": 0.3044958940979374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6302025240225282e-05,
+      "loss": 0.9383,
+      "step": 11087
+    },
+    {
+      "epoch": 0.30452335832578065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6301334567669676e-05,
+      "loss": 1.0042,
+      "step": 11088
+    },
+    {
+      "epoch": 0.3045508225536239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6300643845255468e-05,
+      "loss": 0.9572,
+      "step": 11089
+    },
+    {
+      "epoch": 0.30457828678146714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6299953072988126e-05,
+      "loss": 1.0677,
+      "step": 11090
+    },
+    {
+      "epoch": 0.30460575100931037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6299262250873115e-05,
+      "loss": 0.9185,
+      "step": 11091
+    },
+    {
+      "epoch": 0.3046332152371536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62985713789159e-05,
+      "loss": 0.9713,
+      "step": 11092
+    },
+    {
+      "epoch": 0.30466067946499686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6297880457121947e-05,
+      "loss": 0.9224,
+      "step": 11093
+    },
+    {
+      "epoch": 0.3046881436928401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6297189485496728e-05,
+      "loss": 0.8913,
+      "step": 11094
+    },
+    {
+      "epoch": 0.3047156079206833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6296498464045706e-05,
+      "loss": 0.9114,
+      "step": 11095
+    },
+    {
+      "epoch": 0.30474307214852653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6295807392774348e-05,
+      "loss": 0.9563,
+      "step": 11096
+    },
+    {
+      "epoch": 0.3047705363763698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629511627168812e-05,
+      "loss": 0.8719,
+      "step": 11097
+    },
+    {
+      "epoch": 0.304798000604213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62944251007925e-05,
+      "loss": 1.0135,
+      "step": 11098
+    },
+    {
+      "epoch": 0.30482546483205625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6293733880092945e-05,
+      "loss": 0.9567,
+      "step": 11099
+    },
+    {
+      "epoch": 0.30485292905989947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629304260959493e-05,
+      "loss": 0.9116,
+      "step": 11100
+    },
+    {
+      "epoch": 0.3048803932877427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6292351289303926e-05,
+      "loss": 1.024,
+      "step": 11101
+    },
+    {
+      "epoch": 0.30490785751558597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6291659919225402e-05,
+      "loss": 1.0328,
+      "step": 11102
+    },
+    {
+      "epoch": 0.3049353217434292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6290968499364828e-05,
+      "loss": 0.9633,
+      "step": 11103
+    },
+    {
+      "epoch": 0.3049627859712724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629027702972767e-05,
+      "loss": 1.0247,
+      "step": 11104
+    },
+    {
+      "epoch": 0.30499025019911563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628958551031941e-05,
+      "loss": 0.9542,
+      "step": 11105
+    },
+    {
+      "epoch": 0.3050177144269589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628889394114551e-05,
+      "loss": 1.0162,
+      "step": 11106
+    },
+    {
+      "epoch": 0.30504517865480213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6288202322211445e-05,
+      "loss": 1.0264,
+      "step": 11107
+    },
+    {
+      "epoch": 0.30507264288264535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6287510653522684e-05,
+      "loss": 0.9136,
+      "step": 11108
+    },
+    {
+      "epoch": 0.30510010711048857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6286818935084706e-05,
+      "loss": 0.8819,
+      "step": 11109
+    },
+    {
+      "epoch": 0.30512757133833185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628612716690298e-05,
+      "loss": 0.8855,
+      "step": 11110
+    },
+    {
+      "epoch": 0.30515503556617507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6285435348982982e-05,
+      "loss": 1.0503,
+      "step": 11111
+    },
+    {
+      "epoch": 0.3051824997940183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6284743481330187e-05,
+      "loss": 0.9194,
+      "step": 11112
+    },
+    {
+      "epoch": 0.3052099640218615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6284051563950065e-05,
+      "loss": 0.9921,
+      "step": 11113
+    },
+    {
+      "epoch": 0.30523742824970473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6283359596848094e-05,
+      "loss": 0.9541,
+      "step": 11114
+    },
+    {
+      "epoch": 0.305264892477548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6282667580029747e-05,
+      "loss": 0.9911,
+      "step": 11115
+    },
+    {
+      "epoch": 0.30529235670539123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62819755135005e-05,
+      "loss": 1.0121,
+      "step": 11116
+    },
+    {
+      "epoch": 0.30531982093323445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628128339726583e-05,
+      "loss": 0.9481,
+      "step": 11117
+    },
+    {
+      "epoch": 0.3053472851610777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628059123133121e-05,
+      "loss": 0.974,
+      "step": 11118
+    },
+    {
+      "epoch": 0.30537474938892095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6279899015702124e-05,
+      "loss": 0.9865,
+      "step": 11119
+    },
+    {
+      "epoch": 0.3054022136167642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627920675038404e-05,
+      "loss": 0.9063,
+      "step": 11120
+    },
+    {
+      "epoch": 0.3054296778446074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627851443538244e-05,
+      "loss": 0.991,
+      "step": 11121
+    },
+    {
+      "epoch": 0.3054571420724506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6277822070702804e-05,
+      "loss": 0.9837,
+      "step": 11122
+    },
+    {
+      "epoch": 0.3054846063002939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627712965635061e-05,
+      "loss": 0.931,
+      "step": 11123
+    },
+    {
+      "epoch": 0.3055120705281371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6276437192331333e-05,
+      "loss": 1.0307,
+      "step": 11124
+    },
+    {
+      "epoch": 0.30553953475598034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6275744678650452e-05,
+      "loss": 0.9266,
+      "step": 11125
+    },
+    {
+      "epoch": 0.30556699898382356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6275052115313447e-05,
+      "loss": 0.9314,
+      "step": 11126
+    },
+    {
+      "epoch": 0.3055944632116668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6274359502325804e-05,
+      "loss": 1.0258,
+      "step": 11127
+    },
+    {
+      "epoch": 0.30562192743951005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6273666839692993e-05,
+      "loss": 0.8947,
+      "step": 11128
+    },
+    {
+      "epoch": 0.3056493916673533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6272974127420505e-05,
+      "loss": 0.9291,
+      "step": 11129
+    },
+    {
+      "epoch": 0.3056768558951965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627228136551381e-05,
+      "loss": 0.9346,
+      "step": 11130
+    },
+    {
+      "epoch": 0.3057043201230397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6271588553978398e-05,
+      "loss": 0.9449,
+      "step": 11131
+    },
+    {
+      "epoch": 0.305731784350883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6270895692819747e-05,
+      "loss": 0.9648,
+      "step": 11132
+    },
+    {
+      "epoch": 0.3057592485787262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6270202782043342e-05,
+      "loss": 0.9687,
+      "step": 11133
+    },
+    {
+      "epoch": 0.30578671280656944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6269509821654663e-05,
+      "loss": 1.0848,
+      "step": 11134
+    },
+    {
+      "epoch": 0.30581417703441266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6268816811659194e-05,
+      "loss": 1.0244,
+      "step": 11135
+    },
+    {
+      "epoch": 0.30584164126225594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6268123752062414e-05,
+      "loss": 0.9483,
+      "step": 11136
+    },
+    {
+      "epoch": 0.30586910549009916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6267430642869816e-05,
+      "loss": 1.0078,
+      "step": 11137
+    },
+    {
+      "epoch": 0.3058965697179424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6266737484086877e-05,
+      "loss": 0.9317,
+      "step": 11138
+    },
+    {
+      "epoch": 0.3059240339457856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6266044275719084e-05,
+      "loss": 0.9714,
+      "step": 11139
+    },
+    {
+      "epoch": 0.3059514981736288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626535101777192e-05,
+      "loss": 1.0049,
+      "step": 11140
+    },
+    {
+      "epoch": 0.3059789624014721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6264657710250875e-05,
+      "loss": 0.9166,
+      "step": 11141
+    },
+    {
+      "epoch": 0.3060064266293153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626396435316143e-05,
+      "loss": 0.9986,
+      "step": 11142
+    },
+    {
+      "epoch": 0.30603389085715854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626327094650907e-05,
+      "loss": 1.0599,
+      "step": 11143
+    },
+    {
+      "epoch": 0.30606135508500176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626257749029929e-05,
+      "loss": 0.8276,
+      "step": 11144
+    },
+    {
+      "epoch": 0.30608881931284504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6261883984537567e-05,
+      "loss": 0.9487,
+      "step": 11145
+    },
+    {
+      "epoch": 0.30611628354068826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626119042922939e-05,
+      "loss": 0.929,
+      "step": 11146
+    },
+    {
+      "epoch": 0.3061437477685315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6260496824380254e-05,
+      "loss": 0.9411,
+      "step": 11147
+    },
+    {
+      "epoch": 0.3061712119963747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6259803169995638e-05,
+      "loss": 0.8775,
+      "step": 11148
+    },
+    {
+      "epoch": 0.306198676224218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6259109466081034e-05,
+      "loss": 0.9771,
+      "step": 11149
+    },
+    {
+      "epoch": 0.3062261404520612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6258415712641935e-05,
+      "loss": 0.9434,
+      "step": 11150
+    },
+    {
+      "epoch": 0.3062536046799044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6257721909683827e-05,
+      "loss": 0.992,
+      "step": 11151
+    },
+    {
+      "epoch": 0.30628106890774764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6257028057212197e-05,
+      "loss": 1.0602,
+      "step": 11152
+    },
+    {
+      "epoch": 0.30630853313559087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6256334155232535e-05,
+      "loss": 0.9254,
+      "step": 11153
+    },
+    {
+      "epoch": 0.30633599736343414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6255640203750337e-05,
+      "loss": 1.0057,
+      "step": 11154
+    },
+    {
+      "epoch": 0.30636346159127736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625494620277109e-05,
+      "loss": 1.033,
+      "step": 11155
+    },
+    {
+      "epoch": 0.3063909258191206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6254252152300285e-05,
+      "loss": 1.0211,
+      "step": 11156
+    },
+    {
+      "epoch": 0.3064183900469638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6253558052343416e-05,
+      "loss": 0.8825,
+      "step": 11157
+    },
+    {
+      "epoch": 0.3064458542748071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625286390290597e-05,
+      "loss": 1.0233,
+      "step": 11158
+    },
+    {
+      "epoch": 0.3064733185026503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6252169703993445e-05,
+      "loss": 0.925,
+      "step": 11159
+    },
+    {
+      "epoch": 0.3065007827304935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625147545561133e-05,
+      "loss": 0.9725,
+      "step": 11160
+    },
+    {
+      "epoch": 0.30652824695833675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6250781157765123e-05,
+      "loss": 0.9496,
+      "step": 11161
+    },
+    {
+      "epoch": 0.30655571118618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625008681046031e-05,
+      "loss": 1.02,
+      "step": 11162
+    },
+    {
+      "epoch": 0.30658317541402325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6249392413702393e-05,
+      "loss": 0.9421,
+      "step": 11163
+    },
+    {
+      "epoch": 0.30661063964186647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6248697967496863e-05,
+      "loss": 0.9771,
+      "step": 11164
+    },
+    {
+      "epoch": 0.3066381038697097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624800347184921e-05,
+      "loss": 0.8312,
+      "step": 11165
+    },
+    {
+      "epoch": 0.3066655680975529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6247308926764932e-05,
+      "loss": 0.9456,
+      "step": 11166
+    },
+    {
+      "epoch": 0.3066930323253962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624661433224953e-05,
+      "loss": 0.9587,
+      "step": 11167
+    },
+    {
+      "epoch": 0.3067204965532394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6245919688308495e-05,
+      "loss": 0.8721,
+      "step": 11168
+    },
+    {
+      "epoch": 0.30674796078108263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6245224994947323e-05,
+      "loss": 0.871,
+      "step": 11169
+    },
+    {
+      "epoch": 0.30677542500892585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6244530252171512e-05,
+      "loss": 0.9487,
+      "step": 11170
+    },
+    {
+      "epoch": 0.3068028892367691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6243835459986562e-05,
+      "loss": 0.9712,
+      "step": 11171
+    },
+    {
+      "epoch": 0.30683035346461235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624314061839796e-05,
+      "loss": 0.9357,
+      "step": 11172
+    },
+    {
+      "epoch": 0.30685781769245557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624244572741122e-05,
+      "loss": 0.9225,
+      "step": 11173
+    },
+    {
+      "epoch": 0.3068852819202988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624175078703182e-05,
+      "loss": 1.057,
+      "step": 11174
+    },
+    {
+      "epoch": 0.30691274614814207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624105579726528e-05,
+      "loss": 1.0618,
+      "step": 11175
+    },
+    {
+      "epoch": 0.3069402103759853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6240360758117085e-05,
+      "loss": 1.063,
+      "step": 11176
+    },
+    {
+      "epoch": 0.3069676746038285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623966566959274e-05,
+      "loss": 0.7973,
+      "step": 11177
+    },
+    {
+      "epoch": 0.30699513883167173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623897053169774e-05,
+      "loss": 0.9596,
+      "step": 11178
+    },
+    {
+      "epoch": 0.307022603059515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623827534443759e-05,
+      "loss": 1.0128,
+      "step": 11179
+    },
+    {
+      "epoch": 0.30705006728735823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623758010781779e-05,
+      "loss": 0.9334,
+      "step": 11180
+    },
+    {
+      "epoch": 0.30707753151520145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623688482184384e-05,
+      "loss": 0.946,
+      "step": 11181
+    },
+    {
+      "epoch": 0.3071049957430447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623618948652124e-05,
+      "loss": 1.0693,
+      "step": 11182
+    },
+    {
+      "epoch": 0.3071324599708879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6235494101855493e-05,
+      "loss": 1.0042,
+      "step": 11183
+    },
+    {
+      "epoch": 0.30715992419873117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62347986678521e-05,
+      "loss": 0.9192,
+      "step": 11184
+    },
+    {
+      "epoch": 0.3071873884265744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623410318451657e-05,
+      "loss": 0.9445,
+      "step": 11185
+    },
+    {
+      "epoch": 0.3072148526544176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62334076518544e-05,
+      "loss": 0.9723,
+      "step": 11186
+    },
+    {
+      "epoch": 0.30724231688226084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6232712069871088e-05,
+      "loss": 1.1676,
+      "step": 11187
+    },
+    {
+      "epoch": 0.3072697811101041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623201643857215e-05,
+      "loss": 0.937,
+      "step": 11188
+    },
+    {
+      "epoch": 0.30729724533794733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6231320757963082e-05,
+      "loss": 0.9519,
+      "step": 11189
+    },
+    {
+      "epoch": 0.30732470956579055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6230625028049386e-05,
+      "loss": 0.9874,
+      "step": 11190
+    },
+    {
+      "epoch": 0.3073521737936338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6229929248836577e-05,
+      "loss": 0.9149,
+      "step": 11191
+    },
+    {
+      "epoch": 0.30737963802147705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622923342033015e-05,
+      "loss": 0.9465,
+      "step": 11192
+    },
+    {
+      "epoch": 0.3074071022493203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622853754253562e-05,
+      "loss": 0.9152,
+      "step": 11193
+    },
+    {
+      "epoch": 0.3074345664771635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6227841615458486e-05,
+      "loss": 1.0093,
+      "step": 11194
+    },
+    {
+      "epoch": 0.3074620307050067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6227145639104257e-05,
+      "loss": 1.0242,
+      "step": 11195
+    },
+    {
+      "epoch": 0.30748949493284994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622644961347844e-05,
+      "loss": 0.9874,
+      "step": 11196
+    },
+    {
+      "epoch": 0.3075169591606932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622575353858654e-05,
+      "loss": 0.9564,
+      "step": 11197
+    },
+    {
+      "epoch": 0.30754442338853644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6225057414434065e-05,
+      "loss": 1.0347,
+      "step": 11198
+    },
+    {
+      "epoch": 0.30757188761637966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622436124102653e-05,
+      "loss": 1.0584,
+      "step": 11199
+    },
+    {
+      "epoch": 0.3075993518442229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6223665018369434e-05,
+      "loss": 0.9605,
+      "step": 11200
+    },
+    {
+      "epoch": 0.30762681607206616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622296874646829e-05,
+      "loss": 1.0058,
+      "step": 11201
+    },
+    {
+      "epoch": 0.3076542802999094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6222272425328608e-05,
+      "loss": 1.0973,
+      "step": 11202
+    },
+    {
+      "epoch": 0.3076817445277526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6221576054955896e-05,
+      "loss": 0.9529,
+      "step": 11203
+    },
+    {
+      "epoch": 0.3077092087555958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622087963535566e-05,
+      "loss": 0.9556,
+      "step": 11204
+    },
+    {
+      "epoch": 0.3077366729834391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622018316653342e-05,
+      "loss": 1.0137,
+      "step": 11205
+    },
+    {
+      "epoch": 0.3077641372112823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6219486648494677e-05,
+      "loss": 0.9704,
+      "step": 11206
+    },
+    {
+      "epoch": 0.30779160143912554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621879008124495e-05,
+      "loss": 0.9651,
+      "step": 11207
+    },
+    {
+      "epoch": 0.30781906566696876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6218093464789747e-05,
+      "loss": 1.0305,
+      "step": 11208
+    },
+    {
+      "epoch": 0.307846529894812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621739679913458e-05,
+      "loss": 0.9593,
+      "step": 11209
+    },
+    {
+      "epoch": 0.30787399412265526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621670008428496e-05,
+      "loss": 1.0107,
+      "step": 11210
+    },
+    {
+      "epoch": 0.3079014583504985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62160033202464e-05,
+      "loss": 1.0406,
+      "step": 11211
+    },
+    {
+      "epoch": 0.3079289225783417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6215306507024412e-05,
+      "loss": 0.9159,
+      "step": 11212
+    },
+    {
+      "epoch": 0.3079563868061849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621460964462451e-05,
+      "loss": 0.9778,
+      "step": 11213
+    },
+    {
+      "epoch": 0.3079838510340282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6213912733052216e-05,
+      "loss": 1.0414,
+      "step": 11214
+    },
+    {
+      "epoch": 0.3080113152618714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6213215772313034e-05,
+      "loss": 0.9893,
+      "step": 11215
+    },
+    {
+      "epoch": 0.30803877948971464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6212518762412482e-05,
+      "loss": 0.9446,
+      "step": 11216
+    },
+    {
+      "epoch": 0.30806624371755786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6211821703356075e-05,
+      "loss": 1.0047,
+      "step": 11217
+    },
+    {
+      "epoch": 0.30809370794540114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621112459514933e-05,
+      "loss": 0.953,
+      "step": 11218
+    },
+    {
+      "epoch": 0.30812117217324436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621042743779776e-05,
+      "loss": 1.0403,
+      "step": 11219
+    },
+    {
+      "epoch": 0.3081486364010876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6209730231306883e-05,
+      "loss": 0.8648,
+      "step": 11220
+    },
+    {
+      "epoch": 0.3081761006289308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6209032975682216e-05,
+      "loss": 1.0354,
+      "step": 11221
+    },
+    {
+      "epoch": 0.308203564856774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6208335670929273e-05,
+      "loss": 1.0133,
+      "step": 11222
+    },
+    {
+      "epoch": 0.3082310290846173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6207638317053574e-05,
+      "loss": 0.8642,
+      "step": 11223
+    },
+    {
+      "epoch": 0.3082584933124605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6206940914060635e-05,
+      "loss": 0.9645,
+      "step": 11224
+    },
+    {
+      "epoch": 0.30828595754030375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6206243461955976e-05,
+      "loss": 1.0967,
+      "step": 11225
+    },
+    {
+      "epoch": 0.30831342176814697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6205545960745114e-05,
+      "loss": 0.9734,
+      "step": 11226
+    },
+    {
+      "epoch": 0.30834088599599024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6204848410433568e-05,
+      "loss": 0.943,
+      "step": 11227
+    },
+    {
+      "epoch": 0.30836835022383347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.620415081102686e-05,
+      "loss": 1.0166,
+      "step": 11228
+    },
+    {
+      "epoch": 0.3083958144516767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6203453162530504e-05,
+      "loss": 0.9231,
+      "step": 11229
+    },
+    {
+      "epoch": 0.3084232786795199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6202755464950026e-05,
+      "loss": 0.9345,
+      "step": 11230
+    },
+    {
+      "epoch": 0.3084507429073632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6202057718290944e-05,
+      "loss": 0.8818,
+      "step": 11231
+    },
+    {
+      "epoch": 0.3084782071352064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.620135992255878e-05,
+      "loss": 0.9802,
+      "step": 11232
+    },
+    {
+      "epoch": 0.3085056713630496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6200662077759056e-05,
+      "loss": 0.9736,
+      "step": 11233
+    },
+    {
+      "epoch": 0.30853313559089285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619996418389729e-05,
+      "loss": 1.1013,
+      "step": 11234
+    },
+    {
+      "epoch": 0.30856059981873607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6199266240979003e-05,
+      "loss": 1.0099,
+      "step": 11235
+    },
+    {
+      "epoch": 0.30858806404657935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619856824900972e-05,
+      "loss": 0.9794,
+      "step": 11236
+    },
+    {
+      "epoch": 0.30861552827442257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6197870207994965e-05,
+      "loss": 0.978,
+      "step": 11237
+    },
+    {
+      "epoch": 0.3086429925022658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619717211794026e-05,
+      "loss": 1.0501,
+      "step": 11238
+    },
+    {
+      "epoch": 0.308670456730109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619647397885113e-05,
+      "loss": 1.069,
+      "step": 11239
+    },
+    {
+      "epoch": 0.3086979209579523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6195775790733096e-05,
+      "loss": 0.9144,
+      "step": 11240
+    },
+    {
+      "epoch": 0.3087253851857955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619507755359169e-05,
+      "loss": 0.9612,
+      "step": 11241
+    },
+    {
+      "epoch": 0.30875284941363873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6194379267432423e-05,
+      "loss": 0.8412,
+      "step": 11242
+    },
+    {
+      "epoch": 0.30878031364148195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619368093226083e-05,
+      "loss": 0.9971,
+      "step": 11243
+    },
+    {
+      "epoch": 0.30880777786932523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6192982548082433e-05,
+      "loss": 0.9915,
+      "step": 11244
+    },
+    {
+      "epoch": 0.30883524209716845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6192284114902764e-05,
+      "loss": 0.9235,
+      "step": 11245
+    },
+    {
+      "epoch": 0.30886270632501167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619158563272734e-05,
+      "loss": 1.0143,
+      "step": 11246
+    },
+    {
+      "epoch": 0.3088901705528549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6190887101561692e-05,
+      "loss": 1.0229,
+      "step": 11247
+    },
+    {
+      "epoch": 0.3089176347806981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6190188521411348e-05,
+      "loss": 0.9534,
+      "step": 11248
+    },
+    {
+      "epoch": 0.3089450990085414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6189489892281835e-05,
+      "loss": 0.9963,
+      "step": 11249
+    },
+    {
+      "epoch": 0.3089725632363846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618879121417868e-05,
+      "loss": 1.0319,
+      "step": 11250
+    },
+    {
+      "epoch": 0.30900002746422783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618809248710741e-05,
+      "loss": 0.9098,
+      "step": 11251
+    },
+    {
+      "epoch": 0.30902749169207105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6187393711073562e-05,
+      "loss": 1.0247,
+      "step": 11252
+    },
+    {
+      "epoch": 0.30905495591991433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6186694886082652e-05,
+      "loss": 0.8934,
+      "step": 11253
+    },
+    {
+      "epoch": 0.30908242014775755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6185996012140215e-05,
+      "loss": 0.8684,
+      "step": 11254
+    },
+    {
+      "epoch": 0.3091098843756008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6185297089251785e-05,
+      "loss": 1.0619,
+      "step": 11255
+    },
+    {
+      "epoch": 0.309137348603444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618459811742289e-05,
+      "loss": 1.0052,
+      "step": 11256
+    },
+    {
+      "epoch": 0.3091648128312873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6183899096659053e-05,
+      "loss": 0.9413,
+      "step": 11257
+    },
+    {
+      "epoch": 0.3091922770591305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6183200026965817e-05,
+      "loss": 0.9901,
+      "step": 11258
+    },
+    {
+      "epoch": 0.3092197412869737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61825009083487e-05,
+      "loss": 0.964,
+      "step": 11259
+    },
+    {
+      "epoch": 0.30924720551481694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6181801740813245e-05,
+      "loss": 0.9784,
+      "step": 11260
+    },
+    {
+      "epoch": 0.30927466974266016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618110252436498e-05,
+      "loss": 0.9224,
+      "step": 11261
+    },
+    {
+      "epoch": 0.30930213397050343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6180403259009438e-05,
+      "loss": 0.9109,
+      "step": 11262
+    },
+    {
+      "epoch": 0.30932959819834666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617970394475215e-05,
+      "loss": 0.9682,
+      "step": 11263
+    },
+    {
+      "epoch": 0.3093570624261899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6179004581598652e-05,
+      "loss": 0.8963,
+      "step": 11264
+    },
+    {
+      "epoch": 0.3093845266540331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6178305169554472e-05,
+      "loss": 0.9094,
+      "step": 11265
+    },
+    {
+      "epoch": 0.3094119908818764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6177605708625154e-05,
+      "loss": 0.9818,
+      "step": 11266
+    },
+    {
+      "epoch": 0.3094394551097196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6176906198816225e-05,
+      "loss": 1.014,
+      "step": 11267
+    },
+    {
+      "epoch": 0.3094669193375628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617620664013322e-05,
+      "loss": 0.9651,
+      "step": 11268
+    },
+    {
+      "epoch": 0.30949438356540604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6175507032581675e-05,
+      "loss": 1.0327,
+      "step": 11269
+    },
+    {
+      "epoch": 0.3095218477932493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617480737616713e-05,
+      "loss": 0.8731,
+      "step": 11270
+    },
+    {
+      "epoch": 0.30954931202109254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6174107670895116e-05,
+      "loss": 1.0365,
+      "step": 11271
+    },
+    {
+      "epoch": 0.30957677624893576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6173407916771168e-05,
+      "loss": 0.9865,
+      "step": 11272
+    },
+    {
+      "epoch": 0.309604240476779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617270811380083e-05,
+      "loss": 0.9632,
+      "step": 11273
+    },
+    {
+      "epoch": 0.30963170470462226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6172008261989627e-05,
+      "loss": 1.0013,
+      "step": 11274
+    },
+    {
+      "epoch": 0.3096591689324655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6171308361343107e-05,
+      "loss": 1.0094,
+      "step": 11275
+    },
+    {
+      "epoch": 0.3096866331603087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617060841186681e-05,
+      "loss": 0.9012,
+      "step": 11276
+    },
+    {
+      "epoch": 0.3097140973881519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6169908413566263e-05,
+      "loss": 0.9692,
+      "step": 11277
+    },
+    {
+      "epoch": 0.30974156161599514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6169208366447008e-05,
+      "loss": 0.9157,
+      "step": 11278
+    },
+    {
+      "epoch": 0.3097690258438384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6168508270514593e-05,
+      "loss": 1.0939,
+      "step": 11279
+    },
+    {
+      "epoch": 0.30979649007168164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6167808125774547e-05,
+      "loss": 0.8304,
+      "step": 11280
+    },
+    {
+      "epoch": 0.30982395429952486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616710793223242e-05,
+      "loss": 0.9298,
+      "step": 11281
+    },
+    {
+      "epoch": 0.3098514185273681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616640768989374e-05,
+      "loss": 0.9036,
+      "step": 11282
+    },
+    {
+      "epoch": 0.30987888275521136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6165707398764056e-05,
+      "loss": 0.9325,
+      "step": 11283
+    },
+    {
+      "epoch": 0.3099063469830546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6165007058848906e-05,
+      "loss": 0.9749,
+      "step": 11284
+    },
+    {
+      "epoch": 0.3099338112108978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616430667015383e-05,
+      "loss": 0.8968,
+      "step": 11285
+    },
+    {
+      "epoch": 0.309961275438741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6163606232684378e-05,
+      "loss": 0.8725,
+      "step": 11286
+    },
+    {
+      "epoch": 0.3099887396665843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6162905746446078e-05,
+      "loss": 1.0339,
+      "step": 11287
+    },
+    {
+      "epoch": 0.3100162038944275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616220521144449e-05,
+      "loss": 0.9681,
+      "step": 11288
+    },
+    {
+      "epoch": 0.31004366812227074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6161504627685138e-05,
+      "loss": 0.9535,
+      "step": 11289
+    },
+    {
+      "epoch": 0.31007113235011396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6160803995173576e-05,
+      "loss": 0.9668,
+      "step": 11290
+    },
+    {
+      "epoch": 0.3100985965779572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6160103313915353e-05,
+      "loss": 0.9961,
+      "step": 11291
+    },
+    {
+      "epoch": 0.31012606080580046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6159402583915998e-05,
+      "loss": 0.9134,
+      "step": 11292
+    },
+    {
+      "epoch": 0.3101535250336437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615870180518107e-05,
+      "loss": 0.9964,
+      "step": 11293
+    },
+    {
+      "epoch": 0.3101809892614869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6158000977716108e-05,
+      "loss": 1.0393,
+      "step": 11294
+    },
+    {
+      "epoch": 0.3102084534893301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6157300101526655e-05,
+      "loss": 0.9575,
+      "step": 11295
+    },
+    {
+      "epoch": 0.3102359177171734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6156599176618256e-05,
+      "loss": 1.0601,
+      "step": 11296
+    },
+    {
+      "epoch": 0.3102633819450166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615589820299646e-05,
+      "loss": 0.9324,
+      "step": 11297
+    },
+    {
+      "epoch": 0.31029084617285985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6155197180666818e-05,
+      "loss": 1.092,
+      "step": 11298
+    },
+    {
+      "epoch": 0.31031831040070307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6154496109634864e-05,
+      "loss": 0.9473,
+      "step": 11299
+    },
+    {
+      "epoch": 0.31034577462854634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615379498990616e-05,
+      "loss": 0.9926,
+      "step": 11300
+    },
+    {
+      "epoch": 0.31037323885638957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6153093821486244e-05,
+      "loss": 0.9496,
+      "step": 11301
+    },
+    {
+      "epoch": 0.3104007030842328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6152392604380664e-05,
+      "loss": 0.9023,
+      "step": 11302
+    },
+    {
+      "epoch": 0.310428167312076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615169133859497e-05,
+      "loss": 0.9868,
+      "step": 11303
+    },
+    {
+      "epoch": 0.31045563153991923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6150990024134714e-05,
+      "loss": 0.9581,
+      "step": 11304
+    },
+    {
+      "epoch": 0.3104830957677625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615028866100544e-05,
+      "loss": 1.0422,
+      "step": 11305
+    },
+    {
+      "epoch": 0.31051055999560573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6149587249212705e-05,
+      "loss": 0.9848,
+      "step": 11306
+    },
+    {
+      "epoch": 0.31053802422344895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6148885788762047e-05,
+      "loss": 0.9772,
+      "step": 11307
+    },
+    {
+      "epoch": 0.31056548845129217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6148184279659026e-05,
+      "loss": 0.9432,
+      "step": 11308
+    },
+    {
+      "epoch": 0.31059295267913545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6147482721909188e-05,
+      "loss": 1.0205,
+      "step": 11309
+    },
+    {
+      "epoch": 0.31062041690697867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6146781115518086e-05,
+      "loss": 0.9111,
+      "step": 11310
+    },
+    {
+      "epoch": 0.3106478811348219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614607946049127e-05,
+      "loss": 0.8451,
+      "step": 11311
+    },
+    {
+      "epoch": 0.3106753453626651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6145377756834296e-05,
+      "loss": 0.9668,
+      "step": 11312
+    },
+    {
+      "epoch": 0.3107028095905084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614467600455271e-05,
+      "loss": 0.9528,
+      "step": 11313
+    },
+    {
+      "epoch": 0.3107302738183516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6143974203652066e-05,
+      "loss": 0.9099,
+      "step": 11314
+    },
+    {
+      "epoch": 0.31075773804619483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614327235413792e-05,
+      "loss": 0.9666,
+      "step": 11315
+    },
+    {
+      "epoch": 0.31078520227403805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614257045601582e-05,
+      "loss": 0.9306,
+      "step": 11316
+    },
+    {
+      "epoch": 0.3108126665018813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614186850929133e-05,
+      "loss": 0.9296,
+      "step": 11317
+    },
+    {
+      "epoch": 0.31084013072972455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614116651396999e-05,
+      "loss": 0.9763,
+      "step": 11318
+    },
+    {
+      "epoch": 0.31086759495756777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6140464470057367e-05,
+      "loss": 1.0078,
+      "step": 11319
+    },
+    {
+      "epoch": 0.310895059185411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613976237755901e-05,
+      "loss": 0.965,
+      "step": 11320
+    },
+    {
+      "epoch": 0.3109225234132542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6139060236480472e-05,
+      "loss": 1.074,
+      "step": 11321
+    },
+    {
+      "epoch": 0.3109499876410975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613835804682731e-05,
+      "loss": 0.9651,
+      "step": 11322
+    },
+    {
+      "epoch": 0.3109774518689407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613765580860509e-05,
+      "loss": 1.0231,
+      "step": 11323
+    },
+    {
+      "epoch": 0.31100491609678393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6136953521819353e-05,
+      "loss": 1.1154,
+      "step": 11324
+    },
+    {
+      "epoch": 0.31103238032462716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613625118647566e-05,
+      "loss": 0.9412,
+      "step": 11325
+    },
+    {
+      "epoch": 0.31105984455247043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6135548802579574e-05,
+      "loss": 0.8434,
+      "step": 11326
+    },
+    {
+      "epoch": 0.31108730878031365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6134846370136653e-05,
+      "loss": 0.9125,
+      "step": 11327
+    },
+    {
+      "epoch": 0.3111147730081569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6134143889152446e-05,
+      "loss": 1.0041,
+      "step": 11328
+    },
+    {
+      "epoch": 0.3111422372360001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6133441359632516e-05,
+      "loss": 0.9488,
+      "step": 11329
+    },
+    {
+      "epoch": 0.3111697014638433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6132738781582423e-05,
+      "loss": 0.9356,
+      "step": 11330
+    },
+    {
+      "epoch": 0.3111971656916866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6132036155007726e-05,
+      "loss": 0.9854,
+      "step": 11331
+    },
+    {
+      "epoch": 0.3112246299195298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6131333479913983e-05,
+      "loss": 0.9906,
+      "step": 11332
+    },
+    {
+      "epoch": 0.31125209414737304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6130630756306757e-05,
+      "loss": 1.0441,
+      "step": 11333
+    },
+    {
+      "epoch": 0.31127955837521626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6129927984191603e-05,
+      "loss": 0.9613,
+      "step": 11334
+    },
+    {
+      "epoch": 0.31130702260305954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6129225163574086e-05,
+      "loss": 0.9498,
+      "step": 11335
+    },
+    {
+      "epoch": 0.31133448683090276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6128522294459762e-05,
+      "loss": 0.9711,
+      "step": 11336
+    },
+    {
+      "epoch": 0.311361951058746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6127819376854198e-05,
+      "loss": 0.9805,
+      "step": 11337
+    },
+    {
+      "epoch": 0.3113894152865892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6127116410762957e-05,
+      "loss": 1.012,
+      "step": 11338
+    },
+    {
+      "epoch": 0.3114168795144325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612641339619159e-05,
+      "loss": 1.036,
+      "step": 11339
+    },
+    {
+      "epoch": 0.3114443437422757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6125710333145672e-05,
+      "loss": 0.9464,
+      "step": 11340
+    },
+    {
+      "epoch": 0.3114718079701189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612500722163076e-05,
+      "loss": 0.9565,
+      "step": 11341
+    },
+    {
+      "epoch": 0.31149927219796214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612430406165242e-05,
+      "loss": 0.994,
+      "step": 11342
+    },
+    {
+      "epoch": 0.31152673642580536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612360085321621e-05,
+      "loss": 1.0615,
+      "step": 11343
+    },
+    {
+      "epoch": 0.31155420065364864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61228975963277e-05,
+      "loss": 1.0486,
+      "step": 11344
+    },
+    {
+      "epoch": 0.31158166488149186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6122194290992453e-05,
+      "loss": 1.0112,
+      "step": 11345
+    },
+    {
+      "epoch": 0.3116091291093351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612149093721603e-05,
+      "loss": 0.9485,
+      "step": 11346
+    },
+    {
+      "epoch": 0.3116365933371783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6120787535004005e-05,
+      "loss": 0.9809,
+      "step": 11347
+    },
+    {
+      "epoch": 0.3116640575650216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6120084084361935e-05,
+      "loss": 0.9197,
+      "step": 11348
+    },
+    {
+      "epoch": 0.3116915217928648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611938058529539e-05,
+      "loss": 1.0106,
+      "step": 11349
+    },
+    {
+      "epoch": 0.311718986020708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6118677037809934e-05,
+      "loss": 0.9198,
+      "step": 11350
+    },
+    {
+      "epoch": 0.31174645024855124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6117973441911136e-05,
+      "loss": 0.9737,
+      "step": 11351
+    },
+    {
+      "epoch": 0.3117739144763945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611726979760456e-05,
+      "loss": 0.9196,
+      "step": 11352
+    },
+    {
+      "epoch": 0.31180137870423774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6116566104895782e-05,
+      "loss": 0.9594,
+      "step": 11353
+    },
+    {
+      "epoch": 0.31182884293208096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6115862363790356e-05,
+      "loss": 1.0406,
+      "step": 11354
+    },
+    {
+      "epoch": 0.3118563071599242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611515857429386e-05,
+      "loss": 1.0287,
+      "step": 11355
+    },
+    {
+      "epoch": 0.3118837713877674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6114454736411868e-05,
+      "loss": 0.9529,
+      "step": 11356
+    },
+    {
+      "epoch": 0.3119112356156107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611375085014993e-05,
+      "loss": 1.0767,
+      "step": 11357
+    },
+    {
+      "epoch": 0.3119386998434539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6113046915513635e-05,
+      "loss": 1.0125,
+      "step": 11358
+    },
+    {
+      "epoch": 0.3119661640712971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6112342932508543e-05,
+      "loss": 1.0323,
+      "step": 11359
+    },
+    {
+      "epoch": 0.31199362829914035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6111638901140227e-05,
+      "loss": 1.0375,
+      "step": 11360
+    },
+    {
+      "epoch": 0.3120210925269836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611093482141425e-05,
+      "loss": 1.0132,
+      "step": 11361
+    },
+    {
+      "epoch": 0.31204855675482684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61102306933362e-05,
+      "loss": 0.8691,
+      "step": 11362
+    },
+    {
+      "epoch": 0.31207602098267007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610952651691163e-05,
+      "loss": 0.8967,
+      "step": 11363
+    },
+    {
+      "epoch": 0.3121034852105133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610882229214612e-05,
+      "loss": 0.9207,
+      "step": 11364
+    },
+    {
+      "epoch": 0.31213094943835656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6108118019045246e-05,
+      "loss": 1.0445,
+      "step": 11365
+    },
+    {
+      "epoch": 0.3121584136661998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6107413697614573e-05,
+      "loss": 1.0182,
+      "step": 11366
+    },
+    {
+      "epoch": 0.312185877894043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6106709327859678e-05,
+      "loss": 1.0224,
+      "step": 11367
+    },
+    {
+      "epoch": 0.31221334212188623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6106004909786134e-05,
+      "loss": 0.9321,
+      "step": 11368
+    },
+    {
+      "epoch": 0.3122408063497295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610530044339951e-05,
+      "loss": 1.0108,
+      "step": 11369
+    },
+    {
+      "epoch": 0.3122682705775727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610459592870539e-05,
+      "loss": 1.008,
+      "step": 11370
+    },
+    {
+      "epoch": 0.31229573480541595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610389136570934e-05,
+      "loss": 1.0514,
+      "step": 11371
+    },
+    {
+      "epoch": 0.31232319903325917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610318675441694e-05,
+      "loss": 0.9113,
+      "step": 11372
+    },
+    {
+      "epoch": 0.3123506632611024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6102482094833756e-05,
+      "loss": 0.9958,
+      "step": 11373
+    },
+    {
+      "epoch": 0.31237812748894567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610177738696537e-05,
+      "loss": 1.0408,
+      "step": 11374
+    },
+    {
+      "epoch": 0.3124055917167889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6101072630817363e-05,
+      "loss": 0.9667,
+      "step": 11375
+    },
+    {
+      "epoch": 0.3124330559446321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6100367826395304e-05,
+      "loss": 0.8804,
+      "step": 11376
+    },
+    {
+      "epoch": 0.31246052017247533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6099662973704767e-05,
+      "loss": 0.9885,
+      "step": 11377
+    },
+    {
+      "epoch": 0.3124879844003186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609895807275134e-05,
+      "loss": 0.9073,
+      "step": 11378
+    },
+    {
+      "epoch": 0.31251544862816183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609825312354059e-05,
+      "loss": 0.8885,
+      "step": 11379
+    },
+    {
+      "epoch": 0.31254291285600505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60975481260781e-05,
+      "loss": 0.8924,
+      "step": 11380
+    },
+    {
+      "epoch": 0.31257037708384827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609684308036945e-05,
+      "loss": 1.0347,
+      "step": 11381
+    },
+    {
+      "epoch": 0.31259784131169155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6096137986420216e-05,
+      "loss": 0.9417,
+      "step": 11382
+    },
+    {
+      "epoch": 0.31262530553953477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6095432844235973e-05,
+      "loss": 0.8771,
+      "step": 11383
+    },
+    {
+      "epoch": 0.312652769767378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6094727653822307e-05,
+      "loss": 0.9955,
+      "step": 11384
+    },
+    {
+      "epoch": 0.3126802339952212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6094022415184792e-05,
+      "loss": 0.9787,
+      "step": 11385
+    },
+    {
+      "epoch": 0.31270769822306443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6093317128329016e-05,
+      "loss": 0.8154,
+      "step": 11386
+    },
+    {
+      "epoch": 0.3127351624509077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609261179326055e-05,
+      "loss": 0.9906,
+      "step": 11387
+    },
+    {
+      "epoch": 0.31276262667875093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6091906409984984e-05,
+      "loss": 0.9902,
+      "step": 11388
+    },
+    {
+      "epoch": 0.31279009090659415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609120097850789e-05,
+      "loss": 0.965,
+      "step": 11389
+    },
+    {
+      "epoch": 0.3128175551344374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6090495498834858e-05,
+      "loss": 0.8633,
+      "step": 11390
+    },
+    {
+      "epoch": 0.31284501936228065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6089789970971465e-05,
+      "loss": 1.0848,
+      "step": 11391
+    },
+    {
+      "epoch": 0.3128724835901239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6089084394923292e-05,
+      "loss": 0.957,
+      "step": 11392
+    },
+    {
+      "epoch": 0.3128999478179671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6088378770695926e-05,
+      "loss": 0.9568,
+      "step": 11393
+    },
+    {
+      "epoch": 0.3129274120458103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6087673098294953e-05,
+      "loss": 1.0008,
+      "step": 11394
+    },
+    {
+      "epoch": 0.3129548762736536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6086967377725947e-05,
+      "loss": 0.9206,
+      "step": 11395
+    },
+    {
+      "epoch": 0.3129823405014968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60862616089945e-05,
+      "loss": 0.9667,
+      "step": 11396
+    },
+    {
+      "epoch": 0.31300980472934004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6085555792106192e-05,
+      "loss": 1.0172,
+      "step": 11397
+    },
+    {
+      "epoch": 0.31303726895718326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608484992706661e-05,
+      "loss": 0.9165,
+      "step": 11398
+    },
+    {
+      "epoch": 0.3130647331850265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6084144013881336e-05,
+      "loss": 0.9675,
+      "step": 11399
+    },
+    {
+      "epoch": 0.31309219741286975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608343805255596e-05,
+      "loss": 1.1015,
+      "step": 11400
+    },
+    {
+      "epoch": 0.313119661640713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608273204309607e-05,
+      "loss": 1.0342,
+      "step": 11401
+    },
+    {
+      "epoch": 0.3131471258685562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608202598550724e-05,
+      "loss": 1.0164,
+      "step": 11402
+    },
+    {
+      "epoch": 0.3131745900963994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608131987979507e-05,
+      "loss": 0.9534,
+      "step": 11403
+    },
+    {
+      "epoch": 0.3132020543242427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608061372596514e-05,
+      "loss": 0.9988,
+      "step": 11404
+    },
+    {
+      "epoch": 0.3132295185520859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6079907524023036e-05,
+      "loss": 0.9474,
+      "step": 11405
+    },
+    {
+      "epoch": 0.31325698277992914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6079201273974355e-05,
+      "loss": 1.0624,
+      "step": 11406
+    },
+    {
+      "epoch": 0.31328444700777236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607849497582467e-05,
+      "loss": 0.9011,
+      "step": 11407
+    },
+    {
+      "epoch": 0.31331191123561564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6077788629579584e-05,
+      "loss": 0.9257,
+      "step": 11408
+    },
+    {
+      "epoch": 0.31333937546345886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607708223524468e-05,
+      "loss": 0.9538,
+      "step": 11409
+    },
+    {
+      "epoch": 0.3133668396913021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6076375792825545e-05,
+      "loss": 0.8983,
+      "step": 11410
+    },
+    {
+      "epoch": 0.3133943039191453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6075669302327775e-05,
+      "loss": 1.0507,
+      "step": 11411
+    },
+    {
+      "epoch": 0.3134217681469885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607496276375695e-05,
+      "loss": 0.9009,
+      "step": 11412
+    },
+    {
+      "epoch": 0.3134492323748318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607425617711867e-05,
+      "loss": 0.9587,
+      "step": 11413
+    },
+    {
+      "epoch": 0.313476696602675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6073549542418522e-05,
+      "loss": 0.9211,
+      "step": 11414
+    },
+    {
+      "epoch": 0.31350416083051824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6072842859662098e-05,
+      "loss": 0.9395,
+      "step": 11415
+    },
+    {
+      "epoch": 0.31353162505836146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607213612885499e-05,
+      "loss": 0.9486,
+      "step": 11416
+    },
+    {
+      "epoch": 0.31355908928620474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607142935000278e-05,
+      "loss": 0.9628,
+      "step": 11417
+    },
+    {
+      "epoch": 0.31358655351404796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607072252311108e-05,
+      "loss": 1.0938,
+      "step": 11418
+    },
+    {
+      "epoch": 0.3136140177418912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6070015648185466e-05,
+      "loss": 0.9687,
+      "step": 11419
+    },
+    {
+      "epoch": 0.3136414819697344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606930872523154e-05,
+      "loss": 1.0739,
+      "step": 11420
+    },
+    {
+      "epoch": 0.3136689461975777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606860175425489e-05,
+      "loss": 0.9288,
+      "step": 11421
+    },
+    {
+      "epoch": 0.3136964104254209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6067894735261115e-05,
+      "loss": 0.9751,
+      "step": 11422
+    },
+    {
+      "epoch": 0.3137238746532641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6067187668255804e-05,
+      "loss": 1.0032,
+      "step": 11423
+    },
+    {
+      "epoch": 0.31375133888110734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6066480553244555e-05,
+      "loss": 0.9536,
+      "step": 11424
+    },
+    {
+      "epoch": 0.31377880310895057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606577339023296e-05,
+      "loss": 0.9669,
+      "step": 11425
+    },
+    {
+      "epoch": 0.31380626733679384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6065066179226615e-05,
+      "loss": 0.8853,
+      "step": 11426
+    },
+    {
+      "epoch": 0.31383373156463706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6064358920231122e-05,
+      "loss": 1.0319,
+      "step": 11427
+    },
+    {
+      "epoch": 0.3138611957924803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606365161325207e-05,
+      "loss": 0.9254,
+      "step": 11428
+    },
+    {
+      "epoch": 0.3138886600203235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6062944258295056e-05,
+      "loss": 0.9775,
+      "step": 11429
+    },
+    {
+      "epoch": 0.3139161242481668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606223685536568e-05,
+      "loss": 0.9946,
+      "step": 11430
+    },
+    {
+      "epoch": 0.31394358847601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606152940446954e-05,
+      "loss": 0.963,
+      "step": 11431
+    },
+    {
+      "epoch": 0.3139710527038532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6060821905612226e-05,
+      "loss": 1.0419,
+      "step": 11432
+    },
+    {
+      "epoch": 0.31399851693169645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6060114358799345e-05,
+      "loss": 1.0402,
+      "step": 11433
+    },
+    {
+      "epoch": 0.3140259811595397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6059406764036492e-05,
+      "loss": 1.0462,
+      "step": 11434
+    },
+    {
+      "epoch": 0.31405344538738295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6058699121329263e-05,
+      "loss": 1.009,
+      "step": 11435
+    },
+    {
+      "epoch": 0.31408090961522617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6057991430683263e-05,
+      "loss": 0.9891,
+      "step": 11436
+    },
+    {
+      "epoch": 0.3141083738430694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6057283692104085e-05,
+      "loss": 0.9369,
+      "step": 11437
+    },
+    {
+      "epoch": 0.3141358380709126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6056575905597337e-05,
+      "loss": 0.9459,
+      "step": 11438
+    },
+    {
+      "epoch": 0.3141633022987559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605586807116861e-05,
+      "loss": 1.0699,
+      "step": 11439
+    },
+    {
+      "epoch": 0.3141907665265991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6055160188823507e-05,
+      "loss": 0.9608,
+      "step": 11440
+    },
+    {
+      "epoch": 0.31421823075444233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6054452258567634e-05,
+      "loss": 0.9084,
+      "step": 11441
+    },
+    {
+      "epoch": 0.31424569498228555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605374428040659e-05,
+      "loss": 0.9494,
+      "step": 11442
+    },
+    {
+      "epoch": 0.3142731592101288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605303625434597e-05,
+      "loss": 0.9691,
+      "step": 11443
+    },
+    {
+      "epoch": 0.31430062343797205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6052328180391393e-05,
+      "loss": 0.8782,
+      "step": 11444
+    },
+    {
+      "epoch": 0.31432808766581527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6051620058548444e-05,
+      "loss": 0.9662,
+      "step": 11445
+    },
+    {
+      "epoch": 0.3143555518936585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6050911888822733e-05,
+      "loss": 0.9718,
+      "step": 11446
+    },
+    {
+      "epoch": 0.31438301612150177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6050203671219867e-05,
+      "loss": 0.9409,
+      "step": 11447
+    },
+    {
+      "epoch": 0.314410480349345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6049495405745444e-05,
+      "loss": 1.0103,
+      "step": 11448
+    },
+    {
+      "epoch": 0.3144379445771882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6048787092405073e-05,
+      "loss": 0.9314,
+      "step": 11449
+    },
+    {
+      "epoch": 0.31446540880503143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604807873120435e-05,
+      "loss": 0.8034,
+      "step": 11450
+    },
+    {
+      "epoch": 0.31449287303287465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6047370322148886e-05,
+      "loss": 1.1122,
+      "step": 11451
+    },
+    {
+      "epoch": 0.31452033726071793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6046661865244288e-05,
+      "loss": 0.9533,
+      "step": 11452
+    },
+    {
+      "epoch": 0.31454780148856115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604595336049616e-05,
+      "loss": 0.9695,
+      "step": 11453
+    },
+    {
+      "epoch": 0.3145752657164044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6045244807910102e-05,
+      "loss": 1.116,
+      "step": 11454
+    },
+    {
+      "epoch": 0.3146027299442476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604453620749173e-05,
+      "loss": 0.9074,
+      "step": 11455
+    },
+    {
+      "epoch": 0.31463019417209087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6043827559246647e-05,
+      "loss": 0.9847,
+      "step": 11456
+    },
+    {
+      "epoch": 0.3146576583999341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6043118863180453e-05,
+      "loss": 1.0068,
+      "step": 11457
+    },
+    {
+      "epoch": 0.3146851226277773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6042410119298768e-05,
+      "loss": 0.9363,
+      "step": 11458
+    },
+    {
+      "epoch": 0.31471258685562054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6041701327607193e-05,
+      "loss": 1.004,
+      "step": 11459
+    },
+    {
+      "epoch": 0.3147400510834638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6040992488111334e-05,
+      "loss": 0.8318,
+      "step": 11460
+    },
+    {
+      "epoch": 0.31476751531130703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60402836008168e-05,
+      "loss": 0.9908,
+      "step": 11461
+    },
+    {
+      "epoch": 0.31479497953915025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6039574665729208e-05,
+      "loss": 0.8517,
+      "step": 11462
+    },
+    {
+      "epoch": 0.3148224437669935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6038865682854158e-05,
+      "loss": 0.9325,
+      "step": 11463
+    },
+    {
+      "epoch": 0.3148499079948367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6038156652197265e-05,
+      "loss": 0.9707,
+      "step": 11464
+    },
+    {
+      "epoch": 0.31487737222268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603744757376414e-05,
+      "loss": 1.0003,
+      "step": 11465
+    },
+    {
+      "epoch": 0.3149048364505232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6036738447560387e-05,
+      "loss": 1.0238,
+      "step": 11466
+    },
+    {
+      "epoch": 0.3149323006783664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603602927359162e-05,
+      "loss": 0.9543,
+      "step": 11467
+    },
+    {
+      "epoch": 0.31495976490620964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6035320051863458e-05,
+      "loss": 1.1068,
+      "step": 11468
+    },
+    {
+      "epoch": 0.3149872291340529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60346107823815e-05,
+      "loss": 0.9974,
+      "step": 11469
+    },
+    {
+      "epoch": 0.31501469336189614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6033901465151364e-05,
+      "loss": 0.9449,
+      "step": 11470
+    },
+    {
+      "epoch": 0.31504215758973936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6033192100178664e-05,
+      "loss": 0.8907,
+      "step": 11471
+    },
+    {
+      "epoch": 0.3150696218175826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603248268746901e-05,
+      "loss": 1.0141,
+      "step": 11472
+    },
+    {
+      "epoch": 0.31509708604542586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603177322702802e-05,
+      "loss": 0.9023,
+      "step": 11473
+    },
+    {
+      "epoch": 0.3151245502732691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60310637188613e-05,
+      "loss": 1.016,
+      "step": 11474
+    },
+    {
+      "epoch": 0.3151520145011123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6030354162974468e-05,
+      "loss": 1.0162,
+      "step": 11475
+    },
+    {
+      "epoch": 0.3151794787289555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602964455937314e-05,
+      "loss": 1.0178,
+      "step": 11476
+    },
+    {
+      "epoch": 0.3152069429567988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6028934908062926e-05,
+      "loss": 0.928,
+      "step": 11477
+    },
+    {
+      "epoch": 0.315234407184642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6028225209049444e-05,
+      "loss": 1.0328,
+      "step": 11478
+    },
+    {
+      "epoch": 0.31526187141248524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602751546233831e-05,
+      "loss": 0.9728,
+      "step": 11479
+    },
+    {
+      "epoch": 0.31528933564032846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602680566793514e-05,
+      "loss": 0.8539,
+      "step": 11480
+    },
+    {
+      "epoch": 0.3153167998681717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6026095825845548e-05,
+      "loss": 0.9551,
+      "step": 11481
+    },
+    {
+      "epoch": 0.31534426409601496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6025385936075148e-05,
+      "loss": 0.7907,
+      "step": 11482
+    },
+    {
+      "epoch": 0.3153717283238582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6024675998629565e-05,
+      "loss": 0.866,
+      "step": 11483
+    },
+    {
+      "epoch": 0.3153991925517014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602396601351441e-05,
+      "loss": 0.9227,
+      "step": 11484
+    },
+    {
+      "epoch": 0.3154266567795446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6023255980735305e-05,
+      "loss": 1.0154,
+      "step": 11485
+    },
+    {
+      "epoch": 0.3154541210073879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602254590029786e-05,
+      "loss": 0.8812,
+      "step": 11486
+    },
+    {
+      "epoch": 0.3154815852352311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6021835772207702e-05,
+      "loss": 0.9435,
+      "step": 11487
+    },
+    {
+      "epoch": 0.31550904946307434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6021125596470447e-05,
+      "loss": 1.0278,
+      "step": 11488
+    },
+    {
+      "epoch": 0.31553651369091756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6020415373091712e-05,
+      "loss": 0.8874,
+      "step": 11489
+    },
+    {
+      "epoch": 0.31556397791876084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6019705102077122e-05,
+      "loss": 1.0094,
+      "step": 11490
+    },
+    {
+      "epoch": 0.31559144214660406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6018994783432293e-05,
+      "loss": 0.9206,
+      "step": 11491
+    },
+    {
+      "epoch": 0.3156189063744473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6018284417162844e-05,
+      "loss": 1.0471,
+      "step": 11492
+    },
+    {
+      "epoch": 0.3156463706022905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60175740032744e-05,
+      "loss": 0.9037,
+      "step": 11493
+    },
+    {
+      "epoch": 0.3156738348301337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6016863541772578e-05,
+      "loss": 0.9977,
+      "step": 11494
+    },
+    {
+      "epoch": 0.315701299057977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6016153032663e-05,
+      "loss": 0.9216,
+      "step": 11495
+    },
+    {
+      "epoch": 0.3157287632858202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6015442475951287e-05,
+      "loss": 1.0008,
+      "step": 11496
+    },
+    {
+      "epoch": 0.31575622751366345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6014731871643065e-05,
+      "loss": 1.0439,
+      "step": 11497
+    },
+    {
+      "epoch": 0.31578369174150667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6014021219743955e-05,
+      "loss": 1.0196,
+      "step": 11498
+    },
+    {
+      "epoch": 0.31581115596934994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6013310520259578e-05,
+      "loss": 0.9913,
+      "step": 11499
+    },
+    {
+      "epoch": 0.31583862019719317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6012599773195564e-05,
+      "loss": 0.9709,
+      "step": 11500
+    },
+    {
+      "epoch": 0.3158660844250364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6011888978557525e-05,
+      "loss": 1.0176,
+      "step": 11501
+    },
+    {
+      "epoch": 0.3158935486528796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6011178136351095e-05,
+      "loss": 1.0073,
+      "step": 11502
+    },
+    {
+      "epoch": 0.3159210128807229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6010467246581894e-05,
+      "loss": 0.9939,
+      "step": 11503
+    },
+    {
+      "epoch": 0.3159484771085661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6009756309255548e-05,
+      "loss": 0.9403,
+      "step": 11504
+    },
+    {
+      "epoch": 0.3159759413364093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6009045324377683e-05,
+      "loss": 1.0981,
+      "step": 11505
+    },
+    {
+      "epoch": 0.31600340556425255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600833429195392e-05,
+      "loss": 0.9449,
+      "step": 11506
+    },
+    {
+      "epoch": 0.31603086979209577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6007623211989892e-05,
+      "loss": 1.0633,
+      "step": 11507
+    },
+    {
+      "epoch": 0.31605833401993905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6006912084491222e-05,
+      "loss": 0.9746,
+      "step": 11508
+    },
+    {
+      "epoch": 0.31608579824778227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6006200909463537e-05,
+      "loss": 0.8986,
+      "step": 11509
+    },
+    {
+      "epoch": 0.3161132624756255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6005489686912462e-05,
+      "loss": 0.9615,
+      "step": 11510
+    },
+    {
+      "epoch": 0.3161407267034687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600477841684363e-05,
+      "loss": 0.8021,
+      "step": 11511
+    },
+    {
+      "epoch": 0.316168190931312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600406709926266e-05,
+      "loss": 1.035,
+      "step": 11512
+    },
+    {
+      "epoch": 0.3161956551591552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600335573417519e-05,
+      "loss": 0.8405,
+      "step": 11513
+    },
+    {
+      "epoch": 0.31622311938699843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600264432158684e-05,
+      "loss": 1.0003,
+      "step": 11514
+    },
+    {
+      "epoch": 0.31625058361484165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6001932861503246e-05,
+      "loss": 0.9759,
+      "step": 11515
+    },
+    {
+      "epoch": 0.31627804784268493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6001221353930033e-05,
+      "loss": 0.8973,
+      "step": 11516
+    },
+    {
+      "epoch": 0.31630551207052815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600050979887283e-05,
+      "loss": 0.8869,
+      "step": 11517
+    },
+    {
+      "epoch": 0.31633297629837137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5999798196337274e-05,
+      "loss": 1.0436,
+      "step": 11518
+    },
+    {
+      "epoch": 0.3163604405262146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599908654632899e-05,
+      "loss": 0.9242,
+      "step": 11519
+    },
+    {
+      "epoch": 0.3163879047540578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5998374848853605e-05,
+      "loss": 0.8969,
+      "step": 11520
+    },
+    {
+      "epoch": 0.3164153689819011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5997663103916756e-05,
+      "loss": 0.9057,
+      "step": 11521
+    },
+    {
+      "epoch": 0.3164428332097443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5996951311524073e-05,
+      "loss": 0.992,
+      "step": 11522
+    },
+    {
+      "epoch": 0.31647029743758753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599623947168119e-05,
+      "loss": 0.9696,
+      "step": 11523
+    },
+    {
+      "epoch": 0.31649776166543075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5995527584393736e-05,
+      "loss": 1.0077,
+      "step": 11524
+    },
+    {
+      "epoch": 0.31652522589327403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5994815649667348e-05,
+      "loss": 1.0016,
+      "step": 11525
+    },
+    {
+      "epoch": 0.31655269012111725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5994103667507653e-05,
+      "loss": 0.9343,
+      "step": 11526
+    },
+    {
+      "epoch": 0.3165801543489605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599339163792029e-05,
+      "loss": 0.8783,
+      "step": 11527
+    },
+    {
+      "epoch": 0.3166076185768037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5992679560910892e-05,
+      "loss": 0.9896,
+      "step": 11528
+    },
+    {
+      "epoch": 0.316635082804647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599196743648509e-05,
+      "loss": 1.0105,
+      "step": 11529
+    },
+    {
+      "epoch": 0.3166625470324902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5991255264648523e-05,
+      "loss": 0.991,
+      "step": 11530
+    },
+    {
+      "epoch": 0.3166900112603334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599054304540682e-05,
+      "loss": 0.9286,
+      "step": 11531
+    },
+    {
+      "epoch": 0.31671747548817664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5989830778765618e-05,
+      "loss": 1.0383,
+      "step": 11532
+    },
+    {
+      "epoch": 0.31674493971601986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598911846473056e-05,
+      "loss": 0.9002,
+      "step": 11533
+    },
+    {
+      "epoch": 0.31677240394386313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5988406103307273e-05,
+      "loss": 0.9503,
+      "step": 11534
+    },
+    {
+      "epoch": 0.31679986817170636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59876936945014e-05,
+      "loss": 0.9131,
+      "step": 11535
+    },
+    {
+      "epoch": 0.3168273323995496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5986981238318574e-05,
+      "loss": 0.9908,
+      "step": 11536
+    },
+    {
+      "epoch": 0.3168547966273928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598626873476443e-05,
+      "loss": 0.9169,
+      "step": 11537
+    },
+    {
+      "epoch": 0.3168822608552361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5985556183844616e-05,
+      "loss": 0.951,
+      "step": 11538
+    },
+    {
+      "epoch": 0.3169097250830793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5984843585564757e-05,
+      "loss": 1.0044,
+      "step": 11539
+    },
+    {
+      "epoch": 0.3169371893109225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5984130939930503e-05,
+      "loss": 0.9231,
+      "step": 11540
+    },
+    {
+      "epoch": 0.31696465353876574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598341824694748e-05,
+      "loss": 0.968,
+      "step": 11541
+    },
+    {
+      "epoch": 0.316992117766609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5982705506621338e-05,
+      "loss": 1.0032,
+      "step": 11542
+    },
+    {
+      "epoch": 0.31701958199445224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5981992718957712e-05,
+      "loss": 1.024,
+      "step": 11543
+    },
+    {
+      "epoch": 0.31704704622229546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598127988396224e-05,
+      "loss": 1.0321,
+      "step": 11544
+    },
+    {
+      "epoch": 0.3170745104501387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598056700164057e-05,
+      "loss": 0.897,
+      "step": 11545
+    },
+    {
+      "epoch": 0.3171019746779819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5979854071998336e-05,
+      "loss": 0.9153,
+      "step": 11546
+    },
+    {
+      "epoch": 0.3171294389058252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597914109504118e-05,
+      "loss": 0.9362,
+      "step": 11547
+    },
+    {
+      "epoch": 0.3171569031336684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597842807077474e-05,
+      "loss": 0.9981,
+      "step": 11548
+    },
+    {
+      "epoch": 0.3171843673615116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5977714999204662e-05,
+      "loss": 0.9081,
+      "step": 11549
+    },
+    {
+      "epoch": 0.31721183158935484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5977001880336588e-05,
+      "loss": 0.9119,
+      "step": 11550
+    },
+    {
+      "epoch": 0.3172392958171981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597628871417616e-05,
+      "loss": 1.0253,
+      "step": 11551
+    },
+    {
+      "epoch": 0.31726676004504134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597557550072902e-05,
+      "loss": 1.0665,
+      "step": 11552
+    },
+    {
+      "epoch": 0.31729422427288456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5974862240000817e-05,
+      "loss": 0.9255,
+      "step": 11553
+    },
+    {
+      "epoch": 0.3173216885007278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5974148931997184e-05,
+      "loss": 0.9518,
+      "step": 11554
+    },
+    {
+      "epoch": 0.31734915272857106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5973435576723772e-05,
+      "loss": 0.9327,
+      "step": 11555
+    },
+    {
+      "epoch": 0.3173766169564143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5972722174186228e-05,
+      "loss": 0.9413,
+      "step": 11556
+    },
+    {
+      "epoch": 0.3174040811842575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5972008724390184e-05,
+      "loss": 0.988,
+      "step": 11557
+    },
+    {
+      "epoch": 0.3174315454121007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5971295227341295e-05,
+      "loss": 0.8646,
+      "step": 11558
+    },
+    {
+      "epoch": 0.31745900963994395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597058168304521e-05,
+      "loss": 0.9239,
+      "step": 11559
+    },
+    {
+      "epoch": 0.3174864738677872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596986809150757e-05,
+      "loss": 0.9319,
+      "step": 11560
+    },
+    {
+      "epoch": 0.31751393809563044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5969154452734018e-05,
+      "loss": 0.8276,
+      "step": 11561
+    },
+    {
+      "epoch": 0.31754140232347366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5968440766730206e-05,
+      "loss": 1.0119,
+      "step": 11562
+    },
+    {
+      "epoch": 0.3175688665513169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596772703350178e-05,
+      "loss": 0.9293,
+      "step": 11563
+    },
+    {
+      "epoch": 0.31759633077916016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5967013253054383e-05,
+      "loss": 0.9288,
+      "step": 11564
+    },
+    {
+      "epoch": 0.3176237950070034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5966299425393665e-05,
+      "loss": 0.9684,
+      "step": 11565
+    },
+    {
+      "epoch": 0.3176512592348466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5965585550525277e-05,
+      "loss": 0.8962,
+      "step": 11566
+    },
+    {
+      "epoch": 0.3176787234626898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5964871628454863e-05,
+      "loss": 0.9017,
+      "step": 11567
+    },
+    {
+      "epoch": 0.3177061876905331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5964157659188078e-05,
+      "loss": 0.991,
+      "step": 11568
+    },
+    {
+      "epoch": 0.3177336519183763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5963443642730563e-05,
+      "loss": 0.8449,
+      "step": 11569
+    },
+    {
+      "epoch": 0.31776111614621955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5962729579087974e-05,
+      "loss": 1.0461,
+      "step": 11570
+    },
+    {
+      "epoch": 0.31778858037406277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5962015468265958e-05,
+      "loss": 0.9824,
+      "step": 11571
+    },
+    {
+      "epoch": 0.31781604460190604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596130131027017e-05,
+      "loss": 0.9511,
+      "step": 11572
+    },
+    {
+      "epoch": 0.31784350882974927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596058710510625e-05,
+      "loss": 0.8963,
+      "step": 11573
+    },
+    {
+      "epoch": 0.3178709730575925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5959872852779862e-05,
+      "loss": 0.9866,
+      "step": 11574
+    },
+    {
+      "epoch": 0.3178984372854357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595915855329665e-05,
+      "loss": 0.9576,
+      "step": 11575
+    },
+    {
+      "epoch": 0.31792590151327893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5958444206662266e-05,
+      "loss": 1.0226,
+      "step": 11576
+    },
+    {
+      "epoch": 0.3179533657411222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595772981288236e-05,
+      "loss": 1.04,
+      "step": 11577
+    },
+    {
+      "epoch": 0.31798082996896543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5957015371962593e-05,
+      "loss": 0.9657,
+      "step": 11578
+    },
+    {
+      "epoch": 0.31800829419680865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5956300883908613e-05,
+      "loss": 0.9774,
+      "step": 11579
+    },
+    {
+      "epoch": 0.31803575842465187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5955586348726072e-05,
+      "loss": 0.9215,
+      "step": 11580
+    },
+    {
+      "epoch": 0.31806322265249515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5954871766420624e-05,
+      "loss": 0.9554,
+      "step": 11581
+    },
+    {
+      "epoch": 0.31809068688033837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5954157136997923e-05,
+      "loss": 0.9915,
+      "step": 11582
+    },
+    {
+      "epoch": 0.3181181511081816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5953442460463628e-05,
+      "loss": 0.9882,
+      "step": 11583
+    },
+    {
+      "epoch": 0.3181456153360248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5952727736823384e-05,
+      "loss": 0.9131,
+      "step": 11584
+    },
+    {
+      "epoch": 0.3181730795638681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5952012966082857e-05,
+      "loss": 0.9612,
+      "step": 11585
+    },
+    {
+      "epoch": 0.3182005437917113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59512981482477e-05,
+      "loss": 0.8149,
+      "step": 11586
+    },
+    {
+      "epoch": 0.31822800801955453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595058328332356e-05,
+      "loss": 1.0127,
+      "step": 11587
+    },
+    {
+      "epoch": 0.31825547224739775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5949868371316103e-05,
+      "loss": 0.8491,
+      "step": 11588
+    },
+    {
+      "epoch": 0.318282936475241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5949153412230984e-05,
+      "loss": 0.9394,
+      "step": 11589
+    },
+    {
+      "epoch": 0.31831040070308425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5948438406073858e-05,
+      "loss": 0.9239,
+      "step": 11590
+    },
+    {
+      "epoch": 0.31833786493092747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5947723352850384e-05,
+      "loss": 0.9406,
+      "step": 11591
+    },
+    {
+      "epoch": 0.3183653291587707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5947008252566218e-05,
+      "loss": 1.0112,
+      "step": 11592
+    },
+    {
+      "epoch": 0.3183927933866139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594629310522702e-05,
+      "loss": 1.0294,
+      "step": 11593
+    },
+    {
+      "epoch": 0.3184202576144572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5945577910838444e-05,
+      "loss": 1.0256,
+      "step": 11594
+    },
+    {
+      "epoch": 0.3184477218423004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5944862669406157e-05,
+      "loss": 0.7421,
+      "step": 11595
+    },
+    {
+      "epoch": 0.31847518607014363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594414738093581e-05,
+      "loss": 0.9932,
+      "step": 11596
+    },
+    {
+      "epoch": 0.31850265029798686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5943432045433067e-05,
+      "loss": 0.9753,
+      "step": 11597
+    },
+    {
+      "epoch": 0.31853011452583013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594271666290359e-05,
+      "loss": 0.9214,
+      "step": 11598
+    },
+    {
+      "epoch": 0.31855757875367335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5942001233353034e-05,
+      "loss": 1.0446,
+      "step": 11599
+    },
+    {
+      "epoch": 0.3185850429815166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5941285756787067e-05,
+      "loss": 0.8806,
+      "step": 11600
+    },
+    {
+      "epoch": 0.3186125072093598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594057023321134e-05,
+      "loss": 1.1127,
+      "step": 11601
+    },
+    {
+      "epoch": 0.318639971437203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593985466263152e-05,
+      "loss": 0.9603,
+      "step": 11602
+    },
+    {
+      "epoch": 0.3186674356650463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5939139045053274e-05,
+      "loss": 0.9626,
+      "step": 11603
+    },
+    {
+      "epoch": 0.3186948998928895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593842338048226e-05,
+      "loss": 0.9451,
+      "step": 11604
+    },
+    {
+      "epoch": 0.31872236412073274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5937707668924135e-05,
+      "loss": 0.9649,
+      "step": 11605
+    },
+    {
+      "epoch": 0.31874982834857596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5936991910384566e-05,
+      "loss": 1.0086,
+      "step": 11606
+    },
+    {
+      "epoch": 0.31877729257641924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593627610486922e-05,
+      "loss": 0.9629,
+      "step": 11607
+    },
+    {
+      "epoch": 0.31880475680426246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593556025238376e-05,
+      "loss": 0.9136,
+      "step": 11608
+    },
+    {
+      "epoch": 0.3188322210321057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5934844352933847e-05,
+      "loss": 1.0585,
+      "step": 11609
+    },
+    {
+      "epoch": 0.3188596852599489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5934128406525146e-05,
+      "loss": 0.8413,
+      "step": 11610
+    },
+    {
+      "epoch": 0.3188871494877922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593341241316332e-05,
+      "loss": 0.9769,
+      "step": 11611
+    },
+    {
+      "epoch": 0.3189146137156354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593269637285404e-05,
+      "loss": 1.0056,
+      "step": 11612
+    },
+    {
+      "epoch": 0.3189420779434786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5931980285602967e-05,
+      "loss": 0.8896,
+      "step": 11613
+    },
+    {
+      "epoch": 0.31896954217132184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5931264151415765e-05,
+      "loss": 1.0253,
+      "step": 11614
+    },
+    {
+      "epoch": 0.31899700639916506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593054797029811e-05,
+      "loss": 0.9682,
+      "step": 11615
+    },
+    {
+      "epoch": 0.31902447062700834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5929831742255655e-05,
+      "loss": 1.0305,
+      "step": 11616
+    },
+    {
+      "epoch": 0.31905193485485156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5929115467294078e-05,
+      "loss": 1.0337,
+      "step": 11617
+    },
+    {
+      "epoch": 0.3190793990826948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5928399145419045e-05,
+      "loss": 0.9362,
+      "step": 11618
+    },
+    {
+      "epoch": 0.319106863310538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5927682776636216e-05,
+      "loss": 1.0099,
+      "step": 11619
+    },
+    {
+      "epoch": 0.3191343275383813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592696636095127e-05,
+      "loss": 0.9632,
+      "step": 11620
+    },
+    {
+      "epoch": 0.3191617917662245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5926249898369867e-05,
+      "loss": 0.9281,
+      "step": 11621
+    },
+    {
+      "epoch": 0.3191892559940677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592553338889768e-05,
+      "loss": 1.0345,
+      "step": 11622
+    },
+    {
+      "epoch": 0.31921672022191094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5924816832540377e-05,
+      "loss": 0.9411,
+      "step": 11623
+    },
+    {
+      "epoch": 0.3192441844497542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5924100229303632e-05,
+      "loss": 1.0038,
+      "step": 11624
+    },
+    {
+      "epoch": 0.31927164867759744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5923383579193107e-05,
+      "loss": 1.0799,
+      "step": 11625
+    },
+    {
+      "epoch": 0.31929911290544066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5922666882214477e-05,
+      "loss": 0.9761,
+      "step": 11626
+    },
+    {
+      "epoch": 0.3193265771332839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5921950138373414e-05,
+      "loss": 0.974,
+      "step": 11627
+    },
+    {
+      "epoch": 0.3193540413611271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5921233347675586e-05,
+      "loss": 0.935,
+      "step": 11628
+    },
+    {
+      "epoch": 0.3193815055889704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5920516510126668e-05,
+      "loss": 0.9782,
+      "step": 11629
+    },
+    {
+      "epoch": 0.3194089698168136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591979962573233e-05,
+      "loss": 0.9538,
+      "step": 11630
+    },
+    {
+      "epoch": 0.3194364340446568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591908269449824e-05,
+      "loss": 0.9861,
+      "step": 11631
+    },
+    {
+      "epoch": 0.31946389827250005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5918365716430076e-05,
+      "loss": 1.0067,
+      "step": 11632
+    },
+    {
+      "epoch": 0.3194913625003433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5917648691533513e-05,
+      "loss": 1.0488,
+      "step": 11633
+    },
+    {
+      "epoch": 0.31951882672818654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591693161981422e-05,
+      "loss": 1.0339,
+      "step": 11634
+    },
+    {
+      "epoch": 0.31954629095602977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5916214501277873e-05,
+      "loss": 1.0055,
+      "step": 11635
+    },
+    {
+      "epoch": 0.319573755183873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5915497335930145e-05,
+      "loss": 1.024,
+      "step": 11636
+    },
+    {
+      "epoch": 0.31960121941171626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591478012377671e-05,
+      "loss": 0.9819,
+      "step": 11637
+    },
+    {
+      "epoch": 0.3196286836395595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5914062864823245e-05,
+      "loss": 0.92,
+      "step": 11638
+    },
+    {
+      "epoch": 0.3196561478674027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591334555907542e-05,
+      "loss": 0.9182,
+      "step": 11639
+    },
+    {
+      "epoch": 0.31968361209524593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5912628206538915e-05,
+      "loss": 1.0023,
+      "step": 11640
+    },
+    {
+      "epoch": 0.31971107632308915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5911910807219407e-05,
+      "loss": 0.973,
+      "step": 11641
+    },
+    {
+      "epoch": 0.3197385405509324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5911193361122572e-05,
+      "loss": 0.9179,
+      "step": 11642
+    },
+    {
+      "epoch": 0.31976600477877565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591047586825408e-05,
+      "loss": 0.9702,
+      "step": 11643
+    },
+    {
+      "epoch": 0.31979346900661887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590975832861962e-05,
+      "loss": 0.8354,
+      "step": 11644
+    },
+    {
+      "epoch": 0.3198209332344621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590904074222486e-05,
+      "loss": 1.0109,
+      "step": 11645
+    },
+    {
+      "epoch": 0.31984839746230537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590832310907548e-05,
+      "loss": 0.9933,
+      "step": 11646
+    },
+    {
+      "epoch": 0.3198758616901486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5907605429177158e-05,
+      "loss": 0.9566,
+      "step": 11647
+    },
+    {
+      "epoch": 0.3199033259179918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5906887702535577e-05,
+      "loss": 0.962,
+      "step": 11648
+    },
+    {
+      "epoch": 0.31993079014583503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590616992915641e-05,
+      "loss": 1.0123,
+      "step": 11649
+    },
+    {
+      "epoch": 0.3199582543736783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590545210904534e-05,
+      "loss": 0.956,
+      "step": 11650
+    },
+    {
+      "epoch": 0.31998571860152153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590473424220804e-05,
+      "loss": 0.9139,
+      "step": 11651
+    },
+    {
+      "epoch": 0.32001318282936475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5904016328650206e-05,
+      "loss": 1.0685,
+      "step": 11652
+    },
+    {
+      "epoch": 0.32004064705720797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5903298368377497e-05,
+      "loss": 0.9503,
+      "step": 11653
+    },
+    {
+      "epoch": 0.3200681112850512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590258036139561e-05,
+      "loss": 0.9958,
+      "step": 11654
+    },
+    {
+      "epoch": 0.32009557551289447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590186230771022e-05,
+      "loss": 0.9453,
+      "step": 11655
+    },
+    {
+      "epoch": 0.3201230397407377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590114420732701e-05,
+      "loss": 0.9336,
+      "step": 11656
+    },
+    {
+      "epoch": 0.3201505039685809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590042606025166e-05,
+      "loss": 0.925,
+      "step": 11657
+    },
+    {
+      "epoch": 0.32017796819642413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5899707866489852e-05,
+      "loss": 0.8039,
+      "step": 11658
+    },
+    {
+      "epoch": 0.3202054324242674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5898989626047274e-05,
+      "loss": 0.9356,
+      "step": 11659
+    },
+    {
+      "epoch": 0.32023289665211063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5898271338929606e-05,
+      "loss": 0.965,
+      "step": 11660
+    },
+    {
+      "epoch": 0.32026036087995385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5897553005142527e-05,
+      "loss": 0.978,
+      "step": 11661
+    },
+    {
+      "epoch": 0.3202878251077971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589683462469172e-05,
+      "loss": 1.0428,
+      "step": 11662
+    },
+    {
+      "epoch": 0.32031528933564035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5896116197582882e-05,
+      "loss": 0.9397,
+      "step": 11663
+    },
+    {
+      "epoch": 0.3203427535634836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589539772382168e-05,
+      "loss": 0.9617,
+      "step": 11664
+    },
+    {
+      "epoch": 0.3203702177913268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5894679203413815e-05,
+      "loss": 1.0434,
+      "step": 11665
+    },
+    {
+      "epoch": 0.32039768201917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589396063636496e-05,
+      "loss": 0.9389,
+      "step": 11666
+    },
+    {
+      "epoch": 0.3204251462470133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5893242022680805e-05,
+      "loss": 0.8364,
+      "step": 11667
+    },
+    {
+      "epoch": 0.3204526104748565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589252336236704e-05,
+      "loss": 1.0276,
+      "step": 11668
+    },
+    {
+      "epoch": 0.32048007470269974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5891804655429345e-05,
+      "loss": 0.9413,
+      "step": 11669
+    },
+    {
+      "epoch": 0.32050753893054296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589108590187341e-05,
+      "loss": 0.9397,
+      "step": 11670
+    },
+    {
+      "epoch": 0.3205350031583862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5890367101704926e-05,
+      "loss": 0.9535,
+      "step": 11671
+    },
+    {
+      "epoch": 0.32056246738622945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588964825492957e-05,
+      "loss": 0.9986,
+      "step": 11672
+    },
+    {
+      "epoch": 0.3205899316140727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5888929361553038e-05,
+      "loss": 0.9861,
+      "step": 11673
+    },
+    {
+      "epoch": 0.3206173958419159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5888210421581016e-05,
+      "loss": 1.0038,
+      "step": 11674
+    },
+    {
+      "epoch": 0.3206448600697591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5887491435019186e-05,
+      "loss": 0.9601,
+      "step": 11675
+    },
+    {
+      "epoch": 0.3206723242976024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588677240187325e-05,
+      "loss": 0.9315,
+      "step": 11676
+    },
+    {
+      "epoch": 0.3206997885254456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5886053322148888e-05,
+      "loss": 0.9804,
+      "step": 11677
+    },
+    {
+      "epoch": 0.32072725275328884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5885334195851793e-05,
+      "loss": 1.0317,
+      "step": 11678
+    },
+    {
+      "epoch": 0.32075471698113206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588461502298765e-05,
+      "loss": 0.9296,
+      "step": 11679
+    },
+    {
+      "epoch": 0.32078218120897534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5883895803562158e-05,
+      "loss": 0.9245,
+      "step": 11680
+    },
+    {
+      "epoch": 0.32080964543681856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5883176537581005e-05,
+      "loss": 0.931,
+      "step": 11681
+    },
+    {
+      "epoch": 0.3208371096646618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5882457225049876e-05,
+      "loss": 0.9501,
+      "step": 11682
+    },
+    {
+      "epoch": 0.320864573892505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588173786597447e-05,
+      "loss": 0.9888,
+      "step": 11683
+    },
+    {
+      "epoch": 0.3208920381203482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5881018460360474e-05,
+      "loss": 0.8887,
+      "step": 11684
+    },
+    {
+      "epoch": 0.3209195023481915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588029900821358e-05,
+      "loss": 0.9524,
+      "step": 11685
+    },
+    {
+      "epoch": 0.3209469665760347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5879579509539485e-05,
+      "loss": 0.8885,
+      "step": 11686
+    },
+    {
+      "epoch": 0.32097443080387794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587885996434388e-05,
+      "loss": 0.9163,
+      "step": 11687
+    },
+    {
+      "epoch": 0.32100189503172116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5878140372632456e-05,
+      "loss": 0.9832,
+      "step": 11688
+    },
+    {
+      "epoch": 0.32102935925956444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5877420734410907e-05,
+      "loss": 0.8911,
+      "step": 11689
+    },
+    {
+      "epoch": 0.32105682348740766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5876701049684933e-05,
+      "loss": 1.0126,
+      "step": 11690
+    },
+    {
+      "epoch": 0.3210842877152509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587598131846022e-05,
+      "loss": 0.9976,
+      "step": 11691
+    },
+    {
+      "epoch": 0.3211117519430941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587526154074247e-05,
+      "loss": 1.0348,
+      "step": 11692
+    },
+    {
+      "epoch": 0.3211392161709374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587454171653737e-05,
+      "loss": 0.8272,
+      "step": 11693
+    },
+    {
+      "epoch": 0.3211666803987806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5873821845850623e-05,
+      "loss": 0.9335,
+      "step": 11694
+    },
+    {
+      "epoch": 0.3211941446266238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587310192868792e-05,
+      "loss": 0.9639,
+      "step": 11695
+    },
+    {
+      "epoch": 0.32122160885446704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5872381965054966e-05,
+      "loss": 0.9514,
+      "step": 11696
+    },
+    {
+      "epoch": 0.32124907308231027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5871661954957446e-05,
+      "loss": 0.9677,
+      "step": 11697
+    },
+    {
+      "epoch": 0.32127653731015354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5870941898401063e-05,
+      "loss": 0.9415,
+      "step": 11698
+    },
+    {
+      "epoch": 0.32130400153799676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5870221795391514e-05,
+      "loss": 0.8859,
+      "step": 11699
+    },
+    {
+      "epoch": 0.32133146576584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5869501645934495e-05,
+      "loss": 0.9553,
+      "step": 11700
+    },
+    {
+      "epoch": 0.3213589299936832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5868781450035707e-05,
+      "loss": 0.9354,
+      "step": 11701
+    },
+    {
+      "epoch": 0.3213863942215265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5868061207700845e-05,
+      "loss": 0.9828,
+      "step": 11702
+    },
+    {
+      "epoch": 0.3214138584493697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586734091893561e-05,
+      "loss": 0.939,
+      "step": 11703
+    },
+    {
+      "epoch": 0.3214413226772129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58666205837457e-05,
+      "loss": 0.9059,
+      "step": 11704
+    },
+    {
+      "epoch": 0.32146878690505615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586590020213682e-05,
+      "loss": 1.0069,
+      "step": 11705
+    },
+    {
+      "epoch": 0.3214962511328994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5865179774114664e-05,
+      "loss": 1.0311,
+      "step": 11706
+    },
+    {
+      "epoch": 0.32152371536074265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586445929968493e-05,
+      "loss": 0.968,
+      "step": 11707
+    },
+    {
+      "epoch": 0.32155117958858587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5863738778853327e-05,
+      "loss": 0.9648,
+      "step": 11708
+    },
+    {
+      "epoch": 0.3215786438164291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5863018211625547e-05,
+      "loss": 0.9646,
+      "step": 11709
+    },
+    {
+      "epoch": 0.3216061080442723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5862297598007298e-05,
+      "loss": 1.0526,
+      "step": 11710
+    },
+    {
+      "epoch": 0.3216335722721156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586157693800428e-05,
+      "loss": 0.9011,
+      "step": 11711
+    },
+    {
+      "epoch": 0.3216610364999588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586085623162219e-05,
+      "loss": 0.9832,
+      "step": 11712
+    },
+    {
+      "epoch": 0.32168850072780203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5860135478866743e-05,
+      "loss": 0.9046,
+      "step": 11713
+    },
+    {
+      "epoch": 0.32171596495564525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5859414679743626e-05,
+      "loss": 1.0051,
+      "step": 11714
+    },
+    {
+      "epoch": 0.3217434291834885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5858693834258557e-05,
+      "loss": 1.0789,
+      "step": 11715
+    },
+    {
+      "epoch": 0.32177089341133175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585797294241723e-05,
+      "loss": 0.9843,
+      "step": 11716
+    },
+    {
+      "epoch": 0.32179835763917497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5857252004225353e-05,
+      "loss": 0.958,
+      "step": 11717
+    },
+    {
+      "epoch": 0.3218258218670182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5856531019688628e-05,
+      "loss": 0.9622,
+      "step": 11718
+    },
+    {
+      "epoch": 0.32185328609486147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5855809988812764e-05,
+      "loss": 1.0049,
+      "step": 11719
+    },
+    {
+      "epoch": 0.3218807503227047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585508891160346e-05,
+      "loss": 0.9564,
+      "step": 11720
+    },
+    {
+      "epoch": 0.3219082145505479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585436778806642e-05,
+      "loss": 0.9932,
+      "step": 11721
+    },
+    {
+      "epoch": 0.32193567877839113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5853646618207362e-05,
+      "loss": 0.9586,
+      "step": 11722
+    },
+    {
+      "epoch": 0.32196314300623435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585292540203198e-05,
+      "loss": 0.9206,
+      "step": 11723
+    },
+    {
+      "epoch": 0.32199060723407763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585220413954599e-05,
+      "loss": 0.8911,
+      "step": 11724
+    },
+    {
+      "epoch": 0.32201807146192085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585148283075509e-05,
+      "loss": 0.9425,
+      "step": 11725
+    },
+    {
+      "epoch": 0.3220455356897641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585076147566499e-05,
+      "loss": 0.927,
+      "step": 11726
+    },
+    {
+      "epoch": 0.3220729999176073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58500400742814e-05,
+      "loss": 0.941,
+      "step": 11727
+    },
+    {
+      "epoch": 0.32210046414545057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584931862661003e-05,
+      "loss": 0.9367,
+      "step": 11728
+    },
+    {
+      "epoch": 0.3221279283732938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5848597132656586e-05,
+      "loss": 0.9619,
+      "step": 11729
+    },
+    {
+      "epoch": 0.322155392601137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5847875592426772e-05,
+      "loss": 0.9479,
+      "step": 11730
+    },
+    {
+      "epoch": 0.32218285682898024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58471540059263e-05,
+      "loss": 0.9431,
+      "step": 11731
+    },
+    {
+      "epoch": 0.3222103210568235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5846432373160882e-05,
+      "loss": 1.0104,
+      "step": 11732
+    },
+    {
+      "epoch": 0.32223778528466673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584571069413623e-05,
+      "loss": 0.904,
+      "step": 11733
+    },
+    {
+      "epoch": 0.32226524951250995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5844988968858042e-05,
+      "loss": 1.1237,
+      "step": 11734
+    },
+    {
+      "epoch": 0.3222927137403532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5844267197332042e-05,
+      "loss": 0.915,
+      "step": 11735
+    },
+    {
+      "epoch": 0.3223201779681964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584354537956394e-05,
+      "loss": 0.8423,
+      "step": 11736
+    },
+    {
+      "epoch": 0.3223476421960397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584282351555944e-05,
+      "loss": 0.9544,
+      "step": 11737
+    },
+    {
+      "epoch": 0.3223751064238829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5842101605324257e-05,
+      "loss": 0.9337,
+      "step": 11738
+    },
+    {
+      "epoch": 0.3224025706517261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5841379648864104e-05,
+      "loss": 1.0029,
+      "step": 11739
+    },
+    {
+      "epoch": 0.32243003487956934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5840657646184693e-05,
+      "loss": 0.9911,
+      "step": 11740
+    },
+    {
+      "epoch": 0.3224574991074126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5839935597291733e-05,
+      "loss": 0.9641,
+      "step": 11741
+    },
+    {
+      "epoch": 0.32248496333525584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5839213502190945e-05,
+      "loss": 0.9525,
+      "step": 11742
+    },
+    {
+      "epoch": 0.32251242756309906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5838491360888033e-05,
+      "loss": 0.9484,
+      "step": 11743
+    },
+    {
+      "epoch": 0.3225398917909423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5837769173388716e-05,
+      "loss": 0.8706,
+      "step": 11744
+    },
+    {
+      "epoch": 0.32256735601878556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583704693969871e-05,
+      "loss": 0.9381,
+      "step": 11745
+    },
+    {
+      "epoch": 0.3225948202466288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583632465982373e-05,
+      "loss": 0.947,
+      "step": 11746
+    },
+    {
+      "epoch": 0.322622284474472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5835602333769484e-05,
+      "loss": 1.0742,
+      "step": 11747
+    },
+    {
+      "epoch": 0.3226497487023152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5834879961541694e-05,
+      "loss": 0.9197,
+      "step": 11748
+    },
+    {
+      "epoch": 0.32267721293015844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5834157543146075e-05,
+      "loss": 1.0478,
+      "step": 11749
+    },
+    {
+      "epoch": 0.3227046771580017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5833435078588336e-05,
+      "loss": 0.9207,
+      "step": 11750
+    },
+    {
+      "epoch": 0.32273214138584494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5832712567874206e-05,
+      "loss": 0.9442,
+      "step": 11751
+    },
+    {
+      "epoch": 0.32275960561368816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583199001100939e-05,
+      "loss": 0.803,
+      "step": 11752
+    },
+    {
+      "epoch": 0.3227870698415314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583126740799961e-05,
+      "loss": 0.9333,
+      "step": 11753
+    },
+    {
+      "epoch": 0.32281453406937466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5830544758850587e-05,
+      "loss": 0.8939,
+      "step": 11754
+    },
+    {
+      "epoch": 0.3228419982972179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5829822063568033e-05,
+      "loss": 1.0009,
+      "step": 11755
+    },
+    {
+      "epoch": 0.3228694625250611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582909932215767e-05,
+      "loss": 0.9283,
+      "step": 11756
+    },
+    {
+      "epoch": 0.3228969267529043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5828376534625214e-05,
+      "loss": 1.0048,
+      "step": 11757
+    },
+    {
+      "epoch": 0.3229243909807476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5827653700976385e-05,
+      "loss": 0.9261,
+      "step": 11758
+    },
+    {
+      "epoch": 0.3229518552085908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5826930821216902e-05,
+      "loss": 0.9903,
+      "step": 11759
+    },
+    {
+      "epoch": 0.32297931943643404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5826207895352487e-05,
+      "loss": 1.0533,
+      "step": 11760
+    },
+    {
+      "epoch": 0.32300678366427726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5825484923388857e-05,
+      "loss": 0.9351,
+      "step": 11761
+    },
+    {
+      "epoch": 0.32303424789212054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824761905331733e-05,
+      "loss": 0.9846,
+      "step": 11762
+    },
+    {
+      "epoch": 0.32306171211996376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824038841186838e-05,
+      "loss": 0.9543,
+      "step": 11763
+    },
+    {
+      "epoch": 0.323089176347807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582331573095989e-05,
+      "loss": 1.0311,
+      "step": 11764
+    },
+    {
+      "epoch": 0.3231166405756502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5822592574656616e-05,
+      "loss": 0.863,
+      "step": 11765
+    },
+    {
+      "epoch": 0.3231441048034934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582186937228273e-05,
+      "loss": 1.098,
+      "step": 11766
+    },
+    {
+      "epoch": 0.3231715690313367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582114612384396e-05,
+      "loss": 0.9727,
+      "step": 11767
+    },
+    {
+      "epoch": 0.3231990332591799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5820422829346024e-05,
+      "loss": 1.0021,
+      "step": 11768
+    },
+    {
+      "epoch": 0.32322649748702315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5819699488794654e-05,
+      "loss": 1.0445,
+      "step": 11769
+    },
+    {
+      "epoch": 0.32325396171486637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5818976102195565e-05,
+      "loss": 0.9697,
+      "step": 11770
+    },
+    {
+      "epoch": 0.32328142594270964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581825266955448e-05,
+      "loss": 0.9939,
+      "step": 11771
+    },
+    {
+      "epoch": 0.32330889017055287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581752919087713e-05,
+      "loss": 0.9451,
+      "step": 11772
+    },
+    {
+      "epoch": 0.3233363543983961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5816805666169232e-05,
+      "loss": 1.0625,
+      "step": 11773
+    },
+    {
+      "epoch": 0.3233638186262393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5816082095436517e-05,
+      "loss": 0.9323,
+      "step": 11774
+    },
+    {
+      "epoch": 0.3233912828540826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5815358478684705e-05,
+      "loss": 0.9668,
+      "step": 11775
+    },
+    {
+      "epoch": 0.3234187470819258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581463481591953e-05,
+      "loss": 1.0122,
+      "step": 11776
+    },
+    {
+      "epoch": 0.323446211309769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5813911107146704e-05,
+      "loss": 1.0504,
+      "step": 11777
+    },
+    {
+      "epoch": 0.32347367553761225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5813187352371965e-05,
+      "loss": 0.9634,
+      "step": 11778
+    },
+    {
+      "epoch": 0.32350113976545547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5812463551601035e-05,
+      "loss": 0.965,
+      "step": 11779
+    },
+    {
+      "epoch": 0.32352860399329875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5811739704839642e-05,
+      "loss": 0.9361,
+      "step": 11780
+    },
+    {
+      "epoch": 0.32355606822114197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5811015812093515e-05,
+      "loss": 0.8896,
+      "step": 11781
+    },
+    {
+      "epoch": 0.3235835324489852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5810291873368376e-05,
+      "loss": 0.9331,
+      "step": 11782
+    },
+    {
+      "epoch": 0.3236109966768284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580956788866996e-05,
+      "loss": 1.0029,
+      "step": 11783
+    },
+    {
+      "epoch": 0.3236384609046717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5808843858003994e-05,
+      "loss": 0.933,
+      "step": 11784
+    },
+    {
+      "epoch": 0.3236659251325149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5808119781376202e-05,
+      "loss": 0.9544,
+      "step": 11785
+    },
+    {
+      "epoch": 0.32369338936035813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580739565879232e-05,
+      "loss": 1.0087,
+      "step": 11786
+    },
+    {
+      "epoch": 0.32372085358820135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5806671490258072e-05,
+      "loss": 0.868,
+      "step": 11787
+    },
+    {
+      "epoch": 0.32374831781604463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580594727577919e-05,
+      "loss": 0.9025,
+      "step": 11788
+    },
+    {
+      "epoch": 0.32377578204388785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5805223015361402e-05,
+      "loss": 1.0,
+      "step": 11789
+    },
+    {
+      "epoch": 0.32380324627173107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5804498709010445e-05,
+      "loss": 1.0176,
+      "step": 11790
+    },
+    {
+      "epoch": 0.3238307104995743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5803774356732044e-05,
+      "loss": 0.9647,
+      "step": 11791
+    },
+    {
+      "epoch": 0.3238581747274175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580304995853193e-05,
+      "loss": 1.045,
+      "step": 11792
+    },
+    {
+      "epoch": 0.3238856389552608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5802325514415836e-05,
+      "loss": 1.001,
+      "step": 11793
+    },
+    {
+      "epoch": 0.323913103183104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5801601024389498e-05,
+      "loss": 0.9738,
+      "step": 11794
+    },
+    {
+      "epoch": 0.32394056741094723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5800876488458646e-05,
+      "loss": 1.0202,
+      "step": 11795
+    },
+    {
+      "epoch": 0.32396803163879045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580015190662901e-05,
+      "loss": 0.932,
+      "step": 11796
+    },
+    {
+      "epoch": 0.32399549586663373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5799427278906327e-05,
+      "loss": 0.8655,
+      "step": 11797
+    },
+    {
+      "epoch": 0.32402296009447695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5798702605296328e-05,
+      "loss": 0.9569,
+      "step": 11798
+    },
+    {
+      "epoch": 0.3240504243223202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5797977885804746e-05,
+      "loss": 0.9999,
+      "step": 11799
+    },
+    {
+      "epoch": 0.3240778885501634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.579725312043732e-05,
+      "loss": 1.0196,
+      "step": 11800
+    },
+    {
+      "epoch": 0.3241053527780067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5796528309199782e-05,
+      "loss": 0.9767,
+      "step": 11801
+    },
+    {
+      "epoch": 0.3241328170058499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5795803452097863e-05,
+      "loss": 0.9586,
+      "step": 11802
+    },
+    {
+      "epoch": 0.3241602812336931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5795078549137305e-05,
+      "loss": 0.9534,
+      "step": 11803
+    },
+    {
+      "epoch": 0.32418774546153634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.579435360032384e-05,
+      "loss": 0.935,
+      "step": 11804
+    },
+    {
+      "epoch": 0.32421520968937956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5793628605663205e-05,
+      "loss": 1.0392,
+      "step": 11805
+    },
+    {
+      "epoch": 0.32424267391722283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5792903565161135e-05,
+      "loss": 0.9995,
+      "step": 11806
+    },
+    {
+      "epoch": 0.32427013814506606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5792178478823372e-05,
+      "loss": 0.945,
+      "step": 11807
+    },
+    {
+      "epoch": 0.3242976023729093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5791453346655644e-05,
+      "loss": 0.9638,
+      "step": 11808
+    },
+    {
+      "epoch": 0.3243250666007525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5790728168663695e-05,
+      "loss": 1.017,
+      "step": 11809
+    },
+    {
+      "epoch": 0.3243525308285958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5790002944853264e-05,
+      "loss": 0.948,
+      "step": 11810
+    },
+    {
+      "epoch": 0.324379995056439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5789277675230087e-05,
+      "loss": 0.999,
+      "step": 11811
+    },
+    {
+      "epoch": 0.3244074592842822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57885523597999e-05,
+      "loss": 1.0329,
+      "step": 11812
+    },
+    {
+      "epoch": 0.32443492351212544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5787826998568445e-05,
+      "loss": 0.8154,
+      "step": 11813
+    },
+    {
+      "epoch": 0.3244623877399687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578710159154146e-05,
+      "loss": 0.9157,
+      "step": 11814
+    },
+    {
+      "epoch": 0.32448985196781194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5786376138724687e-05,
+      "loss": 0.9041,
+      "step": 11815
+    },
+    {
+      "epoch": 0.32451731619565516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5785650640123862e-05,
+      "loss": 0.9776,
+      "step": 11816
+    },
+    {
+      "epoch": 0.3245447804234984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5784925095744732e-05,
+      "loss": 0.9682,
+      "step": 11817
+    },
+    {
+      "epoch": 0.3245722446513416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5784199505593032e-05,
+      "loss": 0.9966,
+      "step": 11818
+    },
+    {
+      "epoch": 0.3245997088791849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5783473869674506e-05,
+      "loss": 0.8736,
+      "step": 11819
+    },
+    {
+      "epoch": 0.3246271731070281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5782748187994892e-05,
+      "loss": 1.03,
+      "step": 11820
+    },
+    {
+      "epoch": 0.3246546373348713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5782022460559937e-05,
+      "loss": 1.0265,
+      "step": 11821
+    },
+    {
+      "epoch": 0.32468210156271454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578129668737538e-05,
+      "loss": 0.994,
+      "step": 11822
+    },
+    {
+      "epoch": 0.3247095657905578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5780570868446964e-05,
+      "loss": 0.8897,
+      "step": 11823
+    },
+    {
+      "epoch": 0.32473703001840104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577984500378043e-05,
+      "loss": 0.9873,
+      "step": 11824
+    },
+    {
+      "epoch": 0.32476449424624426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5779119093381525e-05,
+      "loss": 1.0374,
+      "step": 11825
+    },
+    {
+      "epoch": 0.3247919584740875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577839313725599e-05,
+      "loss": 1.0024,
+      "step": 11826
+    },
+    {
+      "epoch": 0.32481942270193076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577766713540957e-05,
+      "loss": 0.923,
+      "step": 11827
+    },
+    {
+      "epoch": 0.324846886929774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776941087848013e-05,
+      "loss": 1.1179,
+      "step": 11828
+    },
+    {
+      "epoch": 0.3248743511576172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776214994577055e-05,
+      "loss": 0.9823,
+      "step": 11829
+    },
+    {
+      "epoch": 0.3249018153854604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577548885560245e-05,
+      "loss": 0.9442,
+      "step": 11830
+    },
+    {
+      "epoch": 0.32492927961330365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5774762670929942e-05,
+      "loss": 0.9963,
+      "step": 11831
+    },
+    {
+      "epoch": 0.3249567438411469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5774036440565272e-05,
+      "loss": 0.9031,
+      "step": 11832
+    },
+    {
+      "epoch": 0.32498420806899014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577331016451419e-05,
+      "loss": 1.0314,
+      "step": 11833
+    },
+    {
+      "epoch": 0.32501167229683336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577258384278244e-05,
+      "loss": 0.8897,
+      "step": 11834
+    },
+    {
+      "epoch": 0.3250391365246766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577185747537577e-05,
+      "loss": 0.9615,
+      "step": 11835
+    },
+    {
+      "epoch": 0.32506660075251986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577113106229993e-05,
+      "loss": 0.9371,
+      "step": 11836
+    },
+    {
+      "epoch": 0.3250940649803631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5770404603560663e-05,
+      "loss": 0.8715,
+      "step": 11837
+    },
+    {
+      "epoch": 0.3251215292082063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5769678099163725e-05,
+      "loss": 1.0456,
+      "step": 11838
+    },
+    {
+      "epoch": 0.3251489934360495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5768951549114855e-05,
+      "loss": 0.9341,
+      "step": 11839
+    },
+    {
+      "epoch": 0.3251764576638928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5768224953419806e-05,
+      "loss": 0.9499,
+      "step": 11840
+    },
+    {
+      "epoch": 0.325203921891736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5767498312084325e-05,
+      "loss": 0.9915,
+      "step": 11841
+    },
+    {
+      "epoch": 0.32523138611957925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5766771625114164e-05,
+      "loss": 0.9895,
+      "step": 11842
+    },
+    {
+      "epoch": 0.32525885034742247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5766044892515074e-05,
+      "loss": 1.0573,
+      "step": 11843
+    },
+    {
+      "epoch": 0.3252863145752657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5765318114292804e-05,
+      "loss": 0.973,
+      "step": 11844
+    },
+    {
+      "epoch": 0.32531377880310897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57645912904531e-05,
+      "loss": 1.0391,
+      "step": 11845
+    },
+    {
+      "epoch": 0.3253412430309522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5763864421001722e-05,
+      "loss": 1.0271,
+      "step": 11846
+    },
+    {
+      "epoch": 0.3253687072587954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5763137505944414e-05,
+      "loss": 0.8317,
+      "step": 11847
+    },
+    {
+      "epoch": 0.32539617148663863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5762410545286927e-05,
+      "loss": 0.9473,
+      "step": 11848
+    },
+    {
+      "epoch": 0.3254236357144819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576168353903502e-05,
+      "loss": 0.8928,
+      "step": 11849
+    },
+    {
+      "epoch": 0.32545109994232513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5760956487194435e-05,
+      "loss": 0.923,
+      "step": 11850
+    },
+    {
+      "epoch": 0.32547856417016835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576022938977093e-05,
+      "loss": 0.8873,
+      "step": 11851
+    },
+    {
+      "epoch": 0.32550602839801157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5759502246770265e-05,
+      "loss": 0.9714,
+      "step": 11852
+    },
+    {
+      "epoch": 0.32553349262585485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5758775058198184e-05,
+      "loss": 1.0339,
+      "step": 11853
+    },
+    {
+      "epoch": 0.32556095685369807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5758047824060447e-05,
+      "loss": 0.9297,
+      "step": 11854
+    },
+    {
+      "epoch": 0.3255884210815413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5757320544362804e-05,
+      "loss": 0.8452,
+      "step": 11855
+    },
+    {
+      "epoch": 0.3256158853093845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575659321911101e-05,
+      "loss": 1.0126,
+      "step": 11856
+    },
+    {
+      "epoch": 0.32564334953722773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575586584831082e-05,
+      "loss": 0.9378,
+      "step": 11857
+    },
+    {
+      "epoch": 0.325670813765071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575513843196799e-05,
+      "loss": 0.8658,
+      "step": 11858
+    },
+    {
+      "epoch": 0.32569827799291423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5754410970088274e-05,
+      "loss": 1.0087,
+      "step": 11859
+    },
+    {
+      "epoch": 0.32572574222075745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575368346267743e-05,
+      "loss": 0.9941,
+      "step": 11860
+    },
+    {
+      "epoch": 0.3257532064486007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5752955909741215e-05,
+      "loss": 0.9687,
+      "step": 11861
+    },
+    {
+      "epoch": 0.32578067067644395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5752228311285382e-05,
+      "loss": 1.0095,
+      "step": 11862
+    },
+    {
+      "epoch": 0.32580813490428717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5751500667315697e-05,
+      "loss": 0.8959,
+      "step": 11863
+    },
+    {
+      "epoch": 0.3258355991321304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5750772977837903e-05,
+      "loss": 0.9461,
+      "step": 11864
+    },
+    {
+      "epoch": 0.3258630633599736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5750045242857766e-05,
+      "loss": 0.954,
+      "step": 11865
+    },
+    {
+      "epoch": 0.3258905275878169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574931746238105e-05,
+      "loss": 0.8973,
+      "step": 11866
+    },
+    {
+      "epoch": 0.3259179918156601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5748589636413502e-05,
+      "loss": 0.9655,
+      "step": 11867
+    },
+    {
+      "epoch": 0.32594545604350333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5747861764960885e-05,
+      "loss": 1.0492,
+      "step": 11868
+    },
+    {
+      "epoch": 0.32597292027134656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5747133848028964e-05,
+      "loss": 0.9804,
+      "step": 11869
+    },
+    {
+      "epoch": 0.32600038449918983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5746405885623493e-05,
+      "loss": 1.0428,
+      "step": 11870
+    },
+    {
+      "epoch": 0.32602784872703305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574567787775023e-05,
+      "loss": 1.0755,
+      "step": 11871
+    },
+    {
+      "epoch": 0.3260553129548763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5744949824414938e-05,
+      "loss": 0.9522,
+      "step": 11872
+    },
+    {
+      "epoch": 0.3260827771827195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574422172562338e-05,
+      "loss": 0.9975,
+      "step": 11873
+    },
+    {
+      "epoch": 0.3261102414105627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574349358138131e-05,
+      "loss": 0.989,
+      "step": 11874
+    },
+    {
+      "epoch": 0.326137705638406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5742765391694498e-05,
+      "loss": 0.9913,
+      "step": 11875
+    },
+    {
+      "epoch": 0.3261651698662492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57420371565687e-05,
+      "loss": 0.9121,
+      "step": 11876
+    },
+    {
+      "epoch": 0.32619263409409244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5741308876009683e-05,
+      "loss": 0.9763,
+      "step": 11877
+    },
+    {
+      "epoch": 0.32622009832193566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5740580550023204e-05,
+      "loss": 0.9765,
+      "step": 11878
+    },
+    {
+      "epoch": 0.32624756254977894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5739852178615027e-05,
+      "loss": 1.0249,
+      "step": 11879
+    },
+    {
+      "epoch": 0.32627502677762216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573912376179092e-05,
+      "loss": 0.9778,
+      "step": 11880
+    },
+    {
+      "epoch": 0.3263024910054654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573839529955664e-05,
+      "loss": 1.0304,
+      "step": 11881
+    },
+    {
+      "epoch": 0.3263299552333086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573766679191795e-05,
+      "loss": 0.9743,
+      "step": 11882
+    },
+    {
+      "epoch": 0.3263574194611519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573693823888062e-05,
+      "loss": 0.9811,
+      "step": 11883
+    },
+    {
+      "epoch": 0.3263848836889951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573620964045042e-05,
+      "loss": 0.8923,
+      "step": 11884
+    },
+    {
+      "epoch": 0.3264123479168383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5735480996633095e-05,
+      "loss": 0.9368,
+      "step": 11885
+    },
+    {
+      "epoch": 0.32643981214468154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573475230743443e-05,
+      "loss": 1.0118,
+      "step": 11886
+    },
+    {
+      "epoch": 0.32646727637252476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573402357286018e-05,
+      "loss": 0.9866,
+      "step": 11887
+    },
+    {
+      "epoch": 0.32649474060036804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5733294792916115e-05,
+      "loss": 0.9032,
+      "step": 11888
+    },
+    {
+      "epoch": 0.32652220482821126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5732565967608003e-05,
+      "loss": 0.9038,
+      "step": 11889
+    },
+    {
+      "epoch": 0.3265496690560545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5731837096941608e-05,
+      "loss": 0.9889,
+      "step": 11890
+    },
+    {
+      "epoch": 0.3265771332838977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5731108180922695e-05,
+      "loss": 1.0426,
+      "step": 11891
+    },
+    {
+      "epoch": 0.326604597511741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5730379219557035e-05,
+      "loss": 1.0471,
+      "step": 11892
+    },
+    {
+      "epoch": 0.3266320617395842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5729650212850397e-05,
+      "loss": 1.0095,
+      "step": 11893
+    },
+    {
+      "epoch": 0.3266595259674274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5728921160808546e-05,
+      "loss": 0.9445,
+      "step": 11894
+    },
+    {
+      "epoch": 0.32668699019527064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5728192063437253e-05,
+      "loss": 0.9398,
+      "step": 11895
+    },
+    {
+      "epoch": 0.3267144544231139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5727462920742284e-05,
+      "loss": 1.0129,
+      "step": 11896
+    },
+    {
+      "epoch": 0.32674191865095714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572673373272941e-05,
+      "loss": 1.0138,
+      "step": 11897
+    },
+    {
+      "epoch": 0.32676938287880036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5726004499404397e-05,
+      "loss": 0.9525,
+      "step": 11898
+    },
+    {
+      "epoch": 0.3267968471066436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5725275220773024e-05,
+      "loss": 0.9119,
+      "step": 11899
+    },
+    {
+      "epoch": 0.3268243113344868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572454589684105e-05,
+      "loss": 1.0199,
+      "step": 11900
+    },
+    {
+      "epoch": 0.3268517755623301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5723816527614255e-05,
+      "loss": 1.0242,
+      "step": 11901
+    },
+    {
+      "epoch": 0.3268792397901733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572308711309841e-05,
+      "loss": 1.027,
+      "step": 11902
+    },
+    {
+      "epoch": 0.3269067040180165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5722357653299276e-05,
+      "loss": 1.0328,
+      "step": 11903
+    },
+    {
+      "epoch": 0.32693416824585975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5721628148222636e-05,
+      "loss": 1.0156,
+      "step": 11904
+    },
+    {
+      "epoch": 0.326961632473703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5720898597874254e-05,
+      "loss": 0.9587,
+      "step": 11905
+    },
+    {
+      "epoch": 0.32698909670154624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572016900225991e-05,
+      "loss": 0.8841,
+      "step": 11906
+    },
+    {
+      "epoch": 0.32701656092938947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571943936138537e-05,
+      "loss": 0.8783,
+      "step": 11907
+    },
+    {
+      "epoch": 0.3270440251572327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5718709675256414e-05,
+      "loss": 0.9252,
+      "step": 11908
+    },
+    {
+      "epoch": 0.32707148938507596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571797994387881e-05,
+      "loss": 0.9652,
+      "step": 11909
+    },
+    {
+      "epoch": 0.3270989536129192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571725016725833e-05,
+      "loss": 1.035,
+      "step": 11910
+    },
+    {
+      "epoch": 0.3271264178407624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5716520345400756e-05,
+      "loss": 0.8818,
+      "step": 11911
+    },
+    {
+      "epoch": 0.32715388206860563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571579047831186e-05,
+      "loss": 1.0002,
+      "step": 11912
+    },
+    {
+      "epoch": 0.32718134629644885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5715060565997412e-05,
+      "loss": 0.9666,
+      "step": 11913
+    },
+    {
+      "epoch": 0.3272088105242921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5714330608463195e-05,
+      "loss": 0.929,
+      "step": 11914
+    },
+    {
+      "epoch": 0.32723627475213535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5713600605714976e-05,
+      "loss": 0.9279,
+      "step": 11915
+    },
+    {
+      "epoch": 0.32726373897997857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571287055775854e-05,
+      "loss": 1.0317,
+      "step": 11916
+    },
+    {
+      "epoch": 0.3272912032078218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5712140464599656e-05,
+      "loss": 1.0012,
+      "step": 11917
+    },
+    {
+      "epoch": 0.32731866743566507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5711410326244105e-05,
+      "loss": 1.0109,
+      "step": 11918
+    },
+    {
+      "epoch": 0.3273461316635083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5710680142697666e-05,
+      "loss": 0.9146,
+      "step": 11919
+    },
+    {
+      "epoch": 0.3273735958913515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5709949913966108e-05,
+      "loss": 0.887,
+      "step": 11920
+    },
+    {
+      "epoch": 0.32740106011919473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570921964005522e-05,
+      "loss": 0.9666,
+      "step": 11921
+    },
+    {
+      "epoch": 0.327428524347038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5708489320970772e-05,
+      "loss": 0.9526,
+      "step": 11922
+    },
+    {
+      "epoch": 0.32745598857488123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5707758956718544e-05,
+      "loss": 0.9644,
+      "step": 11923
+    },
+    {
+      "epoch": 0.32748345280272445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570702854730432e-05,
+      "loss": 0.8984,
+      "step": 11924
+    },
+    {
+      "epoch": 0.32751091703056767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5706298092733872e-05,
+      "loss": 0.9868,
+      "step": 11925
+    },
+    {
+      "epoch": 0.3275383812584109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570556759301298e-05,
+      "loss": 1.0928,
+      "step": 11926
+    },
+    {
+      "epoch": 0.32756584548625417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5704837048147433e-05,
+      "loss": 0.9119,
+      "step": 11927
+    },
+    {
+      "epoch": 0.3275933097140974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5704106458143005e-05,
+      "loss": 0.8904,
+      "step": 11928
+    },
+    {
+      "epoch": 0.3276207739419406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5703375823005476e-05,
+      "loss": 0.8927,
+      "step": 11929
+    },
+    {
+      "epoch": 0.32764823816978383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5702645142740628e-05,
+      "loss": 0.9975,
+      "step": 11930
+    },
+    {
+      "epoch": 0.3276757023976271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5701914417354242e-05,
+      "loss": 0.9322,
+      "step": 11931
+    },
+    {
+      "epoch": 0.32770316662547033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57011836468521e-05,
+      "loss": 0.8918,
+      "step": 11932
+    },
+    {
+      "epoch": 0.32773063085331355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5700452831239986e-05,
+      "loss": 0.9478,
+      "step": 11933
+    },
+    {
+      "epoch": 0.3277580950811568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5699721970523682e-05,
+      "loss": 0.9145,
+      "step": 11934
+    },
+    {
+      "epoch": 0.32778555930900005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5698991064708965e-05,
+      "loss": 0.9055,
+      "step": 11935
+    },
+    {
+      "epoch": 0.3278130235368433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569826011380163e-05,
+      "loss": 0.9347,
+      "step": 11936
+    },
+    {
+      "epoch": 0.3278404877646865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569752911780745e-05,
+      "loss": 0.9514,
+      "step": 11937
+    },
+    {
+      "epoch": 0.3278679519925297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5696798076732213e-05,
+      "loss": 0.9657,
+      "step": 11938
+    },
+    {
+      "epoch": 0.32789541622037294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56960669905817e-05,
+      "loss": 0.9996,
+      "step": 11939
+    },
+    {
+      "epoch": 0.3279228804482162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5695335859361706e-05,
+      "loss": 0.9679,
+      "step": 11940
+    },
+    {
+      "epoch": 0.32795034467605944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5694604683078002e-05,
+      "loss": 1.0612,
+      "step": 11941
+    },
+    {
+      "epoch": 0.32797780890390266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569387346173638e-05,
+      "loss": 0.958,
+      "step": 11942
+    },
+    {
+      "epoch": 0.3280052731317459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569314219534263e-05,
+      "loss": 0.7393,
+      "step": 11943
+    },
+    {
+      "epoch": 0.32803273735958915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569241088390253e-05,
+      "loss": 1.0543,
+      "step": 11944
+    },
+    {
+      "epoch": 0.3280602015874324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5691679527421866e-05,
+      "loss": 1.0478,
+      "step": 11945
+    },
+    {
+      "epoch": 0.3280876658152756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569094812590644e-05,
+      "loss": 0.9051,
+      "step": 11946
+    },
+    {
+      "epoch": 0.3281151300431188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5690216679362018e-05,
+      "loss": 0.972,
+      "step": 11947
+    },
+    {
+      "epoch": 0.3281425942709621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5689485187794403e-05,
+      "loss": 1.071,
+      "step": 11948
+    },
+    {
+      "epoch": 0.3281700584988053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5688753651209375e-05,
+      "loss": 0.9924,
+      "step": 11949
+    },
+    {
+      "epoch": 0.32819752272664854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5688022069612725e-05,
+      "loss": 1.0327,
+      "step": 11950
+    },
+    {
+      "epoch": 0.32822498695449176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568729044301024e-05,
+      "loss": 0.9221,
+      "step": 11951
+    },
+    {
+      "epoch": 0.328252451182335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5686558771407713e-05,
+      "loss": 0.9675,
+      "step": 11952
+    },
+    {
+      "epoch": 0.32827991541017826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5685827054810928e-05,
+      "loss": 0.9202,
+      "step": 11953
+    },
+    {
+      "epoch": 0.3283073796380215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5685095293225676e-05,
+      "loss": 0.8664,
+      "step": 11954
+    },
+    {
+      "epoch": 0.3283348438658647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568436348665775e-05,
+      "loss": 0.8975,
+      "step": 11955
+    },
+    {
+      "epoch": 0.3283623080937079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5683631635112935e-05,
+      "loss": 0.9639,
+      "step": 11956
+    },
+    {
+      "epoch": 0.3283897723215512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568289973859703e-05,
+      "loss": 0.9954,
+      "step": 11957
+    },
+    {
+      "epoch": 0.3284172365493944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568216779711582e-05,
+      "loss": 0.9788,
+      "step": 11958
+    },
+    {
+      "epoch": 0.32844470077723764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5681435810675092e-05,
+      "loss": 0.9454,
+      "step": 11959
+    },
+    {
+      "epoch": 0.32847216500508086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5680703779280647e-05,
+      "loss": 0.9182,
+      "step": 11960
+    },
+    {
+      "epoch": 0.32849962923292414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5679971702938273e-05,
+      "loss": 0.9758,
+      "step": 11961
+    },
+    {
+      "epoch": 0.32852709346076736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567923958165376e-05,
+      "loss": 0.9268,
+      "step": 11962
+    },
+    {
+      "epoch": 0.3285545576886106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5678507415432908e-05,
+      "loss": 0.9562,
+      "step": 11963
+    },
+    {
+      "epoch": 0.3285820219164538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5677775204281503e-05,
+      "loss": 0.9657,
+      "step": 11964
+    },
+    {
+      "epoch": 0.3286094861442971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567704294820534e-05,
+      "loss": 0.9109,
+      "step": 11965
+    },
+    {
+      "epoch": 0.3286369503721403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5676310647210216e-05,
+      "loss": 1.032,
+      "step": 11966
+    },
+    {
+      "epoch": 0.3286644145999835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5675578301301924e-05,
+      "loss": 0.9273,
+      "step": 11967
+    },
+    {
+      "epoch": 0.32869187882782674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567484591048626e-05,
+      "loss": 1.0274,
+      "step": 11968
+    },
+    {
+      "epoch": 0.32871934305566997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567411347476901e-05,
+      "loss": 1.015,
+      "step": 11969
+    },
+    {
+      "epoch": 0.32874680728351324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5673380994155984e-05,
+      "loss": 0.9021,
+      "step": 11970
+    },
+    {
+      "epoch": 0.32877427151135646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5672648468652963e-05,
+      "loss": 0.9567,
+      "step": 11971
+    },
+    {
+      "epoch": 0.3288017357391997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5671915898265755e-05,
+      "loss": 0.9819,
+      "step": 11972
+    },
+    {
+      "epoch": 0.3288291999670429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5671183283000152e-05,
+      "loss": 0.9597,
+      "step": 11973
+    },
+    {
+      "epoch": 0.3288566641948862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5670450622861942e-05,
+      "loss": 0.9177,
+      "step": 11974
+    },
+    {
+      "epoch": 0.3288841284227294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566971791785694e-05,
+      "loss": 0.855,
+      "step": 11975
+    },
+    {
+      "epoch": 0.3289115926505726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5668985167990927e-05,
+      "loss": 1.0446,
+      "step": 11976
+    },
+    {
+      "epoch": 0.32893905687841585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566825237326971e-05,
+      "loss": 0.9773,
+      "step": 11977
+    },
+    {
+      "epoch": 0.3289665211062591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5667519533699086e-05,
+      "loss": 0.8222,
+      "step": 11978
+    },
+    {
+      "epoch": 0.32899398533410235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5666786649284848e-05,
+      "loss": 0.8558,
+      "step": 11979
+    },
+    {
+      "epoch": 0.32902144956194557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5666053720032802e-05,
+      "loss": 0.9619,
+      "step": 11980
+    },
+    {
+      "epoch": 0.3290489137897888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5665320745948745e-05,
+      "loss": 0.8958,
+      "step": 11981
+    },
+    {
+      "epoch": 0.329076378017632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5664587727038472e-05,
+      "loss": 0.9615,
+      "step": 11982
+    },
+    {
+      "epoch": 0.3291038422454753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566385466330779e-05,
+      "loss": 0.9043,
+      "step": 11983
+    },
+    {
+      "epoch": 0.3291313064733185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5663121554762494e-05,
+      "loss": 0.9783,
+      "step": 11984
+    },
+    {
+      "epoch": 0.32915877070116173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5662388401408387e-05,
+      "loss": 0.901,
+      "step": 11985
+    },
+    {
+      "epoch": 0.32918623492900495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5661655203251268e-05,
+      "loss": 1.0017,
+      "step": 11986
+    },
+    {
+      "epoch": 0.3292136991568482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5660921960296948e-05,
+      "loss": 1.007,
+      "step": 11987
+    },
+    {
+      "epoch": 0.32924116338469145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566018867255121e-05,
+      "loss": 0.9081,
+      "step": 11988
+    },
+    {
+      "epoch": 0.32926862761253467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565945534001987e-05,
+      "loss": 1.0186,
+      "step": 11989
+    },
+    {
+      "epoch": 0.3292960918403779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565872196270873e-05,
+      "loss": 0.8763,
+      "step": 11990
+    },
+    {
+      "epoch": 0.32932355606822117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565798854062359e-05,
+      "loss": 0.9429,
+      "step": 11991
+    },
+    {
+      "epoch": 0.3293510202960644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565725507377025e-05,
+      "loss": 0.9583,
+      "step": 11992
+    },
+    {
+      "epoch": 0.3293784845239076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565652156215452e-05,
+      "loss": 0.9572,
+      "step": 11993
+    },
+    {
+      "epoch": 0.32940594875175083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5655788005782193e-05,
+      "loss": 1.0136,
+      "step": 11994
+    },
+    {
+      "epoch": 0.32943341297959405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5655054404659087e-05,
+      "loss": 0.84,
+      "step": 11995
+    },
+    {
+      "epoch": 0.32946087720743733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5654320758790998e-05,
+      "loss": 0.9579,
+      "step": 11996
+    },
+    {
+      "epoch": 0.32948834143528055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565358706818373e-05,
+      "loss": 1.0356,
+      "step": 11997
+    },
+    {
+      "epoch": 0.3295158056631238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5652853332843096e-05,
+      "loss": 0.9994,
+      "step": 11998
+    },
+    {
+      "epoch": 0.329543269890967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5652119552774895e-05,
+      "loss": 1.0015,
+      "step": 11999
+    },
+    {
+      "epoch": 0.32957073411881027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5651385727984936e-05,
+      "loss": 0.8984,
+      "step": 12000
+    },
+    {
+      "epoch": 0.3295981983466535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5650651858479024e-05,
+      "loss": 0.9573,
+      "step": 12001
+    },
+    {
+      "epoch": 0.3296256625744967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564991794426296e-05,
+      "loss": 0.9397,
+      "step": 12002
+    },
+    {
+      "epoch": 0.32965312680233994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564918398534256e-05,
+      "loss": 0.8946,
+      "step": 12003
+    },
+    {
+      "epoch": 0.3296805910301832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564844998172363e-05,
+      "loss": 0.9198,
+      "step": 12004
+    },
+    {
+      "epoch": 0.32970805525802643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5647715933411975e-05,
+      "loss": 0.8992,
+      "step": 12005
+    },
+    {
+      "epoch": 0.32973551948586965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5646981840413404e-05,
+      "loss": 0.9371,
+      "step": 12006
+    },
+    {
+      "epoch": 0.3297629837137129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5646247702733727e-05,
+      "loss": 1.0408,
+      "step": 12007
+    },
+    {
+      "epoch": 0.3297904479415561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5645513520378747e-05,
+      "loss": 1.0412,
+      "step": 12008
+    },
+    {
+      "epoch": 0.3298179121693994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5644779293354278e-05,
+      "loss": 0.9918,
+      "step": 12009
+    },
+    {
+      "epoch": 0.3298453763972426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564404502166613e-05,
+      "loss": 0.9383,
+      "step": 12010
+    },
+    {
+      "epoch": 0.3298728406250858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564331070532011e-05,
+      "loss": 0.9358,
+      "step": 12011
+    },
+    {
+      "epoch": 0.32990030485292904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5642576344322033e-05,
+      "loss": 0.9192,
+      "step": 12012
+    },
+    {
+      "epoch": 0.3299277690807723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5641841938677703e-05,
+      "loss": 0.9477,
+      "step": 12013
+    },
+    {
+      "epoch": 0.32995523330861554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5641107488392937e-05,
+      "loss": 0.9392,
+      "step": 12014
+    },
+    {
+      "epoch": 0.32998269753645876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564037299347354e-05,
+      "loss": 0.9872,
+      "step": 12015
+    },
+    {
+      "epoch": 0.330010161764302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5639638453925328e-05,
+      "loss": 1.0247,
+      "step": 12016
+    },
+    {
+      "epoch": 0.33003762599214526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5638903869754113e-05,
+      "loss": 0.975,
+      "step": 12017
+    },
+    {
+      "epoch": 0.3300650902199885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5638169240965706e-05,
+      "loss": 1.0446,
+      "step": 12018
+    },
+    {
+      "epoch": 0.3300925544478317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563743456756592e-05,
+      "loss": 1.0099,
+      "step": 12019
+    },
+    {
+      "epoch": 0.3301200186756749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5636699849560568e-05,
+      "loss": 0.9654,
+      "step": 12020
+    },
+    {
+      "epoch": 0.33014748290351814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5635965086955462e-05,
+      "loss": 0.897,
+      "step": 12021
+    },
+    {
+      "epoch": 0.3301749471313614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5635230279756417e-05,
+      "loss": 0.9403,
+      "step": 12022
+    },
+    {
+      "epoch": 0.33020241135920464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563449542796925e-05,
+      "loss": 0.9535,
+      "step": 12023
+    },
+    {
+      "epoch": 0.33022987558704786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5633760531599768e-05,
+      "loss": 0.8589,
+      "step": 12024
+    },
+    {
+      "epoch": 0.3302573398148911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5633025590653793e-05,
+      "loss": 0.8548,
+      "step": 12025
+    },
+    {
+      "epoch": 0.33028480404273436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5632290605137138e-05,
+      "loss": 0.9596,
+      "step": 12026
+    },
+    {
+      "epoch": 0.3303122682705776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563155557505562e-05,
+      "loss": 1.004,
+      "step": 12027
+    },
+    {
+      "epoch": 0.3303397324984208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563082050041505e-05,
+      "loss": 0.9451,
+      "step": 12028
+    },
+    {
+      "epoch": 0.330367196726264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5630085381221246e-05,
+      "loss": 0.973,
+      "step": 12029
+    },
+    {
+      "epoch": 0.3303946609541073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5629350217480025e-05,
+      "loss": 0.8985,
+      "step": 12030
+    },
+    {
+      "epoch": 0.3304221251819505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5628615009197208e-05,
+      "loss": 0.9006,
+      "step": 12031
+    },
+    {
+      "epoch": 0.33044958940979374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562787975637861e-05,
+      "loss": 0.9057,
+      "step": 12032
+    },
+    {
+      "epoch": 0.33047705363763696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5627144459030043e-05,
+      "loss": 1.0251,
+      "step": 12033
+    },
+    {
+      "epoch": 0.3305045178654802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562640911715733e-05,
+      "loss": 1.0207,
+      "step": 12034
+    },
+    {
+      "epoch": 0.33053198209332346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5625673730766292e-05,
+      "loss": 0.9919,
+      "step": 12035
+    },
+    {
+      "epoch": 0.3305594463211667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562493829986274e-05,
+      "loss": 0.9727,
+      "step": 12036
+    },
+    {
+      "epoch": 0.3305869105490099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56242028244525e-05,
+      "loss": 1.0011,
+      "step": 12037
+    },
+    {
+      "epoch": 0.3306143747768531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562346730454139e-05,
+      "loss": 0.9228,
+      "step": 12038
+    },
+    {
+      "epoch": 0.3306418390046964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562273174013522e-05,
+      "loss": 1.0186,
+      "step": 12039
+    },
+    {
+      "epoch": 0.3306693032325396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5621996131239827e-05,
+      "loss": 0.9426,
+      "step": 12040
+    },
+    {
+      "epoch": 0.33069676746038285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562126047786102e-05,
+      "loss": 1.0547,
+      "step": 12041
+    },
+    {
+      "epoch": 0.33072423168822607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5620524780004624e-05,
+      "loss": 1.0877,
+      "step": 12042
+    },
+    {
+      "epoch": 0.33075169591606934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5619789037676458e-05,
+      "loss": 0.9658,
+      "step": 12043
+    },
+    {
+      "epoch": 0.33077916014391257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5619053250882345e-05,
+      "loss": 0.9903,
+      "step": 12044
+    },
+    {
+      "epoch": 0.3308066243717558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56183174196281e-05,
+      "loss": 1.0079,
+      "step": 12045
+    },
+    {
+      "epoch": 0.330834088599599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.561758154391956e-05,
+      "loss": 0.9648,
+      "step": 12046
+    },
+    {
+      "epoch": 0.33086155282744223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5616845623762536e-05,
+      "loss": 0.9106,
+      "step": 12047
+    },
+    {
+      "epoch": 0.3308890170552855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.561610965916285e-05,
+      "loss": 0.9598,
+      "step": 12048
+    },
+    {
+      "epoch": 0.3309164812831287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5615373650126334e-05,
+      "loss": 1.0369,
+      "step": 12049
+    },
+    {
+      "epoch": 0.33094394551097195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5614637596658803e-05,
+      "loss": 0.9989,
+      "step": 12050
+    },
+    {
+      "epoch": 0.33097140973881517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5613901498766084e-05,
+      "loss": 0.9497,
+      "step": 12051
+    },
+    {
+      "epoch": 0.33099887396665845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5613165356454005e-05,
+      "loss": 0.8995,
+      "step": 12052
+    },
+    {
+      "epoch": 0.33102633819450167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5612429169728387e-05,
+      "loss": 0.8856,
+      "step": 12053
+    },
+    {
+      "epoch": 0.3310538024223449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5611692938595054e-05,
+      "loss": 0.9746,
+      "step": 12054
+    },
+    {
+      "epoch": 0.3310812666501881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5610956663059834e-05,
+      "loss": 0.913,
+      "step": 12055
+    },
+    {
+      "epoch": 0.3311087308780314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5610220343128547e-05,
+      "loss": 1.0126,
+      "step": 12056
+    },
+    {
+      "epoch": 0.3311361951058746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5609483978807025e-05,
+      "loss": 0.9519,
+      "step": 12057
+    },
+    {
+      "epoch": 0.33116365933371783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608747570101096e-05,
+      "loss": 0.9738,
+      "step": 12058
+    },
+    {
+      "epoch": 0.33119112356156105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608011117016582e-05,
+      "loss": 0.922,
+      "step": 12059
+    },
+    {
+      "epoch": 0.33121858778940433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5607274619559313e-05,
+      "loss": 0.9417,
+      "step": 12060
+    },
+    {
+      "epoch": 0.33124605201724755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5606538077735115e-05,
+      "loss": 0.9843,
+      "step": 12061
+    },
+    {
+      "epoch": 0.33127351624509077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5605801491549813e-05,
+      "loss": 0.9886,
+      "step": 12062
+    },
+    {
+      "epoch": 0.331300980472934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5605064861009237e-05,
+      "loss": 1.0304,
+      "step": 12063
+    },
+    {
+      "epoch": 0.3313284447007772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5604328186119223e-05,
+      "loss": 0.8914,
+      "step": 12064
+    },
+    {
+      "epoch": 0.3313559089286205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560359146688559e-05,
+      "loss": 0.9308,
+      "step": 12065
+    },
+    {
+      "epoch": 0.3313833731564637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560285470331417e-05,
+      "loss": 0.8455,
+      "step": 12066
+    },
+    {
+      "epoch": 0.33141083738430693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5602117895410794e-05,
+      "loss": 0.975,
+      "step": 12067
+    },
+    {
+      "epoch": 0.33143830161215015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5601381043181293e-05,
+      "loss": 0.879,
+      "step": 12068
+    },
+    {
+      "epoch": 0.33146576583999343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560064414663149e-05,
+      "loss": 1.0115,
+      "step": 12069
+    },
+    {
+      "epoch": 0.33149323006783665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559990720576723e-05,
+      "loss": 1.016,
+      "step": 12070
+    },
+    {
+      "epoch": 0.3315206942956799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5599170220594328e-05,
+      "loss": 0.9412,
+      "step": 12071
+    },
+    {
+      "epoch": 0.3315481585235231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5598433191118622e-05,
+      "loss": 0.9233,
+      "step": 12072
+    },
+    {
+      "epoch": 0.3315756227513664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559769611734595e-05,
+      "loss": 0.9464,
+      "step": 12073
+    },
+    {
+      "epoch": 0.3316030869792096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559695899928213e-05,
+      "loss": 0.8542,
+      "step": 12074
+    },
+    {
+      "epoch": 0.3316305512070528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5596221836933007e-05,
+      "loss": 0.8821,
+      "step": 12075
+    },
+    {
+      "epoch": 0.33165801543489604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559548463030441e-05,
+      "loss": 0.9217,
+      "step": 12076
+    },
+    {
+      "epoch": 0.33168547966273926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5594747379402168e-05,
+      "loss": 1.127,
+      "step": 12077
+    },
+    {
+      "epoch": 0.33171294389058253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5594010084232118e-05,
+      "loss": 0.94,
+      "step": 12078
+    },
+    {
+      "epoch": 0.33174040811842576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5593272744800098e-05,
+      "loss": 0.9897,
+      "step": 12079
+    },
+    {
+      "epoch": 0.331767872346269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559253536111193e-05,
+      "loss": 0.9033,
+      "step": 12080
+    },
+    {
+      "epoch": 0.3317953365741122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559179793317346e-05,
+      "loss": 1.0277,
+      "step": 12081
+    },
+    {
+      "epoch": 0.3318228008019555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5591060460990517e-05,
+      "loss": 0.928,
+      "step": 12082
+    },
+    {
+      "epoch": 0.3318502650297987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559032294456894e-05,
+      "loss": 0.9259,
+      "step": 12083
+    },
+    {
+      "epoch": 0.3318777292576419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558958538391456e-05,
+      "loss": 0.9915,
+      "step": 12084
+    },
+    {
+      "epoch": 0.33190519348548514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5588847779033218e-05,
+      "loss": 0.9867,
+      "step": 12085
+    },
+    {
+      "epoch": 0.3319326577133284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5588110129930746e-05,
+      "loss": 1.0081,
+      "step": 12086
+    },
+    {
+      "epoch": 0.33196012194117164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558737243661298e-05,
+      "loss": 0.8909,
+      "step": 12087
+    },
+    {
+      "epoch": 0.33198758616901486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5586634699085762e-05,
+      "loss": 1.0309,
+      "step": 12088
+    },
+    {
+      "epoch": 0.3320150503968581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5585896917354925e-05,
+      "loss": 0.9203,
+      "step": 12089
+    },
+    {
+      "epoch": 0.3320425146247013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558515909142631e-05,
+      "loss": 1.0097,
+      "step": 12090
+    },
+    {
+      "epoch": 0.3320699788525446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5584421221305752e-05,
+      "loss": 0.9241,
+      "step": 12091
+    },
+    {
+      "epoch": 0.3320974430803878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558368330699909e-05,
+      "loss": 0.9766,
+      "step": 12092
+    },
+    {
+      "epoch": 0.332124907308231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5582945348512163e-05,
+      "loss": 0.9695,
+      "step": 12093
+    },
+    {
+      "epoch": 0.33215237153607424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5582207345850807e-05,
+      "loss": 0.8147,
+      "step": 12094
+    },
+    {
+      "epoch": 0.3321798357639175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5581469299020864e-05,
+      "loss": 0.8448,
+      "step": 12095
+    },
+    {
+      "epoch": 0.33220729999176074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5580731208028178e-05,
+      "loss": 0.9226,
+      "step": 12096
+    },
+    {
+      "epoch": 0.33223476421960396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557999307287858e-05,
+      "loss": 0.8548,
+      "step": 12097
+    },
+    {
+      "epoch": 0.3322622284474472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557925489357792e-05,
+      "loss": 0.9401,
+      "step": 12098
+    },
+    {
+      "epoch": 0.33228969267529046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557851667013203e-05,
+      "loss": 0.9501,
+      "step": 12099
+    },
+    {
+      "epoch": 0.3323171569031337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557777840254676e-05,
+      "loss": 1.1099,
+      "step": 12100
+    },
+    {
+      "epoch": 0.3323446211309769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557704009082794e-05,
+      "loss": 1.1087,
+      "step": 12101
+    },
+    {
+      "epoch": 0.3323720853588201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5576301734981423e-05,
+      "loss": 0.9134,
+      "step": 12102
+    },
+    {
+      "epoch": 0.33239954958666335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5575563335013046e-05,
+      "loss": 1.0151,
+      "step": 12103
+    },
+    {
+      "epoch": 0.3324270138145066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5574824890928652e-05,
+      "loss": 1.0555,
+      "step": 12104
+    },
+    {
+      "epoch": 0.33245447804234984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5574086402734085e-05,
+      "loss": 0.9327,
+      "step": 12105
+    },
+    {
+      "epoch": 0.33248194227019306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5573347870435184e-05,
+      "loss": 0.9563,
+      "step": 12106
+    },
+    {
+      "epoch": 0.3325094064980363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55726092940378e-05,
+      "loss": 0.913,
+      "step": 12107
+    },
+    {
+      "epoch": 0.33253687072587956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557187067354777e-05,
+      "loss": 0.9871,
+      "step": 12108
+    },
+    {
+      "epoch": 0.3325643349537228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557113200897094e-05,
+      "loss": 1.0015,
+      "step": 12109
+    },
+    {
+      "epoch": 0.332591799181566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5570393300313156e-05,
+      "loss": 0.9569,
+      "step": 12110
+    },
+    {
+      "epoch": 0.3326192634094092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5569654547580266e-05,
+      "loss": 0.9304,
+      "step": 12111
+    },
+    {
+      "epoch": 0.3326467276372525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5568915750778112e-05,
+      "loss": 0.9411,
+      "step": 12112
+    },
+    {
+      "epoch": 0.3326741918650957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5568176909912535e-05,
+      "loss": 1.0063,
+      "step": 12113
+    },
+    {
+      "epoch": 0.33270165609293895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5567438024989386e-05,
+      "loss": 0.9673,
+      "step": 12114
+    },
+    {
+      "epoch": 0.33272912032078217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556669909601451e-05,
+      "loss": 0.9358,
+      "step": 12115
+    },
+    {
+      "epoch": 0.3327565845486254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5565960122993758e-05,
+      "loss": 0.8439,
+      "step": 12116
+    },
+    {
+      "epoch": 0.33278404877646867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556522110593297e-05,
+      "loss": 0.9081,
+      "step": 12117
+    },
+    {
+      "epoch": 0.3328115130043119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5564482044837998e-05,
+      "loss": 0.9294,
+      "step": 12118
+    },
+    {
+      "epoch": 0.3328389772321551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556374293971469e-05,
+      "loss": 1.0368,
+      "step": 12119
+    },
+    {
+      "epoch": 0.33286644145999833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5563003790568892e-05,
+      "loss": 0.9694,
+      "step": 12120
+    },
+    {
+      "epoch": 0.3328939056878416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5562264597406452e-05,
+      "loss": 0.9846,
+      "step": 12121
+    },
+    {
+      "epoch": 0.33292136991568483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556152536023322e-05,
+      "loss": 1.0235,
+      "step": 12122
+    },
+    {
+      "epoch": 0.33294883414352805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5560786079055045e-05,
+      "loss": 0.9659,
+      "step": 12123
+    },
+    {
+      "epoch": 0.33297629837137127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556004675387778e-05,
+      "loss": 0.9037,
+      "step": 12124
+    },
+    {
+      "epoch": 0.33300376259921455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5559307384707265e-05,
+      "loss": 0.9928,
+      "step": 12125
+    },
+    {
+      "epoch": 0.33303122682705777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5558567971549362e-05,
+      "loss": 0.8161,
+      "step": 12126
+    },
+    {
+      "epoch": 0.333058691054901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5557828514409914e-05,
+      "loss": 1.0293,
+      "step": 12127
+    },
+    {
+      "epoch": 0.3330861552827442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5557089013294772e-05,
+      "loss": 1.0614,
+      "step": 12128
+    },
+    {
+      "epoch": 0.33311361951058743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5556349468209794e-05,
+      "loss": 0.9336,
+      "step": 12129
+    },
+    {
+      "epoch": 0.3331410837384307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.555560987916082e-05,
+      "loss": 0.9659,
+      "step": 12130
+    },
+    {
+      "epoch": 0.33316854796627393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5554870246153714e-05,
+      "loss": 0.9125,
+      "step": 12131
+    },
+    {
+      "epoch": 0.33319601219411715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5554130569194322e-05,
+      "loss": 0.9137,
+      "step": 12132
+    },
+    {
+      "epoch": 0.3332234764219604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5553390848288496e-05,
+      "loss": 0.9489,
+      "step": 12133
+    },
+    {
+      "epoch": 0.33325094064980365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5552651083442093e-05,
+      "loss": 0.8762,
+      "step": 12134
+    },
+    {
+      "epoch": 0.33327840487764687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.555191127466096e-05,
+      "loss": 0.978,
+      "step": 12135
+    },
+    {
+      "epoch": 0.3333058691054901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5551171421950955e-05,
+      "loss": 0.9804,
+      "step": 12136
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5550431525317932e-05,
+      "loss": 1.0744,
+      "step": 12137
+    },
+    {
+      "epoch": 0.3333607975611766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554969158476775e-05,
+      "loss": 1.0142,
+      "step": 12138
+    },
+    {
+      "epoch": 0.3333882617890198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5548951600306252e-05,
+      "loss": 1.0461,
+      "step": 12139
+    },
+    {
+      "epoch": 0.33341572601686303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55482115719393e-05,
+      "loss": 1.0027,
+      "step": 12140
+    },
+    {
+      "epoch": 0.33344319024470626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554747149967275e-05,
+      "loss": 0.9948,
+      "step": 12141
+    },
+    {
+      "epoch": 0.3334706544725495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5546731383512456e-05,
+      "loss": 0.9888,
+      "step": 12142
+    },
+    {
+      "epoch": 0.33349811870039275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5545991223464275e-05,
+      "loss": 1.0461,
+      "step": 12143
+    },
+    {
+      "epoch": 0.333525582928236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5545251019534065e-05,
+      "loss": 0.9849,
+      "step": 12144
+    },
+    {
+      "epoch": 0.3335530471560792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5544510771727677e-05,
+      "loss": 1.0878,
+      "step": 12145
+    },
+    {
+      "epoch": 0.3335805113839224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5543770480050976e-05,
+      "loss": 1.008,
+      "step": 12146
+    },
+    {
+      "epoch": 0.3336079756117657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5543030144509813e-05,
+      "loss": 0.8798,
+      "step": 12147
+    },
+    {
+      "epoch": 0.3336354398396089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554228976511005e-05,
+      "loss": 0.9402,
+      "step": 12148
+    },
+    {
+      "epoch": 0.33366290406745214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554154934185754e-05,
+      "loss": 1.0831,
+      "step": 12149
+    },
+    {
+      "epoch": 0.33369036829529536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5540808874758152e-05,
+      "loss": 0.9415,
+      "step": 12150
+    },
+    {
+      "epoch": 0.33371783252313864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554006836381773e-05,
+      "loss": 1.0178,
+      "step": 12151
+    },
+    {
+      "epoch": 0.33374529675098186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553932780904215e-05,
+      "loss": 0.9996,
+      "step": 12152
+    },
+    {
+      "epoch": 0.3337727609788251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5538587210437253e-05,
+      "loss": 0.8886,
+      "step": 12153
+    },
+    {
+      "epoch": 0.3338002252066683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5537846568008915e-05,
+      "loss": 0.8773,
+      "step": 12154
+    },
+    {
+      "epoch": 0.3338276894345116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553710588176299e-05,
+      "loss": 0.952,
+      "step": 12155
+    },
+    {
+      "epoch": 0.3338551536623548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5536365151705336e-05,
+      "loss": 0.8863,
+      "step": 12156
+    },
+    {
+      "epoch": 0.333882617890198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5535624377841814e-05,
+      "loss": 1.0075,
+      "step": 12157
+    },
+    {
+      "epoch": 0.33391008211804124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553488356017829e-05,
+      "loss": 0.9718,
+      "step": 12158
+    },
+    {
+      "epoch": 0.33393754634588446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553414269872062e-05,
+      "loss": 0.9607,
+      "step": 12159
+    },
+    {
+      "epoch": 0.33396501057372774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5533401793474677e-05,
+      "loss": 0.9415,
+      "step": 12160
+    },
+    {
+      "epoch": 0.33399247480157096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5532660844446314e-05,
+      "loss": 0.9312,
+      "step": 12161
+    },
+    {
+      "epoch": 0.3340199390294142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5531919851641387e-05,
+      "loss": 0.9924,
+      "step": 12162
+    },
+    {
+      "epoch": 0.3340474032572574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5531178815065776e-05,
+      "loss": 1.029,
+      "step": 12163
+    },
+    {
+      "epoch": 0.3340748674851007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553043773472533e-05,
+      "loss": 0.9576,
+      "step": 12164
+    },
+    {
+      "epoch": 0.3341023317129439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5529696610625923e-05,
+      "loss": 0.962,
+      "step": 12165
+    },
+    {
+      "epoch": 0.3341297959407871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552895544277341e-05,
+      "loss": 0.9367,
+      "step": 12166
+    },
+    {
+      "epoch": 0.33415726016863034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5528214231173665e-05,
+      "loss": 0.9139,
+      "step": 12167
+    },
+    {
+      "epoch": 0.3341847243964736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5527472975832543e-05,
+      "loss": 1.0325,
+      "step": 12168
+    },
+    {
+      "epoch": 0.33421218862431684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5526731676755913e-05,
+      "loss": 0.9714,
+      "step": 12169
+    },
+    {
+      "epoch": 0.33423965285216006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5525990333949644e-05,
+      "loss": 0.8233,
+      "step": 12170
+    },
+    {
+      "epoch": 0.3342671170800033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55252489474196e-05,
+      "loss": 0.9436,
+      "step": 12171
+    },
+    {
+      "epoch": 0.3342945813078465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5524507517171642e-05,
+      "loss": 0.9514,
+      "step": 12172
+    },
+    {
+      "epoch": 0.3343220455356898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5523766043211646e-05,
+      "loss": 0.9684,
+      "step": 12173
+    },
+    {
+      "epoch": 0.334349509763533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552302452554547e-05,
+      "loss": 0.9894,
+      "step": 12174
+    },
+    {
+      "epoch": 0.3343769739913762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5522282964178982e-05,
+      "loss": 0.9908,
+      "step": 12175
+    },
+    {
+      "epoch": 0.33440443821921945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552154135911806e-05,
+      "loss": 1.0014,
+      "step": 12176
+    },
+    {
+      "epoch": 0.3344319024470627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5520799710368555e-05,
+      "loss": 1.0197,
+      "step": 12177
+    },
+    {
+      "epoch": 0.33445936667490594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552005801793635e-05,
+      "loss": 1.0347,
+      "step": 12178
+    },
+    {
+      "epoch": 0.33448683090274917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5519316281827302e-05,
+      "loss": 0.9147,
+      "step": 12179
+    },
+    {
+      "epoch": 0.3345142951305924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5518574502047293e-05,
+      "loss": 0.9906,
+      "step": 12180
+    },
+    {
+      "epoch": 0.33454175935843566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5517832678602177e-05,
+      "loss": 0.9714,
+      "step": 12181
+    },
+    {
+      "epoch": 0.3345692235862789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551709081149784e-05,
+      "loss": 0.8882,
+      "step": 12182
+    },
+    {
+      "epoch": 0.3345966878141221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5516348900740136e-05,
+      "loss": 1.0876,
+      "step": 12183
+    },
+    {
+      "epoch": 0.33462415204196533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5515606946334944e-05,
+      "loss": 0.7913,
+      "step": 12184
+    },
+    {
+      "epoch": 0.33465161626980855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5514864948288138e-05,
+      "loss": 1.0115,
+      "step": 12185
+    },
+    {
+      "epoch": 0.3346790804976518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551412290660558e-05,
+      "loss": 0.9443,
+      "step": 12186
+    },
+    {
+      "epoch": 0.33470654472549505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5513380821293146e-05,
+      "loss": 0.9,
+      "step": 12187
+    },
+    {
+      "epoch": 0.33473400895333827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5512638692356706e-05,
+      "loss": 0.9694,
+      "step": 12188
+    },
+    {
+      "epoch": 0.3347614731811815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5511896519802136e-05,
+      "loss": 0.8705,
+      "step": 12189
+    },
+    {
+      "epoch": 0.33478893740902477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5511154303635307e-05,
+      "loss": 0.9955,
+      "step": 12190
+    },
+    {
+      "epoch": 0.334816401636868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5510412043862086e-05,
+      "loss": 0.9055,
+      "step": 12191
+    },
+    {
+      "epoch": 0.3348438658647112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5509669740488352e-05,
+      "loss": 0.8888,
+      "step": 12192
+    },
+    {
+      "epoch": 0.33487133009255443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5508927393519978e-05,
+      "loss": 0.8081,
+      "step": 12193
+    },
+    {
+      "epoch": 0.3348987943203977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5508185002962837e-05,
+      "loss": 0.9535,
+      "step": 12194
+    },
+    {
+      "epoch": 0.33492625854824093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5507442568822802e-05,
+      "loss": 0.9141,
+      "step": 12195
+    },
+    {
+      "epoch": 0.33495372277608415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5506700091105748e-05,
+      "loss": 0.9817,
+      "step": 12196
+    },
+    {
+      "epoch": 0.33498118700392737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5505957569817545e-05,
+      "loss": 0.8816,
+      "step": 12197
+    },
+    {
+      "epoch": 0.3350086512317706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550521500496408e-05,
+      "loss": 0.935,
+      "step": 12198
+    },
+    {
+      "epoch": 0.33503611545961387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5504472396551217e-05,
+      "loss": 1.0989,
+      "step": 12199
+    },
+    {
+      "epoch": 0.3350635796874571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550372974458484e-05,
+      "loss": 1.0048,
+      "step": 12200
+    },
+    {
+      "epoch": 0.3350910439153003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550298704907082e-05,
+      "loss": 1.0107,
+      "step": 12201
+    },
+    {
+      "epoch": 0.33511850814314353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550224431001503e-05,
+      "loss": 0.9865,
+      "step": 12202
+    },
+    {
+      "epoch": 0.3351459723709868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5501501527423357e-05,
+      "loss": 0.8838,
+      "step": 12203
+    },
+    {
+      "epoch": 0.33517343659883003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550075870130167e-05,
+      "loss": 1.0665,
+      "step": 12204
+    },
+    {
+      "epoch": 0.33520090082667325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5500015831655852e-05,
+      "loss": 0.9196,
+      "step": 12205
+    },
+    {
+      "epoch": 0.3352283650545165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5499272918491777e-05,
+      "loss": 0.9455,
+      "step": 12206
+    },
+    {
+      "epoch": 0.33525582928235975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5498529961815324e-05,
+      "loss": 1.0815,
+      "step": 12207
+    },
+    {
+      "epoch": 0.335283293510203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5497786961632372e-05,
+      "loss": 0.854,
+      "step": 12208
+    },
+    {
+      "epoch": 0.3353107577380462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54970439179488e-05,
+      "loss": 0.986,
+      "step": 12209
+    },
+    {
+      "epoch": 0.3353382219658894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549630083077049e-05,
+      "loss": 0.8457,
+      "step": 12210
+    },
+    {
+      "epoch": 0.33536568619373264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5495557700103316e-05,
+      "loss": 1.0657,
+      "step": 12211
+    },
+    {
+      "epoch": 0.3353931504215759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549481452595316e-05,
+      "loss": 0.9681,
+      "step": 12212
+    },
+    {
+      "epoch": 0.33542061464941914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5494071308325904e-05,
+      "loss": 1.0327,
+      "step": 12213
+    },
+    {
+      "epoch": 0.33544807887726236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5493328047227426e-05,
+      "loss": 0.9791,
+      "step": 12214
+    },
+    {
+      "epoch": 0.3354755431051056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549258474266361e-05,
+      "loss": 1.0219,
+      "step": 12215
+    },
+    {
+      "epoch": 0.33550300733294885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5491841394640336e-05,
+      "loss": 0.8878,
+      "step": 12216
+    },
+    {
+      "epoch": 0.3355304715607921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5491098003163488e-05,
+      "loss": 0.9885,
+      "step": 12217
+    },
+    {
+      "epoch": 0.3355579357886353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549035456823894e-05,
+      "loss": 1.0099,
+      "step": 12218
+    },
+    {
+      "epoch": 0.3355854000164785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5489611089872583e-05,
+      "loss": 1.0158,
+      "step": 12219
+    },
+    {
+      "epoch": 0.3356128642443218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54888675680703e-05,
+      "loss": 0.9125,
+      "step": 12220
+    },
+    {
+      "epoch": 0.335640328472165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5488124002837964e-05,
+      "loss": 0.9553,
+      "step": 12221
+    },
+    {
+      "epoch": 0.33566779270000824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548738039418147e-05,
+      "loss": 0.9901,
+      "step": 12222
+    },
+    {
+      "epoch": 0.33569525692785146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5486636742106697e-05,
+      "loss": 1.0521,
+      "step": 12223
+    },
+    {
+      "epoch": 0.3357227211556947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5485893046619526e-05,
+      "loss": 0.9238,
+      "step": 12224
+    },
+    {
+      "epoch": 0.33575018538353796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5485149307725842e-05,
+      "loss": 0.8991,
+      "step": 12225
+    },
+    {
+      "epoch": 0.3357776496113812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5484405525431536e-05,
+      "loss": 0.999,
+      "step": 12226
+    },
+    {
+      "epoch": 0.3358051138392244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548366169974249e-05,
+      "loss": 0.8891,
+      "step": 12227
+    },
+    {
+      "epoch": 0.3358325780670676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5482917830664585e-05,
+      "loss": 0.9883,
+      "step": 12228
+    },
+    {
+      "epoch": 0.3358600422949109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548217391820371e-05,
+      "loss": 0.9279,
+      "step": 12229
+    },
+    {
+      "epoch": 0.3358875065227541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5481429962365753e-05,
+      "loss": 0.9799,
+      "step": 12230
+    },
+    {
+      "epoch": 0.33591497075059734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54806859631566e-05,
+      "loss": 0.8682,
+      "step": 12231
+    },
+    {
+      "epoch": 0.33594243497844056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5479941920582135e-05,
+      "loss": 0.8915,
+      "step": 12232
+    },
+    {
+      "epoch": 0.33596989920628384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5479197834648246e-05,
+      "loss": 0.9232,
+      "step": 12233
+    },
+    {
+      "epoch": 0.33599736343412706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5478453705360824e-05,
+      "loss": 0.9825,
+      "step": 12234
+    },
+    {
+      "epoch": 0.3360248276619703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547770953272575e-05,
+      "loss": 0.8936,
+      "step": 12235
+    },
+    {
+      "epoch": 0.3360522918898135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547696531674892e-05,
+      "loss": 0.9274,
+      "step": 12236
+    },
+    {
+      "epoch": 0.3360797561176567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5476221057436217e-05,
+      "loss": 1.1117,
+      "step": 12237
+    },
+    {
+      "epoch": 0.3361072203455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547547675479353e-05,
+      "loss": 0.9591,
+      "step": 12238
+    },
+    {
+      "epoch": 0.3361346845733432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5474732408826754e-05,
+      "loss": 0.9828,
+      "step": 12239
+    },
+    {
+      "epoch": 0.33616214880118644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5473988019541774e-05,
+      "loss": 1.14,
+      "step": 12240
+    },
+    {
+      "epoch": 0.33618961302902967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547324358694448e-05,
+      "loss": 1.0577,
+      "step": 12241
+    },
+    {
+      "epoch": 0.33621707725687294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5472499111040763e-05,
+      "loss": 0.9463,
+      "step": 12242
+    },
+    {
+      "epoch": 0.33624454148471616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547175459183651e-05,
+      "loss": 0.9691,
+      "step": 12243
+    },
+    {
+      "epoch": 0.3362720057125594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547101002933762e-05,
+      "loss": 1.067,
+      "step": 12244
+    },
+    {
+      "epoch": 0.3362994699404026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5470265423549976e-05,
+      "loss": 0.939,
+      "step": 12245
+    },
+    {
+      "epoch": 0.3363269341682459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5469520774479472e-05,
+      "loss": 0.9443,
+      "step": 12246
+    },
+    {
+      "epoch": 0.3363543983960891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5468776082132002e-05,
+      "loss": 0.9274,
+      "step": 12247
+    },
+    {
+      "epoch": 0.3363818626239323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546803134651346e-05,
+      "loss": 0.9614,
+      "step": 12248
+    },
+    {
+      "epoch": 0.33640932685177555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5467286567629733e-05,
+      "loss": 0.9117,
+      "step": 12249
+    },
+    {
+      "epoch": 0.33643679107961877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5466541745486715e-05,
+      "loss": 0.945,
+      "step": 12250
+    },
+    {
+      "epoch": 0.33646425530746205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5465796880090305e-05,
+      "loss": 0.8819,
+      "step": 12251
+    },
+    {
+      "epoch": 0.33649171953530527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546505197144639e-05,
+      "loss": 0.9789,
+      "step": 12252
+    },
+    {
+      "epoch": 0.3365191837631485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546430701956087e-05,
+      "loss": 0.8681,
+      "step": 12253
+    },
+    {
+      "epoch": 0.3365466479909917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5463562024439633e-05,
+      "loss": 0.8589,
+      "step": 12254
+    },
+    {
+      "epoch": 0.336574112218835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546281698608858e-05,
+      "loss": 0.8693,
+      "step": 12255
+    },
+    {
+      "epoch": 0.3366015764466782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54620719045136e-05,
+      "loss": 0.977,
+      "step": 12256
+    },
+    {
+      "epoch": 0.33662904067452143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5461326779720592e-05,
+      "loss": 0.9161,
+      "step": 12257
+    },
+    {
+      "epoch": 0.33665650490236465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546058161171545e-05,
+      "loss": 0.955,
+      "step": 12258
+    },
+    {
+      "epoch": 0.3366839691302079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5459836400504075e-05,
+      "loss": 0.9315,
+      "step": 12259
+    },
+    {
+      "epoch": 0.33671143335805115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5459091146092358e-05,
+      "loss": 0.885,
+      "step": 12260
+    },
+    {
+      "epoch": 0.33673889758589437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5458345848486195e-05,
+      "loss": 1.0403,
+      "step": 12261
+    },
+    {
+      "epoch": 0.3367663618137376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5457600507691487e-05,
+      "loss": 1.0013,
+      "step": 12262
+    },
+    {
+      "epoch": 0.33679382604158087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545685512371413e-05,
+      "loss": 0.9164,
+      "step": 12263
+    },
+    {
+      "epoch": 0.3368212902694241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545610969656002e-05,
+      "loss": 0.9341,
+      "step": 12264
+    },
+    {
+      "epoch": 0.3368487544972673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545536422623506e-05,
+      "loss": 0.9329,
+      "step": 12265
+    },
+    {
+      "epoch": 0.33687621872511053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545461871274514e-05,
+      "loss": 0.9457,
+      "step": 12266
+    },
+    {
+      "epoch": 0.33690368295295375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5453873156096165e-05,
+      "loss": 0.9428,
+      "step": 12267
+    },
+    {
+      "epoch": 0.33693114718079703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5453127556294035e-05,
+      "loss": 1.0521,
+      "step": 12268
+    },
+    {
+      "epoch": 0.33695861140864025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545238191334465e-05,
+      "loss": 0.8944,
+      "step": 12269
+    },
+    {
+      "epoch": 0.3369860756364835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54516362272539e-05,
+      "loss": 1.0474,
+      "step": 12270
+    },
+    {
+      "epoch": 0.3370135398643267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5450890498027696e-05,
+      "loss": 0.9342,
+      "step": 12271
+    },
+    {
+      "epoch": 0.33704100409216997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5450144725671937e-05,
+      "loss": 0.9095,
+      "step": 12272
+    },
+    {
+      "epoch": 0.3370684683200132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5449398910192517e-05,
+      "loss": 0.8172,
+      "step": 12273
+    },
+    {
+      "epoch": 0.3370959325478564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5448653051595346e-05,
+      "loss": 1.0175,
+      "step": 12274
+    },
+    {
+      "epoch": 0.33712339677569964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544790714988632e-05,
+      "loss": 0.9077,
+      "step": 12275
+    },
+    {
+      "epoch": 0.3371508610035429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5447161205071342e-05,
+      "loss": 1.0154,
+      "step": 12276
+    },
+    {
+      "epoch": 0.33717832523138613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5446415217156316e-05,
+      "loss": 0.8966,
+      "step": 12277
+    },
+    {
+      "epoch": 0.33720578945922935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5445669186147144e-05,
+      "loss": 1.0101,
+      "step": 12278
+    },
+    {
+      "epoch": 0.3372332536870726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5444923112049724e-05,
+      "loss": 0.9231,
+      "step": 12279
+    },
+    {
+      "epoch": 0.3372607179149158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5444176994869966e-05,
+      "loss": 0.914,
+      "step": 12280
+    },
+    {
+      "epoch": 0.3372881821427591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544343083461377e-05,
+      "loss": 0.9594,
+      "step": 12281
+    },
+    {
+      "epoch": 0.3373156463706023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5442684631287042e-05,
+      "loss": 0.9525,
+      "step": 12282
+    },
+    {
+      "epoch": 0.3373431105984455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5441938384895684e-05,
+      "loss": 0.9745,
+      "step": 12283
+    },
+    {
+      "epoch": 0.33737057482628874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54411920954456e-05,
+      "loss": 0.8499,
+      "step": 12284
+    },
+    {
+      "epoch": 0.337398039054132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5440445762942696e-05,
+      "loss": 1.0051,
+      "step": 12285
+    },
+    {
+      "epoch": 0.33742550328197524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5439699387392878e-05,
+      "loss": 0.9286,
+      "step": 12286
+    },
+    {
+      "epoch": 0.33745296750981846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5438952968802056e-05,
+      "loss": 0.7666,
+      "step": 12287
+    },
+    {
+      "epoch": 0.3374804317376617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5438206507176126e-05,
+      "loss": 1.0048,
+      "step": 12288
+    },
+    {
+      "epoch": 0.33750789596550496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5437460002521004e-05,
+      "loss": 1.073,
+      "step": 12289
+    },
+    {
+      "epoch": 0.3375353601933482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5436713454842588e-05,
+      "loss": 0.9758,
+      "step": 12290
+    },
+    {
+      "epoch": 0.3375628244211914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.543596686414679e-05,
+      "loss": 0.9764,
+      "step": 12291
+    },
+    {
+      "epoch": 0.3375902886490346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5435220230439517e-05,
+      "loss": 0.9103,
+      "step": 12292
+    },
+    {
+      "epoch": 0.33761775287687784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5434473553726673e-05,
+      "loss": 0.9443,
+      "step": 12293
+    },
+    {
+      "epoch": 0.3376452171047211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5433726834014175e-05,
+      "loss": 0.9456,
+      "step": 12294
+    },
+    {
+      "epoch": 0.33767268133256434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5432980071307924e-05,
+      "loss": 1.0715,
+      "step": 12295
+    },
+    {
+      "epoch": 0.33770014556040756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5432233265613826e-05,
+      "loss": 0.9145,
+      "step": 12296
+    },
+    {
+      "epoch": 0.3377276097882508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54314864169378e-05,
+      "loss": 0.965,
+      "step": 12297
+    },
+    {
+      "epoch": 0.33775507401609406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5430739525285745e-05,
+      "loss": 0.9708,
+      "step": 12298
+    },
+    {
+      "epoch": 0.3377825382439373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5429992590663577e-05,
+      "loss": 0.8838,
+      "step": 12299
+    },
+    {
+      "epoch": 0.3378100024717805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5429245613077207e-05,
+      "loss": 1.0112,
+      "step": 12300
+    },
+    {
+      "epoch": 0.3378374666996237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5428498592532538e-05,
+      "loss": 0.8735,
+      "step": 12301
+    },
+    {
+      "epoch": 0.337864930927467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542775152903549e-05,
+      "loss": 0.9305,
+      "step": 12302
+    },
+    {
+      "epoch": 0.3378923951553102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5427004422591965e-05,
+      "loss": 0.9886,
+      "step": 12303
+    },
+    {
+      "epoch": 0.33791985938315344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542625727320788e-05,
+      "loss": 1.132,
+      "step": 12304
+    },
+    {
+      "epoch": 0.33794732361099666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5425510080889147e-05,
+      "loss": 0.9588,
+      "step": 12305
+    },
+    {
+      "epoch": 0.3379747878388399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5424762845641673e-05,
+      "loss": 0.9342,
+      "step": 12306
+    },
+    {
+      "epoch": 0.33800225206668316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5424015567471378e-05,
+      "loss": 0.8956,
+      "step": 12307
+    },
+    {
+      "epoch": 0.3380297162945264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542326824638417e-05,
+      "loss": 0.9529,
+      "step": 12308
+    },
+    {
+      "epoch": 0.3380571805223696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542252088238596e-05,
+      "loss": 0.9034,
+      "step": 12309
+    },
+    {
+      "epoch": 0.3380846447502128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5421773475482665e-05,
+      "loss": 0.8727,
+      "step": 12310
+    },
+    {
+      "epoch": 0.3381121089780561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54210260256802e-05,
+      "loss": 0.9476,
+      "step": 12311
+    },
+    {
+      "epoch": 0.3381395732058993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5420278532984476e-05,
+      "loss": 0.9942,
+      "step": 12312
+    },
+    {
+      "epoch": 0.33816703743374255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541953099740141e-05,
+      "loss": 0.9259,
+      "step": 12313
+    },
+    {
+      "epoch": 0.33819450166158577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5418783418936914e-05,
+      "loss": 0.9305,
+      "step": 12314
+    },
+    {
+      "epoch": 0.33822196588942904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541803579759691e-05,
+      "loss": 0.9321,
+      "step": 12315
+    },
+    {
+      "epoch": 0.33824943011727227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54172881333873e-05,
+      "loss": 0.9114,
+      "step": 12316
+    },
+    {
+      "epoch": 0.3382768943451155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541654042631401e-05,
+      "loss": 1.0128,
+      "step": 12317
+    },
+    {
+      "epoch": 0.3383043585729587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5415792676382958e-05,
+      "loss": 1.039,
+      "step": 12318
+    },
+    {
+      "epoch": 0.33833182280080193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5415044883600052e-05,
+      "loss": 0.933,
+      "step": 12319
+    },
+    {
+      "epoch": 0.3383592870286452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5414297047971214e-05,
+      "loss": 0.9052,
+      "step": 12320
+    },
+    {
+      "epoch": 0.3383867512564884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5413549169502363e-05,
+      "loss": 0.8987,
+      "step": 12321
+    },
+    {
+      "epoch": 0.33841421548433165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5412801248199412e-05,
+      "loss": 1.0633,
+      "step": 12322
+    },
+    {
+      "epoch": 0.33844167971217487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5412053284068277e-05,
+      "loss": 1.0055,
+      "step": 12323
+    },
+    {
+      "epoch": 0.33846914394001815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541130527711489e-05,
+      "loss": 1.0065,
+      "step": 12324
+    },
+    {
+      "epoch": 0.33849660816786137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541055722734515e-05,
+      "loss": 0.9609,
+      "step": 12325
+    },
+    {
+      "epoch": 0.3385240723957046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540980913476499e-05,
+      "loss": 0.9625,
+      "step": 12326
+    },
+    {
+      "epoch": 0.3385515366235478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5409060999380322e-05,
+      "loss": 0.9815,
+      "step": 12327
+    },
+    {
+      "epoch": 0.3385790008513911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5408312821197073e-05,
+      "loss": 0.9806,
+      "step": 12328
+    },
+    {
+      "epoch": 0.3386064650792343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5407564600221153e-05,
+      "loss": 0.9504,
+      "step": 12329
+    },
+    {
+      "epoch": 0.33863392930707753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5406816336458485e-05,
+      "loss": 0.8798,
+      "step": 12330
+    },
+    {
+      "epoch": 0.33866139353492075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5406068029915e-05,
+      "loss": 1.0004,
+      "step": 12331
+    },
+    {
+      "epoch": 0.338688857762764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5405319680596603e-05,
+      "loss": 0.9861,
+      "step": 12332
+    },
+    {
+      "epoch": 0.33871632199060725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5404571288509226e-05,
+      "loss": 0.9998,
+      "step": 12333
+    },
+    {
+      "epoch": 0.33874378621845047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5403822853658788e-05,
+      "loss": 0.9111,
+      "step": 12334
+    },
+    {
+      "epoch": 0.3387712504462937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540307437605121e-05,
+      "loss": 0.9868,
+      "step": 12335
+    },
+    {
+      "epoch": 0.3387987146741369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5402325855692414e-05,
+      "loss": 1.041,
+      "step": 12336
+    },
+    {
+      "epoch": 0.3388261789019802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5401577292588323e-05,
+      "loss": 0.99,
+      "step": 12337
+    },
+    {
+      "epoch": 0.3388536431298234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540082868674486e-05,
+      "loss": 0.9008,
+      "step": 12338
+    },
+    {
+      "epoch": 0.33888110735766663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540008003816795e-05,
+      "loss": 0.9186,
+      "step": 12339
+    },
+    {
+      "epoch": 0.33890857158550985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5399331346863514e-05,
+      "loss": 0.9216,
+      "step": 12340
+    },
+    {
+      "epoch": 0.33893603581335313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5398582612837478e-05,
+      "loss": 0.9801,
+      "step": 12341
+    },
+    {
+      "epoch": 0.33896350004119635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539783383609576e-05,
+      "loss": 0.9211,
+      "step": 12342
+    },
+    {
+      "epoch": 0.3389909642690396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5397085016644297e-05,
+      "loss": 0.9696,
+      "step": 12343
+    },
+    {
+      "epoch": 0.3390184284968828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5396336154489003e-05,
+      "loss": 0.8589,
+      "step": 12344
+    },
+    {
+      "epoch": 0.339045892724726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539558724963581e-05,
+      "loss": 1.0074,
+      "step": 12345
+    },
+    {
+      "epoch": 0.3390733569525693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5394838302090636e-05,
+      "loss": 0.9201,
+      "step": 12346
+    },
+    {
+      "epoch": 0.3391008211804125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5394089311859414e-05,
+      "loss": 0.9274,
+      "step": 12347
+    },
+    {
+      "epoch": 0.33912828540825574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5393340278948068e-05,
+      "loss": 0.9847,
+      "step": 12348
+    },
+    {
+      "epoch": 0.33915574963609896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5392591203362522e-05,
+      "loss": 0.8704,
+      "step": 12349
+    },
+    {
+      "epoch": 0.33918321386394223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539184208510871e-05,
+      "loss": 1.0238,
+      "step": 12350
+    },
+    {
+      "epoch": 0.33921067809178546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539109292419255e-05,
+      "loss": 0.94,
+      "step": 12351
+    },
+    {
+      "epoch": 0.3392381423196287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5390343720619976e-05,
+      "loss": 0.9073,
+      "step": 12352
+    },
+    {
+      "epoch": 0.3392656065474719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538959447439692e-05,
+      "loss": 0.9062,
+      "step": 12353
+    },
+    {
+      "epoch": 0.3392930707753152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5388845185529296e-05,
+      "loss": 0.9493,
+      "step": 12354
+    },
+    {
+      "epoch": 0.3393205350031584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5388095854023045e-05,
+      "loss": 0.7879,
+      "step": 12355
+    },
+    {
+      "epoch": 0.3393479992310016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5387346479884094e-05,
+      "loss": 0.8711,
+      "step": 12356
+    },
+    {
+      "epoch": 0.33937546345884484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5386597063118372e-05,
+      "loss": 0.9865,
+      "step": 12357
+    },
+    {
+      "epoch": 0.3394029276866881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5385847603731803e-05,
+      "loss": 0.8927,
+      "step": 12358
+    },
+    {
+      "epoch": 0.33943039191453134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5385098101730325e-05,
+      "loss": 0.9342,
+      "step": 12359
+    },
+    {
+      "epoch": 0.33945785614237456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5384348557119864e-05,
+      "loss": 0.8777,
+      "step": 12360
+    },
+    {
+      "epoch": 0.3394853203702178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538359896990635e-05,
+      "loss": 0.9602,
+      "step": 12361
+    },
+    {
+      "epoch": 0.339512784598061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5382849340095714e-05,
+      "loss": 0.9386,
+      "step": 12362
+    },
+    {
+      "epoch": 0.3395402488259043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5382099667693894e-05,
+      "loss": 0.9233,
+      "step": 12363
+    },
+    {
+      "epoch": 0.3395677130537475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5381349952706815e-05,
+      "loss": 0.8933,
+      "step": 12364
+    },
+    {
+      "epoch": 0.3395951772815907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538060019514041e-05,
+      "loss": 1.0203,
+      "step": 12365
+    },
+    {
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537985039500061e-05,
+      "loss": 0.9539,
+      "step": 12366
+    },
+    {
+      "epoch": 0.3396501057372772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5379100552293352e-05,
+      "loss": 0.9834,
+      "step": 12367
+    },
+    {
+      "epoch": 0.33967756996512044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537835066702457e-05,
+      "loss": 0.9464,
+      "step": 12368
+    },
+    {
+      "epoch": 0.33970503419296366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5377600739200188e-05,
+      "loss": 0.9189,
+      "step": 12369
+    },
+    {
+      "epoch": 0.3397324984208069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537685076882615e-05,
+      "loss": 0.9222,
+      "step": 12370
+    },
+    {
+      "epoch": 0.33975996264865016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537610075590838e-05,
+      "loss": 0.9742,
+      "step": 12371
+    },
+    {
+      "epoch": 0.3397874268764934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5375350700452827e-05,
+      "loss": 0.9319,
+      "step": 12372
+    },
+    {
+      "epoch": 0.3398148911043366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537460060246541e-05,
+      "loss": 0.9407,
+      "step": 12373
+    },
+    {
+      "epoch": 0.3398423553321798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5373850461952073e-05,
+      "loss": 1.0345,
+      "step": 12374
+    },
+    {
+      "epoch": 0.33986981956002305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5373100278918748e-05,
+      "loss": 0.9228,
+      "step": 12375
+    },
+    {
+      "epoch": 0.3398972837878663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537235005337138e-05,
+      "loss": 1.0102,
+      "step": 12376
+    },
+    {
+      "epoch": 0.33992474801570954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537159978531589e-05,
+      "loss": 0.8935,
+      "step": 12377
+    },
+    {
+      "epoch": 0.33995221224355276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5370849474758222e-05,
+      "loss": 0.9631,
+      "step": 12378
+    },
+    {
+      "epoch": 0.339979676471396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5370099121704317e-05,
+      "loss": 0.998,
+      "step": 12379
+    },
+    {
+      "epoch": 0.34000714069923926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5369348726160102e-05,
+      "loss": 0.9995,
+      "step": 12380
+    },
+    {
+      "epoch": 0.3400346049270825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5368598288131523e-05,
+      "loss": 0.956,
+      "step": 12381
+    },
+    {
+      "epoch": 0.3400620691549257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5367847807624513e-05,
+      "loss": 0.9769,
+      "step": 12382
+    },
+    {
+      "epoch": 0.3400895333827689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5367097284645014e-05,
+      "loss": 0.9536,
+      "step": 12383
+    },
+    {
+      "epoch": 0.3401169976106122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5366346719198963e-05,
+      "loss": 0.9649,
+      "step": 12384
+    },
+    {
+      "epoch": 0.3401444618384554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5365596111292296e-05,
+      "loss": 0.9871,
+      "step": 12385
+    },
+    {
+      "epoch": 0.34017192606629865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5364845460930954e-05,
+      "loss": 0.9263,
+      "step": 12386
+    },
+    {
+      "epoch": 0.34019939029414187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5364094768120875e-05,
+      "loss": 1.0261,
+      "step": 12387
+    },
+    {
+      "epoch": 0.3402268545219851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5363344032868005e-05,
+      "loss": 0.8917,
+      "step": 12388
+    },
+    {
+      "epoch": 0.34025431874982837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536259325517828e-05,
+      "loss": 1.0301,
+      "step": 12389
+    },
+    {
+      "epoch": 0.3402817829776716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5361842435057634e-05,
+      "loss": 0.9919,
+      "step": 12390
+    },
+    {
+      "epoch": 0.3403092472055148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536109157251202e-05,
+      "loss": 0.9311,
+      "step": 12391
+    },
+    {
+      "epoch": 0.34033671143335803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536034066754737e-05,
+      "loss": 0.9825,
+      "step": 12392
+    },
+    {
+      "epoch": 0.3403641756612013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5359589720169625e-05,
+      "loss": 0.9692,
+      "step": 12393
+    },
+    {
+      "epoch": 0.34039163988904453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5358838730384733e-05,
+      "loss": 0.9543,
+      "step": 12394
+    },
+    {
+      "epoch": 0.34041910411688775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535808769819863e-05,
+      "loss": 0.9756,
+      "step": 12395
+    },
+    {
+      "epoch": 0.34044656834473097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5357336623617265e-05,
+      "loss": 1.0207,
+      "step": 12396
+    },
+    {
+      "epoch": 0.34047403257257425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5356585506646578e-05,
+      "loss": 0.9626,
+      "step": 12397
+    },
+    {
+      "epoch": 0.34050149680041747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535583434729251e-05,
+      "loss": 0.9651,
+      "step": 12398
+    },
+    {
+      "epoch": 0.3405289610282607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5355083145561008e-05,
+      "loss": 0.9772,
+      "step": 12399
+    },
+    {
+      "epoch": 0.3405564252561039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5354331901458014e-05,
+      "loss": 0.8927,
+      "step": 12400
+    },
+    {
+      "epoch": 0.34058388948394713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5353580614989468e-05,
+      "loss": 0.8136,
+      "step": 12401
+    },
+    {
+      "epoch": 0.3406113537117904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535282928616132e-05,
+      "loss": 1.0574,
+      "step": 12402
+    },
+    {
+      "epoch": 0.34063881793963363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5352077914979513e-05,
+      "loss": 0.9805,
+      "step": 12403
+    },
+    {
+      "epoch": 0.34066628216747685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535132650144999e-05,
+      "loss": 1.0085,
+      "step": 12404
+    },
+    {
+      "epoch": 0.3406937463953201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5350575045578704e-05,
+      "loss": 0.971,
+      "step": 12405
+    },
+    {
+      "epoch": 0.34072121062316335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5349823547371592e-05,
+      "loss": 0.946,
+      "step": 12406
+    },
+    {
+      "epoch": 0.34074867485100657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5349072006834606e-05,
+      "loss": 0.9628,
+      "step": 12407
+    },
+    {
+      "epoch": 0.3407761390788498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534832042397369e-05,
+      "loss": 1.0351,
+      "step": 12408
+    },
+    {
+      "epoch": 0.340803603306693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534756879879479e-05,
+      "loss": 1.0082,
+      "step": 12409
+    },
+    {
+      "epoch": 0.3408310675345363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5346817131303856e-05,
+      "loss": 0.9973,
+      "step": 12410
+    },
+    {
+      "epoch": 0.3408585317623795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5346065421506833e-05,
+      "loss": 0.9807,
+      "step": 12411
+    },
+    {
+      "epoch": 0.34088599599022273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534531366940967e-05,
+      "loss": 0.921,
+      "step": 12412
+    },
+    {
+      "epoch": 0.34091346021806596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534456187501831e-05,
+      "loss": 1.0458,
+      "step": 12413
+    },
+    {
+      "epoch": 0.3409409244459092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5343810038338708e-05,
+      "loss": 1.0645,
+      "step": 12414
+    },
+    {
+      "epoch": 0.34096838867375245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5343058159376815e-05,
+      "loss": 0.989,
+      "step": 12415
+    },
+    {
+      "epoch": 0.3409958529015957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5342306238138577e-05,
+      "loss": 0.8832,
+      "step": 12416
+    },
+    {
+      "epoch": 0.3410233171294389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5341554274629938e-05,
+      "loss": 0.9933,
+      "step": 12417
+    },
+    {
+      "epoch": 0.3410507813572821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5340802268856853e-05,
+      "loss": 0.9772,
+      "step": 12418
+    },
+    {
+      "epoch": 0.3410782455851254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534005022082527e-05,
+      "loss": 0.997,
+      "step": 12419
+    },
+    {
+      "epoch": 0.3411057098129686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533929813054115e-05,
+      "loss": 0.9715,
+      "step": 12420
+    },
+    {
+      "epoch": 0.34113317404081184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5338545998010426e-05,
+      "loss": 0.9605,
+      "step": 12421
+    },
+    {
+      "epoch": 0.34116063826865506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533779382323906e-05,
+      "loss": 0.9931,
+      "step": 12422
+    },
+    {
+      "epoch": 0.34118810249649834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5337041606233005e-05,
+      "loss": 0.9957,
+      "step": 12423
+    },
+    {
+      "epoch": 0.34121556672434156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5336289346998205e-05,
+      "loss": 1.0138,
+      "step": 12424
+    },
+    {
+      "epoch": 0.3412430309521848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533553704554062e-05,
+      "loss": 0.9055,
+      "step": 12425
+    },
+    {
+      "epoch": 0.341270495180028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5334784701866195e-05,
+      "loss": 0.8572,
+      "step": 12426
+    },
+    {
+      "epoch": 0.3412979594078712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5334032315980888e-05,
+      "loss": 1.0057,
+      "step": 12427
+    },
+    {
+      "epoch": 0.3413254236357145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5333279887890652e-05,
+      "loss": 0.899,
+      "step": 12428
+    },
+    {
+      "epoch": 0.3413528878635577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5332527417601443e-05,
+      "loss": 0.9107,
+      "step": 12429
+    },
+    {
+      "epoch": 0.34138035209140094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5331774905119207e-05,
+      "loss": 1.0226,
+      "step": 12430
+    },
+    {
+      "epoch": 0.34140781631924416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5331022350449903e-05,
+      "loss": 1.0032,
+      "step": 12431
+    },
+    {
+      "epoch": 0.34143528054708744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533026975359949e-05,
+      "loss": 1.0866,
+      "step": 12432
+    },
+    {
+      "epoch": 0.34146274477493066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5329517114573914e-05,
+      "loss": 0.9132,
+      "step": 12433
+    },
+    {
+      "epoch": 0.3414902090027739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532876443337913e-05,
+      "loss": 1.0313,
+      "step": 12434
+    },
+    {
+      "epoch": 0.3415176732306171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5328011710021105e-05,
+      "loss": 0.8329,
+      "step": 12435
+    },
+    {
+      "epoch": 0.3415451374584604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532725894450579e-05,
+      "loss": 0.9144,
+      "step": 12436
+    },
+    {
+      "epoch": 0.3415726016863036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5326506136839128e-05,
+      "loss": 0.9893,
+      "step": 12437
+    },
+    {
+      "epoch": 0.3416000659141468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532575328702709e-05,
+      "loss": 0.9043,
+      "step": 12438
+    },
+    {
+      "epoch": 0.34162753014199004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532500039507563e-05,
+      "loss": 0.9551,
+      "step": 12439
+    },
+    {
+      "epoch": 0.34165499436983326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5324247460990705e-05,
+      "loss": 0.936,
+      "step": 12440
+    },
+    {
+      "epoch": 0.34168245859767654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532349448477827e-05,
+      "loss": 0.9784,
+      "step": 12441
+    },
+    {
+      "epoch": 0.34170992282551976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532274146644429e-05,
+      "loss": 0.997,
+      "step": 12442
+    },
+    {
+      "epoch": 0.341737387053363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5321988405994713e-05,
+      "loss": 1.0146,
+      "step": 12443
+    },
+    {
+      "epoch": 0.3417648512812062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53212353034355e-05,
+      "loss": 0.925,
+      "step": 12444
+    },
+    {
+      "epoch": 0.3417923155090495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5320482158772617e-05,
+      "loss": 1.0212,
+      "step": 12445
+    },
+    {
+      "epoch": 0.3418197797368927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5319728972012017e-05,
+      "loss": 0.9442,
+      "step": 12446
+    },
+    {
+      "epoch": 0.3418472439647359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531897574315966e-05,
+      "loss": 0.9868,
+      "step": 12447
+    },
+    {
+      "epoch": 0.34187470819257915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5318222472221504e-05,
+      "loss": 0.9459,
+      "step": 12448
+    },
+    {
+      "epoch": 0.3419021724204224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5317469159203514e-05,
+      "loss": 0.9496,
+      "step": 12449
+    },
+    {
+      "epoch": 0.34192963664826564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5316715804111646e-05,
+      "loss": 0.9886,
+      "step": 12450
+    },
+    {
+      "epoch": 0.34195710087610887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5315962406951867e-05,
+      "loss": 0.9352,
+      "step": 12451
+    },
+    {
+      "epoch": 0.3419845651039521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5315208967730132e-05,
+      "loss": 0.9447,
+      "step": 12452
+    },
+    {
+      "epoch": 0.34201202933179536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5314455486452405e-05,
+      "loss": 1.0152,
+      "step": 12453
+    },
+    {
+      "epoch": 0.3420394935596386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531370196312465e-05,
+      "loss": 0.9657,
+      "step": 12454
+    },
+    {
+      "epoch": 0.3420669577874818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531294839775282e-05,
+      "loss": 0.9213,
+      "step": 12455
+    },
+    {
+      "epoch": 0.34209442201532503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531219479034289e-05,
+      "loss": 0.9242,
+      "step": 12456
+    },
+    {
+      "epoch": 0.34212188624316825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5311441140900816e-05,
+      "loss": 1.0737,
+      "step": 12457
+    },
+    {
+      "epoch": 0.3421493504710115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5310687449432558e-05,
+      "loss": 0.8908,
+      "step": 12458
+    },
+    {
+      "epoch": 0.34217681469885475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5309933715944093e-05,
+      "loss": 0.8907,
+      "step": 12459
+    },
+    {
+      "epoch": 0.34220427892669797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530917994044137e-05,
+      "loss": 1.0076,
+      "step": 12460
+    },
+    {
+      "epoch": 0.3422317431545412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5308426122930358e-05,
+      "loss": 1.0121,
+      "step": 12461
+    },
+    {
+      "epoch": 0.34225920738238447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5307672263417024e-05,
+      "loss": 0.8835,
+      "step": 12462
+    },
+    {
+      "epoch": 0.3422866716102277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530691836190733e-05,
+      "loss": 0.8814,
+      "step": 12463
+    },
+    {
+      "epoch": 0.3423141358380709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530616441840724e-05,
+      "loss": 0.9725,
+      "step": 12464
+    },
+    {
+      "epoch": 0.34234160006591413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5305410432922725e-05,
+      "loss": 0.922,
+      "step": 12465
+    },
+    {
+      "epoch": 0.3423690642937574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5304656405459746e-05,
+      "loss": 0.9728,
+      "step": 12466
+    },
+    {
+      "epoch": 0.34239652852160063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530390233602427e-05,
+      "loss": 0.8812,
+      "step": 12467
+    },
+    {
+      "epoch": 0.34242399274944385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530314822462227e-05,
+      "loss": 1.0057,
+      "step": 12468
+    },
+    {
+      "epoch": 0.34245145697728707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53023940712597e-05,
+      "loss": 0.9899,
+      "step": 12469
+    },
+    {
+      "epoch": 0.3424789212051303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5301639875942537e-05,
+      "loss": 0.9987,
+      "step": 12470
+    },
+    {
+      "epoch": 0.34250638543297357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5300885638676747e-05,
+      "loss": 0.9259,
+      "step": 12471
+    },
+    {
+      "epoch": 0.3425338496608168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530013135946829e-05,
+      "loss": 0.8303,
+      "step": 12472
+    },
+    {
+      "epoch": 0.34256131388866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529937703832315e-05,
+      "loss": 0.9332,
+      "step": 12473
+    },
+    {
+      "epoch": 0.34258877811650323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5298622675247283e-05,
+      "loss": 0.8518,
+      "step": 12474
+    },
+    {
+      "epoch": 0.3426162423443465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5297868270246655e-05,
+      "loss": 0.8763,
+      "step": 12475
+    },
+    {
+      "epoch": 0.34264370657218973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5297113823327248e-05,
+      "loss": 0.9709,
+      "step": 12476
+    },
+    {
+      "epoch": 0.34267117080003295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529635933449502e-05,
+      "loss": 1.0101,
+      "step": 12477
+    },
+    {
+      "epoch": 0.3426986350278762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5295604803755948e-05,
+      "loss": 1.0541,
+      "step": 12478
+    },
+    {
+      "epoch": 0.34272609925571945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5294850231116e-05,
+      "loss": 1.0353,
+      "step": 12479
+    },
+    {
+      "epoch": 0.3427535634835627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5294095616581147e-05,
+      "loss": 1.0086,
+      "step": 12480
+    },
+    {
+      "epoch": 0.3427810277114059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5293340960157356e-05,
+      "loss": 1.0205,
+      "step": 12481
+    },
+    {
+      "epoch": 0.3428084919392491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5292586261850598e-05,
+      "loss": 0.9271,
+      "step": 12482
+    },
+    {
+      "epoch": 0.34283595616709234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529183152166685e-05,
+      "loss": 0.9902,
+      "step": 12483
+    },
+    {
+      "epoch": 0.3428634203949356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5291076739612084e-05,
+      "loss": 1.0225,
+      "step": 12484
+    },
+    {
+      "epoch": 0.34289088462277884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5290321915692265e-05,
+      "loss": 1.0371,
+      "step": 12485
+    },
+    {
+      "epoch": 0.34291834885062206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5289567049913375e-05,
+      "loss": 1.011,
+      "step": 12486
+    },
+    {
+      "epoch": 0.3429458130784653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5288812142281377e-05,
+      "loss": 0.898,
+      "step": 12487
+    },
+    {
+      "epoch": 0.34297327730630855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5288057192802252e-05,
+      "loss": 0.8595,
+      "step": 12488
+    },
+    {
+      "epoch": 0.3430007415341518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5287302201481966e-05,
+      "loss": 0.9562,
+      "step": 12489
+    },
+    {
+      "epoch": 0.343028205761995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52865471683265e-05,
+      "loss": 0.8475,
+      "step": 12490
+    },
+    {
+      "epoch": 0.3430556699898382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528579209334182e-05,
+      "loss": 0.9933,
+      "step": 12491
+    },
+    {
+      "epoch": 0.3430831342176815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5285036976533912e-05,
+      "loss": 0.9795,
+      "step": 12492
+    },
+    {
+      "epoch": 0.3431105984455247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5284281817908743e-05,
+      "loss": 1.0079,
+      "step": 12493
+    },
+    {
+      "epoch": 0.34313806267336794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5283526617472285e-05,
+      "loss": 0.9257,
+      "step": 12494
+    },
+    {
+      "epoch": 0.34316552690121116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5282771375230523e-05,
+      "loss": 0.9634,
+      "step": 12495
+    },
+    {
+      "epoch": 0.3431929911290544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5282016091189423e-05,
+      "loss": 0.8671,
+      "step": 12496
+    },
+    {
+      "epoch": 0.34322045535689766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528126076535497e-05,
+      "loss": 0.9272,
+      "step": 12497
+    },
+    {
+      "epoch": 0.3432479195847409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528050539773313e-05,
+      "loss": 0.9677,
+      "step": 12498
+    },
+    {
+      "epoch": 0.3432753838125841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5279749988329894e-05,
+      "loss": 0.8962,
+      "step": 12499
+    },
+    {
+      "epoch": 0.3433028480404273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5278994537151226e-05,
+      "loss": 0.9159,
+      "step": 12500
+    },
+    {
+      "epoch": 0.3433303122682706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5278239044203108e-05,
+      "loss": 0.9094,
+      "step": 12501
+    },
+    {
+      "epoch": 0.3433577764961138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5277483509491518e-05,
+      "loss": 0.9265,
+      "step": 12502
+    },
+    {
+      "epoch": 0.34338524072395704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5276727933022437e-05,
+      "loss": 0.8927,
+      "step": 12503
+    },
+    {
+      "epoch": 0.34341270495180026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5275972314801842e-05,
+      "loss": 1.0995,
+      "step": 12504
+    },
+    {
+      "epoch": 0.34344016917964354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5275216654835703e-05,
+      "loss": 0.9918,
+      "step": 12505
+    },
+    {
+      "epoch": 0.34346763340748676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.527446095313001e-05,
+      "loss": 1.018,
+      "step": 12506
+    },
+    {
+      "epoch": 0.34349509763533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.527370520969074e-05,
+      "loss": 0.8671,
+      "step": 12507
+    },
+    {
+      "epoch": 0.3435225618631732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5272949424523872e-05,
+      "loss": 1.0227,
+      "step": 12508
+    },
+    {
+      "epoch": 0.3435500260910164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5272193597635382e-05,
+      "loss": 0.8957,
+      "step": 12509
+    },
+    {
+      "epoch": 0.3435774903188597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5271437729031258e-05,
+      "loss": 1.0593,
+      "step": 12510
+    },
+    {
+      "epoch": 0.3436049545467029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5270681818717473e-05,
+      "loss": 0.91,
+      "step": 12511
+    },
+    {
+      "epoch": 0.34363241877454614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5269925866700015e-05,
+      "loss": 0.806,
+      "step": 12512
+    },
+    {
+      "epoch": 0.34365988300238937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5269169872984862e-05,
+      "loss": 1.0986,
+      "step": 12513
+    },
+    {
+      "epoch": 0.34368734723023264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5268413837577996e-05,
+      "loss": 0.986,
+      "step": 12514
+    },
+    {
+      "epoch": 0.34371481145807586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52676577604854e-05,
+      "loss": 0.9652,
+      "step": 12515
+    },
+    {
+      "epoch": 0.3437422756859191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5266901641713052e-05,
+      "loss": 1.0219,
+      "step": 12516
+    },
+    {
+      "epoch": 0.3437697399137623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526614548126694e-05,
+      "loss": 1.0305,
+      "step": 12517
+    },
+    {
+      "epoch": 0.3437972041416056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526538927915304e-05,
+      "loss": 0.9386,
+      "step": 12518
+    },
+    {
+      "epoch": 0.3438246683694488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5264633035377347e-05,
+      "loss": 1.0874,
+      "step": 12519
+    },
+    {
+      "epoch": 0.343852132597292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5263876749945837e-05,
+      "loss": 0.8925,
+      "step": 12520
+    },
+    {
+      "epoch": 0.34387959682513525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5263120422864493e-05,
+      "loss": 0.9281,
+      "step": 12521
+    },
+    {
+      "epoch": 0.34390706105297847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5262364054139306e-05,
+      "loss": 0.9897,
+      "step": 12522
+    },
+    {
+      "epoch": 0.34393452528082175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5261607643776247e-05,
+      "loss": 1.1175,
+      "step": 12523
+    },
+    {
+      "epoch": 0.34396198950866497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526085119178132e-05,
+      "loss": 0.9907,
+      "step": 12524
+    },
+    {
+      "epoch": 0.3439894537365082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5260094698160493e-05,
+      "loss": 0.8317,
+      "step": 12525
+    },
+    {
+      "epoch": 0.3440169179643514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5259338162919762e-05,
+      "loss": 0.963,
+      "step": 12526
+    },
+    {
+      "epoch": 0.3440443821921947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5258581586065113e-05,
+      "loss": 1.0033,
+      "step": 12527
+    },
+    {
+      "epoch": 0.3440718464200379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5257824967602528e-05,
+      "loss": 0.9023,
+      "step": 12528
+    },
+    {
+      "epoch": 0.34409931064788113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5257068307537992e-05,
+      "loss": 1.0123,
+      "step": 12529
+    },
+    {
+      "epoch": 0.34412677487572435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52563116058775e-05,
+      "loss": 0.9479,
+      "step": 12530
+    },
+    {
+      "epoch": 0.3441542391035676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5255554862627028e-05,
+      "loss": 0.9554,
+      "step": 12531
+    },
+    {
+      "epoch": 0.34418170333141085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5254798077792576e-05,
+      "loss": 0.9298,
+      "step": 12532
+    },
+    {
+      "epoch": 0.34420916755925407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5254041251380124e-05,
+      "loss": 1.0305,
+      "step": 12533
+    },
+    {
+      "epoch": 0.3442366317870973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5253284383395663e-05,
+      "loss": 1.0178,
+      "step": 12534
+    },
+    {
+      "epoch": 0.3442640960149405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.525252747384518e-05,
+      "loss": 0.8347,
+      "step": 12535
+    },
+    {
+      "epoch": 0.3442915602427838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5251770522734669e-05,
+      "loss": 1.0016,
+      "step": 12536
+    },
+    {
+      "epoch": 0.344319024470627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.525101353007011e-05,
+      "loss": 0.9622,
+      "step": 12537
+    },
+    {
+      "epoch": 0.34434648869847023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52502564958575e-05,
+      "loss": 0.9348,
+      "step": 12538
+    },
+    {
+      "epoch": 0.34437395292631345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5249499420102827e-05,
+      "loss": 0.8967,
+      "step": 12539
+    },
+    {
+      "epoch": 0.34440141715415673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524874230281208e-05,
+      "loss": 0.9175,
+      "step": 12540
+    },
+    {
+      "epoch": 0.34442888138199995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5247985143991253e-05,
+      "loss": 0.9434,
+      "step": 12541
+    },
+    {
+      "epoch": 0.3444563456098432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5247227943646332e-05,
+      "loss": 0.8425,
+      "step": 12542
+    },
+    {
+      "epoch": 0.3444838098376864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524647070178331e-05,
+      "loss": 1.0023,
+      "step": 12543
+    },
+    {
+      "epoch": 0.34451127406552967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5245713418408183e-05,
+      "loss": 0.9923,
+      "step": 12544
+    },
+    {
+      "epoch": 0.3445387382933729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5244956093526942e-05,
+      "loss": 0.8633,
+      "step": 12545
+    },
+    {
+      "epoch": 0.3445662025212161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524419872714557e-05,
+      "loss": 0.9004,
+      "step": 12546
+    },
+    {
+      "epoch": 0.34459366674905934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5243441319270066e-05,
+      "loss": 0.9307,
+      "step": 12547
+    },
+    {
+      "epoch": 0.3446211309769026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5242683869906429e-05,
+      "loss": 0.9763,
+      "step": 12548
+    },
+    {
+      "epoch": 0.34464859520474583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5241926379060642e-05,
+      "loss": 0.9339,
+      "step": 12549
+    },
+    {
+      "epoch": 0.34467605943258905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5241168846738705e-05,
+      "loss": 0.9634,
+      "step": 12550
+    },
+    {
+      "epoch": 0.3447035236604323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5240411272946612e-05,
+      "loss": 0.9857,
+      "step": 12551
+    },
+    {
+      "epoch": 0.3447309878882755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5239653657690352e-05,
+      "loss": 0.8901,
+      "step": 12552
+    },
+    {
+      "epoch": 0.3447584521161188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5238896000975922e-05,
+      "loss": 0.9279,
+      "step": 12553
+    },
+    {
+      "epoch": 0.344785916343962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5238138302809318e-05,
+      "loss": 0.8659,
+      "step": 12554
+    },
+    {
+      "epoch": 0.3448133805718052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5237380563196533e-05,
+      "loss": 0.9075,
+      "step": 12555
+    },
+    {
+      "epoch": 0.34484084479964844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5236622782143568e-05,
+      "loss": 0.9024,
+      "step": 12556
+    },
+    {
+      "epoch": 0.3448683090274917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5235864959656414e-05,
+      "loss": 0.8936,
+      "step": 12557
+    },
+    {
+      "epoch": 0.34489577325533494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.523510709574107e-05,
+      "loss": 0.8145,
+      "step": 12558
+    },
+    {
+      "epoch": 0.34492323748317816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5234349190403528e-05,
+      "loss": 0.9485,
+      "step": 12559
+    },
+    {
+      "epoch": 0.3449507017110214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5233591243649788e-05,
+      "loss": 1.0333,
+      "step": 12560
+    },
+    {
+      "epoch": 0.34497816593886466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5232833255485846e-05,
+      "loss": 1.0528,
+      "step": 12561
+    },
+    {
+      "epoch": 0.3450056301667079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5232075225917701e-05,
+      "loss": 1.0115,
+      "step": 12562
+    },
+    {
+      "epoch": 0.3450330943945511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5231317154951351e-05,
+      "loss": 0.922,
+      "step": 12563
+    },
+    {
+      "epoch": 0.3450605586223943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5230559042592793e-05,
+      "loss": 1.0162,
+      "step": 12564
+    },
+    {
+      "epoch": 0.34508802285023754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5229800888848022e-05,
+      "loss": 0.968,
+      "step": 12565
+    },
+    {
+      "epoch": 0.3451154870780808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5229042693723046e-05,
+      "loss": 0.9623,
+      "step": 12566
+    },
+    {
+      "epoch": 0.34514295130592404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5228284457223856e-05,
+      "loss": 1.0241,
+      "step": 12567
+    },
+    {
+      "epoch": 0.34517041553376726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5227526179356456e-05,
+      "loss": 0.9474,
+      "step": 12568
+    },
+    {
+      "epoch": 0.3451978797616105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5226767860126843e-05,
+      "loss": 0.9608,
+      "step": 12569
+    },
+    {
+      "epoch": 0.34522534398945376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5226009499541017e-05,
+      "loss": 0.9045,
+      "step": 12570
+    },
+    {
+      "epoch": 0.345252808217297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5225251097604981e-05,
+      "loss": 0.9448,
+      "step": 12571
+    },
+    {
+      "epoch": 0.3452802724451402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5224492654324733e-05,
+      "loss": 0.9774,
+      "step": 12572
+    },
+    {
+      "epoch": 0.3453077366729834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5223734169706274e-05,
+      "loss": 1.0376,
+      "step": 12573
+    },
+    {
+      "epoch": 0.3453352009008267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5222975643755609e-05,
+      "loss": 0.9125,
+      "step": 12574
+    },
+    {
+      "epoch": 0.3453626651286699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5222217076478738e-05,
+      "loss": 1.0116,
+      "step": 12575
+    },
+    {
+      "epoch": 0.34539012935651314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.522145846788166e-05,
+      "loss": 0.9172,
+      "step": 12576
+    },
+    {
+      "epoch": 0.34541759358435636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5220699817970383e-05,
+      "loss": 1.048,
+      "step": 12577
+    },
+    {
+      "epoch": 0.3454450578121996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5219941126750905e-05,
+      "loss": 0.9625,
+      "step": 12578
+    },
+    {
+      "epoch": 0.34547252204004286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5219182394229228e-05,
+      "loss": 1.0121,
+      "step": 12579
+    },
+    {
+      "epoch": 0.3454999862678861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5218423620411367e-05,
+      "loss": 1.0626,
+      "step": 12580
+    },
+    {
+      "epoch": 0.3455274504957293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.521766480530331e-05,
+      "loss": 0.9205,
+      "step": 12581
+    },
+    {
+      "epoch": 0.3455549147235725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5216905948911067e-05,
+      "loss": 1.0142,
+      "step": 12582
+    },
+    {
+      "epoch": 0.3455823789514158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5216147051240647e-05,
+      "loss": 1.0098,
+      "step": 12583
+    },
+    {
+      "epoch": 0.345609843179259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5215388112298048e-05,
+      "loss": 0.9039,
+      "step": 12584
+    },
+    {
+      "epoch": 0.34563730740710225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5214629132089278e-05,
+      "loss": 1.088,
+      "step": 12585
+    },
+    {
+      "epoch": 0.34566477163494547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5213870110620346e-05,
+      "loss": 0.9401,
+      "step": 12586
+    },
+    {
+      "epoch": 0.34569223586278874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.521311104789725e-05,
+      "loss": 0.8984,
+      "step": 12587
+    },
+    {
+      "epoch": 0.34571970009063197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5212351943926003e-05,
+      "loss": 0.9583,
+      "step": 12588
+    },
+    {
+      "epoch": 0.3457471643184752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.521159279871261e-05,
+      "loss": 0.8979,
+      "step": 12589
+    },
+    {
+      "epoch": 0.3457746285463184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5210833612263072e-05,
+      "loss": 0.9857,
+      "step": 12590
+    },
+    {
+      "epoch": 0.34580209277416163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52100743845834e-05,
+      "loss": 0.9452,
+      "step": 12591
+    },
+    {
+      "epoch": 0.3458295570020049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5209315115679604e-05,
+      "loss": 1.0557,
+      "step": 12592
+    },
+    {
+      "epoch": 0.3458570212298481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5208555805557685e-05,
+      "loss": 0.7975,
+      "step": 12593
+    },
+    {
+      "epoch": 0.34588448545769135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5207796454223655e-05,
+      "loss": 0.929,
+      "step": 12594
+    },
+    {
+      "epoch": 0.34591194968553457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5207037061683524e-05,
+      "loss": 0.971,
+      "step": 12595
+    },
+    {
+      "epoch": 0.34593941391337785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5206277627943302e-05,
+      "loss": 0.963,
+      "step": 12596
+    },
+    {
+      "epoch": 0.34596687814122107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5205518153008992e-05,
+      "loss": 0.9214,
+      "step": 12597
+    },
+    {
+      "epoch": 0.3459943423690643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5204758636886605e-05,
+      "loss": 0.9388,
+      "step": 12598
+    },
+    {
+      "epoch": 0.3460218065969075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5203999079582152e-05,
+      "loss": 0.8436,
+      "step": 12599
+    },
+    {
+      "epoch": 0.3460492708247508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5203239481101641e-05,
+      "loss": 0.8602,
+      "step": 12600
+    },
+    {
+      "epoch": 0.346076735052594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5202479841451088e-05,
+      "loss": 0.9362,
+      "step": 12601
+    },
+    {
+      "epoch": 0.34610419928043723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5201720160636497e-05,
+      "loss": 0.9826,
+      "step": 12602
+    },
+    {
+      "epoch": 0.34613166350828045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5200960438663879e-05,
+      "loss": 0.9315,
+      "step": 12603
+    },
+    {
+      "epoch": 0.3461591277361237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5200200675539249e-05,
+      "loss": 0.991,
+      "step": 12604
+    },
+    {
+      "epoch": 0.34618659196396695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5199440871268618e-05,
+      "loss": 0.9369,
+      "step": 12605
+    },
+    {
+      "epoch": 0.34621405619181017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5198681025857993e-05,
+      "loss": 0.9929,
+      "step": 12606
+    },
+    {
+      "epoch": 0.3462415204196534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5197921139313394e-05,
+      "loss": 0.9355,
+      "step": 12607
+    },
+    {
+      "epoch": 0.3462689846474966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519716121164083e-05,
+      "loss": 1.0067,
+      "step": 12608
+    },
+    {
+      "epoch": 0.3462964488753399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519640124284631e-05,
+      "loss": 0.9219,
+      "step": 12609
+    },
+    {
+      "epoch": 0.3463239131031831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5195641232935852e-05,
+      "loss": 0.9706,
+      "step": 12610
+    },
+    {
+      "epoch": 0.34635137733102633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519488118191547e-05,
+      "loss": 0.9304,
+      "step": 12611
+    },
+    {
+      "epoch": 0.34637884155886955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5194121089791174e-05,
+      "loss": 0.9107,
+      "step": 12612
+    },
+    {
+      "epoch": 0.34640630578671283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519336095656898e-05,
+      "loss": 0.907,
+      "step": 12613
+    },
+    {
+      "epoch": 0.34643377001455605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5192600782254899e-05,
+      "loss": 0.9641,
+      "step": 12614
+    },
+    {
+      "epoch": 0.3464612342423993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5191840566854952e-05,
+      "loss": 1.0217,
+      "step": 12615
+    },
+    {
+      "epoch": 0.3464886984702425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5191080310375155e-05,
+      "loss": 1.0596,
+      "step": 12616
+    },
+    {
+      "epoch": 0.3465161626980857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5190320012821516e-05,
+      "loss": 0.9878,
+      "step": 12617
+    },
+    {
+      "epoch": 0.346543626925929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5189559674200053e-05,
+      "loss": 0.9246,
+      "step": 12618
+    },
+    {
+      "epoch": 0.3465710911537722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5188799294516784e-05,
+      "loss": 0.9332,
+      "step": 12619
+    },
+    {
+      "epoch": 0.34659855538161544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5188038873777728e-05,
+      "loss": 1.0812,
+      "step": 12620
+    },
+    {
+      "epoch": 0.34662601960945866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5187278411988895e-05,
+      "loss": 0.8349,
+      "step": 12621
+    },
+    {
+      "epoch": 0.34665348383730193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5186517909156311e-05,
+      "loss": 1.0212,
+      "step": 12622
+    },
+    {
+      "epoch": 0.34668094806514516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5185757365285984e-05,
+      "loss": 0.8912,
+      "step": 12623
+    },
+    {
+      "epoch": 0.3467084122929884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5184996780383938e-05,
+      "loss": 1.0133,
+      "step": 12624
+    },
+    {
+      "epoch": 0.3467358765208316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5184236154456188e-05,
+      "loss": 1.0223,
+      "step": 12625
+    },
+    {
+      "epoch": 0.3467633407486749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5183475487508754e-05,
+      "loss": 0.9566,
+      "step": 12626
+    },
+    {
+      "epoch": 0.3467908049765181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5182714779547652e-05,
+      "loss": 1.0082,
+      "step": 12627
+    },
+    {
+      "epoch": 0.3468182692043613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5181954030578906e-05,
+      "loss": 0.8899,
+      "step": 12628
+    },
+    {
+      "epoch": 0.34684573343220454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.518119324060853e-05,
+      "loss": 0.9898,
+      "step": 12629
+    },
+    {
+      "epoch": 0.34687319766004776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5180432409642547e-05,
+      "loss": 0.9379,
+      "step": 12630
+    },
+    {
+      "epoch": 0.34690066188789104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5179671537686976e-05,
+      "loss": 0.9534,
+      "step": 12631
+    },
+    {
+      "epoch": 0.34692812611573426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5178910624747838e-05,
+      "loss": 0.9203,
+      "step": 12632
+    },
+    {
+      "epoch": 0.3469555903435775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5178149670831153e-05,
+      "loss": 1.0584,
+      "step": 12633
+    },
+    {
+      "epoch": 0.3469830545714207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5177388675942943e-05,
+      "loss": 1.0828,
+      "step": 12634
+    },
+    {
+      "epoch": 0.347010518799264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5176627640089226e-05,
+      "loss": 1.014,
+      "step": 12635
+    },
+    {
+      "epoch": 0.3470379830271072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5175866563276027e-05,
+      "loss": 0.9937,
+      "step": 12636
+    },
+    {
+      "epoch": 0.3470654472549504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5175105445509366e-05,
+      "loss": 0.9667,
+      "step": 12637
+    },
+    {
+      "epoch": 0.34709291148279364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5174344286795266e-05,
+      "loss": 0.9516,
+      "step": 12638
+    },
+    {
+      "epoch": 0.3471203757106369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5173583087139751e-05,
+      "loss": 0.9499,
+      "step": 12639
+    },
+    {
+      "epoch": 0.34714783993848014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5172821846548841e-05,
+      "loss": 0.8977,
+      "step": 12640
+    },
+    {
+      "epoch": 0.34717530416632336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5172060565028563e-05,
+      "loss": 0.9219,
+      "step": 12641
+    },
+    {
+      "epoch": 0.3472027683941666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5171299242584938e-05,
+      "loss": 1.032,
+      "step": 12642
+    },
+    {
+      "epoch": 0.3472302326220098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5170537879223988e-05,
+      "loss": 1.0081,
+      "step": 12643
+    },
+    {
+      "epoch": 0.3472576968498531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5169776474951742e-05,
+      "loss": 0.9915,
+      "step": 12644
+    },
+    {
+      "epoch": 0.3472851610776963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.516901502977422e-05,
+      "loss": 1.086,
+      "step": 12645
+    },
+    {
+      "epoch": 0.3473126253055395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5168253543697448e-05,
+      "loss": 0.9791,
+      "step": 12646
+    },
+    {
+      "epoch": 0.34734008953338275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5167492016727452e-05,
+      "loss": 0.9299,
+      "step": 12647
+    },
+    {
+      "epoch": 0.347367553761226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5166730448870262e-05,
+      "loss": 0.9525,
+      "step": 12648
+    },
+    {
+      "epoch": 0.34739501798906924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5165968840131896e-05,
+      "loss": 0.9367,
+      "step": 12649
+    },
+    {
+      "epoch": 0.34742248221691246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5165207190518383e-05,
+      "loss": 0.9662,
+      "step": 12650
+    },
+    {
+      "epoch": 0.3474499464447557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5164445500035751e-05,
+      "loss": 0.9672,
+      "step": 12651
+    },
+    {
+      "epoch": 0.34747741067259896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5163683768690027e-05,
+      "loss": 0.9542,
+      "step": 12652
+    },
+    {
+      "epoch": 0.3475048749004422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5162921996487234e-05,
+      "loss": 0.9669,
+      "step": 12653
+    },
+    {
+      "epoch": 0.3475323391282854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5162160183433403e-05,
+      "loss": 1.0528,
+      "step": 12654
+    },
+    {
+      "epoch": 0.3475598033561286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.516139832953456e-05,
+      "loss": 0.903,
+      "step": 12655
+    },
+    {
+      "epoch": 0.3475872675839719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5160636434796735e-05,
+      "loss": 0.9699,
+      "step": 12656
+    },
+    {
+      "epoch": 0.3476147318118151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5159874499225956e-05,
+      "loss": 0.9441,
+      "step": 12657
+    },
+    {
+      "epoch": 0.34764219603965835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515911252282825e-05,
+      "loss": 0.9424,
+      "step": 12658
+    },
+    {
+      "epoch": 0.34766966026750157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5158350505609648e-05,
+      "loss": 0.9588,
+      "step": 12659
+    },
+    {
+      "epoch": 0.3476971244953448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5157588447576181e-05,
+      "loss": 0.8735,
+      "step": 12660
+    },
+    {
+      "epoch": 0.34772458872318807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5156826348733872e-05,
+      "loss": 0.8906,
+      "step": 12661
+    },
+    {
+      "epoch": 0.3477520529510313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5156064209088759e-05,
+      "loss": 0.936,
+      "step": 12662
+    },
+    {
+      "epoch": 0.3477795171788745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5155302028646867e-05,
+      "loss": 1.0149,
+      "step": 12663
+    },
+    {
+      "epoch": 0.34780698140671773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515453980741423e-05,
+      "loss": 0.9403,
+      "step": 12664
+    },
+    {
+      "epoch": 0.347834445634561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5153777545396877e-05,
+      "loss": 0.9958,
+      "step": 12665
+    },
+    {
+      "epoch": 0.34786190986240423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5153015242600839e-05,
+      "loss": 0.9191,
+      "step": 12666
+    },
+    {
+      "epoch": 0.34788937409024745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5152252899032145e-05,
+      "loss": 0.944,
+      "step": 12667
+    },
+    {
+      "epoch": 0.34791683831809067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5151490514696834e-05,
+      "loss": 0.9233,
+      "step": 12668
+    },
+    {
+      "epoch": 0.34794430254593395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5150728089600936e-05,
+      "loss": 0.9313,
+      "step": 12669
+    },
+    {
+      "epoch": 0.34797176677377717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514996562375048e-05,
+      "loss": 0.9586,
+      "step": 12670
+    },
+    {
+      "epoch": 0.3479992310016204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.51492031171515e-05,
+      "loss": 1.0102,
+      "step": 12671
+    },
+    {
+      "epoch": 0.3480266952294636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5148440569810033e-05,
+      "loss": 0.9481,
+      "step": 12672
+    },
+    {
+      "epoch": 0.34805415945730683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5147677981732107e-05,
+      "loss": 0.9578,
+      "step": 12673
+    },
+    {
+      "epoch": 0.3480816236851501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5146915352923758e-05,
+      "loss": 0.9173,
+      "step": 12674
+    },
+    {
+      "epoch": 0.34810908791299333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5146152683391026e-05,
+      "loss": 0.9267,
+      "step": 12675
+    },
+    {
+      "epoch": 0.34813655214083655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5145389973139936e-05,
+      "loss": 0.8773,
+      "step": 12676
+    },
+    {
+      "epoch": 0.3481640163686798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5144627222176528e-05,
+      "loss": 0.8877,
+      "step": 12677
+    },
+    {
+      "epoch": 0.34819148059652305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514386443050684e-05,
+      "loss": 0.9476,
+      "step": 12678
+    },
+    {
+      "epoch": 0.34821894482436627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.51431015981369e-05,
+      "loss": 0.9487,
+      "step": 12679
+    },
+    {
+      "epoch": 0.3482464090522095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5142338725072752e-05,
+      "loss": 0.9412,
+      "step": 12680
+    },
+    {
+      "epoch": 0.3482738732800527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5141575811320426e-05,
+      "loss": 1.025,
+      "step": 12681
+    },
+    {
+      "epoch": 0.348301337507896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514081285688596e-05,
+      "loss": 1.0734,
+      "step": 12682
+    },
+    {
+      "epoch": 0.3483288017357392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514004986177539e-05,
+      "loss": 0.9178,
+      "step": 12683
+    },
+    {
+      "epoch": 0.34835626596358243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5139286825994757e-05,
+      "loss": 1.0615,
+      "step": 12684
+    },
+    {
+      "epoch": 0.34838373019142566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5138523749550094e-05,
+      "loss": 0.9703,
+      "step": 12685
+    },
+    {
+      "epoch": 0.3484111944192689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5137760632447443e-05,
+      "loss": 0.9917,
+      "step": 12686
+    },
+    {
+      "epoch": 0.34843865864711215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5136997474692838e-05,
+      "loss": 0.9114,
+      "step": 12687
+    },
+    {
+      "epoch": 0.3484661228749554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513623427629232e-05,
+      "loss": 1.0245,
+      "step": 12688
+    },
+    {
+      "epoch": 0.3484935871027986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5135471037251925e-05,
+      "loss": 0.8681,
+      "step": 12689
+    },
+    {
+      "epoch": 0.3485210513306418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5134707757577696e-05,
+      "loss": 0.9738,
+      "step": 12690
+    },
+    {
+      "epoch": 0.3485485155584851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5133944437275668e-05,
+      "loss": 0.986,
+      "step": 12691
+    },
+    {
+      "epoch": 0.3485759797863283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5133181076351884e-05,
+      "loss": 0.9013,
+      "step": 12692
+    },
+    {
+      "epoch": 0.34860344401417154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5132417674812385e-05,
+      "loss": 0.9737,
+      "step": 12693
+    },
+    {
+      "epoch": 0.34863090824201476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5131654232663205e-05,
+      "loss": 0.9968,
+      "step": 12694
+    },
+    {
+      "epoch": 0.34865837246985804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5130890749910391e-05,
+      "loss": 1.0389,
+      "step": 12695
+    },
+    {
+      "epoch": 0.34868583669770126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5130127226559985e-05,
+      "loss": 0.9815,
+      "step": 12696
+    },
+    {
+      "epoch": 0.3487133009255445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.512936366261802e-05,
+      "loss": 0.9662,
+      "step": 12697
+    },
+    {
+      "epoch": 0.3487407651533877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5128600058090546e-05,
+      "loss": 0.9194,
+      "step": 12698
+    },
+    {
+      "epoch": 0.3487682293812309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5127836412983598e-05,
+      "loss": 1.0078,
+      "step": 12699
+    },
+    {
+      "epoch": 0.3487956936090742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5127072727303227e-05,
+      "loss": 0.9381,
+      "step": 12700
+    },
+    {
+      "epoch": 0.3488231578369174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5126309001055468e-05,
+      "loss": 0.9753,
+      "step": 12701
+    },
+    {
+      "epoch": 0.34885062206476064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5125545234246366e-05,
+      "loss": 0.9909,
+      "step": 12702
+    },
+    {
+      "epoch": 0.34887808629260386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5124781426881962e-05,
+      "loss": 0.9441,
+      "step": 12703
+    },
+    {
+      "epoch": 0.34890555052044714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5124017578968312e-05,
+      "loss": 0.9612,
+      "step": 12704
+    },
+    {
+      "epoch": 0.34893301474829036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5123253690511441e-05,
+      "loss": 0.9297,
+      "step": 12705
+    },
+    {
+      "epoch": 0.3489604789761336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5122489761517405e-05,
+      "loss": 0.9714,
+      "step": 12706
+    },
+    {
+      "epoch": 0.3489879432039768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5121725791992243e-05,
+      "loss": 0.9863,
+      "step": 12707
+    },
+    {
+      "epoch": 0.3490154074318201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5120961781942007e-05,
+      "loss": 1.0384,
+      "step": 12708
+    },
+    {
+      "epoch": 0.3490428716596633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5120197731372734e-05,
+      "loss": 0.9692,
+      "step": 12709
+    },
+    {
+      "epoch": 0.3490703358875065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5119433640290477e-05,
+      "loss": 1.0045,
+      "step": 12710
+    },
+    {
+      "epoch": 0.34909780011534974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5118669508701276e-05,
+      "loss": 0.9313,
+      "step": 12711
+    },
+    {
+      "epoch": 0.34912526434319296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5117905336611178e-05,
+      "loss": 0.964,
+      "step": 12712
+    },
+    {
+      "epoch": 0.34915272857103624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5117141124026234e-05,
+      "loss": 0.9344,
+      "step": 12713
+    },
+    {
+      "epoch": 0.34918019279887946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5116376870952483e-05,
+      "loss": 0.9486,
+      "step": 12714
+    },
+    {
+      "epoch": 0.3492076570267227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5115612577395978e-05,
+      "loss": 0.8608,
+      "step": 12715
+    },
+    {
+      "epoch": 0.3492351212545659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5114848243362765e-05,
+      "loss": 0.9394,
+      "step": 12716
+    },
+    {
+      "epoch": 0.3492625854824092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.511408386885889e-05,
+      "loss": 0.9859,
+      "step": 12717
+    },
+    {
+      "epoch": 0.3492900497102524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5113319453890403e-05,
+      "loss": 0.9487,
+      "step": 12718
+    },
+    {
+      "epoch": 0.3493175139380956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5112554998463354e-05,
+      "loss": 0.806,
+      "step": 12719
+    },
+    {
+      "epoch": 0.34934497816593885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5111790502583788e-05,
+      "loss": 0.8623,
+      "step": 12720
+    },
+    {
+      "epoch": 0.3493724423937821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5111025966257752e-05,
+      "loss": 0.8535,
+      "step": 12721
+    },
+    {
+      "epoch": 0.34939990662162534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5110261389491306e-05,
+      "loss": 0.9041,
+      "step": 12722
+    },
+    {
+      "epoch": 0.34942737084946857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5109496772290486e-05,
+      "loss": 0.9191,
+      "step": 12723
+    },
+    {
+      "epoch": 0.3494548350773118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5108732114661352e-05,
+      "loss": 0.9729,
+      "step": 12724
+    },
+    {
+      "epoch": 0.349482299305155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.510796741660995e-05,
+      "loss": 0.953,
+      "step": 12725
+    },
+    {
+      "epoch": 0.3495097635329983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5107202678142328e-05,
+      "loss": 0.8888,
+      "step": 12726
+    },
+    {
+      "epoch": 0.3495372277608415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.510643789926454e-05,
+      "loss": 0.9998,
+      "step": 12727
+    },
+    {
+      "epoch": 0.34956469198868473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5105673079982644e-05,
+      "loss": 0.9901,
+      "step": 12728
+    },
+    {
+      "epoch": 0.34959215621652795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5104908220302679e-05,
+      "loss": 0.9452,
+      "step": 12729
+    },
+    {
+      "epoch": 0.3496196204443712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5104143320230703e-05,
+      "loss": 0.9021,
+      "step": 12730
+    },
+    {
+      "epoch": 0.34964708467221445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.510337837977277e-05,
+      "loss": 0.9102,
+      "step": 12731
+    },
+    {
+      "epoch": 0.34967454890005767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5102613398934931e-05,
+      "loss": 0.9858,
+      "step": 12732
+    },
+    {
+      "epoch": 0.3497020131279009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5101848377723234e-05,
+      "loss": 0.9595,
+      "step": 12733
+    },
+    {
+      "epoch": 0.34972947735574417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5101083316143742e-05,
+      "loss": 0.9424,
+      "step": 12734
+    },
+    {
+      "epoch": 0.3497569415835874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5100318214202499e-05,
+      "loss": 0.9969,
+      "step": 12735
+    },
+    {
+      "epoch": 0.3497844058114306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5099553071905564e-05,
+      "loss": 0.9917,
+      "step": 12736
+    },
+    {
+      "epoch": 0.34981187003927383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509878788925899e-05,
+      "loss": 0.9498,
+      "step": 12737
+    },
+    {
+      "epoch": 0.34983933426711705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509802266626883e-05,
+      "loss": 0.8472,
+      "step": 12738
+    },
+    {
+      "epoch": 0.34986679849496033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5097257402941136e-05,
+      "loss": 0.9547,
+      "step": 12739
+    },
+    {
+      "epoch": 0.34989426272280355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5096492099281977e-05,
+      "loss": 1.0443,
+      "step": 12740
+    },
+    {
+      "epoch": 0.34992172695064677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5095726755297391e-05,
+      "loss": 0.9524,
+      "step": 12741
+    },
+    {
+      "epoch": 0.34994919117849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509496137099344e-05,
+      "loss": 0.9722,
+      "step": 12742
+    },
+    {
+      "epoch": 0.34997665540633327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5094195946376187e-05,
+      "loss": 0.9225,
+      "step": 12743
+    },
+    {
+      "epoch": 0.3500041196341765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5093430481451678e-05,
+      "loss": 0.9719,
+      "step": 12744
+    },
+    {
+      "epoch": 0.3500315838620197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5092664976225972e-05,
+      "loss": 0.9604,
+      "step": 12745
+    },
+    {
+      "epoch": 0.35005904808986293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5091899430705135e-05,
+      "loss": 1.0701,
+      "step": 12746
+    },
+    {
+      "epoch": 0.3500865123177062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509113384489521e-05,
+      "loss": 0.9815,
+      "step": 12747
+    },
+    {
+      "epoch": 0.35011397654554943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5090368218802267e-05,
+      "loss": 0.9682,
+      "step": 12748
+    },
+    {
+      "epoch": 0.35014144077339265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5089602552432356e-05,
+      "loss": 0.9811,
+      "step": 12749
+    },
+    {
+      "epoch": 0.3501689050012359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.508883684579154e-05,
+      "loss": 1.0132,
+      "step": 12750
+    },
+    {
+      "epoch": 0.35019636922907915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5088071098885873e-05,
+      "loss": 0.9841,
+      "step": 12751
+    },
+    {
+      "epoch": 0.3502238334569224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5087305311721424e-05,
+      "loss": 0.9705,
+      "step": 12752
+    },
+    {
+      "epoch": 0.3502512976847656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5086539484304234e-05,
+      "loss": 0.8966,
+      "step": 12753
+    },
+    {
+      "epoch": 0.3502787619126088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5085773616640379e-05,
+      "loss": 0.9224,
+      "step": 12754
+    },
+    {
+      "epoch": 0.35030622614045204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5085007708735912e-05,
+      "loss": 1.0056,
+      "step": 12755
+    },
+    {
+      "epoch": 0.3503336903682953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5084241760596897e-05,
+      "loss": 0.9624,
+      "step": 12756
+    },
+    {
+      "epoch": 0.35036115459613854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.508347577222939e-05,
+      "loss": 1.03,
+      "step": 12757
+    },
+    {
+      "epoch": 0.35038861882398176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5082709743639452e-05,
+      "loss": 1.0651,
+      "step": 12758
+    },
+    {
+      "epoch": 0.350416083051825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5081943674833146e-05,
+      "loss": 0.9249,
+      "step": 12759
+    },
+    {
+      "epoch": 0.35044354727966825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5081177565816534e-05,
+      "loss": 1.0473,
+      "step": 12760
+    },
+    {
+      "epoch": 0.3504710115075115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5080411416595678e-05,
+      "loss": 0.8316,
+      "step": 12761
+    },
+    {
+      "epoch": 0.3504984757353547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5079645227176634e-05,
+      "loss": 0.9505,
+      "step": 12762
+    },
+    {
+      "epoch": 0.3505259399631979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5078878997565473e-05,
+      "loss": 0.9731,
+      "step": 12763
+    },
+    {
+      "epoch": 0.3505534041910412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5078112727768255e-05,
+      "loss": 0.9173,
+      "step": 12764
+    },
+    {
+      "epoch": 0.3505808684188844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5077346417791039e-05,
+      "loss": 0.9322,
+      "step": 12765
+    },
+    {
+      "epoch": 0.35060833264672764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5076580067639897e-05,
+      "loss": 0.8954,
+      "step": 12766
+    },
+    {
+      "epoch": 0.35063579687457086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5075813677320882e-05,
+      "loss": 0.9104,
+      "step": 12767
+    },
+    {
+      "epoch": 0.3506632611024141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5075047246840062e-05,
+      "loss": 0.9156,
+      "step": 12768
+    },
+    {
+      "epoch": 0.35069072533025736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5074280776203508e-05,
+      "loss": 0.9531,
+      "step": 12769
+    },
+    {
+      "epoch": 0.3507181895581006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5073514265417273e-05,
+      "loss": 1.0158,
+      "step": 12770
+    },
+    {
+      "epoch": 0.3507456537859438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.507274771448743e-05,
+      "loss": 0.9773,
+      "step": 12771
+    },
+    {
+      "epoch": 0.350773118013787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5071981123420045e-05,
+      "loss": 0.882,
+      "step": 12772
+    },
+    {
+      "epoch": 0.3508005822416303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5071214492221175e-05,
+      "loss": 1.0281,
+      "step": 12773
+    },
+    {
+      "epoch": 0.3508280464694735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5070447820896894e-05,
+      "loss": 0.8836,
+      "step": 12774
+    },
+    {
+      "epoch": 0.35085551069731674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5069681109453268e-05,
+      "loss": 0.8999,
+      "step": 12775
+    },
+    {
+      "epoch": 0.35088297492515996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5068914357896361e-05,
+      "loss": 0.9959,
+      "step": 12776
+    },
+    {
+      "epoch": 0.35091043915300324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5068147566232236e-05,
+      "loss": 0.9253,
+      "step": 12777
+    },
+    {
+      "epoch": 0.35093790338084646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5067380734466967e-05,
+      "loss": 0.9761,
+      "step": 12778
+    },
+    {
+      "epoch": 0.3509653676086897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5066613862606618e-05,
+      "loss": 0.9329,
+      "step": 12779
+    },
+    {
+      "epoch": 0.3509928318365329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5065846950657259e-05,
+      "loss": 0.9097,
+      "step": 12780
+    },
+    {
+      "epoch": 0.3510202960643761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5065079998624957e-05,
+      "loss": 0.8973,
+      "step": 12781
+    },
+    {
+      "epoch": 0.3510477602922194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5064313006515777e-05,
+      "loss": 0.9164,
+      "step": 12782
+    },
+    {
+      "epoch": 0.3510752245200626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5063545974335791e-05,
+      "loss": 0.9887,
+      "step": 12783
+    },
+    {
+      "epoch": 0.35110268874790584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5062778902091074e-05,
+      "loss": 0.9493,
+      "step": 12784
+    },
+    {
+      "epoch": 0.35113015297574907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5062011789787683e-05,
+      "loss": 0.922,
+      "step": 12785
+    },
+    {
+      "epoch": 0.35115761720359234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5061244637431691e-05,
+      "loss": 0.9749,
+      "step": 12786
+    },
+    {
+      "epoch": 0.35118508143143556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5060477445029177e-05,
+      "loss": 0.951,
+      "step": 12787
+    },
+    {
+      "epoch": 0.3512125456592788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5059710212586205e-05,
+      "loss": 0.991,
+      "step": 12788
+    },
+    {
+      "epoch": 0.351240009887122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5058942940108843e-05,
+      "loss": 0.9136,
+      "step": 12789
+    },
+    {
+      "epoch": 0.3512674741149653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5058175627603166e-05,
+      "loss": 1.0114,
+      "step": 12790
+    },
+    {
+      "epoch": 0.3512949383428085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5057408275075243e-05,
+      "loss": 1.0388,
+      "step": 12791
+    },
+    {
+      "epoch": 0.3513224025706517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5056640882531151e-05,
+      "loss": 0.9313,
+      "step": 12792
+    },
+    {
+      "epoch": 0.35134986679849495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5055873449976952e-05,
+      "loss": 0.953,
+      "step": 12793
+    },
+    {
+      "epoch": 0.35137733102633817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5055105977418727e-05,
+      "loss": 0.8833,
+      "step": 12794
+    },
+    {
+      "epoch": 0.35140479525418145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5054338464862544e-05,
+      "loss": 0.9061,
+      "step": 12795
+    },
+    {
+      "epoch": 0.35143225948202467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5053570912314478e-05,
+      "loss": 0.8757,
+      "step": 12796
+    },
+    {
+      "epoch": 0.3514597237098679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5052803319780602e-05,
+      "loss": 0.9024,
+      "step": 12797
+    },
+    {
+      "epoch": 0.3514871879377111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5052035687266985e-05,
+      "loss": 0.9308,
+      "step": 12798
+    },
+    {
+      "epoch": 0.3515146521655544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.505126801477971e-05,
+      "loss": 0.9058,
+      "step": 12799
+    },
+    {
+      "epoch": 0.3515421163933976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5050500302324844e-05,
+      "loss": 1.0307,
+      "step": 12800
+    },
+    {
+      "epoch": 0.35156958062124083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5049732549908462e-05,
+      "loss": 0.9987,
+      "step": 12801
+    },
+    {
+      "epoch": 0.35159704484908405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5048964757536641e-05,
+      "loss": 0.9578,
+      "step": 12802
+    },
+    {
+      "epoch": 0.3516245090769273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5048196925215455e-05,
+      "loss": 0.9663,
+      "step": 12803
+    },
+    {
+      "epoch": 0.35165197330477055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504742905295098e-05,
+      "loss": 0.9711,
+      "step": 12804
+    },
+    {
+      "epoch": 0.35167943753261377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504666114074929e-05,
+      "loss": 1.0176,
+      "step": 12805
+    },
+    {
+      "epoch": 0.351706901760457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5045893188616463e-05,
+      "loss": 0.913,
+      "step": 12806
+    },
+    {
+      "epoch": 0.3517343659883002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5045125196558572e-05,
+      "loss": 1.0621,
+      "step": 12807
+    },
+    {
+      "epoch": 0.3517618302161435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.50443571645817e-05,
+      "loss": 0.993,
+      "step": 12808
+    },
+    {
+      "epoch": 0.3517892944439867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504358909269192e-05,
+      "loss": 0.9954,
+      "step": 12809
+    },
+    {
+      "epoch": 0.35181675867182993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5042820980895302e-05,
+      "loss": 1.0054,
+      "step": 12810
+    },
+    {
+      "epoch": 0.35184422289967315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504205282919794e-05,
+      "loss": 0.9208,
+      "step": 12811
+    },
+    {
+      "epoch": 0.35187168712751643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.50412846376059e-05,
+      "loss": 0.983,
+      "step": 12812
+    },
+    {
+      "epoch": 0.35189915135535965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5040516406125262e-05,
+      "loss": 0.8635,
+      "step": 12813
+    },
+    {
+      "epoch": 0.3519266155832029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5039748134762107e-05,
+      "loss": 0.8863,
+      "step": 12814
+    },
+    {
+      "epoch": 0.3519540798110461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503897982352251e-05,
+      "loss": 1.0261,
+      "step": 12815
+    },
+    {
+      "epoch": 0.35198154403888937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5038211472412556e-05,
+      "loss": 0.9834,
+      "step": 12816
+    },
+    {
+      "epoch": 0.3520090082667326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503744308143832e-05,
+      "loss": 0.9424,
+      "step": 12817
+    },
+    {
+      "epoch": 0.3520364724945758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5036674650605883e-05,
+      "loss": 0.9886,
+      "step": 12818
+    },
+    {
+      "epoch": 0.35206393672241904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5035906179921327e-05,
+      "loss": 0.8769,
+      "step": 12819
+    },
+    {
+      "epoch": 0.35209140095026226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5035137669390727e-05,
+      "loss": 1.0379,
+      "step": 12820
+    },
+    {
+      "epoch": 0.35211886517810553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5034369119020169e-05,
+      "loss": 0.9442,
+      "step": 12821
+    },
+    {
+      "epoch": 0.35214632940594875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5033600528815733e-05,
+      "loss": 0.8899,
+      "step": 12822
+    },
+    {
+      "epoch": 0.352173793633792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5032831898783497e-05,
+      "loss": 0.8478,
+      "step": 12823
+    },
+    {
+      "epoch": 0.3522012578616352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503206322892955e-05,
+      "loss": 0.9973,
+      "step": 12824
+    },
+    {
+      "epoch": 0.3522287220894785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5031294519259967e-05,
+      "loss": 0.9515,
+      "step": 12825
+    },
+    {
+      "epoch": 0.3522561863173217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5030525769780833e-05,
+      "loss": 1.0958,
+      "step": 12826
+    },
+    {
+      "epoch": 0.3522836505451649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5029756980498229e-05,
+      "loss": 0.9487,
+      "step": 12827
+    },
+    {
+      "epoch": 0.35231111477300814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5028988151418244e-05,
+      "loss": 0.9202,
+      "step": 12828
+    },
+    {
+      "epoch": 0.3523385790008514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5028219282546952e-05,
+      "loss": 0.858,
+      "step": 12829
+    },
+    {
+      "epoch": 0.35236604322869464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5027450373890439e-05,
+      "loss": 0.9132,
+      "step": 12830
+    },
+    {
+      "epoch": 0.35239350745653786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.50266814254548e-05,
+      "loss": 0.9804,
+      "step": 12831
+    },
+    {
+      "epoch": 0.3524209716843811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5025912437246104e-05,
+      "loss": 0.9486,
+      "step": 12832
+    },
+    {
+      "epoch": 0.3524484359122243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5025143409270442e-05,
+      "loss": 0.947,
+      "step": 12833
+    },
+    {
+      "epoch": 0.3524759001400676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5024374341533903e-05,
+      "loss": 0.9304,
+      "step": 12834
+    },
+    {
+      "epoch": 0.3525033643679108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5023605234042564e-05,
+      "loss": 0.8474,
+      "step": 12835
+    },
+    {
+      "epoch": 0.352530828595754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5022836086802515e-05,
+      "loss": 0.8557,
+      "step": 12836
+    },
+    {
+      "epoch": 0.35255829282359724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5022066899819842e-05,
+      "loss": 1.0408,
+      "step": 12837
+    },
+    {
+      "epoch": 0.3525857570514405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502129767310063e-05,
+      "loss": 0.9702,
+      "step": 12838
+    },
+    {
+      "epoch": 0.35261322127928374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5020528406650965e-05,
+      "loss": 0.8569,
+      "step": 12839
+    },
+    {
+      "epoch": 0.35264068550712696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5019759100476936e-05,
+      "loss": 1.0089,
+      "step": 12840
+    },
+    {
+      "epoch": 0.3526681497349702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5018989754584627e-05,
+      "loss": 0.9182,
+      "step": 12841
+    },
+    {
+      "epoch": 0.35269561396281346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5018220368980128e-05,
+      "loss": 0.8865,
+      "step": 12842
+    },
+    {
+      "epoch": 0.3527230781906567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5017450943669526e-05,
+      "loss": 0.9176,
+      "step": 12843
+    },
+    {
+      "epoch": 0.3527505424184999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5016681478658907e-05,
+      "loss": 0.9518,
+      "step": 12844
+    },
+    {
+      "epoch": 0.3527780066463431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5015911973954363e-05,
+      "loss": 0.9013,
+      "step": 12845
+    },
+    {
+      "epoch": 0.3528054708741864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.501514242956198e-05,
+      "loss": 1.0042,
+      "step": 12846
+    },
+    {
+      "epoch": 0.3528329351020296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5014372845487844e-05,
+      "loss": 0.9243,
+      "step": 12847
+    },
+    {
+      "epoch": 0.35286039932987284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5013603221738051e-05,
+      "loss": 0.9096,
+      "step": 12848
+    },
+    {
+      "epoch": 0.35288786355771606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5012833558318689e-05,
+      "loss": 0.9426,
+      "step": 12849
+    },
+    {
+      "epoch": 0.3529153277855593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5012063855235842e-05,
+      "loss": 0.9111,
+      "step": 12850
+    },
+    {
+      "epoch": 0.35294279201340256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5011294112495604e-05,
+      "loss": 0.9879,
+      "step": 12851
+    },
+    {
+      "epoch": 0.3529702562412458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5010524330104073e-05,
+      "loss": 0.8293,
+      "step": 12852
+    },
+    {
+      "epoch": 0.352997720469089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5009754508067325e-05,
+      "loss": 0.8696,
+      "step": 12853
+    },
+    {
+      "epoch": 0.3530251846969322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500898464639146e-05,
+      "loss": 0.9554,
+      "step": 12854
+    },
+    {
+      "epoch": 0.3530526489247755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500821474508257e-05,
+      "loss": 0.9622,
+      "step": 12855
+    },
+    {
+      "epoch": 0.3530801131526187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5007444804146744e-05,
+      "loss": 0.8835,
+      "step": 12856
+    },
+    {
+      "epoch": 0.35310757738046195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5006674823590074e-05,
+      "loss": 1.0244,
+      "step": 12857
+    },
+    {
+      "epoch": 0.35313504160830517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5005904803418656e-05,
+      "loss": 1.0393,
+      "step": 12858
+    },
+    {
+      "epoch": 0.35316250583614844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5005134743638579e-05,
+      "loss": 0.9185,
+      "step": 12859
+    },
+    {
+      "epoch": 0.35318997006399167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5004364644255938e-05,
+      "loss": 0.9769,
+      "step": 12860
+    },
+    {
+      "epoch": 0.3532174342918349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5003594505276824e-05,
+      "loss": 0.92,
+      "step": 12861
+    },
+    {
+      "epoch": 0.3532448985196781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500282432670733e-05,
+      "loss": 0.8875,
+      "step": 12862
+    },
+    {
+      "epoch": 0.35327236274752133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5002054108553555e-05,
+      "loss": 0.9681,
+      "step": 12863
+    },
+    {
+      "epoch": 0.3532998269753646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500128385082159e-05,
+      "loss": 0.9614,
+      "step": 12864
+    },
+    {
+      "epoch": 0.3533272912032078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500051355351753e-05,
+      "loss": 0.9308,
+      "step": 12865
+    },
+    {
+      "epoch": 0.35335475543105105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.499974321664747e-05,
+      "loss": 0.8932,
+      "step": 12866
+    },
+    {
+      "epoch": 0.35338221965889427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4998972840217503e-05,
+      "loss": 0.9926,
+      "step": 12867
+    },
+    {
+      "epoch": 0.35340968388673755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4998202424233728e-05,
+      "loss": 0.8654,
+      "step": 12868
+    },
+    {
+      "epoch": 0.35343714811458077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4997431968702238e-05,
+      "loss": 0.9656,
+      "step": 12869
+    },
+    {
+      "epoch": 0.353464612342424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4996661473629133e-05,
+      "loss": 0.9362,
+      "step": 12870
+    },
+    {
+      "epoch": 0.3534920765702672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4995890939020503e-05,
+      "loss": 1.0607,
+      "step": 12871
+    },
+    {
+      "epoch": 0.3535195407981105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4995120364882452e-05,
+      "loss": 0.8753,
+      "step": 12872
+    },
+    {
+      "epoch": 0.3535470050259537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4994349751221073e-05,
+      "loss": 0.9733,
+      "step": 12873
+    },
+    {
+      "epoch": 0.35357446925379693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4993579098042464e-05,
+      "loss": 0.9642,
+      "step": 12874
+    },
+    {
+      "epoch": 0.35360193348164015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4992808405352721e-05,
+      "loss": 0.9296,
+      "step": 12875
+    },
+    {
+      "epoch": 0.3536293977094834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4992037673157948e-05,
+      "loss": 1.0112,
+      "step": 12876
+    },
+    {
+      "epoch": 0.35365686193732665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4991266901464237e-05,
+      "loss": 0.9569,
+      "step": 12877
+    },
+    {
+      "epoch": 0.35368432616516987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4990496090277689e-05,
+      "loss": 0.9502,
+      "step": 12878
+    },
+    {
+      "epoch": 0.3537117903930131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.49897252396044e-05,
+      "loss": 0.774,
+      "step": 12879
+    },
+    {
+      "epoch": 0.3537392546208563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4988954349450475e-05,
+      "loss": 0.9969,
+      "step": 12880
+    },
+    {
+      "epoch": 0.3537667188486996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.498818341982201e-05,
+      "loss": 0.9884,
+      "step": 12881
+    },
+    {
+      "epoch": 0.3537941830765428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4987412450725108e-05,
+      "loss": 0.936,
+      "step": 12882
+    },
+    {
+      "epoch": 0.35382164730438603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4986641442165861e-05,
+      "loss": 0.8885,
+      "step": 12883
+    },
+    {
+      "epoch": 0.35384911153222925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4985870394150381e-05,
+      "loss": 0.9256,
+      "step": 12884
+    },
+    {
+      "epoch": 0.35387657576007253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.498509930668476e-05,
+      "loss": 1.0298,
+      "step": 12885
+    },
+    {
+      "epoch": 0.35390403998791575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4984328179775103e-05,
+      "loss": 1.0306,
+      "step": 12886
+    },
+    {
+      "epoch": 0.353931504215759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4983557013427513e-05,
+      "loss": 0.9309,
+      "step": 12887
+    },
+    {
+      "epoch": 0.3539589684436022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4982785807648086e-05,
+      "loss": 0.9932,
+      "step": 12888
+    },
+    {
+      "epoch": 0.3539864326714454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4982014562442928e-05,
+      "loss": 0.9854,
+      "step": 12889
+    },
+    {
+      "epoch": 0.3540138968992887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4981243277818142e-05,
+      "loss": 0.9741,
+      "step": 12890
+    },
+    {
+      "epoch": 0.3540413611271319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.498047195377983e-05,
+      "loss": 0.976,
+      "step": 12891
+    },
+    {
+      "epoch": 0.35406882535497514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4979700590334095e-05,
+      "loss": 0.9465,
+      "step": 12892
+    },
+    {
+      "epoch": 0.35409628958281836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4978929187487037e-05,
+      "loss": 0.9409,
+      "step": 12893
+    },
+    {
+      "epoch": 0.35412375381066163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4978157745244763e-05,
+      "loss": 1.0025,
+      "step": 12894
+    },
+    {
+      "epoch": 0.35415121803850486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4977386263613379e-05,
+      "loss": 0.9289,
+      "step": 12895
+    },
+    {
+      "epoch": 0.3541786822663481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4976614742598985e-05,
+      "loss": 0.9842,
+      "step": 12896
+    },
+    {
+      "epoch": 0.3542061464941913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.497584318220769e-05,
+      "loss": 0.8193,
+      "step": 12897
+    },
+    {
+      "epoch": 0.3542336107220346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4975071582445595e-05,
+      "loss": 0.8986,
+      "step": 12898
+    },
+    {
+      "epoch": 0.3542610749498778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4974299943318807e-05,
+      "loss": 0.9809,
+      "step": 12899
+    },
+    {
+      "epoch": 0.354288539177721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.497352826483343e-05,
+      "loss": 0.943,
+      "step": 12900
+    },
+    {
+      "epoch": 0.35431600340556424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4972756546995571e-05,
+      "loss": 0.9821,
+      "step": 12901
+    },
+    {
+      "epoch": 0.35434346763340746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4971984789811336e-05,
+      "loss": 1.0411,
+      "step": 12902
+    },
+    {
+      "epoch": 0.35437093186125074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4971212993286827e-05,
+      "loss": 0.9308,
+      "step": 12903
+    },
+    {
+      "epoch": 0.35439839608909396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4970441157428162e-05,
+      "loss": 0.9606,
+      "step": 12904
+    },
+    {
+      "epoch": 0.3544258603169372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496966928224144e-05,
+      "loss": 1.027,
+      "step": 12905
+    },
+    {
+      "epoch": 0.3544533245447804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4968897367732766e-05,
+      "loss": 0.9744,
+      "step": 12906
+    },
+    {
+      "epoch": 0.3544807887726237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4968125413908252e-05,
+      "loss": 0.8672,
+      "step": 12907
+    },
+    {
+      "epoch": 0.3545082530004669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496735342077401e-05,
+      "loss": 0.9609,
+      "step": 12908
+    },
+    {
+      "epoch": 0.3545357172283101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4966581388336139e-05,
+      "loss": 0.8896,
+      "step": 12909
+    },
+    {
+      "epoch": 0.35456318145615334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4965809316600752e-05,
+      "loss": 0.9811,
+      "step": 12910
+    },
+    {
+      "epoch": 0.3545906456839966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4965037205573962e-05,
+      "loss": 0.9338,
+      "step": 12911
+    },
+    {
+      "epoch": 0.35461810991183984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4964265055261872e-05,
+      "loss": 0.9401,
+      "step": 12912
+    },
+    {
+      "epoch": 0.35464557413968306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4963492865670592e-05,
+      "loss": 1.0274,
+      "step": 12913
+    },
+    {
+      "epoch": 0.3546730383675263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4962720636806235e-05,
+      "loss": 0.9179,
+      "step": 12914
+    },
+    {
+      "epoch": 0.3547005025953695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496194836867491e-05,
+      "loss": 0.9691,
+      "step": 12915
+    },
+    {
+      "epoch": 0.3547279668232128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496117606128273e-05,
+      "loss": 0.9171,
+      "step": 12916
+    },
+    {
+      "epoch": 0.354755431051056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4960403714635802e-05,
+      "loss": 0.9506,
+      "step": 12917
+    },
+    {
+      "epoch": 0.3547828952788992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4959631328740236e-05,
+      "loss": 0.9108,
+      "step": 12918
+    },
+    {
+      "epoch": 0.35481035950674245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4958858903602148e-05,
+      "loss": 0.9058,
+      "step": 12919
+    },
+    {
+      "epoch": 0.3548378237345857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4958086439227647e-05,
+      "loss": 0.9512,
+      "step": 12920
+    },
+    {
+      "epoch": 0.35486528796242894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4957313935622843e-05,
+      "loss": 0.928,
+      "step": 12921
+    },
+    {
+      "epoch": 0.35489275219027216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4956541392793855e-05,
+      "loss": 1.0304,
+      "step": 12922
+    },
+    {
+      "epoch": 0.3549202164181154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4955768810746788e-05,
+      "loss": 0.9499,
+      "step": 12923
+    },
+    {
+      "epoch": 0.35494768064595866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4954996189487763e-05,
+      "loss": 0.9791,
+      "step": 12924
+    },
+    {
+      "epoch": 0.3549751448738019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4954223529022885e-05,
+      "loss": 0.9684,
+      "step": 12925
+    },
+    {
+      "epoch": 0.3550026091016451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4953450829358272e-05,
+      "loss": 1.0215,
+      "step": 12926
+    },
+    {
+      "epoch": 0.3550300733294883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4952678090500037e-05,
+      "loss": 1.0183,
+      "step": 12927
+    },
+    {
+      "epoch": 0.35505753755733155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4951905312454295e-05,
+      "loss": 0.9309,
+      "step": 12928
+    },
+    {
+      "epoch": 0.3550850017851748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4951132495227161e-05,
+      "loss": 0.9636,
+      "step": 12929
+    },
+    {
+      "epoch": 0.35511246601301805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4950359638824747e-05,
+      "loss": 0.9896,
+      "step": 12930
+    },
+    {
+      "epoch": 0.35513993024086127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4949586743253171e-05,
+      "loss": 1.0344,
+      "step": 12931
+    },
+    {
+      "epoch": 0.3551673944687045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4948813808518548e-05,
+      "loss": 0.9094,
+      "step": 12932
+    },
+    {
+      "epoch": 0.35519485869654777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4948040834626992e-05,
+      "loss": 0.832,
+      "step": 12933
+    },
+    {
+      "epoch": 0.355222322924391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494726782158462e-05,
+      "loss": 0.9139,
+      "step": 12934
+    },
+    {
+      "epoch": 0.3552497871522342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4946494769397548e-05,
+      "loss": 1.0232,
+      "step": 12935
+    },
+    {
+      "epoch": 0.35527725138007743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4945721678071896e-05,
+      "loss": 0.996,
+      "step": 12936
+    },
+    {
+      "epoch": 0.3553047156079207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4944948547613778e-05,
+      "loss": 0.9632,
+      "step": 12937
+    },
+    {
+      "epoch": 0.35533217983576393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4944175378029308e-05,
+      "loss": 0.9988,
+      "step": 12938
+    },
+    {
+      "epoch": 0.35535964406360715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4943402169324607e-05,
+      "loss": 0.9813,
+      "step": 12939
+    },
+    {
+      "epoch": 0.35538710829145037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4942628921505798e-05,
+      "loss": 1.0043,
+      "step": 12940
+    },
+    {
+      "epoch": 0.35541457251929365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4941855634578992e-05,
+      "loss": 0.9362,
+      "step": 12941
+    },
+    {
+      "epoch": 0.35544203674713687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494108230855031e-05,
+      "loss": 0.95,
+      "step": 12942
+    },
+    {
+      "epoch": 0.3554695009749801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4940308943425871e-05,
+      "loss": 0.9214,
+      "step": 12943
+    },
+    {
+      "epoch": 0.3554969652028233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4939535539211796e-05,
+      "loss": 0.963,
+      "step": 12944
+    },
+    {
+      "epoch": 0.35552442943066653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4938762095914197e-05,
+      "loss": 0.8552,
+      "step": 12945
+    },
+    {
+      "epoch": 0.3555518936585098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4937988613539203e-05,
+      "loss": 0.9412,
+      "step": 12946
+    },
+    {
+      "epoch": 0.35557935788635303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.493721509209293e-05,
+      "loss": 0.957,
+      "step": 12947
+    },
+    {
+      "epoch": 0.35560682211419625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4936441531581498e-05,
+      "loss": 1.035,
+      "step": 12948
+    },
+    {
+      "epoch": 0.3556342863420395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4935667932011031e-05,
+      "loss": 0.9863,
+      "step": 12949
+    },
+    {
+      "epoch": 0.35566175056988275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4934894293387642e-05,
+      "loss": 0.9323,
+      "step": 12950
+    },
+    {
+      "epoch": 0.35568921479772597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.493412061571746e-05,
+      "loss": 1.0311,
+      "step": 12951
+    },
+    {
+      "epoch": 0.3557166790255692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4933346899006607e-05,
+      "loss": 0.8786,
+      "step": 12952
+    },
+    {
+      "epoch": 0.3557441432534124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4932573143261203e-05,
+      "loss": 0.96,
+      "step": 12953
+    },
+    {
+      "epoch": 0.3557716074812557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4931799348487363e-05,
+      "loss": 0.9921,
+      "step": 12954
+    },
+    {
+      "epoch": 0.3557990717090989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4931025514691222e-05,
+      "loss": 0.8665,
+      "step": 12955
+    },
+    {
+      "epoch": 0.35582653593694213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4930251641878897e-05,
+      "loss": 0.9977,
+      "step": 12956
+    },
+    {
+      "epoch": 0.35585400016478536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4929477730056507e-05,
+      "loss": 0.9523,
+      "step": 12957
+    },
+    {
+      "epoch": 0.3558814643926286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4928703779230186e-05,
+      "loss": 1.0115,
+      "step": 12958
+    },
+    {
+      "epoch": 0.35590892862047185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4927929789406046e-05,
+      "loss": 0.9223,
+      "step": 12959
+    },
+    {
+      "epoch": 0.3559363928483151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492715576059022e-05,
+      "loss": 0.969,
+      "step": 12960
+    },
+    {
+      "epoch": 0.3559638570761583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4926381692788827e-05,
+      "loss": 0.9747,
+      "step": 12961
+    },
+    {
+      "epoch": 0.3559913213040015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4925607586007994e-05,
+      "loss": 0.9166,
+      "step": 12962
+    },
+    {
+      "epoch": 0.3560187855318448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4924833440253843e-05,
+      "loss": 0.8377,
+      "step": 12963
+    },
+    {
+      "epoch": 0.356046249759688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4924059255532506e-05,
+      "loss": 0.9489,
+      "step": 12964
+    },
+    {
+      "epoch": 0.35607371398753124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4923285031850105e-05,
+      "loss": 1.018,
+      "step": 12965
+    },
+    {
+      "epoch": 0.35610117821537446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4922510769212763e-05,
+      "loss": 1.0149,
+      "step": 12966
+    },
+    {
+      "epoch": 0.35612864244321774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4921736467626612e-05,
+      "loss": 0.976,
+      "step": 12967
+    },
+    {
+      "epoch": 0.35615610667106096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4920962127097776e-05,
+      "loss": 0.9302,
+      "step": 12968
+    },
+    {
+      "epoch": 0.3561835708989042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4920187747632378e-05,
+      "loss": 0.9911,
+      "step": 12969
+    },
+    {
+      "epoch": 0.3562110351267474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4919413329236552e-05,
+      "loss": 1.0149,
+      "step": 12970
+    },
+    {
+      "epoch": 0.3562384993545906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4918638871916418e-05,
+      "loss": 0.9814,
+      "step": 12971
+    },
+    {
+      "epoch": 0.3562659635824339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491786437567811e-05,
+      "loss": 0.8374,
+      "step": 12972
+    },
+    {
+      "epoch": 0.3562934278102771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4917089840527756e-05,
+      "loss": 0.9355,
+      "step": 12973
+    },
+    {
+      "epoch": 0.35632089203812034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4916315266471482e-05,
+      "loss": 0.9073,
+      "step": 12974
+    },
+    {
+      "epoch": 0.35634835626596356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4915540653515415e-05,
+      "loss": 0.8962,
+      "step": 12975
+    },
+    {
+      "epoch": 0.35637582049380684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4914766001665685e-05,
+      "loss": 0.9781,
+      "step": 12976
+    },
+    {
+      "epoch": 0.35640328472165006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4913991310928428e-05,
+      "loss": 0.9888,
+      "step": 12977
+    },
+    {
+      "epoch": 0.3564307489494933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4913216581309764e-05,
+      "loss": 0.9103,
+      "step": 12978
+    },
+    {
+      "epoch": 0.3564582131773365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4912441812815828e-05,
+      "loss": 0.9212,
+      "step": 12979
+    },
+    {
+      "epoch": 0.3564856774051798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491166700545275e-05,
+      "loss": 0.9683,
+      "step": 12980
+    },
+    {
+      "epoch": 0.356513141633023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4910892159226659e-05,
+      "loss": 0.9575,
+      "step": 12981
+    },
+    {
+      "epoch": 0.3565406058608662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491011727414369e-05,
+      "loss": 0.9193,
+      "step": 12982
+    },
+    {
+      "epoch": 0.35656807008870944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4909342350209968e-05,
+      "loss": 0.9188,
+      "step": 12983
+    },
+    {
+      "epoch": 0.35659553431655266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490856738743163e-05,
+      "loss": 0.8236,
+      "step": 12984
+    },
+    {
+      "epoch": 0.35662299854439594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4907792385814803e-05,
+      "loss": 0.9547,
+      "step": 12985
+    },
+    {
+      "epoch": 0.35665046277223916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4907017345365622e-05,
+      "loss": 0.9655,
+      "step": 12986
+    },
+    {
+      "epoch": 0.3566779270000824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4906242266090218e-05,
+      "loss": 0.954,
+      "step": 12987
+    },
+    {
+      "epoch": 0.3567053912279256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490546714799473e-05,
+      "loss": 0.9609,
+      "step": 12988
+    },
+    {
+      "epoch": 0.3567328554557689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4904691991085281e-05,
+      "loss": 0.9893,
+      "step": 12989
+    },
+    {
+      "epoch": 0.3567603196836121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490391679536801e-05,
+      "loss": 0.9857,
+      "step": 12990
+    },
+    {
+      "epoch": 0.3567877839114553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4903141560849051e-05,
+      "loss": 0.9384,
+      "step": 12991
+    },
+    {
+      "epoch": 0.35681524813929855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4902366287534535e-05,
+      "loss": 0.9308,
+      "step": 12992
+    },
+    {
+      "epoch": 0.3568427123671418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4901590975430598e-05,
+      "loss": 0.971,
+      "step": 12993
+    },
+    {
+      "epoch": 0.35687017659498504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4900815624543375e-05,
+      "loss": 0.9789,
+      "step": 12994
+    },
+    {
+      "epoch": 0.35689764082282827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4900040234878998e-05,
+      "loss": 0.9624,
+      "step": 12995
+    },
+    {
+      "epoch": 0.3569251050506715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489926480644361e-05,
+      "loss": 0.9739,
+      "step": 12996
+    },
+    {
+      "epoch": 0.3569525692785147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4898489339243337e-05,
+      "loss": 0.9558,
+      "step": 12997
+    },
+    {
+      "epoch": 0.356980033506358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489771383328432e-05,
+      "loss": 0.9736,
+      "step": 12998
+    },
+    {
+      "epoch": 0.3570074977342012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4896938288572694e-05,
+      "loss": 0.9915,
+      "step": 12999
+    },
+    {
+      "epoch": 0.35703496196204443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4896162705114597e-05,
+      "loss": 0.912,
+      "step": 13000
+    },
+    {
+      "epoch": 0.35706242618988765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489538708291616e-05,
+      "loss": 0.9741,
+      "step": 13001
+    },
+    {
+      "epoch": 0.3570898904177309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4894611421983527e-05,
+      "loss": 1.1011,
+      "step": 13002
+    },
+    {
+      "epoch": 0.35711735464557415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4893835722322832e-05,
+      "loss": 0.9186,
+      "step": 13003
+    },
+    {
+      "epoch": 0.35714481887341737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4893059983940212e-05,
+      "loss": 0.956,
+      "step": 13004
+    },
+    {
+      "epoch": 0.3571722831012606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4892284206841808e-05,
+      "loss": 0.9238,
+      "step": 13005
+    },
+    {
+      "epoch": 0.35719974732910387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4891508391033753e-05,
+      "loss": 0.9534,
+      "step": 13006
+    },
+    {
+      "epoch": 0.3572272115569471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4890732536522192e-05,
+      "loss": 1.0237,
+      "step": 13007
+    },
+    {
+      "epoch": 0.3572546757847903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.488995664331326e-05,
+      "loss": 0.9072,
+      "step": 13008
+    },
+    {
+      "epoch": 0.35728214001263353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4889180711413096e-05,
+      "loss": 0.9768,
+      "step": 13009
+    },
+    {
+      "epoch": 0.35730960424047675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.488840474082784e-05,
+      "loss": 0.9603,
+      "step": 13010
+    },
+    {
+      "epoch": 0.35733706846832003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4887628731563635e-05,
+      "loss": 0.9186,
+      "step": 13011
+    },
+    {
+      "epoch": 0.35736453269616325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4886852683626616e-05,
+      "loss": 1.0516,
+      "step": 13012
+    },
+    {
+      "epoch": 0.35739199692400647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4886076597022925e-05,
+      "loss": 1.0165,
+      "step": 13013
+    },
+    {
+      "epoch": 0.3574194611518497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4885300471758703e-05,
+      "loss": 0.8605,
+      "step": 13014
+    },
+    {
+      "epoch": 0.35744692537969297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4884524307840091e-05,
+      "loss": 0.9846,
+      "step": 13015
+    },
+    {
+      "epoch": 0.3574743896075362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4883748105273235e-05,
+      "loss": 1.0132,
+      "step": 13016
+    },
+    {
+      "epoch": 0.3575018538353794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4882971864064268e-05,
+      "loss": 0.9629,
+      "step": 13017
+    },
+    {
+      "epoch": 0.35752931806322263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4882195584219336e-05,
+      "loss": 1.0499,
+      "step": 13018
+    },
+    {
+      "epoch": 0.3575567822910659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4881419265744583e-05,
+      "loss": 1.0327,
+      "step": 13019
+    },
+    {
+      "epoch": 0.35758424651890913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.488064290864615e-05,
+      "loss": 1.0131,
+      "step": 13020
+    },
+    {
+      "epoch": 0.35761171074675235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.487986651293018e-05,
+      "loss": 1.0566,
+      "step": 13021
+    },
+    {
+      "epoch": 0.3576391749745956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4879090078602813e-05,
+      "loss": 0.9278,
+      "step": 13022
+    },
+    {
+      "epoch": 0.3576666392024388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4878313605670197e-05,
+      "loss": 1.0382,
+      "step": 13023
+    },
+    {
+      "epoch": 0.3576941034302821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4877537094138477e-05,
+      "loss": 0.959,
+      "step": 13024
+    },
+    {
+      "epoch": 0.3577215676581253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.487676054401379e-05,
+      "loss": 0.8762,
+      "step": 13025
+    },
+    {
+      "epoch": 0.3577490318859685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4875983955302287e-05,
+      "loss": 0.9019,
+      "step": 13026
+    },
+    {
+      "epoch": 0.35777649611381174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4875207328010109e-05,
+      "loss": 1.0485,
+      "step": 13027
+    },
+    {
+      "epoch": 0.357803960341655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4874430662143402e-05,
+      "loss": 0.8835,
+      "step": 13028
+    },
+    {
+      "epoch": 0.35783142456949824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4873653957708315e-05,
+      "loss": 0.9232,
+      "step": 13029
+    },
+    {
+      "epoch": 0.35785888879734146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4872877214710983e-05,
+      "loss": 0.9666,
+      "step": 13030
+    },
+    {
+      "epoch": 0.3578863530251847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4872100433157561e-05,
+      "loss": 0.9249,
+      "step": 13031
+    },
+    {
+      "epoch": 0.35791381725302795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4871323613054196e-05,
+      "loss": 0.9405,
+      "step": 13032
+    },
+    {
+      "epoch": 0.3579412814808712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4870546754407032e-05,
+      "loss": 1.0369,
+      "step": 13033
+    },
+    {
+      "epoch": 0.3579687457087144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4869769857222211e-05,
+      "loss": 1.0044,
+      "step": 13034
+    },
+    {
+      "epoch": 0.3579962099365576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4868992921505889e-05,
+      "loss": 0.9591,
+      "step": 13035
+    },
+    {
+      "epoch": 0.35802367416440084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4868215947264208e-05,
+      "loss": 0.94,
+      "step": 13036
+    },
+    {
+      "epoch": 0.3580511383922441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4867438934503316e-05,
+      "loss": 0.909,
+      "step": 13037
+    },
+    {
+      "epoch": 0.35807860262008734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4866661883229361e-05,
+      "loss": 1.0135,
+      "step": 13038
+    },
+    {
+      "epoch": 0.35810606684793056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4865884793448493e-05,
+      "loss": 0.8961,
+      "step": 13039
+    },
+    {
+      "epoch": 0.3581335310757738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486510766516686e-05,
+      "loss": 0.9749,
+      "step": 13040
+    },
+    {
+      "epoch": 0.35816099530361706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486433049839061e-05,
+      "loss": 0.9234,
+      "step": 13041
+    },
+    {
+      "epoch": 0.3581884595314603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4863553293125891e-05,
+      "loss": 0.9016,
+      "step": 13042
+    },
+    {
+      "epoch": 0.3582159237593035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4862776049378855e-05,
+      "loss": 0.9083,
+      "step": 13043
+    },
+    {
+      "epoch": 0.3582433879871467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4861998767155654e-05,
+      "loss": 0.9462,
+      "step": 13044
+    },
+    {
+      "epoch": 0.35827085221499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4861221446462435e-05,
+      "loss": 0.9116,
+      "step": 13045
+    },
+    {
+      "epoch": 0.3582983164428332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4860444087305346e-05,
+      "loss": 0.9638,
+      "step": 13046
+    },
+    {
+      "epoch": 0.35832578067067644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4859666689690544e-05,
+      "loss": 0.8876,
+      "step": 13047
+    },
+    {
+      "epoch": 0.35835324489851966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4858889253624176e-05,
+      "loss": 0.9173,
+      "step": 13048
+    },
+    {
+      "epoch": 0.35838070912636294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.485811177911239e-05,
+      "loss": 1.0649,
+      "step": 13049
+    },
+    {
+      "epoch": 0.35840817335420616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4857334266161346e-05,
+      "loss": 0.9616,
+      "step": 13050
+    },
+    {
+      "epoch": 0.3584356375820494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4856556714777188e-05,
+      "loss": 0.8471,
+      "step": 13051
+    },
+    {
+      "epoch": 0.3584631018098926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4855779124966076e-05,
+      "loss": 0.9352,
+      "step": 13052
+    },
+    {
+      "epoch": 0.3584905660377358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4855001496734158e-05,
+      "loss": 0.8929,
+      "step": 13053
+    },
+    {
+      "epoch": 0.3585180302655791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4854223830087586e-05,
+      "loss": 0.9681,
+      "step": 13054
+    },
+    {
+      "epoch": 0.3585454944934223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4853446125032514e-05,
+      "loss": 0.8083,
+      "step": 13055
+    },
+    {
+      "epoch": 0.35857295872126554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4852668381575099e-05,
+      "loss": 0.9618,
+      "step": 13056
+    },
+    {
+      "epoch": 0.35860042294910877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.485189059972149e-05,
+      "loss": 0.9851,
+      "step": 13057
+    },
+    {
+      "epoch": 0.35862788717695204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4851112779477845e-05,
+      "loss": 0.9254,
+      "step": 13058
+    },
+    {
+      "epoch": 0.35865535140479526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4850334920850316e-05,
+      "loss": 1.0097,
+      "step": 13059
+    },
+    {
+      "epoch": 0.3586828156326385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4849557023845057e-05,
+      "loss": 0.9484,
+      "step": 13060
+    },
+    {
+      "epoch": 0.3587102798604817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4848779088468226e-05,
+      "loss": 0.9214,
+      "step": 13061
+    },
+    {
+      "epoch": 0.358737744088325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4848001114725975e-05,
+      "loss": 1.0078,
+      "step": 13062
+    },
+    {
+      "epoch": 0.3587652083161682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4847223102624459e-05,
+      "loss": 0.8717,
+      "step": 13063
+    },
+    {
+      "epoch": 0.3587926725440114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.484644505216984e-05,
+      "loss": 0.8854,
+      "step": 13064
+    },
+    {
+      "epoch": 0.35882013677185465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.484566696336827e-05,
+      "loss": 0.8698,
+      "step": 13065
+    },
+    {
+      "epoch": 0.35884760099969787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4844888836225904e-05,
+      "loss": 1.0529,
+      "step": 13066
+    },
+    {
+      "epoch": 0.35887506522754115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4844110670748904e-05,
+      "loss": 0.9988,
+      "step": 13067
+    },
+    {
+      "epoch": 0.35890252945538437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.484333246694342e-05,
+      "loss": 0.9245,
+      "step": 13068
+    },
+    {
+      "epoch": 0.3589299936832276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4842554224815616e-05,
+      "loss": 0.977,
+      "step": 13069
+    },
+    {
+      "epoch": 0.3589574579110708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4841775944371648e-05,
+      "loss": 0.9592,
+      "step": 13070
+    },
+    {
+      "epoch": 0.3589849221389141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4840997625617668e-05,
+      "loss": 1.0094,
+      "step": 13071
+    },
+    {
+      "epoch": 0.3590123863667573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4840219268559845e-05,
+      "loss": 0.8089,
+      "step": 13072
+    },
+    {
+      "epoch": 0.35903985059460053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4839440873204332e-05,
+      "loss": 0.9324,
+      "step": 13073
+    },
+    {
+      "epoch": 0.35906731482244375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4838662439557286e-05,
+      "loss": 0.8829,
+      "step": 13074
+    },
+    {
+      "epoch": 0.359094779050287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4837883967624866e-05,
+      "loss": 1.0155,
+      "step": 13075
+    },
+    {
+      "epoch": 0.35912224327813025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4837105457413239e-05,
+      "loss": 1.0006,
+      "step": 13076
+    },
+    {
+      "epoch": 0.35914970750597347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.483632690892856e-05,
+      "loss": 0.9265,
+      "step": 13077
+    },
+    {
+      "epoch": 0.3591771717338167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4835548322176984e-05,
+      "loss": 0.8701,
+      "step": 13078
+    },
+    {
+      "epoch": 0.3592046359616599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4834769697164681e-05,
+      "loss": 0.9662,
+      "step": 13079
+    },
+    {
+      "epoch": 0.3592321001895032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4833991033897807e-05,
+      "loss": 0.9261,
+      "step": 13080
+    },
+    {
+      "epoch": 0.3592595644173464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.483321233238252e-05,
+      "loss": 0.9587,
+      "step": 13081
+    },
+    {
+      "epoch": 0.35928702864518963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4832433592624988e-05,
+      "loss": 0.9723,
+      "step": 13082
+    },
+    {
+      "epoch": 0.35931449287303285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4831654814631369e-05,
+      "loss": 1.0461,
+      "step": 13083
+    },
+    {
+      "epoch": 0.35934195710087613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4830875998407827e-05,
+      "loss": 0.9679,
+      "step": 13084
+    },
+    {
+      "epoch": 0.35936942132871935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4830097143960523e-05,
+      "loss": 0.9322,
+      "step": 13085
+    },
+    {
+      "epoch": 0.3593968855565626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4829318251295615e-05,
+      "loss": 0.9377,
+      "step": 13086
+    },
+    {
+      "epoch": 0.3594243497844058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4828539320419271e-05,
+      "loss": 0.9702,
+      "step": 13087
+    },
+    {
+      "epoch": 0.35945181401224907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4827760351337657e-05,
+      "loss": 0.9008,
+      "step": 13088
+    },
+    {
+      "epoch": 0.3594792782400923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4826981344056931e-05,
+      "loss": 0.9279,
+      "step": 13089
+    },
+    {
+      "epoch": 0.3595067424679355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4826202298583259e-05,
+      "loss": 0.9108,
+      "step": 13090
+    },
+    {
+      "epoch": 0.35953420669577874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4825423214922804e-05,
+      "loss": 0.8512,
+      "step": 13091
+    },
+    {
+      "epoch": 0.35956167092362196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4824644093081735e-05,
+      "loss": 0.9685,
+      "step": 13092
+    },
+    {
+      "epoch": 0.35958913515146523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.482386493306621e-05,
+      "loss": 0.9054,
+      "step": 13093
+    },
+    {
+      "epoch": 0.35961659937930845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4823085734882398e-05,
+      "loss": 0.9175,
+      "step": 13094
+    },
+    {
+      "epoch": 0.3596440636071517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.482230649853646e-05,
+      "loss": 0.9248,
+      "step": 13095
+    },
+    {
+      "epoch": 0.3596715278349949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4821527224034569e-05,
+      "loss": 0.8781,
+      "step": 13096
+    },
+    {
+      "epoch": 0.3596989920628382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4820747911382886e-05,
+      "loss": 0.9012,
+      "step": 13097
+    },
+    {
+      "epoch": 0.3597264562906814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4819968560587578e-05,
+      "loss": 0.9382,
+      "step": 13098
+    },
+    {
+      "epoch": 0.3597539205185246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4819189171654812e-05,
+      "loss": 0.9079,
+      "step": 13099
+    },
+    {
+      "epoch": 0.35978138474636784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4818409744590753e-05,
+      "loss": 0.8701,
+      "step": 13100
+    },
+    {
+      "epoch": 0.3598088489742111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4817630279401572e-05,
+      "loss": 0.8688,
+      "step": 13101
+    },
+    {
+      "epoch": 0.35983631320205434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4816850776093432e-05,
+      "loss": 0.9918,
+      "step": 13102
+    },
+    {
+      "epoch": 0.35986377742989756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4816071234672505e-05,
+      "loss": 0.8832,
+      "step": 13103
+    },
+    {
+      "epoch": 0.3598912416577408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4815291655144957e-05,
+      "loss": 0.9727,
+      "step": 13104
+    },
+    {
+      "epoch": 0.359918705885584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.481451203751695e-05,
+      "loss": 0.9779,
+      "step": 13105
+    },
+    {
+      "epoch": 0.3599461701134273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4813732381794666e-05,
+      "loss": 0.9263,
+      "step": 13106
+    },
+    {
+      "epoch": 0.3599736343412705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812952687984264e-05,
+      "loss": 0.9659,
+      "step": 13107
+    },
+    {
+      "epoch": 0.3600010985691137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812172956091914e-05,
+      "loss": 0.8662,
+      "step": 13108
+    },
+    {
+      "epoch": 0.36002856279695694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4811393186123792e-05,
+      "loss": 1.0089,
+      "step": 13109
+    },
+    {
+      "epoch": 0.3600560270248002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.481061337808606e-05,
+      "loss": 1.0014,
+      "step": 13110
+    },
+    {
+      "epoch": 0.36008349125264344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4809833531984892e-05,
+      "loss": 1.0354,
+      "step": 13111
+    },
+    {
+      "epoch": 0.36011095548048666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480905364782646e-05,
+      "loss": 0.9455,
+      "step": 13112
+    },
+    {
+      "epoch": 0.3601384197083299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480827372561693e-05,
+      "loss": 1.0004,
+      "step": 13113
+    },
+    {
+      "epoch": 0.36016588393617316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4807493765362475e-05,
+      "loss": 1.0462,
+      "step": 13114
+    },
+    {
+      "epoch": 0.3601933481640164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4806713767069272e-05,
+      "loss": 1.0399,
+      "step": 13115
+    },
+    {
+      "epoch": 0.3602208123918596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4805933730743484e-05,
+      "loss": 0.8733,
+      "step": 13116
+    },
+    {
+      "epoch": 0.3602482766197028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480515365639129e-05,
+      "loss": 1.0079,
+      "step": 13117
+    },
+    {
+      "epoch": 0.36027574084754604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4804373544018853e-05,
+      "loss": 0.9456,
+      "step": 13118
+    },
+    {
+      "epoch": 0.3603032050753893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4803593393632354e-05,
+      "loss": 1.0068,
+      "step": 13119
+    },
+    {
+      "epoch": 0.36033066930323254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4802813205237964e-05,
+      "loss": 0.9389,
+      "step": 13120
+    },
+    {
+      "epoch": 0.36035813353107576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4802032978841857e-05,
+      "loss": 0.8569,
+      "step": 13121
+    },
+    {
+      "epoch": 0.360385597758919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4801252714450203e-05,
+      "loss": 0.8794,
+      "step": 13122
+    },
+    {
+      "epoch": 0.36041306198676226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4800472412069177e-05,
+      "loss": 0.8732,
+      "step": 13123
+    },
+    {
+      "epoch": 0.3604405262146055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4799692071704954e-05,
+      "loss": 0.9995,
+      "step": 13124
+    },
+    {
+      "epoch": 0.3604679904424487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4798911693363712e-05,
+      "loss": 0.8535,
+      "step": 13125
+    },
+    {
+      "epoch": 0.3604954546702919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4798131277051617e-05,
+      "loss": 1.0095,
+      "step": 13126
+    },
+    {
+      "epoch": 0.3605229188981352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4797350822774852e-05,
+      "loss": 0.9725,
+      "step": 13127
+    },
+    {
+      "epoch": 0.3605503831259784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4796570330539588e-05,
+      "loss": 0.8848,
+      "step": 13128
+    },
+    {
+      "epoch": 0.36057784735382165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4795789800352e-05,
+      "loss": 0.9041,
+      "step": 13129
+    },
+    {
+      "epoch": 0.36060531158166487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4795009232218265e-05,
+      "loss": 0.8288,
+      "step": 13130
+    },
+    {
+      "epoch": 0.3606327758095081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.479422862614456e-05,
+      "loss": 0.9589,
+      "step": 13131
+    },
+    {
+      "epoch": 0.36066024003735137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4793447982137064e-05,
+      "loss": 0.9783,
+      "step": 13132
+    },
+    {
+      "epoch": 0.3606877042651946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.479266730020195e-05,
+      "loss": 0.9362,
+      "step": 13133
+    },
+    {
+      "epoch": 0.3607151684930378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4791886580345391e-05,
+      "loss": 1.0508,
+      "step": 13134
+    },
+    {
+      "epoch": 0.36074263272088103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4791105822573575e-05,
+      "loss": 0.9414,
+      "step": 13135
+    },
+    {
+      "epoch": 0.3607700969487243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.479032502689267e-05,
+      "loss": 0.9538,
+      "step": 13136
+    },
+    {
+      "epoch": 0.3607975611765675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478954419330886e-05,
+      "loss": 0.9161,
+      "step": 13137
+    },
+    {
+      "epoch": 0.36082502540441075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478876332182832e-05,
+      "loss": 0.9588,
+      "step": 13138
+    },
+    {
+      "epoch": 0.36085248963225397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478798241245723e-05,
+      "loss": 0.8827,
+      "step": 13139
+    },
+    {
+      "epoch": 0.36087995386009725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4787201465201771e-05,
+      "loss": 0.9492,
+      "step": 13140
+    },
+    {
+      "epoch": 0.36090741808794047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4786420480068118e-05,
+      "loss": 0.9398,
+      "step": 13141
+    },
+    {
+      "epoch": 0.3609348823157837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4785639457062449e-05,
+      "loss": 0.9045,
+      "step": 13142
+    },
+    {
+      "epoch": 0.3609623465436269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4784858396190948e-05,
+      "loss": 1.048,
+      "step": 13143
+    },
+    {
+      "epoch": 0.3609898107714702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4784077297459795e-05,
+      "loss": 0.9343,
+      "step": 13144
+    },
+    {
+      "epoch": 0.3610172749993134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478329616087517e-05,
+      "loss": 0.9053,
+      "step": 13145
+    },
+    {
+      "epoch": 0.36104473922715663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4782514986443253e-05,
+      "loss": 0.938,
+      "step": 13146
+    },
+    {
+      "epoch": 0.36107220345499985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4781733774170224e-05,
+      "loss": 0.9992,
+      "step": 13147
+    },
+    {
+      "epoch": 0.3610996676828431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4780952524062264e-05,
+      "loss": 0.9277,
+      "step": 13148
+    },
+    {
+      "epoch": 0.36112713191068635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4780171236125555e-05,
+      "loss": 0.8121,
+      "step": 13149
+    },
+    {
+      "epoch": 0.36115459613852957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4779389910366285e-05,
+      "loss": 0.9949,
+      "step": 13150
+    },
+    {
+      "epoch": 0.3611820603663728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4778608546790625e-05,
+      "loss": 0.8555,
+      "step": 13151
+    },
+    {
+      "epoch": 0.361209524594216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4777827145404766e-05,
+      "loss": 0.9438,
+      "step": 13152
+    },
+    {
+      "epoch": 0.3612369888220593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4777045706214886e-05,
+      "loss": 1.0432,
+      "step": 13153
+    },
+    {
+      "epoch": 0.3612644530499025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4776264229227171e-05,
+      "loss": 0.9172,
+      "step": 13154
+    },
+    {
+      "epoch": 0.36129191727774573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4775482714447802e-05,
+      "loss": 0.9648,
+      "step": 13155
+    },
+    {
+      "epoch": 0.36131938150558895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4774701161882965e-05,
+      "loss": 0.9985,
+      "step": 13156
+    },
+    {
+      "epoch": 0.36134684573343223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4773919571538843e-05,
+      "loss": 1.0206,
+      "step": 13157
+    },
+    {
+      "epoch": 0.36137430996127545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4773137943421621e-05,
+      "loss": 1.0445,
+      "step": 13158
+    },
+    {
+      "epoch": 0.3614017741891187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4772356277537481e-05,
+      "loss": 0.8601,
+      "step": 13159
+    },
+    {
+      "epoch": 0.3614292384169619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4771574573892611e-05,
+      "loss": 0.843,
+      "step": 13160
+    },
+    {
+      "epoch": 0.3614567026448051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4770792832493192e-05,
+      "loss": 0.9354,
+      "step": 13161
+    },
+    {
+      "epoch": 0.3614841668726484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4770011053345414e-05,
+      "loss": 0.8882,
+      "step": 13162
+    },
+    {
+      "epoch": 0.3615116311004916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476922923645546e-05,
+      "loss": 1.0208,
+      "step": 13163
+    },
+    {
+      "epoch": 0.36153909532833484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4768447381829518e-05,
+      "loss": 0.9558,
+      "step": 13164
+    },
+    {
+      "epoch": 0.36156655955617806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4767665489473774e-05,
+      "loss": 0.9895,
+      "step": 13165
+    },
+    {
+      "epoch": 0.36159402378402133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4766883559394412e-05,
+      "loss": 0.9844,
+      "step": 13166
+    },
+    {
+      "epoch": 0.36162148801186456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476610159159762e-05,
+      "loss": 0.9937,
+      "step": 13167
+    },
+    {
+      "epoch": 0.3616489522397078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4765319586089589e-05,
+      "loss": 1.0185,
+      "step": 13168
+    },
+    {
+      "epoch": 0.361676416467551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4764537542876501e-05,
+      "loss": 0.9518,
+      "step": 13169
+    },
+    {
+      "epoch": 0.3617038806953943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4763755461964547e-05,
+      "loss": 0.7294,
+      "step": 13170
+    },
+    {
+      "epoch": 0.3617313449232375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4762973343359915e-05,
+      "loss": 1.004,
+      "step": 13171
+    },
+    {
+      "epoch": 0.3617588091510807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4762191187068793e-05,
+      "loss": 0.9619,
+      "step": 13172
+    },
+    {
+      "epoch": 0.36178627337892394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4761408993097368e-05,
+      "loss": 1.0457,
+      "step": 13173
+    },
+    {
+      "epoch": 0.36181373760676716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4760626761451833e-05,
+      "loss": 0.8338,
+      "step": 13174
+    },
+    {
+      "epoch": 0.36184120183461044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4759844492138373e-05,
+      "loss": 0.864,
+      "step": 13175
+    },
+    {
+      "epoch": 0.36186866606245366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4759062185163183e-05,
+      "loss": 0.8999,
+      "step": 13176
+    },
+    {
+      "epoch": 0.3618961302902969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4758279840532445e-05,
+      "loss": 0.959,
+      "step": 13177
+    },
+    {
+      "epoch": 0.3619235945181401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4757497458252355e-05,
+      "loss": 0.9425,
+      "step": 13178
+    },
+    {
+      "epoch": 0.3619510587459834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4756715038329103e-05,
+      "loss": 1.0032,
+      "step": 13179
+    },
+    {
+      "epoch": 0.3619785229738266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4755932580768879e-05,
+      "loss": 0.8805,
+      "step": 13180
+    },
+    {
+      "epoch": 0.3620059872016698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4755150085577874e-05,
+      "loss": 0.8941,
+      "step": 13181
+    },
+    {
+      "epoch": 0.36203345142951304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.475436755276228e-05,
+      "loss": 1.0487,
+      "step": 13182
+    },
+    {
+      "epoch": 0.3620609156573563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4753584982328288e-05,
+      "loss": 0.9436,
+      "step": 13183
+    },
+    {
+      "epoch": 0.36208837988519954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.475280237428209e-05,
+      "loss": 1.0327,
+      "step": 13184
+    },
+    {
+      "epoch": 0.36211584411304276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.475201972862988e-05,
+      "loss": 0.9572,
+      "step": 13185
+    },
+    {
+      "epoch": 0.362143308340886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4751237045377847e-05,
+      "loss": 0.8877,
+      "step": 13186
+    },
+    {
+      "epoch": 0.3621707725687292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4750454324532184e-05,
+      "loss": 0.9288,
+      "step": 13187
+    },
+    {
+      "epoch": 0.3621982367965725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4749671566099091e-05,
+      "loss": 0.9631,
+      "step": 13188
+    },
+    {
+      "epoch": 0.3622257010244157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4748888770084757e-05,
+      "loss": 0.9309,
+      "step": 13189
+    },
+    {
+      "epoch": 0.3622531652522589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4748105936495372e-05,
+      "loss": 1.0027,
+      "step": 13190
+    },
+    {
+      "epoch": 0.36228062948010215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4747323065337136e-05,
+      "loss": 0.9325,
+      "step": 13191
+    },
+    {
+      "epoch": 0.3623080937079454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4746540156616237e-05,
+      "loss": 0.8602,
+      "step": 13192
+    },
+    {
+      "epoch": 0.36233555793578864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4745757210338878e-05,
+      "loss": 0.9792,
+      "step": 13193
+    },
+    {
+      "epoch": 0.36236302216363186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4744974226511248e-05,
+      "loss": 0.9067,
+      "step": 13194
+    },
+    {
+      "epoch": 0.3623904863914751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4744191205139544e-05,
+      "loss": 0.9258,
+      "step": 13195
+    },
+    {
+      "epoch": 0.36241795061931836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4743408146229963e-05,
+      "loss": 0.9615,
+      "step": 13196
+    },
+    {
+      "epoch": 0.3624454148471616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4742625049788698e-05,
+      "loss": 1.0418,
+      "step": 13197
+    },
+    {
+      "epoch": 0.3624728790750048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4741841915821944e-05,
+      "loss": 0.929,
+      "step": 13198
+    },
+    {
+      "epoch": 0.362500343302848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4741058744335903e-05,
+      "loss": 0.9877,
+      "step": 13199
+    },
+    {
+      "epoch": 0.36252780753069125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4740275535336768e-05,
+      "loss": 0.9602,
+      "step": 13200
+    },
+    {
+      "epoch": 0.3625552717585345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4739492288830737e-05,
+      "loss": 0.9785,
+      "step": 13201
+    },
+    {
+      "epoch": 0.36258273598637775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4738709004824005e-05,
+      "loss": 0.9993,
+      "step": 13202
+    },
+    {
+      "epoch": 0.36261020021422097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4737925683322776e-05,
+      "loss": 0.8841,
+      "step": 13203
+    },
+    {
+      "epoch": 0.3626376644420642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473714232433324e-05,
+      "loss": 1.0576,
+      "step": 13204
+    },
+    {
+      "epoch": 0.36266512866990747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.47363589278616e-05,
+      "loss": 0.8837,
+      "step": 13205
+    },
+    {
+      "epoch": 0.3626925928977507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4735575493914056e-05,
+      "loss": 0.9369,
+      "step": 13206
+    },
+    {
+      "epoch": 0.3627200571255939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.47347920224968e-05,
+      "loss": 1.0503,
+      "step": 13207
+    },
+    {
+      "epoch": 0.36274752135343713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4734008513616037e-05,
+      "loss": 0.8972,
+      "step": 13208
+    },
+    {
+      "epoch": 0.3627749855812804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4733224967277969e-05,
+      "loss": 0.9293,
+      "step": 13209
+    },
+    {
+      "epoch": 0.36280244980912363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4732441383488785e-05,
+      "loss": 0.9582,
+      "step": 13210
+    },
+    {
+      "epoch": 0.36282991403696685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4731657762254694e-05,
+      "loss": 0.8823,
+      "step": 13211
+    },
+    {
+      "epoch": 0.36285737826481007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4730874103581894e-05,
+      "loss": 0.8608,
+      "step": 13212
+    },
+    {
+      "epoch": 0.3628848424926533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4730090407476588e-05,
+      "loss": 0.9548,
+      "step": 13213
+    },
+    {
+      "epoch": 0.36291230672049657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4729306673944969e-05,
+      "loss": 0.9325,
+      "step": 13214
+    },
+    {
+      "epoch": 0.3629397709483398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.472852290299325e-05,
+      "loss": 1.0004,
+      "step": 13215
+    },
+    {
+      "epoch": 0.362967235176183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4727739094627622e-05,
+      "loss": 0.8652,
+      "step": 13216
+    },
+    {
+      "epoch": 0.36299469940402623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4726955248854294e-05,
+      "loss": 0.9153,
+      "step": 13217
+    },
+    {
+      "epoch": 0.3630221636318695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4726171365679462e-05,
+      "loss": 0.9725,
+      "step": 13218
+    },
+    {
+      "epoch": 0.36304962785971273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4725387445109331e-05,
+      "loss": 0.9163,
+      "step": 13219
+    },
+    {
+      "epoch": 0.36307709208755595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4724603487150106e-05,
+      "loss": 0.9044,
+      "step": 13220
+    },
+    {
+      "epoch": 0.3631045563153992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.472381949180799e-05,
+      "loss": 0.9432,
+      "step": 13221
+    },
+    {
+      "epoch": 0.36313202054324245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4723035459089181e-05,
+      "loss": 0.9076,
+      "step": 13222
+    },
+    {
+      "epoch": 0.36315948477108567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4722251388999883e-05,
+      "loss": 0.9872,
+      "step": 13223
+    },
+    {
+      "epoch": 0.3631869489989289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4721467281546309e-05,
+      "loss": 0.9637,
+      "step": 13224
+    },
+    {
+      "epoch": 0.3632144132267721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4720683136734656e-05,
+      "loss": 0.951,
+      "step": 13225
+    },
+    {
+      "epoch": 0.36324187745461534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4719898954571128e-05,
+      "loss": 0.8983,
+      "step": 13226
+    },
+    {
+      "epoch": 0.3632693416824586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4719114735061932e-05,
+      "loss": 0.974,
+      "step": 13227
+    },
+    {
+      "epoch": 0.36329680591030183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4718330478213272e-05,
+      "loss": 0.9154,
+      "step": 13228
+    },
+    {
+      "epoch": 0.36332427013814506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4717546184031352e-05,
+      "loss": 0.8896,
+      "step": 13229
+    },
+    {
+      "epoch": 0.3633517343659883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4716761852522381e-05,
+      "loss": 0.9373,
+      "step": 13230
+    },
+    {
+      "epoch": 0.36337919859383155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471597748369256e-05,
+      "loss": 1.0141,
+      "step": 13231
+    },
+    {
+      "epoch": 0.3634066628216748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4715193077548103e-05,
+      "loss": 0.8502,
+      "step": 13232
+    },
+    {
+      "epoch": 0.363434127049518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471440863409521e-05,
+      "loss": 0.9826,
+      "step": 13233
+    },
+    {
+      "epoch": 0.3634615912773612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4713624153340087e-05,
+      "loss": 0.9574,
+      "step": 13234
+    },
+    {
+      "epoch": 0.3634890555052045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4712839635288947e-05,
+      "loss": 0.9173,
+      "step": 13235
+    },
+    {
+      "epoch": 0.3635165197330477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4712055079947994e-05,
+      "loss": 0.961,
+      "step": 13236
+    },
+    {
+      "epoch": 0.36354398396089094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4711270487323434e-05,
+      "loss": 0.8796,
+      "step": 13237
+    },
+    {
+      "epoch": 0.36357144818873416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471048585742148e-05,
+      "loss": 1.028,
+      "step": 13238
+    },
+    {
+      "epoch": 0.36359891241657744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4709701190248333e-05,
+      "loss": 1.0184,
+      "step": 13239
+    },
+    {
+      "epoch": 0.36362637664442066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.470891648581021e-05,
+      "loss": 0.8635,
+      "step": 13240
+    },
+    {
+      "epoch": 0.3636538408722639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4708131744113315e-05,
+      "loss": 1.0062,
+      "step": 13241
+    },
+    {
+      "epoch": 0.3636813051001071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4707346965163855e-05,
+      "loss": 0.9937,
+      "step": 13242
+    },
+    {
+      "epoch": 0.3637087693279503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4706562148968044e-05,
+      "loss": 0.9993,
+      "step": 13243
+    },
+    {
+      "epoch": 0.3637362335557936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.470577729553209e-05,
+      "loss": 0.9465,
+      "step": 13244
+    },
+    {
+      "epoch": 0.3637636977836368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4704992404862203e-05,
+      "loss": 0.9772,
+      "step": 13245
+    },
+    {
+      "epoch": 0.36379116201148004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4704207476964595e-05,
+      "loss": 0.8877,
+      "step": 13246
+    },
+    {
+      "epoch": 0.36381862623932326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4703422511845475e-05,
+      "loss": 1.0604,
+      "step": 13247
+    },
+    {
+      "epoch": 0.36384609046716654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4702637509511051e-05,
+      "loss": 0.9368,
+      "step": 13248
+    },
+    {
+      "epoch": 0.36387355469500976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4701852469967541e-05,
+      "loss": 0.9531,
+      "step": 13249
+    },
+    {
+      "epoch": 0.363901018922853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4701067393221153e-05,
+      "loss": 0.8871,
+      "step": 13250
+    },
+    {
+      "epoch": 0.3639284831506962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4700282279278097e-05,
+      "loss": 0.8985,
+      "step": 13251
+    },
+    {
+      "epoch": 0.3639559473785395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4699497128144588e-05,
+      "loss": 0.993,
+      "step": 13252
+    },
+    {
+      "epoch": 0.3639834116063827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4698711939826837e-05,
+      "loss": 0.8511,
+      "step": 13253
+    },
+    {
+      "epoch": 0.3640108758342259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4697926714331057e-05,
+      "loss": 0.8826,
+      "step": 13254
+    },
+    {
+      "epoch": 0.36403834006206914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4697141451663459e-05,
+      "loss": 0.9779,
+      "step": 13255
+    },
+    {
+      "epoch": 0.36406580428991236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.469635615183026e-05,
+      "loss": 0.9706,
+      "step": 13256
+    },
+    {
+      "epoch": 0.36409326851775564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4695570814837674e-05,
+      "loss": 0.9843,
+      "step": 13257
+    },
+    {
+      "epoch": 0.36412073274559886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4694785440691907e-05,
+      "loss": 0.9224,
+      "step": 13258
+    },
+    {
+      "epoch": 0.3641481969734421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4694000029399184e-05,
+      "loss": 0.8082,
+      "step": 13259
+    },
+    {
+      "epoch": 0.3641756612012853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4693214580965713e-05,
+      "loss": 1.077,
+      "step": 13260
+    },
+    {
+      "epoch": 0.3642031254291286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4692429095397712e-05,
+      "loss": 0.9836,
+      "step": 13261
+    },
+    {
+      "epoch": 0.3642305896569718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4691643572701393e-05,
+      "loss": 1.0016,
+      "step": 13262
+    },
+    {
+      "epoch": 0.364258053884815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4690858012882972e-05,
+      "loss": 0.8801,
+      "step": 13263
+    },
+    {
+      "epoch": 0.36428551811265825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4690072415948666e-05,
+      "loss": 0.9717,
+      "step": 13264
+    },
+    {
+      "epoch": 0.3643129823405015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4689286781904693e-05,
+      "loss": 0.9506,
+      "step": 13265
+    },
+    {
+      "epoch": 0.36434044656834474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.468850111075726e-05,
+      "loss": 0.9238,
+      "step": 13266
+    },
+    {
+      "epoch": 0.36436791079618797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4687715402512595e-05,
+      "loss": 0.9385,
+      "step": 13267
+    },
+    {
+      "epoch": 0.3643953750240312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4686929657176906e-05,
+      "loss": 1.0224,
+      "step": 13268
+    },
+    {
+      "epoch": 0.3644228392518744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.468614387475642e-05,
+      "loss": 0.9084,
+      "step": 13269
+    },
+    {
+      "epoch": 0.3644503034797177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4685358055257343e-05,
+      "loss": 0.8907,
+      "step": 13270
+    },
+    {
+      "epoch": 0.3644777677075609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4684572198685903e-05,
+      "loss": 1.0548,
+      "step": 13271
+    },
+    {
+      "epoch": 0.36450523193540413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.468378630504831e-05,
+      "loss": 1.0566,
+      "step": 13272
+    },
+    {
+      "epoch": 0.36453269616324735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4683000374350785e-05,
+      "loss": 0.9856,
+      "step": 13273
+    },
+    {
+      "epoch": 0.3645601603910906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4682214406599548e-05,
+      "loss": 0.9299,
+      "step": 13274
+    },
+    {
+      "epoch": 0.36458762461893385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4681428401800817e-05,
+      "loss": 0.871,
+      "step": 13275
+    },
+    {
+      "epoch": 0.36461508884677707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4680642359960811e-05,
+      "loss": 0.9664,
+      "step": 13276
+    },
+    {
+      "epoch": 0.3646425530746203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4679856281085748e-05,
+      "loss": 0.9621,
+      "step": 13277
+    },
+    {
+      "epoch": 0.36467001730246357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467907016518185e-05,
+      "loss": 0.9373,
+      "step": 13278
+    },
+    {
+      "epoch": 0.3646974815303068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4678284012255338e-05,
+      "loss": 0.8037,
+      "step": 13279
+    },
+    {
+      "epoch": 0.36472494575815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467749782231243e-05,
+      "loss": 0.9494,
+      "step": 13280
+    },
+    {
+      "epoch": 0.36475240998599323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4676711595359348e-05,
+      "loss": 0.9984,
+      "step": 13281
+    },
+    {
+      "epoch": 0.36477987421383645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467592533140231e-05,
+      "loss": 1.0189,
+      "step": 13282
+    },
+    {
+      "epoch": 0.36480733844167973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467513903044754e-05,
+      "loss": 0.9048,
+      "step": 13283
+    },
+    {
+      "epoch": 0.36483480266952295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467435269250126e-05,
+      "loss": 0.9302,
+      "step": 13284
+    },
+    {
+      "epoch": 0.36486226689736617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4673566317569692e-05,
+      "loss": 0.9945,
+      "step": 13285
+    },
+    {
+      "epoch": 0.3648897311252094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4672779905659056e-05,
+      "loss": 0.8217,
+      "step": 13286
+    },
+    {
+      "epoch": 0.36491719535305267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4671993456775573e-05,
+      "loss": 0.9743,
+      "step": 13287
+    },
+    {
+      "epoch": 0.3649446595808959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467120697092547e-05,
+      "loss": 0.9538,
+      "step": 13288
+    },
+    {
+      "epoch": 0.3649721238087391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467042044811497e-05,
+      "loss": 0.9789,
+      "step": 13289
+    },
+    {
+      "epoch": 0.36499958803658233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4669633888350292e-05,
+      "loss": 0.9659,
+      "step": 13290
+    },
+    {
+      "epoch": 0.3650270522644256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.466884729163766e-05,
+      "loss": 0.8666,
+      "step": 13291
+    },
+    {
+      "epoch": 0.36505451649226883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4668060657983303e-05,
+      "loss": 1.0615,
+      "step": 13292
+    },
+    {
+      "epoch": 0.36508198072011205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.466727398739344e-05,
+      "loss": 0.966,
+      "step": 13293
+    },
+    {
+      "epoch": 0.3651094449479553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4666487279874297e-05,
+      "loss": 1.0081,
+      "step": 13294
+    },
+    {
+      "epoch": 0.3651369091757985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4665700535432102e-05,
+      "loss": 0.8554,
+      "step": 13295
+    },
+    {
+      "epoch": 0.3651643734036418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4664913754073075e-05,
+      "loss": 1.0063,
+      "step": 13296
+    },
+    {
+      "epoch": 0.365191837631485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4664126935803442e-05,
+      "loss": 0.9862,
+      "step": 13297
+    },
+    {
+      "epoch": 0.3652193018593282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.466334008062943e-05,
+      "loss": 0.9431,
+      "step": 13298
+    },
+    {
+      "epoch": 0.36524676608717144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4662553188557266e-05,
+      "loss": 1.0504,
+      "step": 13299
+    },
+    {
+      "epoch": 0.3652742303150147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4661766259593177e-05,
+      "loss": 1.082,
+      "step": 13300
+    },
+    {
+      "epoch": 0.36530169454285794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4660979293743388e-05,
+      "loss": 0.7763,
+      "step": 13301
+    },
+    {
+      "epoch": 0.36532915877070116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4660192291014119e-05,
+      "loss": 0.9619,
+      "step": 13302
+    },
+    {
+      "epoch": 0.3653566229985444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.465940525141161e-05,
+      "loss": 0.9829,
+      "step": 13303
+    },
+    {
+      "epoch": 0.36538408722638765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4658618174942078e-05,
+      "loss": 0.9522,
+      "step": 13304
+    },
+    {
+      "epoch": 0.3654115514542309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4657831061611756e-05,
+      "loss": 0.9989,
+      "step": 13305
+    },
+    {
+      "epoch": 0.3654390156820741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4657043911426872e-05,
+      "loss": 0.9553,
+      "step": 13306
+    },
+    {
+      "epoch": 0.3654664799099173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4656256724393649e-05,
+      "loss": 0.9897,
+      "step": 13307
+    },
+    {
+      "epoch": 0.36549394413776054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4655469500518322e-05,
+      "loss": 0.9576,
+      "step": 13308
+    },
+    {
+      "epoch": 0.3655214083656038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4654682239807116e-05,
+      "loss": 0.8484,
+      "step": 13309
+    },
+    {
+      "epoch": 0.36554887259344704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.465389494226626e-05,
+      "loss": 0.9486,
+      "step": 13310
+    },
+    {
+      "epoch": 0.36557633682129026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4653107607901985e-05,
+      "loss": 0.9494,
+      "step": 13311
+    },
+    {
+      "epoch": 0.3656038010491335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4652320236720518e-05,
+      "loss": 0.8554,
+      "step": 13312
+    },
+    {
+      "epoch": 0.36563126527697676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4651532828728097e-05,
+      "loss": 0.9483,
+      "step": 13313
+    },
+    {
+      "epoch": 0.36565872950482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4650745383930943e-05,
+      "loss": 1.0288,
+      "step": 13314
+    },
+    {
+      "epoch": 0.3656861937326632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464995790233529e-05,
+      "loss": 0.9642,
+      "step": 13315
+    },
+    {
+      "epoch": 0.3657136579605064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4649170383947367e-05,
+      "loss": 0.8293,
+      "step": 13316
+    },
+    {
+      "epoch": 0.3657411221883497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4648382828773408e-05,
+      "loss": 0.9618,
+      "step": 13317
+    },
+    {
+      "epoch": 0.3657685864161929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4647595236819648e-05,
+      "loss": 0.8994,
+      "step": 13318
+    },
+    {
+      "epoch": 0.36579605064403614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4646807608092308e-05,
+      "loss": 0.8939,
+      "step": 13319
+    },
+    {
+      "epoch": 0.36582351487187936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4646019942597629e-05,
+      "loss": 0.9421,
+      "step": 13320
+    },
+    {
+      "epoch": 0.3658509790997226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4645232240341842e-05,
+      "loss": 1.0718,
+      "step": 13321
+    },
+    {
+      "epoch": 0.36587844332756586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4644444501331175e-05,
+      "loss": 1.0609,
+      "step": 13322
+    },
+    {
+      "epoch": 0.3659059075554091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4643656725571865e-05,
+      "loss": 1.043,
+      "step": 13323
+    },
+    {
+      "epoch": 0.3659333717832523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4642868913070146e-05,
+      "loss": 1.0443,
+      "step": 13324
+    },
+    {
+      "epoch": 0.3659608360110955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4642081063832247e-05,
+      "loss": 0.9039,
+      "step": 13325
+    },
+    {
+      "epoch": 0.3659883002389388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4641293177864406e-05,
+      "loss": 0.9673,
+      "step": 13326
+    },
+    {
+      "epoch": 0.366015764466782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4640505255172855e-05,
+      "loss": 0.9284,
+      "step": 13327
+    },
+    {
+      "epoch": 0.36604322869462524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4639717295763828e-05,
+      "loss": 0.962,
+      "step": 13328
+    },
+    {
+      "epoch": 0.36607069292246847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4638929299643562e-05,
+      "loss": 0.9276,
+      "step": 13329
+    },
+    {
+      "epoch": 0.36609815715031174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.463814126681829e-05,
+      "loss": 0.9972,
+      "step": 13330
+    },
+    {
+      "epoch": 0.36612562137815496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4637353197294248e-05,
+      "loss": 0.9777,
+      "step": 13331
+    },
+    {
+      "epoch": 0.3661530856059982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.463656509107767e-05,
+      "loss": 1.0294,
+      "step": 13332
+    },
+    {
+      "epoch": 0.3661805498338414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4635776948174796e-05,
+      "loss": 0.913,
+      "step": 13333
+    },
+    {
+      "epoch": 0.36620801406168463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4634988768591855e-05,
+      "loss": 0.9633,
+      "step": 13334
+    },
+    {
+      "epoch": 0.3662354782895279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4634200552335089e-05,
+      "loss": 0.9121,
+      "step": 13335
+    },
+    {
+      "epoch": 0.3662629425173711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4633412299410735e-05,
+      "loss": 1.0243,
+      "step": 13336
+    },
+    {
+      "epoch": 0.36629040674521435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4632624009825027e-05,
+      "loss": 0.9056,
+      "step": 13337
+    },
+    {
+      "epoch": 0.36631787097305757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4631835683584201e-05,
+      "loss": 1.0101,
+      "step": 13338
+    },
+    {
+      "epoch": 0.36634533520090085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46310473206945e-05,
+      "loss": 0.9897,
+      "step": 13339
+    },
+    {
+      "epoch": 0.36637279942874407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4630258921162157e-05,
+      "loss": 0.8912,
+      "step": 13340
+    },
+    {
+      "epoch": 0.3664002636565873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.462947048499341e-05,
+      "loss": 0.9712,
+      "step": 13341
+    },
+    {
+      "epoch": 0.3664277278844305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4628682012194503e-05,
+      "loss": 0.9369,
+      "step": 13342
+    },
+    {
+      "epoch": 0.3664551921122738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4627893502771668e-05,
+      "loss": 0.9379,
+      "step": 13343
+    },
+    {
+      "epoch": 0.366482656340117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4627104956731151e-05,
+      "loss": 0.9075,
+      "step": 13344
+    },
+    {
+      "epoch": 0.36651012056796023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4626316374079185e-05,
+      "loss": 0.8734,
+      "step": 13345
+    },
+    {
+      "epoch": 0.36653758479580345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4625527754822008e-05,
+      "loss": 0.9791,
+      "step": 13346
+    },
+    {
+      "epoch": 0.3665650490236467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4624739098965868e-05,
+      "loss": 0.9471,
+      "step": 13347
+    },
+    {
+      "epoch": 0.36659251325148995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4623950406517001e-05,
+      "loss": 0.9721,
+      "step": 13348
+    },
+    {
+      "epoch": 0.36661997747933317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4623161677481645e-05,
+      "loss": 0.942,
+      "step": 13349
+    },
+    {
+      "epoch": 0.3666474417071764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4622372911866043e-05,
+      "loss": 0.9858,
+      "step": 13350
+    },
+    {
+      "epoch": 0.3666749059350196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4621584109676439e-05,
+      "loss": 0.9298,
+      "step": 13351
+    },
+    {
+      "epoch": 0.3667023701628629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4620795270919069e-05,
+      "loss": 0.9628,
+      "step": 13352
+    },
+    {
+      "epoch": 0.3667298343907061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4620006395600177e-05,
+      "loss": 1.0508,
+      "step": 13353
+    },
+    {
+      "epoch": 0.36675729861854933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4619217483726005e-05,
+      "loss": 0.9005,
+      "step": 13354
+    },
+    {
+      "epoch": 0.36678476284639255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4618428535302793e-05,
+      "loss": 0.8707,
+      "step": 13355
+    },
+    {
+      "epoch": 0.36681222707423583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4617639550336788e-05,
+      "loss": 0.9863,
+      "step": 13356
+    },
+    {
+      "epoch": 0.36683969130207905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4616850528834232e-05,
+      "loss": 1.0051,
+      "step": 13357
+    },
+    {
+      "epoch": 0.3668671555299223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4616061470801362e-05,
+      "loss": 0.8853,
+      "step": 13358
+    },
+    {
+      "epoch": 0.3668946197577655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4615272376244428e-05,
+      "loss": 0.856,
+      "step": 13359
+    },
+    {
+      "epoch": 0.36692208398560877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.461448324516967e-05,
+      "loss": 0.9479,
+      "step": 13360
+    },
+    {
+      "epoch": 0.366949548213452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4613694077583333e-05,
+      "loss": 0.9703,
+      "step": 13361
+    },
+    {
+      "epoch": 0.3669770124412952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4612904873491664e-05,
+      "loss": 0.9203,
+      "step": 13362
+    },
+    {
+      "epoch": 0.36700447666913844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46121156329009e-05,
+      "loss": 1.0039,
+      "step": 13363
+    },
+    {
+      "epoch": 0.36703194089698166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4611326355817292e-05,
+      "loss": 0.9225,
+      "step": 13364
+    },
+    {
+      "epoch": 0.36705940512482493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4610537042247088e-05,
+      "loss": 0.891,
+      "step": 13365
+    },
+    {
+      "epoch": 0.36708686935266815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4609747692196526e-05,
+      "loss": 0.8724,
+      "step": 13366
+    },
+    {
+      "epoch": 0.3671143335805114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4608958305671851e-05,
+      "loss": 0.9697,
+      "step": 13367
+    },
+    {
+      "epoch": 0.3671417978083546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4608168882679317e-05,
+      "loss": 0.8465,
+      "step": 13368
+    },
+    {
+      "epoch": 0.3671692620361979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4607379423225165e-05,
+      "loss": 1.0149,
+      "step": 13369
+    },
+    {
+      "epoch": 0.3671967262640411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4606589927315641e-05,
+      "loss": 0.9081,
+      "step": 13370
+    },
+    {
+      "epoch": 0.3672241904918843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4605800394956994e-05,
+      "loss": 0.9744,
+      "step": 13371
+    },
+    {
+      "epoch": 0.36725165471972754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460501082615547e-05,
+      "loss": 0.9253,
+      "step": 13372
+    },
+    {
+      "epoch": 0.3672791189475708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4604221220917318e-05,
+      "loss": 0.9471,
+      "step": 13373
+    },
+    {
+      "epoch": 0.36730658317541404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4603431579248783e-05,
+      "loss": 0.9772,
+      "step": 13374
+    },
+    {
+      "epoch": 0.36733404740325726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4602641901156113e-05,
+      "loss": 1.0036,
+      "step": 13375
+    },
+    {
+      "epoch": 0.3673615116311005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460185218664556e-05,
+      "loss": 0.9775,
+      "step": 13376
+    },
+    {
+      "epoch": 0.3673889758589437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4601062435723368e-05,
+      "loss": 0.8921,
+      "step": 13377
+    },
+    {
+      "epoch": 0.367416440086787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4600272648395788e-05,
+      "loss": 0.9615,
+      "step": 13378
+    },
+    {
+      "epoch": 0.3674439043146302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4599482824669067e-05,
+      "loss": 0.9254,
+      "step": 13379
+    },
+    {
+      "epoch": 0.3674713685424734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.459869296454946e-05,
+      "loss": 0.9723,
+      "step": 13380
+    },
+    {
+      "epoch": 0.36749883277031664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4597903068043212e-05,
+      "loss": 0.9764,
+      "step": 13381
+    },
+    {
+      "epoch": 0.3675262969981599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.459711313515657e-05,
+      "loss": 1.0426,
+      "step": 13382
+    },
+    {
+      "epoch": 0.36755376122600314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4596323165895791e-05,
+      "loss": 1.0216,
+      "step": 13383
+    },
+    {
+      "epoch": 0.36758122545384636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4595533160267124e-05,
+      "loss": 0.9327,
+      "step": 13384
+    },
+    {
+      "epoch": 0.3676086896816896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4594743118276818e-05,
+      "loss": 0.8882,
+      "step": 13385
+    },
+    {
+      "epoch": 0.36763615390953286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4593953039931123e-05,
+      "loss": 0.9233,
+      "step": 13386
+    },
+    {
+      "epoch": 0.3676636181373761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4593162925236293e-05,
+      "loss": 1.0809,
+      "step": 13387
+    },
+    {
+      "epoch": 0.3676910823652193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4592372774198578e-05,
+      "loss": 1.0386,
+      "step": 13388
+    },
+    {
+      "epoch": 0.3677185465930625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4591582586824231e-05,
+      "loss": 0.8477,
+      "step": 13389
+    },
+    {
+      "epoch": 0.36774601082090574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4590792363119503e-05,
+      "loss": 0.8839,
+      "step": 13390
+    },
+    {
+      "epoch": 0.367773475048749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4590002103090648e-05,
+      "loss": 0.9026,
+      "step": 13391
+    },
+    {
+      "epoch": 0.36780093927659224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4589211806743921e-05,
+      "loss": 1.0415,
+      "step": 13392
+    },
+    {
+      "epoch": 0.36782840350443546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.458842147408557e-05,
+      "loss": 0.907,
+      "step": 13393
+    },
+    {
+      "epoch": 0.3678558677322787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4587631105121851e-05,
+      "loss": 0.8806,
+      "step": 13394
+    },
+    {
+      "epoch": 0.36788333196012196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.458684069985902e-05,
+      "loss": 0.9811,
+      "step": 13395
+    },
+    {
+      "epoch": 0.3679107961879652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4586050258303323e-05,
+      "loss": 0.8968,
+      "step": 13396
+    },
+    {
+      "epoch": 0.3679382604158084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4585259780461025e-05,
+      "loss": 0.9354,
+      "step": 13397
+    },
+    {
+      "epoch": 0.3679657246436516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4584469266338374e-05,
+      "loss": 0.829,
+      "step": 13398
+    },
+    {
+      "epoch": 0.3679931888714949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4583678715941628e-05,
+      "loss": 1.0419,
+      "step": 13399
+    },
+    {
+      "epoch": 0.3680206530993381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4582888129277038e-05,
+      "loss": 0.9165,
+      "step": 13400
+    },
+    {
+      "epoch": 0.36804811732718135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4582097506350864e-05,
+      "loss": 0.951,
+      "step": 13401
+    },
+    {
+      "epoch": 0.36807558155502457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4581306847169357e-05,
+      "loss": 1.0117,
+      "step": 13402
+    },
+    {
+      "epoch": 0.3681030457828678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4580516151738776e-05,
+      "loss": 0.9907,
+      "step": 13403
+    },
+    {
+      "epoch": 0.36813051001071107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4579725420065379e-05,
+      "loss": 0.8642,
+      "step": 13404
+    },
+    {
+      "epoch": 0.3681579742385543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4578934652155418e-05,
+      "loss": 1.0525,
+      "step": 13405
+    },
+    {
+      "epoch": 0.3681854384663975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4578143848015154e-05,
+      "loss": 0.9564,
+      "step": 13406
+    },
+    {
+      "epoch": 0.36821290269424073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4577353007650842e-05,
+      "loss": 0.952,
+      "step": 13407
+    },
+    {
+      "epoch": 0.368240366922084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457656213106874e-05,
+      "loss": 0.9042,
+      "step": 13408
+    },
+    {
+      "epoch": 0.3682678311499272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4575771218275106e-05,
+      "loss": 0.9386,
+      "step": 13409
+    },
+    {
+      "epoch": 0.36829529537777045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4574980269276198e-05,
+      "loss": 1.0194,
+      "step": 13410
+    },
+    {
+      "epoch": 0.36832275960561367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457418928407827e-05,
+      "loss": 0.9584,
+      "step": 13411
+    },
+    {
+      "epoch": 0.36835022383345695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457339826268759e-05,
+      "loss": 0.9575,
+      "step": 13412
+    },
+    {
+      "epoch": 0.36837768806130017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457260720511041e-05,
+      "loss": 1.0956,
+      "step": 13413
+    },
+    {
+      "epoch": 0.3684051522891434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4571816111352988e-05,
+      "loss": 0.8942,
+      "step": 13414
+    },
+    {
+      "epoch": 0.3684326165169866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4571024981421586e-05,
+      "loss": 1.0018,
+      "step": 13415
+    },
+    {
+      "epoch": 0.36846008074482983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4570233815322465e-05,
+      "loss": 1.0192,
+      "step": 13416
+    },
+    {
+      "epoch": 0.3684875449726731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4569442613061886e-05,
+      "loss": 0.917,
+      "step": 13417
+    },
+    {
+      "epoch": 0.36851500920051633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4568651374646104e-05,
+      "loss": 0.909,
+      "step": 13418
+    },
+    {
+      "epoch": 0.36854247342835955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4567860100081383e-05,
+      "loss": 1.0178,
+      "step": 13419
+    },
+    {
+      "epoch": 0.3685699376562028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4567068789373984e-05,
+      "loss": 0.8527,
+      "step": 13420
+    },
+    {
+      "epoch": 0.36859740188404605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.456627744253017e-05,
+      "loss": 0.9363,
+      "step": 13421
+    },
+    {
+      "epoch": 0.36862486611188927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4565486059556197e-05,
+      "loss": 0.9254,
+      "step": 13422
+    },
+    {
+      "epoch": 0.3686523303397325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4564694640458328e-05,
+      "loss": 0.9287,
+      "step": 13423
+    },
+    {
+      "epoch": 0.3686797945675757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4563903185242832e-05,
+      "loss": 0.9219,
+      "step": 13424
+    },
+    {
+      "epoch": 0.368707258795419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4563111693915963e-05,
+      "loss": 0.9368,
+      "step": 13425
+    },
+    {
+      "epoch": 0.3687347230232622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4562320166483984e-05,
+      "loss": 0.9585,
+      "step": 13426
+    },
+    {
+      "epoch": 0.36876218725110543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4561528602953167e-05,
+      "loss": 1.0052,
+      "step": 13427
+    },
+    {
+      "epoch": 0.36878965147894865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4560737003329763e-05,
+      "loss": 0.9598,
+      "step": 13428
+    },
+    {
+      "epoch": 0.3688171157067919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4559945367620045e-05,
+      "loss": 0.906,
+      "step": 13429
+    },
+    {
+      "epoch": 0.36884457993463515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4559153695830272e-05,
+      "loss": 0.9663,
+      "step": 13430
+    },
+    {
+      "epoch": 0.3688720441624784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4558361987966707e-05,
+      "loss": 1.0021,
+      "step": 13431
+    },
+    {
+      "epoch": 0.3688995083903216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4557570244035617e-05,
+      "loss": 0.8699,
+      "step": 13432
+    },
+    {
+      "epoch": 0.3689269726181648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4556778464043269e-05,
+      "loss": 0.8936,
+      "step": 13433
+    },
+    {
+      "epoch": 0.3689544368460081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4555986647995921e-05,
+      "loss": 0.8975,
+      "step": 13434
+    },
+    {
+      "epoch": 0.3689819010738513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4555194795899841e-05,
+      "loss": 0.9748,
+      "step": 13435
+    },
+    {
+      "epoch": 0.36900936530169454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4554402907761298e-05,
+      "loss": 0.923,
+      "step": 13436
+    },
+    {
+      "epoch": 0.36903682952953776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4553610983586556e-05,
+      "loss": 0.9038,
+      "step": 13437
+    },
+    {
+      "epoch": 0.36906429375738103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4552819023381876e-05,
+      "loss": 0.93,
+      "step": 13438
+    },
+    {
+      "epoch": 0.36909175798522426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.455202702715353e-05,
+      "loss": 0.9453,
+      "step": 13439
+    },
+    {
+      "epoch": 0.3691192222130675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4551234994907783e-05,
+      "loss": 1.0102,
+      "step": 13440
+    },
+    {
+      "epoch": 0.3691466864409107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4550442926650902e-05,
+      "loss": 0.9907,
+      "step": 13441
+    },
+    {
+      "epoch": 0.369174150668754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4549650822389156e-05,
+      "loss": 0.8892,
+      "step": 13442
+    },
+    {
+      "epoch": 0.3692016148965972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4548858682128808e-05,
+      "loss": 0.9412,
+      "step": 13443
+    },
+    {
+      "epoch": 0.3692290791244404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4548066505876126e-05,
+      "loss": 0.8539,
+      "step": 13444
+    },
+    {
+      "epoch": 0.36925654335228364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4547274293637386e-05,
+      "loss": 0.938,
+      "step": 13445
+    },
+    {
+      "epoch": 0.36928400758012686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4546482045418846e-05,
+      "loss": 0.9302,
+      "step": 13446
+    },
+    {
+      "epoch": 0.36931147180797014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4545689761226779e-05,
+      "loss": 0.9461,
+      "step": 13447
+    },
+    {
+      "epoch": 0.36933893603581336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4544897441067454e-05,
+      "loss": 1.0147,
+      "step": 13448
+    },
+    {
+      "epoch": 0.3693664002636566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4544105084947139e-05,
+      "loss": 0.8229,
+      "step": 13449
+    },
+    {
+      "epoch": 0.3693938644914998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4543312692872107e-05,
+      "loss": 0.967,
+      "step": 13450
+    },
+    {
+      "epoch": 0.3694213287193431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4542520264848624e-05,
+      "loss": 1.0072,
+      "step": 13451
+    },
+    {
+      "epoch": 0.3694487929471863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4541727800882959e-05,
+      "loss": 0.9232,
+      "step": 13452
+    },
+    {
+      "epoch": 0.3694762571750295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4540935300981386e-05,
+      "loss": 0.9097,
+      "step": 13453
+    },
+    {
+      "epoch": 0.36950372140287274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4540142765150175e-05,
+      "loss": 0.8918,
+      "step": 13454
+    },
+    {
+      "epoch": 0.369531185630716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4539350193395594e-05,
+      "loss": 0.8529,
+      "step": 13455
+    },
+    {
+      "epoch": 0.36955864985855924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4538557585723917e-05,
+      "loss": 0.9175,
+      "step": 13456
+    },
+    {
+      "epoch": 0.36958611408640246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4537764942141416e-05,
+      "loss": 0.8673,
+      "step": 13457
+    },
+    {
+      "epoch": 0.3696135783142457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4536972262654358e-05,
+      "loss": 0.9246,
+      "step": 13458
+    },
+    {
+      "epoch": 0.3696410425420889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4536179547269018e-05,
+      "loss": 0.9465,
+      "step": 13459
+    },
+    {
+      "epoch": 0.3696685067699322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4535386795991668e-05,
+      "loss": 0.8691,
+      "step": 13460
+    },
+    {
+      "epoch": 0.3696959709977754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4534594008828584e-05,
+      "loss": 0.9214,
+      "step": 13461
+    },
+    {
+      "epoch": 0.3697234352256186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4533801185786032e-05,
+      "loss": 0.8939,
+      "step": 13462
+    },
+    {
+      "epoch": 0.36975089945346185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4533008326870292e-05,
+      "loss": 0.9388,
+      "step": 13463
+    },
+    {
+      "epoch": 0.3697783636813051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4532215432087631e-05,
+      "loss": 0.911,
+      "step": 13464
+    },
+    {
+      "epoch": 0.36980582790914834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.453142250144433e-05,
+      "loss": 1.0588,
+      "step": 13465
+    },
+    {
+      "epoch": 0.36983329213699156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4530629534946657e-05,
+      "loss": 0.9634,
+      "step": 13466
+    },
+    {
+      "epoch": 0.3698607563648348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4529836532600884e-05,
+      "loss": 1.0103,
+      "step": 13467
+    },
+    {
+      "epoch": 0.36988822059267806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4529043494413296e-05,
+      "loss": 0.9958,
+      "step": 13468
+    },
+    {
+      "epoch": 0.3699156848205213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4528250420390159e-05,
+      "loss": 0.9659,
+      "step": 13469
+    },
+    {
+      "epoch": 0.3699431490483645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4527457310537752e-05,
+      "loss": 0.9316,
+      "step": 13470
+    },
+    {
+      "epoch": 0.3699706132762077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4526664164862345e-05,
+      "loss": 1.0284,
+      "step": 13471
+    },
+    {
+      "epoch": 0.36999807750405095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.452587098337022e-05,
+      "loss": 0.835,
+      "step": 13472
+    },
+    {
+      "epoch": 0.3700255417318942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4525077766067653e-05,
+      "loss": 1.037,
+      "step": 13473
+    },
+    {
+      "epoch": 0.37005300595973745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4524284512960914e-05,
+      "loss": 0.9693,
+      "step": 13474
+    },
+    {
+      "epoch": 0.37008047018758067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4523491224056286e-05,
+      "loss": 0.9288,
+      "step": 13475
+    },
+    {
+      "epoch": 0.3701079344154239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4522697899360042e-05,
+      "loss": 0.8883,
+      "step": 13476
+    },
+    {
+      "epoch": 0.37013539864326717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.452190453887846e-05,
+      "loss": 0.9192,
+      "step": 13477
+    },
+    {
+      "epoch": 0.3701628628711104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4521111142617818e-05,
+      "loss": 0.7604,
+      "step": 13478
+    },
+    {
+      "epoch": 0.3701903270989536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4520317710584393e-05,
+      "loss": 0.8912,
+      "step": 13479
+    },
+    {
+      "epoch": 0.37021779132679683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4519524242784464e-05,
+      "loss": 0.9632,
+      "step": 13480
+    },
+    {
+      "epoch": 0.3702452555546401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4518730739224309e-05,
+      "loss": 0.9245,
+      "step": 13481
+    },
+    {
+      "epoch": 0.37027271978248333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4517937199910204e-05,
+      "loss": 0.9748,
+      "step": 13482
+    },
+    {
+      "epoch": 0.37030018401032655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451714362484843e-05,
+      "loss": 0.9974,
+      "step": 13483
+    },
+    {
+      "epoch": 0.37032764823816977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4516350014045269e-05,
+      "loss": 0.9402,
+      "step": 13484
+    },
+    {
+      "epoch": 0.370355112466013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4515556367506993e-05,
+      "loss": 0.8539,
+      "step": 13485
+    },
+    {
+      "epoch": 0.37038257669385627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451476268523989e-05,
+      "loss": 1.0234,
+      "step": 13486
+    },
+    {
+      "epoch": 0.3704100409216995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4513968967250232e-05,
+      "loss": 0.9017,
+      "step": 13487
+    },
+    {
+      "epoch": 0.3704375051495427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4513175213544304e-05,
+      "loss": 0.9696,
+      "step": 13488
+    },
+    {
+      "epoch": 0.37046496937738593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451238142412839e-05,
+      "loss": 1.0109,
+      "step": 13489
+    },
+    {
+      "epoch": 0.3704924336052292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451158759900876e-05,
+      "loss": 0.9399,
+      "step": 13490
+    },
+    {
+      "epoch": 0.37051989783307243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4510793738191701e-05,
+      "loss": 0.9034,
+      "step": 13491
+    },
+    {
+      "epoch": 0.37054736206091565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.45099998416835e-05,
+      "loss": 0.9167,
+      "step": 13492
+    },
+    {
+      "epoch": 0.3705748262887589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4509205909490431e-05,
+      "loss": 0.9805,
+      "step": 13493
+    },
+    {
+      "epoch": 0.37060229051660215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4508411941618777e-05,
+      "loss": 0.985,
+      "step": 13494
+    },
+    {
+      "epoch": 0.37062975474444537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4507617938074823e-05,
+      "loss": 0.8615,
+      "step": 13495
+    },
+    {
+      "epoch": 0.3706572189722886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4506823898864848e-05,
+      "loss": 0.975,
+      "step": 13496
+    },
+    {
+      "epoch": 0.3706846832001318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4506029823995137e-05,
+      "loss": 0.9761,
+      "step": 13497
+    },
+    {
+      "epoch": 0.37071214742797504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4505235713471974e-05,
+      "loss": 0.8319,
+      "step": 13498
+    },
+    {
+      "epoch": 0.3707396116558183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.450444156730164e-05,
+      "loss": 0.8626,
+      "step": 13499
+    },
+    {
+      "epoch": 0.37076707588366153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4503647385490417e-05,
+      "loss": 1.0295,
+      "step": 13500
+    },
+    {
+      "epoch": 0.37079454011150476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4502853168044597e-05,
+      "loss": 0.9494,
+      "step": 13501
+    },
+    {
+      "epoch": 0.370822004339348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4502058914970454e-05,
+      "loss": 0.9387,
+      "step": 13502
+    },
+    {
+      "epoch": 0.37084946856719125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4501264626274277e-05,
+      "loss": 1.0142,
+      "step": 13503
+    },
+    {
+      "epoch": 0.3708769327950345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.450047030196235e-05,
+      "loss": 0.9456,
+      "step": 13504
+    },
+    {
+      "epoch": 0.3709043970228777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4499675942040963e-05,
+      "loss": 0.9333,
+      "step": 13505
+    },
+    {
+      "epoch": 0.3709318612507209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4498881546516392e-05,
+      "loss": 0.9418,
+      "step": 13506
+    },
+    {
+      "epoch": 0.3709593254785642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4498087115394933e-05,
+      "loss": 0.9348,
+      "step": 13507
+    },
+    {
+      "epoch": 0.3709867897064074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4497292648682862e-05,
+      "loss": 0.9331,
+      "step": 13508
+    },
+    {
+      "epoch": 0.37101425393425064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449649814638647e-05,
+      "loss": 0.8949,
+      "step": 13509
+    },
+    {
+      "epoch": 0.37104171816209386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4495703608512044e-05,
+      "loss": 0.9137,
+      "step": 13510
+    },
+    {
+      "epoch": 0.3710691823899371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449490903506587e-05,
+      "loss": 0.8201,
+      "step": 13511
+    },
+    {
+      "epoch": 0.37109664661778036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449411442605423e-05,
+      "loss": 1.0051,
+      "step": 13512
+    },
+    {
+      "epoch": 0.3711241108456236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4493319781483423e-05,
+      "loss": 0.8668,
+      "step": 13513
+    },
+    {
+      "epoch": 0.3711515750734668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4492525101359724e-05,
+      "loss": 0.9605,
+      "step": 13514
+    },
+    {
+      "epoch": 0.37117903930131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4491730385689428e-05,
+      "loss": 0.9591,
+      "step": 13515
+    },
+    {
+      "epoch": 0.3712065035291533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4490935634478823e-05,
+      "loss": 0.9116,
+      "step": 13516
+    },
+    {
+      "epoch": 0.3712339677569965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4490140847734194e-05,
+      "loss": 1.0106,
+      "step": 13517
+    },
+    {
+      "epoch": 0.37126143198483974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4489346025461829e-05,
+      "loss": 0.9129,
+      "step": 13518
+    },
+    {
+      "epoch": 0.37128889621268296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4488551167668022e-05,
+      "loss": 0.9902,
+      "step": 13519
+    },
+    {
+      "epoch": 0.37131636044052624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4487756274359059e-05,
+      "loss": 0.9781,
+      "step": 13520
+    },
+    {
+      "epoch": 0.37134382466836946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.448696134554123e-05,
+      "loss": 0.8862,
+      "step": 13521
+    },
+    {
+      "epoch": 0.3713712888962127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4486166381220826e-05,
+      "loss": 1.0545,
+      "step": 13522
+    },
+    {
+      "epoch": 0.3713987531240559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4485371381404133e-05,
+      "loss": 0.9612,
+      "step": 13523
+    },
+    {
+      "epoch": 0.3714262173518991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4484576346097443e-05,
+      "loss": 0.8753,
+      "step": 13524
+    },
+    {
+      "epoch": 0.3714536815797424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4483781275307055e-05,
+      "loss": 0.8906,
+      "step": 13525
+    },
+    {
+      "epoch": 0.3714811458075856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4482986169039247e-05,
+      "loss": 0.9369,
+      "step": 13526
+    },
+    {
+      "epoch": 0.37150861003542884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4482191027300314e-05,
+      "loss": 0.9914,
+      "step": 13527
+    },
+    {
+      "epoch": 0.37153607426327206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4481395850096553e-05,
+      "loss": 0.9852,
+      "step": 13528
+    },
+    {
+      "epoch": 0.37156353849111534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4480600637434253e-05,
+      "loss": 0.9565,
+      "step": 13529
+    },
+    {
+      "epoch": 0.37159100271895856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44798053893197e-05,
+      "loss": 0.9352,
+      "step": 13530
+    },
+    {
+      "epoch": 0.3716184669468018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4479010105759198e-05,
+      "loss": 1.0106,
+      "step": 13531
+    },
+    {
+      "epoch": 0.371645931174645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4478214786759028e-05,
+      "loss": 0.9714,
+      "step": 13532
+    },
+    {
+      "epoch": 0.3716733954024883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4477419432325493e-05,
+      "loss": 0.8684,
+      "step": 13533
+    },
+    {
+      "epoch": 0.3717008596303315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4476624042464877e-05,
+      "loss": 0.8974,
+      "step": 13534
+    },
+    {
+      "epoch": 0.3717283238581747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4475828617183479e-05,
+      "loss": 0.7864,
+      "step": 13535
+    },
+    {
+      "epoch": 0.37175578808601795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447503315648759e-05,
+      "loss": 0.9958,
+      "step": 13536
+    },
+    {
+      "epoch": 0.3717832523138612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4474237660383507e-05,
+      "loss": 0.9744,
+      "step": 13537
+    },
+    {
+      "epoch": 0.37181071654170444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4473442128877519e-05,
+      "loss": 0.9451,
+      "step": 13538
+    },
+    {
+      "epoch": 0.37183818076954767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447264656197593e-05,
+      "loss": 0.941,
+      "step": 13539
+    },
+    {
+      "epoch": 0.3718656449973909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4471850959685023e-05,
+      "loss": 0.97,
+      "step": 13540
+    },
+    {
+      "epoch": 0.3718931092252341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4471055322011101e-05,
+      "loss": 0.9229,
+      "step": 13541
+    },
+    {
+      "epoch": 0.3719205734530774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4470259648960458e-05,
+      "loss": 0.908,
+      "step": 13542
+    },
+    {
+      "epoch": 0.3719480376809206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.446946394053939e-05,
+      "loss": 0.8859,
+      "step": 13543
+    },
+    {
+      "epoch": 0.37197550190876383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4468668196754188e-05,
+      "loss": 0.9867,
+      "step": 13544
+    },
+    {
+      "epoch": 0.37200296613660705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4467872417611157e-05,
+      "loss": 0.8806,
+      "step": 13545
+    },
+    {
+      "epoch": 0.3720304303644503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4467076603116588e-05,
+      "loss": 0.9255,
+      "step": 13546
+    },
+    {
+      "epoch": 0.37205789459229355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4466280753276775e-05,
+      "loss": 0.8688,
+      "step": 13547
+    },
+    {
+      "epoch": 0.37208535882013677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4465484868098022e-05,
+      "loss": 0.8446,
+      "step": 13548
+    },
+    {
+      "epoch": 0.37211282304798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4464688947586623e-05,
+      "loss": 0.9054,
+      "step": 13549
+    },
+    {
+      "epoch": 0.37214028727582327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.446389299174887e-05,
+      "loss": 0.9297,
+      "step": 13550
+    },
+    {
+      "epoch": 0.3721677515036665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4463097000591074e-05,
+      "loss": 0.9293,
+      "step": 13551
+    },
+    {
+      "epoch": 0.3721952157315097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4462300974119521e-05,
+      "loss": 0.943,
+      "step": 13552
+    },
+    {
+      "epoch": 0.37222267995935293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4461504912340515e-05,
+      "loss": 0.8929,
+      "step": 13553
+    },
+    {
+      "epoch": 0.37225014418719615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4460708815260356e-05,
+      "loss": 0.8692,
+      "step": 13554
+    },
+    {
+      "epoch": 0.37227760841503943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445991268288534e-05,
+      "loss": 0.8864,
+      "step": 13555
+    },
+    {
+      "epoch": 0.37230507264288265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4459116515221765e-05,
+      "loss": 0.9831,
+      "step": 13556
+    },
+    {
+      "epoch": 0.37233253687072587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4458320312275938e-05,
+      "loss": 0.9056,
+      "step": 13557
+    },
+    {
+      "epoch": 0.3723600010985691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4457524074054148e-05,
+      "loss": 0.8985,
+      "step": 13558
+    },
+    {
+      "epoch": 0.37238746532641237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4456727800562703e-05,
+      "loss": 1.0489,
+      "step": 13559
+    },
+    {
+      "epoch": 0.3724149295542556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455931491807902e-05,
+      "loss": 0.9248,
+      "step": 13560
+    },
+    {
+      "epoch": 0.3724423937820988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455135147796044e-05,
+      "loss": 1.0721,
+      "step": 13561
+    },
+    {
+      "epoch": 0.37246985800994203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445433876853343e-05,
+      "loss": 0.9653,
+      "step": 13562
+    },
+    {
+      "epoch": 0.3724973222377853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4453542354026366e-05,
+      "loss": 0.8929,
+      "step": 13563
+    },
+    {
+      "epoch": 0.37252478646562853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4452745904281144e-05,
+      "loss": 0.8827,
+      "step": 13564
+    },
+    {
+      "epoch": 0.37255225069347175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445194941930408e-05,
+      "loss": 0.9994,
+      "step": 13565
+    },
+    {
+      "epoch": 0.372579714921315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4451152899101462e-05,
+      "loss": 0.9967,
+      "step": 13566
+    },
+    {
+      "epoch": 0.3726071791491582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4450356343679596e-05,
+      "loss": 0.9775,
+      "step": 13567
+    },
+    {
+      "epoch": 0.3726346433770015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4449559753044788e-05,
+      "loss": 0.9922,
+      "step": 13568
+    },
+    {
+      "epoch": 0.3726621076048447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4448763127203344e-05,
+      "loss": 0.8862,
+      "step": 13569
+    },
+    {
+      "epoch": 0.3726895718326879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444796646616156e-05,
+      "loss": 0.8676,
+      "step": 13570
+    },
+    {
+      "epoch": 0.37271703606053114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444716976992574e-05,
+      "loss": 0.9272,
+      "step": 13571
+    },
+    {
+      "epoch": 0.3727445002883744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4446373038502193e-05,
+      "loss": 0.9391,
+      "step": 13572
+    },
+    {
+      "epoch": 0.37277196451621764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444557627189722e-05,
+      "loss": 0.8798,
+      "step": 13573
+    },
+    {
+      "epoch": 0.37279942874406086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4444779470117125e-05,
+      "loss": 0.9177,
+      "step": 13574
+    },
+    {
+      "epoch": 0.3728268929719041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4443982633168214e-05,
+      "loss": 1.0055,
+      "step": 13575
+    },
+    {
+      "epoch": 0.37285435719974735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4443185761056789e-05,
+      "loss": 0.8742,
+      "step": 13576
+    },
+    {
+      "epoch": 0.3728818214275906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4442388853789158e-05,
+      "loss": 0.9334,
+      "step": 13577
+    },
+    {
+      "epoch": 0.3729092856554338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4441591911371626e-05,
+      "loss": 0.9608,
+      "step": 13578
+    },
+    {
+      "epoch": 0.372936749883277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4440794933810498e-05,
+      "loss": 0.9684,
+      "step": 13579
+    },
+    {
+      "epoch": 0.37296421411112024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.443999792111208e-05,
+      "loss": 0.9293,
+      "step": 13580
+    },
+    {
+      "epoch": 0.3729916783389635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4439200873282682e-05,
+      "loss": 0.9257,
+      "step": 13581
+    },
+    {
+      "epoch": 0.37301914256680674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4438403790328601e-05,
+      "loss": 0.8794,
+      "step": 13582
+    },
+    {
+      "epoch": 0.37304660679464996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4437606672256152e-05,
+      "loss": 0.9111,
+      "step": 13583
+    },
+    {
+      "epoch": 0.3730740710224932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4436809519071641e-05,
+      "loss": 0.8565,
+      "step": 13584
+    },
+    {
+      "epoch": 0.37310153525033646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4436012330781375e-05,
+      "loss": 0.8698,
+      "step": 13585
+    },
+    {
+      "epoch": 0.3731289994781797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.443521510739166e-05,
+      "loss": 0.9961,
+      "step": 13586
+    },
+    {
+      "epoch": 0.3731564637060229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4434417848908807e-05,
+      "loss": 0.9297,
+      "step": 13587
+    },
+    {
+      "epoch": 0.3731839279338661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4433620555339118e-05,
+      "loss": 0.9903,
+      "step": 13588
+    },
+    {
+      "epoch": 0.3732113921617094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.443282322668891e-05,
+      "loss": 0.8801,
+      "step": 13589
+    },
+    {
+      "epoch": 0.3732388563895526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4432025862964487e-05,
+      "loss": 1.0107,
+      "step": 13590
+    },
+    {
+      "epoch": 0.37326632061739584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4431228464172158e-05,
+      "loss": 0.9066,
+      "step": 13591
+    },
+    {
+      "epoch": 0.37329378484523906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4430431030318231e-05,
+      "loss": 0.9565,
+      "step": 13592
+    },
+    {
+      "epoch": 0.3733212490730823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442963356140902e-05,
+      "loss": 0.9006,
+      "step": 13593
+    },
+    {
+      "epoch": 0.37334871330092556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4428836057450833e-05,
+      "loss": 1.0181,
+      "step": 13594
+    },
+    {
+      "epoch": 0.3733761775287688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4428038518449973e-05,
+      "loss": 0.8868,
+      "step": 13595
+    },
+    {
+      "epoch": 0.373403641756612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4427240944412763e-05,
+      "loss": 1.0329,
+      "step": 13596
+    },
+    {
+      "epoch": 0.3734311059844552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4426443335345508e-05,
+      "loss": 0.9264,
+      "step": 13597
+    },
+    {
+      "epoch": 0.3734585702122985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4425645691254518e-05,
+      "loss": 0.9392,
+      "step": 13598
+    },
+    {
+      "epoch": 0.3734860344401417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4424848012146104e-05,
+      "loss": 0.8284,
+      "step": 13599
+    },
+    {
+      "epoch": 0.37351349866798494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4424050298026577e-05,
+      "loss": 1.0126,
+      "step": 13600
+    },
+    {
+      "epoch": 0.37354096289582817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4423252548902256e-05,
+      "loss": 0.9331,
+      "step": 13601
+    },
+    {
+      "epoch": 0.37356842712367144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4422454764779442e-05,
+      "loss": 1.0387,
+      "step": 13602
+    },
+    {
+      "epoch": 0.37359589135151466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4421656945664453e-05,
+      "loss": 0.9558,
+      "step": 13603
+    },
+    {
+      "epoch": 0.3736233555793579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4420859091563603e-05,
+      "loss": 0.9903,
+      "step": 13604
+    },
+    {
+      "epoch": 0.3736508198072011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4420061202483205e-05,
+      "loss": 0.8553,
+      "step": 13605
+    },
+    {
+      "epoch": 0.37367828403504433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4419263278429567e-05,
+      "loss": 1.0146,
+      "step": 13606
+    },
+    {
+      "epoch": 0.3737057482628876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.441846531940901e-05,
+      "loss": 1.0627,
+      "step": 13607
+    },
+    {
+      "epoch": 0.3737332124907308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.441766732542784e-05,
+      "loss": 1.0013,
+      "step": 13608
+    },
+    {
+      "epoch": 0.37376067671857405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4416869296492378e-05,
+      "loss": 0.9889,
+      "step": 13609
+    },
+    {
+      "epoch": 0.37378814094641727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4416071232608935e-05,
+      "loss": 0.8421,
+      "step": 13610
+    },
+    {
+      "epoch": 0.37381560517426055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4415273133783823e-05,
+      "loss": 0.8901,
+      "step": 13611
+    },
+    {
+      "epoch": 0.37384306940210377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4414475000023358e-05,
+      "loss": 0.8162,
+      "step": 13612
+    },
+    {
+      "epoch": 0.373870533629947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4413676831333864e-05,
+      "loss": 1.0159,
+      "step": 13613
+    },
+    {
+      "epoch": 0.3738979978577902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4412878627721645e-05,
+      "loss": 0.9923,
+      "step": 13614
+    },
+    {
+      "epoch": 0.3739254620856335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4412080389193019e-05,
+      "loss": 0.9208,
+      "step": 13615
+    },
+    {
+      "epoch": 0.3739529263134767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4411282115754306e-05,
+      "loss": 0.9261,
+      "step": 13616
+    },
+    {
+      "epoch": 0.37398039054131993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4410483807411821e-05,
+      "loss": 0.8703,
+      "step": 13617
+    },
+    {
+      "epoch": 0.37400785476916315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440968546417188e-05,
+      "loss": 0.8948,
+      "step": 13618
+    },
+    {
+      "epoch": 0.37403531899700637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4408887086040797e-05,
+      "loss": 0.9398,
+      "step": 13619
+    },
+    {
+      "epoch": 0.37406278322484965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4408088673024891e-05,
+      "loss": 0.8709,
+      "step": 13620
+    },
+    {
+      "epoch": 0.37409024745269287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4407290225130484e-05,
+      "loss": 0.8905,
+      "step": 13621
+    },
+    {
+      "epoch": 0.3741177116805361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4406491742363889e-05,
+      "loss": 0.9577,
+      "step": 13622
+    },
+    {
+      "epoch": 0.3741451759083793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4405693224731421e-05,
+      "loss": 1.0444,
+      "step": 13623
+    },
+    {
+      "epoch": 0.3741726401362226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44048946722394e-05,
+      "loss": 0.9577,
+      "step": 13624
+    },
+    {
+      "epoch": 0.3742001043640658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4404096084894152e-05,
+      "loss": 1.0257,
+      "step": 13625
+    },
+    {
+      "epoch": 0.37422756859190903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4403297462701988e-05,
+      "loss": 1.0095,
+      "step": 13626
+    },
+    {
+      "epoch": 0.37425503281975225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4402498805669228e-05,
+      "loss": 0.8608,
+      "step": 13627
+    },
+    {
+      "epoch": 0.37428249704759553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440170011380219e-05,
+      "loss": 0.9445,
+      "step": 13628
+    },
+    {
+      "epoch": 0.37430996127543875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44009013871072e-05,
+      "loss": 0.8683,
+      "step": 13629
+    },
+    {
+      "epoch": 0.374337425503282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440010262559057e-05,
+      "loss": 0.9339,
+      "step": 13630
+    },
+    {
+      "epoch": 0.3743648897311252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4399303829258624e-05,
+      "loss": 1.0873,
+      "step": 13631
+    },
+    {
+      "epoch": 0.37439235395896847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439850499811768e-05,
+      "loss": 0.9004,
+      "step": 13632
+    },
+    {
+      "epoch": 0.3744198181868117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4397706132174064e-05,
+      "loss": 0.9613,
+      "step": 13633
+    },
+    {
+      "epoch": 0.3744472824146549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4396907231434096e-05,
+      "loss": 0.9798,
+      "step": 13634
+    },
+    {
+      "epoch": 0.37447474664249814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439610829590409e-05,
+      "loss": 0.9875,
+      "step": 13635
+    },
+    {
+      "epoch": 0.37450221087034136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4395309325590372e-05,
+      "loss": 0.8741,
+      "step": 13636
+    },
+    {
+      "epoch": 0.37452967509818463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439451032049927e-05,
+      "loss": 0.8965,
+      "step": 13637
+    },
+    {
+      "epoch": 0.37455713932602785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4393711280637095e-05,
+      "loss": 1.0764,
+      "step": 13638
+    },
+    {
+      "epoch": 0.3745846035538711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4392912206010175e-05,
+      "loss": 0.929,
+      "step": 13639
+    },
+    {
+      "epoch": 0.3746120677817143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439211309662483e-05,
+      "loss": 0.8836,
+      "step": 13640
+    },
+    {
+      "epoch": 0.3746395320095576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4391313952487388e-05,
+      "loss": 1.0279,
+      "step": 13641
+    },
+    {
+      "epoch": 0.3746669962374008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439051477360417e-05,
+      "loss": 0.8849,
+      "step": 13642
+    },
+    {
+      "epoch": 0.374694460465244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4389715559981496e-05,
+      "loss": 0.9012,
+      "step": 13643
+    },
+    {
+      "epoch": 0.37472192469308724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4388916311625694e-05,
+      "loss": 0.8832,
+      "step": 13644
+    },
+    {
+      "epoch": 0.3747493889209305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4388117028543084e-05,
+      "loss": 0.9424,
+      "step": 13645
+    },
+    {
+      "epoch": 0.37477685314877374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4387317710739995e-05,
+      "loss": 1.0297,
+      "step": 13646
+    },
+    {
+      "epoch": 0.37480431737661696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4386518358222744e-05,
+      "loss": 0.8283,
+      "step": 13647
+    },
+    {
+      "epoch": 0.3748317816044602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4385718970997662e-05,
+      "loss": 0.9502,
+      "step": 13648
+    },
+    {
+      "epoch": 0.3748592458323034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4384919549071077e-05,
+      "loss": 0.9976,
+      "step": 13649
+    },
+    {
+      "epoch": 0.3748867100601467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4384120092449307e-05,
+      "loss": 0.9392,
+      "step": 13650
+    },
+    {
+      "epoch": 0.3749141742879899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4383320601138678e-05,
+      "loss": 0.8414,
+      "step": 13651
+    },
+    {
+      "epoch": 0.3749416385158331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4382521075145523e-05,
+      "loss": 0.9496,
+      "step": 13652
+    },
+    {
+      "epoch": 0.37496910274367634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4381721514476161e-05,
+      "loss": 1.0459,
+      "step": 13653
+    },
+    {
+      "epoch": 0.3749965669715196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.438092191913692e-05,
+      "loss": 0.8518,
+      "step": 13654
+    },
+    {
+      "epoch": 0.37502403119936284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.438012228913413e-05,
+      "loss": 0.9988,
+      "step": 13655
+    },
+    {
+      "epoch": 0.37505149542720606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4379322624474115e-05,
+      "loss": 0.9776,
+      "step": 13656
+    },
+    {
+      "epoch": 0.3750789596550493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4378522925163205e-05,
+      "loss": 0.8981,
+      "step": 13657
+    },
+    {
+      "epoch": 0.37510642388289256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4377723191207724e-05,
+      "loss": 0.9613,
+      "step": 13658
+    },
+    {
+      "epoch": 0.3751338881107358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4376923422614e-05,
+      "loss": 0.9088,
+      "step": 13659
+    },
+    {
+      "epoch": 0.375161352338579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4376123619388364e-05,
+      "loss": 0.8142,
+      "step": 13660
+    },
+    {
+      "epoch": 0.3751888165664222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4375323781537143e-05,
+      "loss": 1.0228,
+      "step": 13661
+    },
+    {
+      "epoch": 0.37521628079426544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4374523909066664e-05,
+      "loss": 0.9691,
+      "step": 13662
+    },
+    {
+      "epoch": 0.3752437450221087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.437372400198326e-05,
+      "loss": 0.8181,
+      "step": 13663
+    },
+    {
+      "epoch": 0.37527120924995194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4372924060293253e-05,
+      "loss": 0.9235,
+      "step": 13664
+    },
+    {
+      "epoch": 0.37529867347779516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4372124084002981e-05,
+      "loss": 0.9155,
+      "step": 13665
+    },
+    {
+      "epoch": 0.3753261377056384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4371324073118769e-05,
+      "loss": 1.0142,
+      "step": 13666
+    },
+    {
+      "epoch": 0.37535360193348166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4370524027646947e-05,
+      "loss": 1.1219,
+      "step": 13667
+    },
+    {
+      "epoch": 0.3753810661613249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4369723947593842e-05,
+      "loss": 1.004,
+      "step": 13668
+    },
+    {
+      "epoch": 0.3754085303891681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4368923832965795e-05,
+      "loss": 0.9917,
+      "step": 13669
+    },
+    {
+      "epoch": 0.3754359946170113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4368123683769126e-05,
+      "loss": 1.0059,
+      "step": 13670
+    },
+    {
+      "epoch": 0.3754634588448546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4367323500010171e-05,
+      "loss": 0.9231,
+      "step": 13671
+    },
+    {
+      "epoch": 0.3754909230726978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4366523281695264e-05,
+      "loss": 0.9722,
+      "step": 13672
+    },
+    {
+      "epoch": 0.37551838730054105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.436572302883073e-05,
+      "loss": 0.8311,
+      "step": 13673
+    },
+    {
+      "epoch": 0.37554585152838427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4364922741422905e-05,
+      "loss": 1.0215,
+      "step": 13674
+    },
+    {
+      "epoch": 0.3755733157562275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4364122419478119e-05,
+      "loss": 0.8746,
+      "step": 13675
+    },
+    {
+      "epoch": 0.37560077998407077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4363322063002707e-05,
+      "loss": 0.9859,
+      "step": 13676
+    },
+    {
+      "epoch": 0.375628244211914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4362521672003002e-05,
+      "loss": 0.9793,
+      "step": 13677
+    },
+    {
+      "epoch": 0.3756557084397572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4361721246485337e-05,
+      "loss": 0.9799,
+      "step": 13678
+    },
+    {
+      "epoch": 0.37568317266760043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4360920786456039e-05,
+      "loss": 0.8838,
+      "step": 13679
+    },
+    {
+      "epoch": 0.3757106368954437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4360120291921448e-05,
+      "loss": 0.9485,
+      "step": 13680
+    },
+    {
+      "epoch": 0.3757381011232869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.43593197628879e-05,
+      "loss": 1.0138,
+      "step": 13681
+    },
+    {
+      "epoch": 0.37576556535113015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4358519199361722e-05,
+      "loss": 0.8661,
+      "step": 13682
+    },
+    {
+      "epoch": 0.37579302957897337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4357718601349253e-05,
+      "loss": 0.8545,
+      "step": 13683
+    },
+    {
+      "epoch": 0.37582049380681665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4356917968856827e-05,
+      "loss": 1.0344,
+      "step": 13684
+    },
+    {
+      "epoch": 0.37584795803465987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4356117301890777e-05,
+      "loss": 0.987,
+      "step": 13685
+    },
+    {
+      "epoch": 0.3758754222625031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4355316600457438e-05,
+      "loss": 1.0057,
+      "step": 13686
+    },
+    {
+      "epoch": 0.3759028864903463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4354515864563152e-05,
+      "loss": 0.9914,
+      "step": 13687
+    },
+    {
+      "epoch": 0.37593035071818953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4353715094214244e-05,
+      "loss": 0.9173,
+      "step": 13688
+    },
+    {
+      "epoch": 0.3759578149460328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.435291428941706e-05,
+      "loss": 0.9547,
+      "step": 13689
+    },
+    {
+      "epoch": 0.37598527917387603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.435211345017793e-05,
+      "loss": 0.9468,
+      "step": 13690
+    },
+    {
+      "epoch": 0.37601274340171925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4351312576503191e-05,
+      "loss": 0.9283,
+      "step": 13691
+    },
+    {
+      "epoch": 0.3760402076295625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4350511668399183e-05,
+      "loss": 0.9276,
+      "step": 13692
+    },
+    {
+      "epoch": 0.37606767185740575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4349710725872243e-05,
+      "loss": 0.8864,
+      "step": 13693
+    },
+    {
+      "epoch": 0.37609513608524897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4348909748928706e-05,
+      "loss": 1.0128,
+      "step": 13694
+    },
+    {
+      "epoch": 0.3761226003130922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4348108737574909e-05,
+      "loss": 1.0208,
+      "step": 13695
+    },
+    {
+      "epoch": 0.3761500645409354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4347307691817192e-05,
+      "loss": 0.9135,
+      "step": 13696
+    },
+    {
+      "epoch": 0.3761775287687787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4346506611661894e-05,
+      "loss": 0.9944,
+      "step": 13697
+    },
+    {
+      "epoch": 0.3762049929966219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.434570549711535e-05,
+      "loss": 1.0146,
+      "step": 13698
+    },
+    {
+      "epoch": 0.37623245722446513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.43449043481839e-05,
+      "loss": 0.9526,
+      "step": 13699
+    },
+    {
+      "epoch": 0.37625992145230835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4344103164873886e-05,
+      "loss": 0.9647,
+      "step": 13700
+    },
+    {
+      "epoch": 0.3762873856801516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4343301947191645e-05,
+      "loss": 1.0143,
+      "step": 13701
+    },
+    {
+      "epoch": 0.37631484990799485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4342500695143518e-05,
+      "loss": 1.0013,
+      "step": 13702
+    },
+    {
+      "epoch": 0.3763423141358381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.434169940873584e-05,
+      "loss": 0.936,
+      "step": 13703
+    },
+    {
+      "epoch": 0.3763697783636813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4340898087974954e-05,
+      "loss": 0.8735,
+      "step": 13704
+    },
+    {
+      "epoch": 0.3763972425915245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4340096732867203e-05,
+      "loss": 0.8837,
+      "step": 13705
+    },
+    {
+      "epoch": 0.3764247068193678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4339295343418927e-05,
+      "loss": 1.095,
+      "step": 13706
+    },
+    {
+      "epoch": 0.376452171047211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433849391963646e-05,
+      "loss": 0.9109,
+      "step": 13707
+    },
+    {
+      "epoch": 0.37647963527505424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4337692461526154e-05,
+      "loss": 0.9162,
+      "step": 13708
+    },
+    {
+      "epoch": 0.37650709950289746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4336890969094343e-05,
+      "loss": 0.9605,
+      "step": 13709
+    },
+    {
+      "epoch": 0.37653456373074073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433608944234737e-05,
+      "loss": 0.9261,
+      "step": 13710
+    },
+    {
+      "epoch": 0.37656202795858396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433528788129158e-05,
+      "loss": 1.0111,
+      "step": 13711
+    },
+    {
+      "epoch": 0.3765894921864272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4334486285933308e-05,
+      "loss": 0.9585,
+      "step": 13712
+    },
+    {
+      "epoch": 0.3766169564142704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4333684656278906e-05,
+      "loss": 0.9055,
+      "step": 13713
+    },
+    {
+      "epoch": 0.3766444206421136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433288299233471e-05,
+      "loss": 0.8267,
+      "step": 13714
+    },
+    {
+      "epoch": 0.3766718848699569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4332081294107067e-05,
+      "loss": 0.9335,
+      "step": 13715
+    },
+    {
+      "epoch": 0.3766993490978001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4331279561602318e-05,
+      "loss": 1.0213,
+      "step": 13716
+    },
+    {
+      "epoch": 0.37672681332564334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433047779482681e-05,
+      "loss": 0.9211,
+      "step": 13717
+    },
+    {
+      "epoch": 0.37675427755348656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432967599378688e-05,
+      "loss": 0.997,
+      "step": 13718
+    },
+    {
+      "epoch": 0.37678174178132984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4328874158488877e-05,
+      "loss": 1.0509,
+      "step": 13719
+    },
+    {
+      "epoch": 0.37680920600917306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4328072288939145e-05,
+      "loss": 0.9125,
+      "step": 13720
+    },
+    {
+      "epoch": 0.3768366702370163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4327270385144032e-05,
+      "loss": 0.9071,
+      "step": 13721
+    },
+    {
+      "epoch": 0.3768641344648595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4326468447109877e-05,
+      "loss": 0.9369,
+      "step": 13722
+    },
+    {
+      "epoch": 0.3768915986927028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4325666474843027e-05,
+      "loss": 0.9793,
+      "step": 13723
+    },
+    {
+      "epoch": 0.376919062920546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4324864468349828e-05,
+      "loss": 0.834,
+      "step": 13724
+    },
+    {
+      "epoch": 0.3769465271483892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432406242763663e-05,
+      "loss": 0.8747,
+      "step": 13725
+    },
+    {
+      "epoch": 0.37697399137623244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4323260352709769e-05,
+      "loss": 1.0284,
+      "step": 13726
+    },
+    {
+      "epoch": 0.37700145560407566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.43224582435756e-05,
+      "loss": 0.9052,
+      "step": 13727
+    },
+    {
+      "epoch": 0.37702891983191894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432165610024047e-05,
+      "loss": 0.8546,
+      "step": 13728
+    },
+    {
+      "epoch": 0.37705638405976216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432085392271072e-05,
+      "loss": 1.0389,
+      "step": 13729
+    },
+    {
+      "epoch": 0.3770838482876054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4320051710992698e-05,
+      "loss": 0.7949,
+      "step": 13730
+    },
+    {
+      "epoch": 0.3771113125154486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4319249465092758e-05,
+      "loss": 0.8576,
+      "step": 13731
+    },
+    {
+      "epoch": 0.3771387767432919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431844718501724e-05,
+      "loss": 0.9313,
+      "step": 13732
+    },
+    {
+      "epoch": 0.3771662409711351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4317644870772496e-05,
+      "loss": 0.9365,
+      "step": 13733
+    },
+    {
+      "epoch": 0.3771937051989783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4316842522364873e-05,
+      "loss": 0.9288,
+      "step": 13734
+    },
+    {
+      "epoch": 0.37722116942682155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431604013980072e-05,
+      "loss": 1.0498,
+      "step": 13735
+    },
+    {
+      "epoch": 0.3772486336546648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4315237723086383e-05,
+      "loss": 1.0694,
+      "step": 13736
+    },
+    {
+      "epoch": 0.37727609788250804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431443527222822e-05,
+      "loss": 0.9624,
+      "step": 13737
+    },
+    {
+      "epoch": 0.37730356211035126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4313632787232567e-05,
+      "loss": 0.9247,
+      "step": 13738
+    },
+    {
+      "epoch": 0.3773310263381945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4312830268105784e-05,
+      "loss": 0.9111,
+      "step": 13739
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4312027714854218e-05,
+      "loss": 0.8962,
+      "step": 13740
+    },
+    {
+      "epoch": 0.377385954793881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4311225127484216e-05,
+      "loss": 0.9134,
+      "step": 13741
+    },
+    {
+      "epoch": 0.3774134190217242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431042250600213e-05,
+      "loss": 0.8992,
+      "step": 13742
+    },
+    {
+      "epoch": 0.3774408832495674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4309619850414316e-05,
+      "loss": 1.0038,
+      "step": 13743
+    },
+    {
+      "epoch": 0.37746834747741065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4308817160727115e-05,
+      "loss": 0.8632,
+      "step": 13744
+    },
+    {
+      "epoch": 0.3774958117052539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4308014436946888e-05,
+      "loss": 0.9903,
+      "step": 13745
+    },
+    {
+      "epoch": 0.37752327593309715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4307211679079979e-05,
+      "loss": 0.9885,
+      "step": 13746
+    },
+    {
+      "epoch": 0.37755074016094037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4306408887132745e-05,
+      "loss": 0.968,
+      "step": 13747
+    },
+    {
+      "epoch": 0.3775782043887836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4305606061111532e-05,
+      "loss": 0.9839,
+      "step": 13748
+    },
+    {
+      "epoch": 0.37760566861662687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4304803201022703e-05,
+      "loss": 0.9324,
+      "step": 13749
+    },
+    {
+      "epoch": 0.3776331328444701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4304000306872597e-05,
+      "loss": 0.9978,
+      "step": 13750
+    },
+    {
+      "epoch": 0.3776605970723133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4303197378667576e-05,
+      "loss": 0.9305,
+      "step": 13751
+    },
+    {
+      "epoch": 0.37768806130015653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.430239441641399e-05,
+      "loss": 1.0347,
+      "step": 13752
+    },
+    {
+      "epoch": 0.3777155255279998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4301591420118195e-05,
+      "loss": 0.97,
+      "step": 13753
+    },
+    {
+      "epoch": 0.37774298975584303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4300788389786537e-05,
+      "loss": 0.9953,
+      "step": 13754
+    },
+    {
+      "epoch": 0.37777045398368625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429998532542538e-05,
+      "loss": 0.8936,
+      "step": 13755
+    },
+    {
+      "epoch": 0.37779791821152947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429918222704107e-05,
+      "loss": 0.9716,
+      "step": 13756
+    },
+    {
+      "epoch": 0.3778253824393727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429837909463997e-05,
+      "loss": 0.9152,
+      "step": 13757
+    },
+    {
+      "epoch": 0.37785284666721597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4297575928228426e-05,
+      "loss": 1.0751,
+      "step": 13758
+    },
+    {
+      "epoch": 0.3778803108950592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4296772727812796e-05,
+      "loss": 0.8522,
+      "step": 13759
+    },
+    {
+      "epoch": 0.3779077751229024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4295969493399435e-05,
+      "loss": 1.0094,
+      "step": 13760
+    },
+    {
+      "epoch": 0.37793523935074563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4295166224994702e-05,
+      "loss": 0.8779,
+      "step": 13761
+    },
+    {
+      "epoch": 0.3779627035785889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429436292260495e-05,
+      "loss": 0.8969,
+      "step": 13762
+    },
+    {
+      "epoch": 0.37799016780643213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4293559586236532e-05,
+      "loss": 0.9748,
+      "step": 13763
+    },
+    {
+      "epoch": 0.37801763203427535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4292756215895811e-05,
+      "loss": 0.9268,
+      "step": 13764
+    },
+    {
+      "epoch": 0.3780450962621186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429195281158914e-05,
+      "loss": 0.8985,
+      "step": 13765
+    },
+    {
+      "epoch": 0.37807256048996185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4291149373322872e-05,
+      "loss": 1.0288,
+      "step": 13766
+    },
+    {
+      "epoch": 0.37810002471780507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4290345901103372e-05,
+      "loss": 0.9415,
+      "step": 13767
+    },
+    {
+      "epoch": 0.3781274889456483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.428954239493699e-05,
+      "loss": 1.0238,
+      "step": 13768
+    },
+    {
+      "epoch": 0.3781549531734915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4288738854830089e-05,
+      "loss": 1.038,
+      "step": 13769
+    },
+    {
+      "epoch": 0.37818241740133474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4287935280789025e-05,
+      "loss": 1.0087,
+      "step": 13770
+    },
+    {
+      "epoch": 0.378209881629178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4287131672820156e-05,
+      "loss": 0.9523,
+      "step": 13771
+    },
+    {
+      "epoch": 0.37823734585702123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4286328030929837e-05,
+      "loss": 0.8809,
+      "step": 13772
+    },
+    {
+      "epoch": 0.37826481008486446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4285524355124436e-05,
+      "loss": 1.0651,
+      "step": 13773
+    },
+    {
+      "epoch": 0.3782922743127077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4284720645410304e-05,
+      "loss": 0.961,
+      "step": 13774
+    },
+    {
+      "epoch": 0.37831973854055095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.42839169017938e-05,
+      "loss": 0.9016,
+      "step": 13775
+    },
+    {
+      "epoch": 0.3783472027683942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.428311312428129e-05,
+      "loss": 0.8303,
+      "step": 13776
+    },
+    {
+      "epoch": 0.3783746669962374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4282309312879127e-05,
+      "loss": 1.0355,
+      "step": 13777
+    },
+    {
+      "epoch": 0.3784021312240806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4281505467593672e-05,
+      "loss": 0.893,
+      "step": 13778
+    },
+    {
+      "epoch": 0.3784295954519239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4280701588431295e-05,
+      "loss": 0.8148,
+      "step": 13779
+    },
+    {
+      "epoch": 0.3784570596797671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427989767539834e-05,
+      "loss": 0.9377,
+      "step": 13780
+    },
+    {
+      "epoch": 0.37848452390761034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427909372850118e-05,
+      "loss": 1.0248,
+      "step": 13781
+    },
+    {
+      "epoch": 0.37851198813545356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4278289747746176e-05,
+      "loss": 0.9009,
+      "step": 13782
+    },
+    {
+      "epoch": 0.3785394523632968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4277485733139682e-05,
+      "loss": 1.0137,
+      "step": 13783
+    },
+    {
+      "epoch": 0.37856691659114006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4276681684688068e-05,
+      "loss": 0.9892,
+      "step": 13784
+    },
+    {
+      "epoch": 0.3785943808189833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427587760239769e-05,
+      "loss": 0.9032,
+      "step": 13785
+    },
+    {
+      "epoch": 0.3786218450468265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4275073486274911e-05,
+      "loss": 0.9359,
+      "step": 13786
+    },
+    {
+      "epoch": 0.3786493092746697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4274269336326094e-05,
+      "loss": 0.9511,
+      "step": 13787
+    },
+    {
+      "epoch": 0.378676773502513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4273465152557604e-05,
+      "loss": 0.9593,
+      "step": 13788
+    },
+    {
+      "epoch": 0.3787042377303562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4272660934975801e-05,
+      "loss": 0.9851,
+      "step": 13789
+    },
+    {
+      "epoch": 0.37873170195819944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4271856683587053e-05,
+      "loss": 1.0348,
+      "step": 13790
+    },
+    {
+      "epoch": 0.37875916618604266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4271052398397715e-05,
+      "loss": 0.9634,
+      "step": 13791
+    },
+    {
+      "epoch": 0.37878663041388594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4270248079414159e-05,
+      "loss": 1.0085,
+      "step": 13792
+    },
+    {
+      "epoch": 0.37881409464172916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4269443726642744e-05,
+      "loss": 0.9139,
+      "step": 13793
+    },
+    {
+      "epoch": 0.3788415588695724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.426863934008984e-05,
+      "loss": 0.9797,
+      "step": 13794
+    },
+    {
+      "epoch": 0.3788690230974156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4267834919761805e-05,
+      "loss": 0.8887,
+      "step": 13795
+    },
+    {
+      "epoch": 0.3788964873252588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4267030465665009e-05,
+      "loss": 0.9791,
+      "step": 13796
+    },
+    {
+      "epoch": 0.3789239515531021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4266225977805816e-05,
+      "loss": 0.9845,
+      "step": 13797
+    },
+    {
+      "epoch": 0.3789514157809453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4265421456190586e-05,
+      "loss": 0.9402,
+      "step": 13798
+    },
+    {
+      "epoch": 0.37897888000878854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4264616900825692e-05,
+      "loss": 0.9828,
+      "step": 13799
+    },
+    {
+      "epoch": 0.37900634423663176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4263812311717498e-05,
+      "loss": 1.015,
+      "step": 13800
+    },
+    {
+      "epoch": 0.37903380846447504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4263007688872368e-05,
+      "loss": 0.9238,
+      "step": 13801
+    },
+    {
+      "epoch": 0.37906127269231826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.426220303229667e-05,
+      "loss": 0.9453,
+      "step": 13802
+    },
+    {
+      "epoch": 0.3790887369201615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.426139834199677e-05,
+      "loss": 1.0266,
+      "step": 13803
+    },
+    {
+      "epoch": 0.3791162011480047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4260593617979036e-05,
+      "loss": 1.0382,
+      "step": 13804
+    },
+    {
+      "epoch": 0.379143665375848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4259788860249837e-05,
+      "loss": 1.02,
+      "step": 13805
+    },
+    {
+      "epoch": 0.3791711296036912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4258984068815536e-05,
+      "loss": 0.9795,
+      "step": 13806
+    },
+    {
+      "epoch": 0.3791985938315344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4258179243682501e-05,
+      "loss": 0.9642,
+      "step": 13807
+    },
+    {
+      "epoch": 0.37922605805937765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4257374384857108e-05,
+      "loss": 0.9239,
+      "step": 13808
+    },
+    {
+      "epoch": 0.37925352228722087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4256569492345717e-05,
+      "loss": 0.9235,
+      "step": 13809
+    },
+    {
+      "epoch": 0.37928098651506414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4255764566154699e-05,
+      "loss": 0.8871,
+      "step": 13810
+    },
+    {
+      "epoch": 0.37930845074290737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4254959606290423e-05,
+      "loss": 0.9358,
+      "step": 13811
+    },
+    {
+      "epoch": 0.3793359149707506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4254154612759259e-05,
+      "loss": 0.9467,
+      "step": 13812
+    },
+    {
+      "epoch": 0.3793633791985938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4253349585567575e-05,
+      "loss": 0.8943,
+      "step": 13813
+    },
+    {
+      "epoch": 0.3793908434264371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4252544524721744e-05,
+      "loss": 0.9282,
+      "step": 13814
+    },
+    {
+      "epoch": 0.3794183076542803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.425173943022813e-05,
+      "loss": 0.9102,
+      "step": 13815
+    },
+    {
+      "epoch": 0.37944577188212353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4250934302093107e-05,
+      "loss": 0.961,
+      "step": 13816
+    },
+    {
+      "epoch": 0.37947323610996675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.425012914032305e-05,
+      "loss": 0.8835,
+      "step": 13817
+    },
+    {
+      "epoch": 0.37950070033781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.424932394492432e-05,
+      "loss": 0.9989,
+      "step": 13818
+    },
+    {
+      "epoch": 0.37952816456565325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4248518715903292e-05,
+      "loss": 0.8503,
+      "step": 13819
+    },
+    {
+      "epoch": 0.37955562879349647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4247713453266342e-05,
+      "loss": 0.8767,
+      "step": 13820
+    },
+    {
+      "epoch": 0.3795830930213397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4246908157019836e-05,
+      "loss": 0.948,
+      "step": 13821
+    },
+    {
+      "epoch": 0.3796105572491829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4246102827170148e-05,
+      "loss": 0.901,
+      "step": 13822
+    },
+    {
+      "epoch": 0.3796380214770262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4245297463723651e-05,
+      "loss": 0.9224,
+      "step": 13823
+    },
+    {
+      "epoch": 0.3796654857048694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4244492066686713e-05,
+      "loss": 0.8845,
+      "step": 13824
+    },
+    {
+      "epoch": 0.37969294993271263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.424368663606571e-05,
+      "loss": 0.942,
+      "step": 13825
+    },
+    {
+      "epoch": 0.37972041416055585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4242881171867019e-05,
+      "loss": 0.9487,
+      "step": 13826
+    },
+    {
+      "epoch": 0.37974787838839913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4242075674097002e-05,
+      "loss": 1.073,
+      "step": 13827
+    },
+    {
+      "epoch": 0.37977534261624235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4241270142762043e-05,
+      "loss": 0.9247,
+      "step": 13828
+    },
+    {
+      "epoch": 0.37980280684408557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4240464577868513e-05,
+      "loss": 0.9942,
+      "step": 13829
+    },
+    {
+      "epoch": 0.3798302710719288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4239658979422781e-05,
+      "loss": 0.9225,
+      "step": 13830
+    },
+    {
+      "epoch": 0.37985773529977207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4238853347431226e-05,
+      "loss": 0.8495,
+      "step": 13831
+    },
+    {
+      "epoch": 0.3798851995276153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4238047681900222e-05,
+      "loss": 1.0511,
+      "step": 13832
+    },
+    {
+      "epoch": 0.3799126637554585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4237241982836143e-05,
+      "loss": 0.9393,
+      "step": 13833
+    },
+    {
+      "epoch": 0.37994012798330173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4236436250245362e-05,
+      "loss": 0.9165,
+      "step": 13834
+    },
+    {
+      "epoch": 0.379967592211145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.423563048413426e-05,
+      "loss": 1.0674,
+      "step": 13835
+    },
+    {
+      "epoch": 0.37999505643898823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4234824684509206e-05,
+      "loss": 0.9942,
+      "step": 13836
+    },
+    {
+      "epoch": 0.38002252066683145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.423401885137658e-05,
+      "loss": 0.8732,
+      "step": 13837
+    },
+    {
+      "epoch": 0.3800499848946747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4233212984742756e-05,
+      "loss": 0.9581,
+      "step": 13838
+    },
+    {
+      "epoch": 0.3800774491225179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4232407084614113e-05,
+      "loss": 1.0465,
+      "step": 13839
+    },
+    {
+      "epoch": 0.3801049133503612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4231601150997023e-05,
+      "loss": 0.9592,
+      "step": 13840
+    },
+    {
+      "epoch": 0.3801323775782044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.423079518389787e-05,
+      "loss": 0.9101,
+      "step": 13841
+    },
+    {
+      "epoch": 0.3801598418060476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4229989183323021e-05,
+      "loss": 0.9841,
+      "step": 13842
+    },
+    {
+      "epoch": 0.38018730603389084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4229183149278862e-05,
+      "loss": 0.9739,
+      "step": 13843
+    },
+    {
+      "epoch": 0.3802147702617341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4228377081771767e-05,
+      "loss": 0.9531,
+      "step": 13844
+    },
+    {
+      "epoch": 0.38024223448957734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4227570980808117e-05,
+      "loss": 0.9174,
+      "step": 13845
+    },
+    {
+      "epoch": 0.38026969871742056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4226764846394285e-05,
+      "loss": 0.9499,
+      "step": 13846
+    },
+    {
+      "epoch": 0.3802971629452638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4225958678536653e-05,
+      "loss": 0.8723,
+      "step": 13847
+    },
+    {
+      "epoch": 0.38032462717310705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4225152477241598e-05,
+      "loss": 0.9058,
+      "step": 13848
+    },
+    {
+      "epoch": 0.3803520914009503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.42243462425155e-05,
+      "loss": 1.0463,
+      "step": 13849
+    },
+    {
+      "epoch": 0.3803795556287935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4223539974364739e-05,
+      "loss": 0.8912,
+      "step": 13850
+    },
+    {
+      "epoch": 0.3804070198566367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4222733672795694e-05,
+      "loss": 0.8998,
+      "step": 13851
+    },
+    {
+      "epoch": 0.38043448408447994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4221927337814745e-05,
+      "loss": 0.8714,
+      "step": 13852
+    },
+    {
+      "epoch": 0.3804619483123232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4221120969428271e-05,
+      "loss": 0.941,
+      "step": 13853
+    },
+    {
+      "epoch": 0.38048941254016644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4220314567642652e-05,
+      "loss": 0.9357,
+      "step": 13854
+    },
+    {
+      "epoch": 0.38051687676800966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4219508132464271e-05,
+      "loss": 0.9782,
+      "step": 13855
+    },
+    {
+      "epoch": 0.3805443409958529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4218701663899504e-05,
+      "loss": 1.0042,
+      "step": 13856
+    },
+    {
+      "epoch": 0.38057180522369616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4217895161954739e-05,
+      "loss": 0.9285,
+      "step": 13857
+    },
+    {
+      "epoch": 0.3805992694515394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.421708862663635e-05,
+      "loss": 1.0099,
+      "step": 13858
+    },
+    {
+      "epoch": 0.3806267336793826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4216282057950724e-05,
+      "loss": 0.9225,
+      "step": 13859
+    },
+    {
+      "epoch": 0.3806541979072258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.421547545590424e-05,
+      "loss": 0.902,
+      "step": 13860
+    },
+    {
+      "epoch": 0.3806816621350691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4214668820503283e-05,
+      "loss": 0.941,
+      "step": 13861
+    },
+    {
+      "epoch": 0.3807091263629123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4213862151754234e-05,
+      "loss": 0.9906,
+      "step": 13862
+    },
+    {
+      "epoch": 0.38073659059075554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4213055449663472e-05,
+      "loss": 0.9956,
+      "step": 13863
+    },
+    {
+      "epoch": 0.38076405481859876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4212248714237389e-05,
+      "loss": 0.801,
+      "step": 13864
+    },
+    {
+      "epoch": 0.380791519046442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4211441945482358e-05,
+      "loss": 0.9485,
+      "step": 13865
+    },
+    {
+      "epoch": 0.38081898327428526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4210635143404767e-05,
+      "loss": 0.995,
+      "step": 13866
+    },
+    {
+      "epoch": 0.3808464475021285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4209828308011003e-05,
+      "loss": 0.9198,
+      "step": 13867
+    },
+    {
+      "epoch": 0.3808739117299717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4209021439307443e-05,
+      "loss": 0.955,
+      "step": 13868
+    },
+    {
+      "epoch": 0.3809013759578149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4208214537300478e-05,
+      "loss": 0.9219,
+      "step": 13869
+    },
+    {
+      "epoch": 0.3809288401856582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4207407601996488e-05,
+      "loss": 1.0207,
+      "step": 13870
+    },
+    {
+      "epoch": 0.3809563044135014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4206600633401858e-05,
+      "loss": 0.8659,
+      "step": 13871
+    },
+    {
+      "epoch": 0.38098376864134464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4205793631522976e-05,
+      "loss": 0.9745,
+      "step": 13872
+    },
+    {
+      "epoch": 0.38101123286918787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4204986596366225e-05,
+      "loss": 0.9385,
+      "step": 13873
+    },
+    {
+      "epoch": 0.38103869709703114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4204179527937994e-05,
+      "loss": 0.8908,
+      "step": 13874
+    },
+    {
+      "epoch": 0.38106616132487436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4203372426244662e-05,
+      "loss": 0.997,
+      "step": 13875
+    },
+    {
+      "epoch": 0.3810936255527176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420256529129262e-05,
+      "loss": 0.9746,
+      "step": 13876
+    },
+    {
+      "epoch": 0.3811210897805608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4201758123088257e-05,
+      "loss": 0.8051,
+      "step": 13877
+    },
+    {
+      "epoch": 0.38114855400840403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420095092163795e-05,
+      "loss": 0.9403,
+      "step": 13878
+    },
+    {
+      "epoch": 0.3811760182362473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4200143686948095e-05,
+      "loss": 0.9442,
+      "step": 13879
+    },
+    {
+      "epoch": 0.3812034824640905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4199336419025076e-05,
+      "loss": 0.9745,
+      "step": 13880
+    },
+    {
+      "epoch": 0.38123094669193375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4198529117875284e-05,
+      "loss": 0.9068,
+      "step": 13881
+    },
+    {
+      "epoch": 0.38125841091977697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.41977217835051e-05,
+      "loss": 0.8964,
+      "step": 13882
+    },
+    {
+      "epoch": 0.38128587514762025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4196914415920912e-05,
+      "loss": 0.9608,
+      "step": 13883
+    },
+    {
+      "epoch": 0.38131333937546347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4196107015129115e-05,
+      "loss": 0.8823,
+      "step": 13884
+    },
+    {
+      "epoch": 0.3813408036033067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4195299581136097e-05,
+      "loss": 0.9602,
+      "step": 13885
+    },
+    {
+      "epoch": 0.3813682678311499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.419449211394824e-05,
+      "loss": 1.0608,
+      "step": 13886
+    },
+    {
+      "epoch": 0.3813957320589932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4193684613571935e-05,
+      "loss": 0.9975,
+      "step": 13887
+    },
+    {
+      "epoch": 0.3814231962868364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4192877080013576e-05,
+      "loss": 1.0059,
+      "step": 13888
+    },
+    {
+      "epoch": 0.38145066051467963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.419206951327955e-05,
+      "loss": 0.8925,
+      "step": 13889
+    },
+    {
+      "epoch": 0.38147812474252285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4191261913376241e-05,
+      "loss": 0.8905,
+      "step": 13890
+    },
+    {
+      "epoch": 0.38150558897036607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4190454280310051e-05,
+      "loss": 0.9073,
+      "step": 13891
+    },
+    {
+      "epoch": 0.38153305319820935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4189646614087358e-05,
+      "loss": 0.786,
+      "step": 13892
+    },
+    {
+      "epoch": 0.38156051742605257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.418883891471456e-05,
+      "loss": 1.0441,
+      "step": 13893
+    },
+    {
+      "epoch": 0.3815879816538958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4188031182198047e-05,
+      "loss": 0.9191,
+      "step": 13894
+    },
+    {
+      "epoch": 0.381615445881739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4187223416544207e-05,
+      "loss": 0.9368,
+      "step": 13895
+    },
+    {
+      "epoch": 0.3816429101095823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4186415617759432e-05,
+      "loss": 0.9623,
+      "step": 13896
+    },
+    {
+      "epoch": 0.3816703743374255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4185607785850122e-05,
+      "loss": 0.8603,
+      "step": 13897
+    },
+    {
+      "epoch": 0.38169783856526873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4184799920822657e-05,
+      "loss": 0.9096,
+      "step": 13898
+    },
+    {
+      "epoch": 0.38172530279311195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.418399202268343e-05,
+      "loss": 0.9867,
+      "step": 13899
+    },
+    {
+      "epoch": 0.38175276702095523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4183184091438844e-05,
+      "loss": 0.9086,
+      "step": 13900
+    },
+    {
+      "epoch": 0.38178023124879845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4182376127095284e-05,
+      "loss": 0.8774,
+      "step": 13901
+    },
+    {
+      "epoch": 0.3818076954766417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4181568129659144e-05,
+      "loss": 0.9176,
+      "step": 13902
+    },
+    {
+      "epoch": 0.3818351597044849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4180760099136814e-05,
+      "loss": 0.9676,
+      "step": 13903
+    },
+    {
+      "epoch": 0.3818626239323281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4179952035534692e-05,
+      "loss": 0.8889,
+      "step": 13904
+    },
+    {
+      "epoch": 0.3818900881601714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.417914393885917e-05,
+      "loss": 0.9927,
+      "step": 13905
+    },
+    {
+      "epoch": 0.3819175523880146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4178335809116646e-05,
+      "loss": 0.9893,
+      "step": 13906
+    },
+    {
+      "epoch": 0.38194501661585784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4177527646313505e-05,
+      "loss": 0.8236,
+      "step": 13907
+    },
+    {
+      "epoch": 0.38197248084370106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4176719450456151e-05,
+      "loss": 0.9191,
+      "step": 13908
+    },
+    {
+      "epoch": 0.38199994507154433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4175911221550975e-05,
+      "loss": 0.8383,
+      "step": 13909
+    },
+    {
+      "epoch": 0.38202740929938755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4175102959604368e-05,
+      "loss": 1.1057,
+      "step": 13910
+    },
+    {
+      "epoch": 0.3820548735272308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4174294664622733e-05,
+      "loss": 0.8749,
+      "step": 13911
+    },
+    {
+      "epoch": 0.382082337755074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.417348633661246e-05,
+      "loss": 0.9722,
+      "step": 13912
+    },
+    {
+      "epoch": 0.3821098019829173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4172677975579944e-05,
+      "loss": 0.9432,
+      "step": 13913
+    },
+    {
+      "epoch": 0.3821372662107605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4171869581531587e-05,
+      "loss": 1.0178,
+      "step": 13914
+    },
+    {
+      "epoch": 0.3821647304386037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4171061154473782e-05,
+      "loss": 0.8738,
+      "step": 13915
+    },
+    {
+      "epoch": 0.38219219466644694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4170252694412922e-05,
+      "loss": 0.934,
+      "step": 13916
+    },
+    {
+      "epoch": 0.38221965889429016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416944420135541e-05,
+      "loss": 1.0438,
+      "step": 13917
+    },
+    {
+      "epoch": 0.38224712312213344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416863567530764e-05,
+      "loss": 0.9482,
+      "step": 13918
+    },
+    {
+      "epoch": 0.38227458734997666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416782711627601e-05,
+      "loss": 0.9037,
+      "step": 13919
+    },
+    {
+      "epoch": 0.3823020515778199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4167018524266919e-05,
+      "loss": 0.9199,
+      "step": 13920
+    },
+    {
+      "epoch": 0.3823295158056631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416620989928676e-05,
+      "loss": 1.0271,
+      "step": 13921
+    },
+    {
+      "epoch": 0.3823569800335064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4165401241341936e-05,
+      "loss": 1.0502,
+      "step": 13922
+    },
+    {
+      "epoch": 0.3823844442613496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4164592550438845e-05,
+      "loss": 0.9498,
+      "step": 13923
+    },
+    {
+      "epoch": 0.3824119084891928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4163783826583881e-05,
+      "loss": 0.8157,
+      "step": 13924
+    },
+    {
+      "epoch": 0.38243937271703604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416297506978345e-05,
+      "loss": 0.8952,
+      "step": 13925
+    },
+    {
+      "epoch": 0.3824668369448793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4162166280043948e-05,
+      "loss": 0.978,
+      "step": 13926
+    },
+    {
+      "epoch": 0.38249430117272254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4161357457371771e-05,
+      "loss": 0.9232,
+      "step": 13927
+    },
+    {
+      "epoch": 0.38252176540056576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4160548601773321e-05,
+      "loss": 0.9834,
+      "step": 13928
+    },
+    {
+      "epoch": 0.382549229628409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4159739713255002e-05,
+      "loss": 0.984,
+      "step": 13929
+    },
+    {
+      "epoch": 0.38257669385625226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4158930791823214e-05,
+      "loss": 0.8873,
+      "step": 13930
+    },
+    {
+      "epoch": 0.3826041580840955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4158121837484346e-05,
+      "loss": 0.9252,
+      "step": 13931
+    },
+    {
+      "epoch": 0.3826316223119387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4157312850244815e-05,
+      "loss": 0.9031,
+      "step": 13932
+    },
+    {
+      "epoch": 0.3826590865397819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4156503830111012e-05,
+      "loss": 0.9141,
+      "step": 13933
+    },
+    {
+      "epoch": 0.38268655076762514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415569477708934e-05,
+      "loss": 0.9761,
+      "step": 13934
+    },
+    {
+      "epoch": 0.3827140149954684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4154885691186201e-05,
+      "loss": 0.9682,
+      "step": 13935
+    },
+    {
+      "epoch": 0.38274147922331164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4154076572407996e-05,
+      "loss": 1.02,
+      "step": 13936
+    },
+    {
+      "epoch": 0.38276894345115486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415326742076113e-05,
+      "loss": 0.888,
+      "step": 13937
+    },
+    {
+      "epoch": 0.3827964076789981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4152458236252003e-05,
+      "loss": 0.8723,
+      "step": 13938
+    },
+    {
+      "epoch": 0.38282387190684136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4151649018887018e-05,
+      "loss": 0.8826,
+      "step": 13939
+    },
+    {
+      "epoch": 0.3828513361346846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4150839768672576e-05,
+      "loss": 0.9527,
+      "step": 13940
+    },
+    {
+      "epoch": 0.3828788003625278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4150030485615084e-05,
+      "loss": 0.9013,
+      "step": 13941
+    },
+    {
+      "epoch": 0.382906264590371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4149221169720941e-05,
+      "loss": 0.9401,
+      "step": 13942
+    },
+    {
+      "epoch": 0.3829337288182143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4148411820996555e-05,
+      "loss": 0.855,
+      "step": 13943
+    },
+    {
+      "epoch": 0.3829611930460575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4147602439448327e-05,
+      "loss": 0.9016,
+      "step": 13944
+    },
+    {
+      "epoch": 0.38298865727390075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414679302508266e-05,
+      "loss": 0.8845,
+      "step": 13945
+    },
+    {
+      "epoch": 0.38301612150174397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4145983577905962e-05,
+      "loss": 0.9993,
+      "step": 13946
+    },
+    {
+      "epoch": 0.3830435857295872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4145174097924637e-05,
+      "loss": 0.9317,
+      "step": 13947
+    },
+    {
+      "epoch": 0.38307104995743047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4144364585145088e-05,
+      "loss": 0.8882,
+      "step": 13948
+    },
+    {
+      "epoch": 0.3830985141852737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4143555039573721e-05,
+      "loss": 0.9663,
+      "step": 13949
+    },
+    {
+      "epoch": 0.3831259784131169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4142745461216942e-05,
+      "loss": 0.9811,
+      "step": 13950
+    },
+    {
+      "epoch": 0.38315344264096013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4141935850081155e-05,
+      "loss": 0.9279,
+      "step": 13951
+    },
+    {
+      "epoch": 0.3831809068688034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4141126206172766e-05,
+      "loss": 0.9693,
+      "step": 13952
+    },
+    {
+      "epoch": 0.3832083710966466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4140316529498186e-05,
+      "loss": 0.9194,
+      "step": 13953
+    },
+    {
+      "epoch": 0.38323583532448985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4139506820063815e-05,
+      "loss": 1.0642,
+      "step": 13954
+    },
+    {
+      "epoch": 0.38326329955233307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4138697077876063e-05,
+      "loss": 0.9754,
+      "step": 13955
+    },
+    {
+      "epoch": 0.38329076378017635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4137887302941337e-05,
+      "loss": 0.9637,
+      "step": 13956
+    },
+    {
+      "epoch": 0.38331822800801957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4137077495266045e-05,
+      "loss": 0.8972,
+      "step": 13957
+    },
+    {
+      "epoch": 0.3833456922358628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.413626765485659e-05,
+      "loss": 0.8549,
+      "step": 13958
+    },
+    {
+      "epoch": 0.383373156463706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4135457781719386e-05,
+      "loss": 0.9442,
+      "step": 13959
+    },
+    {
+      "epoch": 0.38340062069154923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4134647875860836e-05,
+      "loss": 0.9543,
+      "step": 13960
+    },
+    {
+      "epoch": 0.3834280849193925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4133837937287353e-05,
+      "loss": 0.9751,
+      "step": 13961
+    },
+    {
+      "epoch": 0.38345554914723573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4133027966005344e-05,
+      "loss": 0.9492,
+      "step": 13962
+    },
+    {
+      "epoch": 0.38348301337507895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4132217962021214e-05,
+      "loss": 1.0018,
+      "step": 13963
+    },
+    {
+      "epoch": 0.3835104776029222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4131407925341374e-05,
+      "loss": 1.0012,
+      "step": 13964
+    },
+    {
+      "epoch": 0.38353794183076545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4130597855972235e-05,
+      "loss": 0.8956,
+      "step": 13965
+    },
+    {
+      "epoch": 0.38356540605860867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4129787753920205e-05,
+      "loss": 0.843,
+      "step": 13966
+    },
+    {
+      "epoch": 0.3835928702864519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4128977619191698e-05,
+      "loss": 0.9778,
+      "step": 13967
+    },
+    {
+      "epoch": 0.3836203345142951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4128167451793116e-05,
+      "loss": 0.9211,
+      "step": 13968
+    },
+    {
+      "epoch": 0.3836477987421384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4127357251730876e-05,
+      "loss": 0.9672,
+      "step": 13969
+    },
+    {
+      "epoch": 0.3836752629699816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412654701901139e-05,
+      "loss": 0.9449,
+      "step": 13970
+    },
+    {
+      "epoch": 0.38370272719782483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4125736753641059e-05,
+      "loss": 0.9355,
+      "step": 13971
+    },
+    {
+      "epoch": 0.38373019142566805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4124926455626304e-05,
+      "loss": 0.8538,
+      "step": 13972
+    },
+    {
+      "epoch": 0.3837576556535113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4124116124973533e-05,
+      "loss": 0.9429,
+      "step": 13973
+    },
+    {
+      "epoch": 0.38378511988135455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412330576168916e-05,
+      "loss": 0.9206,
+      "step": 13974
+    },
+    {
+      "epoch": 0.3838125841091978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412249536577959e-05,
+      "loss": 0.9819,
+      "step": 13975
+    },
+    {
+      "epoch": 0.383840048337041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4121684937251243e-05,
+      "loss": 0.8614,
+      "step": 13976
+    },
+    {
+      "epoch": 0.3838675125648842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412087447611053e-05,
+      "loss": 1.0022,
+      "step": 13977
+    },
+    {
+      "epoch": 0.3838949767927275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4120063982363856e-05,
+      "loss": 1.0404,
+      "step": 13978
+    },
+    {
+      "epoch": 0.3839224410205707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4119253456017643e-05,
+      "loss": 1.0292,
+      "step": 13979
+    },
+    {
+      "epoch": 0.38394990524841394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4118442897078302e-05,
+      "loss": 0.9077,
+      "step": 13980
+    },
+    {
+      "epoch": 0.38397736947625716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4117632305552244e-05,
+      "loss": 0.9985,
+      "step": 13981
+    },
+    {
+      "epoch": 0.38400483370410043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4116821681445884e-05,
+      "loss": 0.9929,
+      "step": 13982
+    },
+    {
+      "epoch": 0.38403229793194366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4116011024765638e-05,
+      "loss": 1.038,
+      "step": 13983
+    },
+    {
+      "epoch": 0.3840597621597869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4115200335517916e-05,
+      "loss": 0.9286,
+      "step": 13984
+    },
+    {
+      "epoch": 0.3840872263876301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4114389613709136e-05,
+      "loss": 0.9152,
+      "step": 13985
+    },
+    {
+      "epoch": 0.3841146906154733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4113578859345711e-05,
+      "loss": 0.9431,
+      "step": 13986
+    },
+    {
+      "epoch": 0.3841421548433166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4112768072434057e-05,
+      "loss": 0.9866,
+      "step": 13987
+    },
+    {
+      "epoch": 0.3841696190711598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.411195725298059e-05,
+      "loss": 0.9651,
+      "step": 13988
+    },
+    {
+      "epoch": 0.38419708329900304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4111146400991725e-05,
+      "loss": 0.9738,
+      "step": 13989
+    },
+    {
+      "epoch": 0.38422454752684626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4110335516473874e-05,
+      "loss": 0.8731,
+      "step": 13990
+    },
+    {
+      "epoch": 0.38425201175468954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4109524599433457e-05,
+      "loss": 0.9794,
+      "step": 13991
+    },
+    {
+      "epoch": 0.38427947598253276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.410871364987689e-05,
+      "loss": 0.8629,
+      "step": 13992
+    },
+    {
+      "epoch": 0.384306940210376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.410790266781059e-05,
+      "loss": 0.962,
+      "step": 13993
+    },
+    {
+      "epoch": 0.3843344044382192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4107091653240973e-05,
+      "loss": 1.0034,
+      "step": 13994
+    },
+    {
+      "epoch": 0.3843618686660625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4106280606174454e-05,
+      "loss": 0.9327,
+      "step": 13995
+    },
+    {
+      "epoch": 0.3843893328939057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4105469526617452e-05,
+      "loss": 0.9601,
+      "step": 13996
+    },
+    {
+      "epoch": 0.3844167971217489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4104658414576389e-05,
+      "loss": 0.8935,
+      "step": 13997
+    },
+    {
+      "epoch": 0.38444426134959214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4103847270057675e-05,
+      "loss": 0.9166,
+      "step": 13998
+    },
+    {
+      "epoch": 0.38447172557743536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4103036093067732e-05,
+      "loss": 0.9027,
+      "step": 13999
+    },
+    {
+      "epoch": 0.38449918980527864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.410222488361298e-05,
+      "loss": 0.8943,
+      "step": 14000
+    },
+    {
+      "epoch": 0.38452665403312186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4101413641699836e-05,
+      "loss": 0.909,
+      "step": 14001
+    },
+    {
+      "epoch": 0.3845541182609651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4100602367334715e-05,
+      "loss": 0.8781,
+      "step": 14002
+    },
+    {
+      "epoch": 0.3845815824888083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4099791060524042e-05,
+      "loss": 0.8573,
+      "step": 14003
+    },
+    {
+      "epoch": 0.3846090467166516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4098979721274232e-05,
+      "loss": 0.8886,
+      "step": 14004
+    },
+    {
+      "epoch": 0.3846365109444948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.409816834959171e-05,
+      "loss": 0.932,
+      "step": 14005
+    },
+    {
+      "epoch": 0.384663975172338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4097356945482892e-05,
+      "loss": 0.8866,
+      "step": 14006
+    },
+    {
+      "epoch": 0.38469143940018125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4096545508954194e-05,
+      "loss": 0.9621,
+      "step": 14007
+    },
+    {
+      "epoch": 0.3847189036280245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4095734040012043e-05,
+      "loss": 0.9367,
+      "step": 14008
+    },
+    {
+      "epoch": 0.38474636785586774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.409492253866286e-05,
+      "loss": 0.9254,
+      "step": 14009
+    },
+    {
+      "epoch": 0.38477383208371096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4094111004913065e-05,
+      "loss": 0.8817,
+      "step": 14010
+    },
+    {
+      "epoch": 0.3848012963115542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4093299438769076e-05,
+      "loss": 0.8901,
+      "step": 14011
+    },
+    {
+      "epoch": 0.3848287605393974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4092487840237314e-05,
+      "loss": 1.0035,
+      "step": 14012
+    },
+    {
+      "epoch": 0.3848562247672407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4091676209324208e-05,
+      "loss": 0.8691,
+      "step": 14013
+    },
+    {
+      "epoch": 0.3848836889950839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.409086454603617e-05,
+      "loss": 1.079,
+      "step": 14014
+    },
+    {
+      "epoch": 0.3849111532229271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4090052850379628e-05,
+      "loss": 0.9423,
+      "step": 14015
+    },
+    {
+      "epoch": 0.38493861745077035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4089241122361003e-05,
+      "loss": 0.884,
+      "step": 14016
+    },
+    {
+      "epoch": 0.3849660816786136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4088429361986721e-05,
+      "loss": 0.8647,
+      "step": 14017
+    },
+    {
+      "epoch": 0.38499354590645685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40876175692632e-05,
+      "loss": 0.9361,
+      "step": 14018
+    },
+    {
+      "epoch": 0.38502101013430007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4086805744196865e-05,
+      "loss": 0.9311,
+      "step": 14019
+    },
+    {
+      "epoch": 0.3850484743621433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4085993886794142e-05,
+      "loss": 0.9242,
+      "step": 14020
+    },
+    {
+      "epoch": 0.38507593858998657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4085181997061452e-05,
+      "loss": 1.006,
+      "step": 14021
+    },
+    {
+      "epoch": 0.3851034028178298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.408437007500522e-05,
+      "loss": 0.9555,
+      "step": 14022
+    },
+    {
+      "epoch": 0.385130867045673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4083558120631868e-05,
+      "loss": 0.9306,
+      "step": 14023
+    },
+    {
+      "epoch": 0.38515833127351623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4082746133947824e-05,
+      "loss": 0.9188,
+      "step": 14024
+    },
+    {
+      "epoch": 0.3851857955013595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4081934114959511e-05,
+      "loss": 0.9457,
+      "step": 14025
+    },
+    {
+      "epoch": 0.38521325972920273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4081122063673354e-05,
+      "loss": 0.8671,
+      "step": 14026
+    },
+    {
+      "epoch": 0.38524072395704595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4080309980095776e-05,
+      "loss": 0.9747,
+      "step": 14027
+    },
+    {
+      "epoch": 0.38526818818488917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4079497864233206e-05,
+      "loss": 1.0161,
+      "step": 14028
+    },
+    {
+      "epoch": 0.3852956524127324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.407868571609207e-05,
+      "loss": 0.8425,
+      "step": 14029
+    },
+    {
+      "epoch": 0.38532311664057567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4077873535678795e-05,
+      "loss": 0.9379,
+      "step": 14030
+    },
+    {
+      "epoch": 0.3853505808684189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4077061322999798e-05,
+      "loss": 0.9543,
+      "step": 14031
+    },
+    {
+      "epoch": 0.3853780450962621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.407624907806152e-05,
+      "loss": 0.9658,
+      "step": 14032
+    },
+    {
+      "epoch": 0.38540550932410533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4075436800870374e-05,
+      "loss": 0.9866,
+      "step": 14033
+    },
+    {
+      "epoch": 0.3854329735519486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4074624491432798e-05,
+      "loss": 0.9625,
+      "step": 14034
+    },
+    {
+      "epoch": 0.38546043777979183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4073812149755213e-05,
+      "loss": 0.9547,
+      "step": 14035
+    },
+    {
+      "epoch": 0.38548790200763505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4072999775844047e-05,
+      "loss": 0.9887,
+      "step": 14036
+    },
+    {
+      "epoch": 0.3855153662354783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.407218736970573e-05,
+      "loss": 0.8937,
+      "step": 14037
+    },
+    {
+      "epoch": 0.38554283046332155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4071374931346691e-05,
+      "loss": 0.9564,
+      "step": 14038
+    },
+    {
+      "epoch": 0.38557029469116477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4070562460773353e-05,
+      "loss": 0.9788,
+      "step": 14039
+    },
+    {
+      "epoch": 0.385597758919008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.406974995799215e-05,
+      "loss": 0.8959,
+      "step": 14040
+    },
+    {
+      "epoch": 0.3856252231468512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4068937423009507e-05,
+      "loss": 0.8828,
+      "step": 14041
+    },
+    {
+      "epoch": 0.38565268737469444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.406812485583186e-05,
+      "loss": 0.8846,
+      "step": 14042
+    },
+    {
+      "epoch": 0.3856801516025377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4067312256465626e-05,
+      "loss": 0.8518,
+      "step": 14043
+    },
+    {
+      "epoch": 0.38570761583038093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.406649962491725e-05,
+      "loss": 0.9816,
+      "step": 14044
+    },
+    {
+      "epoch": 0.38573508005822416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4065686961193148e-05,
+      "loss": 0.8251,
+      "step": 14045
+    },
+    {
+      "epoch": 0.3857625442860674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4064874265299757e-05,
+      "loss": 1.0096,
+      "step": 14046
+    },
+    {
+      "epoch": 0.38579000851391065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4064061537243508e-05,
+      "loss": 0.9754,
+      "step": 14047
+    },
+    {
+      "epoch": 0.3858174727417539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4063248777030827e-05,
+      "loss": 0.9666,
+      "step": 14048
+    },
+    {
+      "epoch": 0.3858449369695971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4062435984668148e-05,
+      "loss": 0.9419,
+      "step": 14049
+    },
+    {
+      "epoch": 0.3858724011974403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4061623160161906e-05,
+      "loss": 1.0092,
+      "step": 14050
+    },
+    {
+      "epoch": 0.3858998654252836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4060810303518522e-05,
+      "loss": 0.8815,
+      "step": 14051
+    },
+    {
+      "epoch": 0.3859273296531268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4059997414744437e-05,
+      "loss": 0.9349,
+      "step": 14052
+    },
+    {
+      "epoch": 0.38595479388097004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.405918449384608e-05,
+      "loss": 1.0408,
+      "step": 14053
+    },
+    {
+      "epoch": 0.38598225810881326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4058371540829884e-05,
+      "loss": 0.9406,
+      "step": 14054
+    },
+    {
+      "epoch": 0.3860097223366565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4057558555702278e-05,
+      "loss": 0.9666,
+      "step": 14055
+    },
+    {
+      "epoch": 0.38603718656449976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40567455384697e-05,
+      "loss": 1.0057,
+      "step": 14056
+    },
+    {
+      "epoch": 0.386064650792343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4055932489138577e-05,
+      "loss": 0.9346,
+      "step": 14057
+    },
+    {
+      "epoch": 0.3860921150201862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4055119407715346e-05,
+      "loss": 0.8399,
+      "step": 14058
+    },
+    {
+      "epoch": 0.3861195792480294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4054306294206442e-05,
+      "loss": 0.9644,
+      "step": 14059
+    },
+    {
+      "epoch": 0.3861470434758727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.405349314861829e-05,
+      "loss": 0.9067,
+      "step": 14060
+    },
+    {
+      "epoch": 0.3861745077037159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4052679970957335e-05,
+      "loss": 1.0042,
+      "step": 14061
+    },
+    {
+      "epoch": 0.38620197193155914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4051866761230004e-05,
+      "loss": 0.939,
+      "step": 14062
+    },
+    {
+      "epoch": 0.38622943615940236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4051053519442734e-05,
+      "loss": 0.8399,
+      "step": 14063
+    },
+    {
+      "epoch": 0.38625690038724564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4050240245601957e-05,
+      "loss": 0.9235,
+      "step": 14064
+    },
+    {
+      "epoch": 0.38628436461508886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4049426939714117e-05,
+      "loss": 1.0975,
+      "step": 14065
+    },
+    {
+      "epoch": 0.3863118288429321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4048613601785638e-05,
+      "loss": 0.8965,
+      "step": 14066
+    },
+    {
+      "epoch": 0.3863392930707753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.404780023182296e-05,
+      "loss": 0.9797,
+      "step": 14067
+    },
+    {
+      "epoch": 0.3863667572986185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4046986829832518e-05,
+      "loss": 0.9332,
+      "step": 14068
+    },
+    {
+      "epoch": 0.3863942215264618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.404617339582075e-05,
+      "loss": 0.9619,
+      "step": 14069
+    },
+    {
+      "epoch": 0.386421685754305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4045359929794091e-05,
+      "loss": 0.9816,
+      "step": 14070
+    },
+    {
+      "epoch": 0.38644914998214824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4044546431758977e-05,
+      "loss": 1.0289,
+      "step": 14071
+    },
+    {
+      "epoch": 0.38647661420999146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4043732901721843e-05,
+      "loss": 0.9519,
+      "step": 14072
+    },
+    {
+      "epoch": 0.38650407843783474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4042919339689129e-05,
+      "loss": 0.9195,
+      "step": 14073
+    },
+    {
+      "epoch": 0.38653154266567796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4042105745667274e-05,
+      "loss": 0.9927,
+      "step": 14074
+    },
+    {
+      "epoch": 0.3865590068935212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4041292119662708e-05,
+      "loss": 0.9641,
+      "step": 14075
+    },
+    {
+      "epoch": 0.3865864711213644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4040478461681874e-05,
+      "loss": 1.0503,
+      "step": 14076
+    },
+    {
+      "epoch": 0.3866139353492077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4039664771731212e-05,
+      "loss": 0.933,
+      "step": 14077
+    },
+    {
+      "epoch": 0.3866413995770509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4038851049817157e-05,
+      "loss": 1.0475,
+      "step": 14078
+    },
+    {
+      "epoch": 0.3866688638048941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4038037295946146e-05,
+      "loss": 0.9279,
+      "step": 14079
+    },
+    {
+      "epoch": 0.38669632803273735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.403722351012462e-05,
+      "loss": 0.902,
+      "step": 14080
+    },
+    {
+      "epoch": 0.38672379226058057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4036409692359022e-05,
+      "loss": 0.9487,
+      "step": 14081
+    },
+    {
+      "epoch": 0.38675125648842384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4035595842655782e-05,
+      "loss": 0.9516,
+      "step": 14082
+    },
+    {
+      "epoch": 0.38677872071626707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4034781961021344e-05,
+      "loss": 0.894,
+      "step": 14083
+    },
+    {
+      "epoch": 0.3868061849441103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.403396804746215e-05,
+      "loss": 0.9342,
+      "step": 14084
+    },
+    {
+      "epoch": 0.3868336491719535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4033154101984638e-05,
+      "loss": 0.9729,
+      "step": 14085
+    },
+    {
+      "epoch": 0.3868611133997968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4032340124595249e-05,
+      "loss": 0.8642,
+      "step": 14086
+    },
+    {
+      "epoch": 0.38688857762764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4031526115300422e-05,
+      "loss": 0.9973,
+      "step": 14087
+    },
+    {
+      "epoch": 0.38691604185548323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40307120741066e-05,
+      "loss": 0.9099,
+      "step": 14088
+    },
+    {
+      "epoch": 0.38694350608332645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4029898001020223e-05,
+      "loss": 0.9738,
+      "step": 14089
+    },
+    {
+      "epoch": 0.3869709703111697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4029083896047731e-05,
+      "loss": 0.8722,
+      "step": 14090
+    },
+    {
+      "epoch": 0.38699843453901295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4028269759195567e-05,
+      "loss": 0.995,
+      "step": 14091
+    },
+    {
+      "epoch": 0.38702589876685617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4027455590470172e-05,
+      "loss": 0.9075,
+      "step": 14092
+    },
+    {
+      "epoch": 0.3870533629946994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402664138987799e-05,
+      "loss": 0.9221,
+      "step": 14093
+    },
+    {
+      "epoch": 0.3870808272225426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402582715742546e-05,
+      "loss": 0.9808,
+      "step": 14094
+    },
+    {
+      "epoch": 0.3871082914503859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4025012893119024e-05,
+      "loss": 0.8576,
+      "step": 14095
+    },
+    {
+      "epoch": 0.3871357556782291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4024198596965126e-05,
+      "loss": 0.9112,
+      "step": 14096
+    },
+    {
+      "epoch": 0.38716321990607233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4023384268970215e-05,
+      "loss": 0.9712,
+      "step": 14097
+    },
+    {
+      "epoch": 0.38719068413391555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4022569909140727e-05,
+      "loss": 0.9273,
+      "step": 14098
+    },
+    {
+      "epoch": 0.38721814836175883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4021755517483107e-05,
+      "loss": 0.9924,
+      "step": 14099
+    },
+    {
+      "epoch": 0.38724561258960205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40209410940038e-05,
+      "loss": 0.8904,
+      "step": 14100
+    },
+    {
+      "epoch": 0.38727307681744527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4020126638709249e-05,
+      "loss": 0.9439,
+      "step": 14101
+    },
+    {
+      "epoch": 0.3873005410452885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4019312151605897e-05,
+      "loss": 0.9528,
+      "step": 14102
+    },
+    {
+      "epoch": 0.38732800527313177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4018497632700194e-05,
+      "loss": 0.9529,
+      "step": 14103
+    },
+    {
+      "epoch": 0.387355469500975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.401768308199858e-05,
+      "loss": 0.9399,
+      "step": 14104
+    },
+    {
+      "epoch": 0.3873829337288182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4016868499507501e-05,
+      "loss": 0.8958,
+      "step": 14105
+    },
+    {
+      "epoch": 0.38741039795666143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4016053885233402e-05,
+      "loss": 0.919,
+      "step": 14106
+    },
+    {
+      "epoch": 0.38743786218450466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4015239239182727e-05,
+      "loss": 0.9137,
+      "step": 14107
+    },
+    {
+      "epoch": 0.38746532641234793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4014424561361926e-05,
+      "loss": 0.9785,
+      "step": 14108
+    },
+    {
+      "epoch": 0.38749279064019115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4013609851777443e-05,
+      "loss": 1.0148,
+      "step": 14109
+    },
+    {
+      "epoch": 0.3875202548680344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4012795110435726e-05,
+      "loss": 0.8795,
+      "step": 14110
+    },
+    {
+      "epoch": 0.3875477190958776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4011980337343217e-05,
+      "loss": 0.9963,
+      "step": 14111
+    },
+    {
+      "epoch": 0.3875751833237209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4011165532506366e-05,
+      "loss": 0.919,
+      "step": 14112
+    },
+    {
+      "epoch": 0.3876026475515641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4010350695931618e-05,
+      "loss": 0.9614,
+      "step": 14113
+    },
+    {
+      "epoch": 0.3876301117794073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4009535827625424e-05,
+      "loss": 1.043,
+      "step": 14114
+    },
+    {
+      "epoch": 0.38765757600725054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400872092759423e-05,
+      "loss": 0.8881,
+      "step": 14115
+    },
+    {
+      "epoch": 0.3876850402350938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4007905995844482e-05,
+      "loss": 0.9999,
+      "step": 14116
+    },
+    {
+      "epoch": 0.38771250446293704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4007091032382629e-05,
+      "loss": 0.8707,
+      "step": 14117
+    },
+    {
+      "epoch": 0.38773996869078026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400627603721512e-05,
+      "loss": 0.8285,
+      "step": 14118
+    },
+    {
+      "epoch": 0.3877674329186235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4005461010348403e-05,
+      "loss": 1.006,
+      "step": 14119
+    },
+    {
+      "epoch": 0.3877948971464667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4004645951788925e-05,
+      "loss": 0.9082,
+      "step": 14120
+    },
+    {
+      "epoch": 0.38782236137431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400383086154314e-05,
+      "loss": 0.7762,
+      "step": 14121
+    },
+    {
+      "epoch": 0.3878498256021532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4003015739617493e-05,
+      "loss": 0.9436,
+      "step": 14122
+    },
+    {
+      "epoch": 0.3878772898299964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4002200586018435e-05,
+      "loss": 1.0111,
+      "step": 14123
+    },
+    {
+      "epoch": 0.38790475405783964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4001385400752416e-05,
+      "loss": 0.8556,
+      "step": 14124
+    },
+    {
+      "epoch": 0.3879322182856829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4000570183825886e-05,
+      "loss": 0.9448,
+      "step": 14125
+    },
+    {
+      "epoch": 0.38795968251352614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3999754935245293e-05,
+      "loss": 0.8557,
+      "step": 14126
+    },
+    {
+      "epoch": 0.38798714674136936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3998939655017093e-05,
+      "loss": 0.9548,
+      "step": 14127
+    },
+    {
+      "epoch": 0.3880146109692126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3998124343147727e-05,
+      "loss": 0.9222,
+      "step": 14128
+    },
+    {
+      "epoch": 0.38804207519705586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.399730899964366e-05,
+      "loss": 0.9702,
+      "step": 14129
+    },
+    {
+      "epoch": 0.3880695394248991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3996493624511334e-05,
+      "loss": 0.865,
+      "step": 14130
+    },
+    {
+      "epoch": 0.3880970036527423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3995678217757199e-05,
+      "loss": 0.9243,
+      "step": 14131
+    },
+    {
+      "epoch": 0.3881244678805855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.399486277938771e-05,
+      "loss": 0.9646,
+      "step": 14132
+    },
+    {
+      "epoch": 0.3881519321084288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3994047309409322e-05,
+      "loss": 1.0714,
+      "step": 14133
+    },
+    {
+      "epoch": 0.388179396336272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3993231807828485e-05,
+      "loss": 0.9776,
+      "step": 14134
+    },
+    {
+      "epoch": 0.38820686056411524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3992416274651648e-05,
+      "loss": 0.9386,
+      "step": 14135
+    },
+    {
+      "epoch": 0.38823432479195846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3991600709885269e-05,
+      "loss": 0.9651,
+      "step": 14136
+    },
+    {
+      "epoch": 0.3882617890198017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.39907851135358e-05,
+      "loss": 1.0338,
+      "step": 14137
+    },
+    {
+      "epoch": 0.38828925324764496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3989969485609691e-05,
+      "loss": 0.9347,
+      "step": 14138
+    },
+    {
+      "epoch": 0.3883167174754882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3989153826113401e-05,
+      "loss": 0.9741,
+      "step": 14139
+    },
+    {
+      "epoch": 0.3883441817033314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3988338135053376e-05,
+      "loss": 0.8979,
+      "step": 14140
+    },
+    {
+      "epoch": 0.3883716459311746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3987522412436078e-05,
+      "loss": 0.9952,
+      "step": 14141
+    },
+    {
+      "epoch": 0.3883991101590179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3986706658267961e-05,
+      "loss": 0.9122,
+      "step": 14142
+    },
+    {
+      "epoch": 0.3884265743868611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3985890872555472e-05,
+      "loss": 0.8898,
+      "step": 14143
+    },
+    {
+      "epoch": 0.38845403861470434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.398507505530507e-05,
+      "loss": 1.0193,
+      "step": 14144
+    },
+    {
+      "epoch": 0.38848150284254757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3984259206523211e-05,
+      "loss": 0.8979,
+      "step": 14145
+    },
+    {
+      "epoch": 0.38850896707039084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3983443326216352e-05,
+      "loss": 0.9181,
+      "step": 14146
+    },
+    {
+      "epoch": 0.38853643129823406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3982627414390945e-05,
+      "loss": 0.9151,
+      "step": 14147
+    },
+    {
+      "epoch": 0.3885638955260773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3981811471053447e-05,
+      "loss": 1.0038,
+      "step": 14148
+    },
+    {
+      "epoch": 0.3885913597539205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3980995496210317e-05,
+      "loss": 0.9796,
+      "step": 14149
+    },
+    {
+      "epoch": 0.38861882398176373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3980179489868006e-05,
+      "loss": 0.8884,
+      "step": 14150
+    },
+    {
+      "epoch": 0.388646288209607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3979363452032972e-05,
+      "loss": 0.9146,
+      "step": 14151
+    },
+    {
+      "epoch": 0.3886737524374502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3978547382711673e-05,
+      "loss": 0.8918,
+      "step": 14152
+    },
+    {
+      "epoch": 0.38870121666529345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.397773128191057e-05,
+      "loss": 0.9928,
+      "step": 14153
+    },
+    {
+      "epoch": 0.38872868089313667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3976915149636115e-05,
+      "loss": 0.9401,
+      "step": 14154
+    },
+    {
+      "epoch": 0.38875614512097995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3976098985894762e-05,
+      "loss": 0.9551,
+      "step": 14155
+    },
+    {
+      "epoch": 0.38878360934882317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3975282790692978e-05,
+      "loss": 0.9559,
+      "step": 14156
+    },
+    {
+      "epoch": 0.3888110735766664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3974466564037215e-05,
+      "loss": 1.0079,
+      "step": 14157
+    },
+    {
+      "epoch": 0.3888385378045096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3973650305933934e-05,
+      "loss": 0.9382,
+      "step": 14158
+    },
+    {
+      "epoch": 0.3888660020323529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3972834016389594e-05,
+      "loss": 1.0042,
+      "step": 14159
+    },
+    {
+      "epoch": 0.3888934662601961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3972017695410647e-05,
+      "loss": 1.0201,
+      "step": 14160
+    },
+    {
+      "epoch": 0.38892093048803933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3971201343003561e-05,
+      "loss": 0.9848,
+      "step": 14161
+    },
+    {
+      "epoch": 0.38894839471588255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3970384959174794e-05,
+      "loss": 0.9353,
+      "step": 14162
+    },
+    {
+      "epoch": 0.38897585894372577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3969568543930798e-05,
+      "loss": 0.879,
+      "step": 14163
+    },
+    {
+      "epoch": 0.38900332317156905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.396875209727804e-05,
+      "loss": 0.9579,
+      "step": 14164
+    },
+    {
+      "epoch": 0.38903078739941227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3967935619222982e-05,
+      "loss": 0.8967,
+      "step": 14165
+    },
+    {
+      "epoch": 0.3890582516272555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3967119109772076e-05,
+      "loss": 0.8802,
+      "step": 14166
+    },
+    {
+      "epoch": 0.3890857158550987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3966302568931786e-05,
+      "loss": 1.0112,
+      "step": 14167
+    },
+    {
+      "epoch": 0.389113180082942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3965485996708576e-05,
+      "loss": 0.937,
+      "step": 14168
+    },
+    {
+      "epoch": 0.3891406443107852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3964669393108904e-05,
+      "loss": 0.9794,
+      "step": 14169
+    },
+    {
+      "epoch": 0.38916810853862843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3963852758139234e-05,
+      "loss": 0.8973,
+      "step": 14170
+    },
+    {
+      "epoch": 0.38919557276647165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3963036091806025e-05,
+      "loss": 0.9798,
+      "step": 14171
+    },
+    {
+      "epoch": 0.38922303699431493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3962219394115737e-05,
+      "loss": 0.9537,
+      "step": 14172
+    },
+    {
+      "epoch": 0.38925050122215815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3961402665074838e-05,
+      "loss": 0.913,
+      "step": 14173
+    },
+    {
+      "epoch": 0.3892779654500014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3960585904689782e-05,
+      "loss": 0.8946,
+      "step": 14174
+    },
+    {
+      "epoch": 0.3893054296778446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.395976911296704e-05,
+      "loss": 0.9432,
+      "step": 14175
+    },
+    {
+      "epoch": 0.3893328939056878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3958952289913066e-05,
+      "loss": 0.9715,
+      "step": 14176
+    },
+    {
+      "epoch": 0.3893603581335311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3958135435534333e-05,
+      "loss": 0.9236,
+      "step": 14177
+    },
+    {
+      "epoch": 0.3893878223613743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3957318549837298e-05,
+      "loss": 0.8611,
+      "step": 14178
+    },
+    {
+      "epoch": 0.38941528658921754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3956501632828424e-05,
+      "loss": 0.8589,
+      "step": 14179
+    },
+    {
+      "epoch": 0.38944275081706076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3955684684514176e-05,
+      "loss": 0.9307,
+      "step": 14180
+    },
+    {
+      "epoch": 0.38947021504490403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3954867704901021e-05,
+      "loss": 0.9351,
+      "step": 14181
+    },
+    {
+      "epoch": 0.38949767927274725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3954050693995421e-05,
+      "loss": 0.9608,
+      "step": 14182
+    },
+    {
+      "epoch": 0.3895251435005905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.395323365180384e-05,
+      "loss": 0.9324,
+      "step": 14183
+    },
+    {
+      "epoch": 0.3895526077284337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3952416578332738e-05,
+      "loss": 0.8518,
+      "step": 14184
+    },
+    {
+      "epoch": 0.389580071956277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.395159947358859e-05,
+      "loss": 0.9437,
+      "step": 14185
+    },
+    {
+      "epoch": 0.3896075361841202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3950782337577857e-05,
+      "loss": 0.9156,
+      "step": 14186
+    },
+    {
+      "epoch": 0.3896350004119634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3949965170307002e-05,
+      "loss": 0.9262,
+      "step": 14187
+    },
+    {
+      "epoch": 0.38966246463980664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3949147971782491e-05,
+      "loss": 0.9363,
+      "step": 14188
+    },
+    {
+      "epoch": 0.38968992886764986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3948330742010794e-05,
+      "loss": 0.9983,
+      "step": 14189
+    },
+    {
+      "epoch": 0.38971739309549314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3947513480998373e-05,
+      "loss": 0.9526,
+      "step": 14190
+    },
+    {
+      "epoch": 0.38974485732333636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3946696188751698e-05,
+      "loss": 0.8541,
+      "step": 14191
+    },
+    {
+      "epoch": 0.3897723215511796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3945878865277234e-05,
+      "loss": 0.9515,
+      "step": 14192
+    },
+    {
+      "epoch": 0.3897997857790228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3945061510581448e-05,
+      "loss": 0.9055,
+      "step": 14193
+    },
+    {
+      "epoch": 0.3898272500068661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3944244124670805e-05,
+      "loss": 0.9635,
+      "step": 14194
+    },
+    {
+      "epoch": 0.3898547142347093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3943426707551777e-05,
+      "loss": 0.9562,
+      "step": 14195
+    },
+    {
+      "epoch": 0.3898821784625525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3942609259230826e-05,
+      "loss": 0.939,
+      "step": 14196
+    },
+    {
+      "epoch": 0.38990964269039574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3941791779714426e-05,
+      "loss": 0.8706,
+      "step": 14197
+    },
+    {
+      "epoch": 0.389937106918239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3940974269009043e-05,
+      "loss": 0.9758,
+      "step": 14198
+    },
+    {
+      "epoch": 0.38996457114608224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3940156727121141e-05,
+      "loss": 0.8829,
+      "step": 14199
+    },
+    {
+      "epoch": 0.38999203537392546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3939339154057194e-05,
+      "loss": 0.9904,
+      "step": 14200
+    },
+    {
+      "epoch": 0.3900194996017687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3938521549823673e-05,
+      "loss": 0.9213,
+      "step": 14201
+    },
+    {
+      "epoch": 0.3900469638296119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3937703914427043e-05,
+      "loss": 0.8448,
+      "step": 14202
+    },
+    {
+      "epoch": 0.3900744280574552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3936886247873776e-05,
+      "loss": 0.8678,
+      "step": 14203
+    },
+    {
+      "epoch": 0.3901018922852984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3936068550170335e-05,
+      "loss": 0.9671,
+      "step": 14204
+    },
+    {
+      "epoch": 0.3901293565131416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3935250821323199e-05,
+      "loss": 0.888,
+      "step": 14205
+    },
+    {
+      "epoch": 0.39015682074098484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3934433061338833e-05,
+      "loss": 0.8644,
+      "step": 14206
+    },
+    {
+      "epoch": 0.3901842849688281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3933615270223706e-05,
+      "loss": 0.9649,
+      "step": 14207
+    },
+    {
+      "epoch": 0.39021174919667134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3932797447984293e-05,
+      "loss": 0.8658,
+      "step": 14208
+    },
+    {
+      "epoch": 0.39023921342451456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3931979594627065e-05,
+      "loss": 1.0267,
+      "step": 14209
+    },
+    {
+      "epoch": 0.3902666776523578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3931161710158489e-05,
+      "loss": 0.9545,
+      "step": 14210
+    },
+    {
+      "epoch": 0.39029414188020106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.393034379458504e-05,
+      "loss": 0.9111,
+      "step": 14211
+    },
+    {
+      "epoch": 0.3903216061080443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3929525847913188e-05,
+      "loss": 0.8563,
+      "step": 14212
+    },
+    {
+      "epoch": 0.3903490703358875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3928707870149405e-05,
+      "loss": 1.0289,
+      "step": 14213
+    },
+    {
+      "epoch": 0.3903765345637307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3927889861300163e-05,
+      "loss": 0.8802,
+      "step": 14214
+    },
+    {
+      "epoch": 0.39040399879157395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3927071821371938e-05,
+      "loss": 0.8727,
+      "step": 14215
+    },
+    {
+      "epoch": 0.3904314630194172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3926253750371195e-05,
+      "loss": 0.952,
+      "step": 14216
+    },
+    {
+      "epoch": 0.39045892724726045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3925435648304415e-05,
+      "loss": 0.8821,
+      "step": 14217
+    },
+    {
+      "epoch": 0.39048639147510367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3924617515178066e-05,
+      "loss": 0.8953,
+      "step": 14218
+    },
+    {
+      "epoch": 0.3905138557029469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3923799350998623e-05,
+      "loss": 0.8882,
+      "step": 14219
+    },
+    {
+      "epoch": 0.39054131993079017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3922981155772559e-05,
+      "loss": 0.9512,
+      "step": 14220
+    },
+    {
+      "epoch": 0.3905687841586334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3922162929506351e-05,
+      "loss": 0.8397,
+      "step": 14221
+    },
+    {
+      "epoch": 0.3905962483864766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3921344672206469e-05,
+      "loss": 0.82,
+      "step": 14222
+    },
+    {
+      "epoch": 0.39062371261431983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3920526383879387e-05,
+      "loss": 0.9428,
+      "step": 14223
+    },
+    {
+      "epoch": 0.3906511768421631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3919708064531585e-05,
+      "loss": 0.8122,
+      "step": 14224
+    },
+    {
+      "epoch": 0.3906786410700063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3918889714169532e-05,
+      "loss": 0.9226,
+      "step": 14225
+    },
+    {
+      "epoch": 0.39070610529784955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3918071332799708e-05,
+      "loss": 1.0233,
+      "step": 14226
+    },
+    {
+      "epoch": 0.39073356952569277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3917252920428586e-05,
+      "loss": 0.891,
+      "step": 14227
+    },
+    {
+      "epoch": 0.39076103375353605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3916434477062636e-05,
+      "loss": 0.9163,
+      "step": 14228
+    },
+    {
+      "epoch": 0.39078849798137927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3915616002708345e-05,
+      "loss": 0.9252,
+      "step": 14229
+    },
+    {
+      "epoch": 0.3908159622092225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3914797497372183e-05,
+      "loss": 1.0187,
+      "step": 14230
+    },
+    {
+      "epoch": 0.3908434264370657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3913978961060623e-05,
+      "loss": 0.8686,
+      "step": 14231
+    },
+    {
+      "epoch": 0.39087089066490893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3913160393780147e-05,
+      "loss": 0.8706,
+      "step": 14232
+    },
+    {
+      "epoch": 0.3908983548927522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3912341795537235e-05,
+      "loss": 0.9854,
+      "step": 14233
+    },
+    {
+      "epoch": 0.39092581912059543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3911523166338355e-05,
+      "loss": 1.0407,
+      "step": 14234
+    },
+    {
+      "epoch": 0.39095328334843865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3910704506189987e-05,
+      "loss": 0.9108,
+      "step": 14235
+    },
+    {
+      "epoch": 0.3909807475762819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3909885815098612e-05,
+      "loss": 0.9424,
+      "step": 14236
+    },
+    {
+      "epoch": 0.39100821180412515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3909067093070705e-05,
+      "loss": 1.0258,
+      "step": 14237
+    },
+    {
+      "epoch": 0.39103567603196837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3908248340112747e-05,
+      "loss": 0.905,
+      "step": 14238
+    },
+    {
+      "epoch": 0.3910631402598116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3907429556231211e-05,
+      "loss": 0.9261,
+      "step": 14239
+    },
+    {
+      "epoch": 0.3910906044876548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.390661074143258e-05,
+      "loss": 0.8622,
+      "step": 14240
+    },
+    {
+      "epoch": 0.3911180687154981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3905791895723332e-05,
+      "loss": 0.893,
+      "step": 14241
+    },
+    {
+      "epoch": 0.3911455329433413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3904973019109949e-05,
+      "loss": 0.9319,
+      "step": 14242
+    },
+    {
+      "epoch": 0.39117299717118453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3904154111598902e-05,
+      "loss": 0.9085,
+      "step": 14243
+    },
+    {
+      "epoch": 0.39120046139902775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3903335173196674e-05,
+      "loss": 0.9493,
+      "step": 14244
+    },
+    {
+      "epoch": 0.391227925626871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.390251620390975e-05,
+      "loss": 0.96,
+      "step": 14245
+    },
+    {
+      "epoch": 0.39125538985471425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3901697203744607e-05,
+      "loss": 1.0282,
+      "step": 14246
+    },
+    {
+      "epoch": 0.3912828540825575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3900878172707722e-05,
+      "loss": 1.0225,
+      "step": 14247
+    },
+    {
+      "epoch": 0.3913103183104007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3900059110805578e-05,
+      "loss": 1.0501,
+      "step": 14248
+    },
+    {
+      "epoch": 0.3913377825382439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3899240018044655e-05,
+      "loss": 0.9392,
+      "step": 14249
+    },
+    {
+      "epoch": 0.3913652467660872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3898420894431434e-05,
+      "loss": 0.8994,
+      "step": 14250
+    },
+    {
+      "epoch": 0.3913927109939304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3897601739972402e-05,
+      "loss": 0.9654,
+      "step": 14251
+    },
+    {
+      "epoch": 0.39142017522177364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.389678255467403e-05,
+      "loss": 0.9082,
+      "step": 14252
+    },
+    {
+      "epoch": 0.39144763944961686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3895963338542807e-05,
+      "loss": 0.9,
+      "step": 14253
+    },
+    {
+      "epoch": 0.39147510367746013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3895144091585211e-05,
+      "loss": 0.9734,
+      "step": 14254
+    },
+    {
+      "epoch": 0.39150256790530336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3894324813807727e-05,
+      "loss": 0.9921,
+      "step": 14255
+    },
+    {
+      "epoch": 0.3915300321331466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3893505505216835e-05,
+      "loss": 0.948,
+      "step": 14256
+    },
+    {
+      "epoch": 0.3915574963609898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3892686165819022e-05,
+      "loss": 0.9844,
+      "step": 14257
+    },
+    {
+      "epoch": 0.391584960588833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.389186679562077e-05,
+      "loss": 0.9204,
+      "step": 14258
+    },
+    {
+      "epoch": 0.3916124248166763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3891047394628555e-05,
+      "loss": 0.9477,
+      "step": 14259
+    },
+    {
+      "epoch": 0.3916398890445195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.389022796284887e-05,
+      "loss": 1.0564,
+      "step": 14260
+    },
+    {
+      "epoch": 0.39166735327236274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3889408500288192e-05,
+      "loss": 0.8843,
+      "step": 14261
+    },
+    {
+      "epoch": 0.39169481750020596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388858900695301e-05,
+      "loss": 1.0024,
+      "step": 14262
+    },
+    {
+      "epoch": 0.39172228172804924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3887769482849805e-05,
+      "loss": 0.8853,
+      "step": 14263
+    },
+    {
+      "epoch": 0.39174974595589246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3886949927985058e-05,
+      "loss": 0.9483,
+      "step": 14264
+    },
+    {
+      "epoch": 0.3917772101837357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388613034236526e-05,
+      "loss": 1.025,
+      "step": 14265
+    },
+    {
+      "epoch": 0.3918046744115789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3885310725996894e-05,
+      "loss": 0.8849,
+      "step": 14266
+    },
+    {
+      "epoch": 0.3918321386394222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3884491078886444e-05,
+      "loss": 0.8456,
+      "step": 14267
+    },
+    {
+      "epoch": 0.3918596028672654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3883671401040398e-05,
+      "loss": 0.9001,
+      "step": 14268
+    },
+    {
+      "epoch": 0.3918870670951086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3882851692465233e-05,
+      "loss": 0.981,
+      "step": 14269
+    },
+    {
+      "epoch": 0.39191453132295184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3882031953167448e-05,
+      "loss": 0.9356,
+      "step": 14270
+    },
+    {
+      "epoch": 0.39194199555079506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388121218315352e-05,
+      "loss": 0.8767,
+      "step": 14271
+    },
+    {
+      "epoch": 0.39196945977863834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3880392382429938e-05,
+      "loss": 1.058,
+      "step": 14272
+    },
+    {
+      "epoch": 0.39199692400648156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387957255100319e-05,
+      "loss": 0.9541,
+      "step": 14273
+    },
+    {
+      "epoch": 0.3920243882343248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387875268887976e-05,
+      "loss": 0.8524,
+      "step": 14274
+    },
+    {
+      "epoch": 0.392051852462168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3877932796066133e-05,
+      "loss": 0.8275,
+      "step": 14275
+    },
+    {
+      "epoch": 0.3920793166900113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3877112872568801e-05,
+      "loss": 1.0688,
+      "step": 14276
+    },
+    {
+      "epoch": 0.3921067809178545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3876292918394251e-05,
+      "loss": 0.9994,
+      "step": 14277
+    },
+    {
+      "epoch": 0.3921342451456977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3875472933548972e-05,
+      "loss": 0.9472,
+      "step": 14278
+    },
+    {
+      "epoch": 0.39216170937354095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3874652918039448e-05,
+      "loss": 0.9655,
+      "step": 14279
+    },
+    {
+      "epoch": 0.3921891736013842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387383287187217e-05,
+      "loss": 0.9918,
+      "step": 14280
+    },
+    {
+      "epoch": 0.39221663782922744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3873012795053624e-05,
+      "loss": 0.9803,
+      "step": 14281
+    },
+    {
+      "epoch": 0.39224410205707066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3872192687590302e-05,
+      "loss": 0.9555,
+      "step": 14282
+    },
+    {
+      "epoch": 0.3922715662849139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3871372549488693e-05,
+      "loss": 0.9141,
+      "step": 14283
+    },
+    {
+      "epoch": 0.3922990305127571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387055238075528e-05,
+      "loss": 0.8507,
+      "step": 14284
+    },
+    {
+      "epoch": 0.3923264947406004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386973218139656e-05,
+      "loss": 0.8582,
+      "step": 14285
+    },
+    {
+      "epoch": 0.3923539589684436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3868911951419021e-05,
+      "loss": 0.9514,
+      "step": 14286
+    },
+    {
+      "epoch": 0.3923814231962868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386809169082915e-05,
+      "loss": 0.8635,
+      "step": 14287
+    },
+    {
+      "epoch": 0.39240888742413005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386727139963344e-05,
+      "loss": 1.0367,
+      "step": 14288
+    },
+    {
+      "epoch": 0.3924363516519733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3866451077838382e-05,
+      "loss": 0.9356,
+      "step": 14289
+    },
+    {
+      "epoch": 0.39246381587981655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3865630725450465e-05,
+      "loss": 0.9918,
+      "step": 14290
+    },
+    {
+      "epoch": 0.39249128010765977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3864810342476178e-05,
+      "loss": 0.9109,
+      "step": 14291
+    },
+    {
+      "epoch": 0.392518744335503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3863989928922017e-05,
+      "loss": 0.9635,
+      "step": 14292
+    },
+    {
+      "epoch": 0.39254620856334627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3863169484794469e-05,
+      "loss": 1.0135,
+      "step": 14293
+    },
+    {
+      "epoch": 0.3925736727911895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386234901010003e-05,
+      "loss": 0.8876,
+      "step": 14294
+    },
+    {
+      "epoch": 0.3926011370190327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3861528504845188e-05,
+      "loss": 0.9464,
+      "step": 14295
+    },
+    {
+      "epoch": 0.39262860124687593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3860707969036436e-05,
+      "loss": 0.997,
+      "step": 14296
+    },
+    {
+      "epoch": 0.39265606547471915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3859887402680268e-05,
+      "loss": 0.8466,
+      "step": 14297
+    },
+    {
+      "epoch": 0.39268352970256243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3859066805783177e-05,
+      "loss": 0.8638,
+      "step": 14298
+    },
+    {
+      "epoch": 0.39271099393040565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.385824617835165e-05,
+      "loss": 0.9427,
+      "step": 14299
+    },
+    {
+      "epoch": 0.39273845815824887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.385742552039219e-05,
+      "loss": 0.8336,
+      "step": 14300
+    },
+    {
+      "epoch": 0.3927659223860921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3856604831911283e-05,
+      "loss": 0.9069,
+      "step": 14301
+    },
+    {
+      "epoch": 0.39279338661393537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3855784112915424e-05,
+      "loss": 1.0085,
+      "step": 14302
+    },
+    {
+      "epoch": 0.3928208508417786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3854963363411106e-05,
+      "loss": 0.9554,
+      "step": 14303
+    },
+    {
+      "epoch": 0.3928483150696218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3854142583404828e-05,
+      "loss": 1.0367,
+      "step": 14304
+    },
+    {
+      "epoch": 0.39287577929746503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.385332177290308e-05,
+      "loss": 0.9539,
+      "step": 14305
+    },
+    {
+      "epoch": 0.3929032435253083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3852500931912355e-05,
+      "loss": 0.9997,
+      "step": 14306
+    },
+    {
+      "epoch": 0.39293070775315153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3851680060439155e-05,
+      "loss": 1.0168,
+      "step": 14307
+    },
+    {
+      "epoch": 0.39295817198099475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3850859158489964e-05,
+      "loss": 0.8463,
+      "step": 14308
+    },
+    {
+      "epoch": 0.392985636208838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.385003822607129e-05,
+      "loss": 0.8978,
+      "step": 14309
+    },
+    {
+      "epoch": 0.3930131004366812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3849217263189621e-05,
+      "loss": 0.9292,
+      "step": 14310
+    },
+    {
+      "epoch": 0.39304056466452447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.384839626985145e-05,
+      "loss": 0.8941,
+      "step": 14311
+    },
+    {
+      "epoch": 0.3930680288923677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3847575246063279e-05,
+      "loss": 0.9609,
+      "step": 14312
+    },
+    {
+      "epoch": 0.3930954931202109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3846754191831604e-05,
+      "loss": 0.9468,
+      "step": 14313
+    },
+    {
+      "epoch": 0.39312295734805414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3845933107162918e-05,
+      "loss": 0.9684,
+      "step": 14314
+    },
+    {
+      "epoch": 0.3931504215758974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.384511199206372e-05,
+      "loss": 1.0346,
+      "step": 14315
+    },
+    {
+      "epoch": 0.39317788580374063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3844290846540508e-05,
+      "loss": 0.9066,
+      "step": 14316
+    },
+    {
+      "epoch": 0.39320535003158386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3843469670599775e-05,
+      "loss": 0.884,
+      "step": 14317
+    },
+    {
+      "epoch": 0.3932328142594271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3842648464248021e-05,
+      "loss": 0.9332,
+      "step": 14318
+    },
+    {
+      "epoch": 0.39326027848727035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3841827227491746e-05,
+      "loss": 0.834,
+      "step": 14319
+    },
+    {
+      "epoch": 0.3932877427151136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3841005960337443e-05,
+      "loss": 0.8203,
+      "step": 14320
+    },
+    {
+      "epoch": 0.3933152069429568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3840184662791616e-05,
+      "loss": 1.1285,
+      "step": 14321
+    },
+    {
+      "epoch": 0.3933426711708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3839363334860761e-05,
+      "loss": 0.9594,
+      "step": 14322
+    },
+    {
+      "epoch": 0.3933701353986433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3838541976551371e-05,
+      "loss": 1.0318,
+      "step": 14323
+    },
+    {
+      "epoch": 0.3933975996264865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3837720587869955e-05,
+      "loss": 0.9803,
+      "step": 14324
+    },
+    {
+      "epoch": 0.39342506385432974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3836899168823006e-05,
+      "loss": 0.8929,
+      "step": 14325
+    },
+    {
+      "epoch": 0.39345252808217296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3836077719417022e-05,
+      "loss": 0.9835,
+      "step": 14326
+    },
+    {
+      "epoch": 0.3934799923100162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.383525623965851e-05,
+      "loss": 0.9711,
+      "step": 14327
+    },
+    {
+      "epoch": 0.39350745653785946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.383443472955396e-05,
+      "loss": 0.9258,
+      "step": 14328
+    },
+    {
+      "epoch": 0.3935349207657027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3833613189109878e-05,
+      "loss": 0.8111,
+      "step": 14329
+    },
+    {
+      "epoch": 0.3935623849935459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3832791618332769e-05,
+      "loss": 0.9945,
+      "step": 14330
+    },
+    {
+      "epoch": 0.3935898492213891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3831970017229121e-05,
+      "loss": 0.9984,
+      "step": 14331
+    },
+    {
+      "epoch": 0.3936173134492324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3831148385805444e-05,
+      "loss": 1.0606,
+      "step": 14332
+    },
+    {
+      "epoch": 0.3936447776770756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3830326724068237e-05,
+      "loss": 0.8658,
+      "step": 14333
+    },
+    {
+      "epoch": 0.39367224190491884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3829505032024004e-05,
+      "loss": 1.0029,
+      "step": 14334
+    },
+    {
+      "epoch": 0.39369970613276206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.382868330967924e-05,
+      "loss": 0.849,
+      "step": 14335
+    },
+    {
+      "epoch": 0.39372717036060534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3827861557040452e-05,
+      "loss": 0.9847,
+      "step": 14336
+    },
+    {
+      "epoch": 0.39375463458844856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3827039774114139e-05,
+      "loss": 0.9658,
+      "step": 14337
+    },
+    {
+      "epoch": 0.3937820988162918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.382621796090681e-05,
+      "loss": 1.0105,
+      "step": 14338
+    },
+    {
+      "epoch": 0.393809563044135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3825396117424959e-05,
+      "loss": 0.8725,
+      "step": 14339
+    },
+    {
+      "epoch": 0.3938370272719782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3824574243675089e-05,
+      "loss": 0.9327,
+      "step": 14340
+    },
+    {
+      "epoch": 0.3938644914998215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.382375233966371e-05,
+      "loss": 0.9923,
+      "step": 14341
+    },
+    {
+      "epoch": 0.3938919557276647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.382293040539732e-05,
+      "loss": 0.9324,
+      "step": 14342
+    },
+    {
+      "epoch": 0.39391941995550794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3822108440882424e-05,
+      "loss": 0.9125,
+      "step": 14343
+    },
+    {
+      "epoch": 0.39394688418335116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3821286446125526e-05,
+      "loss": 0.9957,
+      "step": 14344
+    },
+    {
+      "epoch": 0.39397434841119444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3820464421133129e-05,
+      "loss": 0.8488,
+      "step": 14345
+    },
+    {
+      "epoch": 0.39400181263903766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3819642365911739e-05,
+      "loss": 0.9487,
+      "step": 14346
+    },
+    {
+      "epoch": 0.3940292768668809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3818820280467856e-05,
+      "loss": 0.8592,
+      "step": 14347
+    },
+    {
+      "epoch": 0.3940567410947241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3817998164807988e-05,
+      "loss": 0.968,
+      "step": 14348
+    },
+    {
+      "epoch": 0.3940842053225674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381717601893864e-05,
+      "loss": 0.8843,
+      "step": 14349
+    },
+    {
+      "epoch": 0.3941116695504106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381635384286632e-05,
+      "loss": 0.9243,
+      "step": 14350
+    },
+    {
+      "epoch": 0.3941391337782538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3815531636597528e-05,
+      "loss": 0.8843,
+      "step": 14351
+    },
+    {
+      "epoch": 0.39416659800609705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3814709400138769e-05,
+      "loss": 0.8886,
+      "step": 14352
+    },
+    {
+      "epoch": 0.39419406223394027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3813887133496556e-05,
+      "loss": 0.8997,
+      "step": 14353
+    },
+    {
+      "epoch": 0.39422152646178354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3813064836677388e-05,
+      "loss": 0.7921,
+      "step": 14354
+    },
+    {
+      "epoch": 0.39424899068962677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3812242509687774e-05,
+      "loss": 0.9558,
+      "step": 14355
+    },
+    {
+      "epoch": 0.39427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3811420152534218e-05,
+      "loss": 1.0431,
+      "step": 14356
+    },
+    {
+      "epoch": 0.3943039191453132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3810597765223235e-05,
+      "loss": 0.85,
+      "step": 14357
+    },
+    {
+      "epoch": 0.3943313833731565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3809775347761324e-05,
+      "loss": 0.9158,
+      "step": 14358
+    },
+    {
+      "epoch": 0.3943588476009997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3808952900154993e-05,
+      "loss": 0.9171,
+      "step": 14359
+    },
+    {
+      "epoch": 0.39438631182884293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3808130422410755e-05,
+      "loss": 0.8716,
+      "step": 14360
+    },
+    {
+      "epoch": 0.39441377605668615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3807307914535108e-05,
+      "loss": 0.9979,
+      "step": 14361
+    },
+    {
+      "epoch": 0.3944412402845294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3806485376534573e-05,
+      "loss": 0.9519,
+      "step": 14362
+    },
+    {
+      "epoch": 0.39446870451237265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3805662808415648e-05,
+      "loss": 0.9464,
+      "step": 14363
+    },
+    {
+      "epoch": 0.39449616874021587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3804840210184842e-05,
+      "loss": 1.0079,
+      "step": 14364
+    },
+    {
+      "epoch": 0.3945236329680591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.380401758184867e-05,
+      "loss": 0.782,
+      "step": 14365
+    },
+    {
+      "epoch": 0.3945510971959023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3803194923413639e-05,
+      "loss": 0.9802,
+      "step": 14366
+    },
+    {
+      "epoch": 0.3945785614237456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3802372234886251e-05,
+      "loss": 0.8481,
+      "step": 14367
+    },
+    {
+      "epoch": 0.3946060256515888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3801549516273023e-05,
+      "loss": 0.9943,
+      "step": 14368
+    },
+    {
+      "epoch": 0.39463348987943203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3800726767580463e-05,
+      "loss": 1.0332,
+      "step": 14369
+    },
+    {
+      "epoch": 0.39466095410727525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3799903988815081e-05,
+      "loss": 0.9434,
+      "step": 14370
+    },
+    {
+      "epoch": 0.39468841833511853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3799081179983388e-05,
+      "loss": 1.0112,
+      "step": 14371
+    },
+    {
+      "epoch": 0.39471588256296175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3798258341091892e-05,
+      "loss": 1.0333,
+      "step": 14372
+    },
+    {
+      "epoch": 0.39474334679080497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3797435472147102e-05,
+      "loss": 0.9711,
+      "step": 14373
+    },
+    {
+      "epoch": 0.3947708110186482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3796612573155534e-05,
+      "loss": 1.0335,
+      "step": 14374
+    },
+    {
+      "epoch": 0.39479827524649147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3795789644123698e-05,
+      "loss": 0.9583,
+      "step": 14375
+    },
+    {
+      "epoch": 0.3948257394743347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3794966685058102e-05,
+      "loss": 0.9159,
+      "step": 14376
+    },
+    {
+      "epoch": 0.3948532037021779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3794143695965261e-05,
+      "loss": 0.9906,
+      "step": 14377
+    },
+    {
+      "epoch": 0.39488066793002113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3793320676851685e-05,
+      "loss": 0.9885,
+      "step": 14378
+    },
+    {
+      "epoch": 0.39490813215786436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3792497627723887e-05,
+      "loss": 0.9222,
+      "step": 14379
+    },
+    {
+      "epoch": 0.39493559638570763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3791674548588374e-05,
+      "loss": 1.0453,
+      "step": 14380
+    },
+    {
+      "epoch": 0.39496306061355085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3790851439451669e-05,
+      "loss": 0.9204,
+      "step": 14381
+    },
+    {
+      "epoch": 0.3949905248413941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3790028300320277e-05,
+      "loss": 0.9471,
+      "step": 14382
+    },
+    {
+      "epoch": 0.3950179890692373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3789205131200713e-05,
+      "loss": 0.8821,
+      "step": 14383
+    },
+    {
+      "epoch": 0.3950454532970806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.378838193209949e-05,
+      "loss": 0.9872,
+      "step": 14384
+    },
+    {
+      "epoch": 0.3950729175249238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3787558703023121e-05,
+      "loss": 0.8899,
+      "step": 14385
+    },
+    {
+      "epoch": 0.395100381752767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3786735443978124e-05,
+      "loss": 0.9681,
+      "step": 14386
+    },
+    {
+      "epoch": 0.39512784598061024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3785912154971004e-05,
+      "loss": 0.889,
+      "step": 14387
+    },
+    {
+      "epoch": 0.3951553102084535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3785088836008283e-05,
+      "loss": 0.9063,
+      "step": 14388
+    },
+    {
+      "epoch": 0.39518277443629674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3784265487096473e-05,
+      "loss": 0.9411,
+      "step": 14389
+    },
+    {
+      "epoch": 0.39521023866413996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3783442108242089e-05,
+      "loss": 0.9319,
+      "step": 14390
+    },
+    {
+      "epoch": 0.3952377028919832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3782618699451644e-05,
+      "loss": 0.9274,
+      "step": 14391
+    },
+    {
+      "epoch": 0.3952651671198264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3781795260731656e-05,
+      "loss": 1.0229,
+      "step": 14392
+    },
+    {
+      "epoch": 0.3952926313476697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3780971792088637e-05,
+      "loss": 0.9548,
+      "step": 14393
+    },
+    {
+      "epoch": 0.3953200955755129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3780148293529107e-05,
+      "loss": 0.9619,
+      "step": 14394
+    },
+    {
+      "epoch": 0.3953475598033561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3779324765059578e-05,
+      "loss": 0.9803,
+      "step": 14395
+    },
+    {
+      "epoch": 0.39537502403119934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3778501206686567e-05,
+      "loss": 0.8765,
+      "step": 14396
+    },
+    {
+      "epoch": 0.3954024882590426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3777677618416592e-05,
+      "loss": 0.8907,
+      "step": 14397
+    },
+    {
+      "epoch": 0.39542995248688584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3776854000256171e-05,
+      "loss": 0.8216,
+      "step": 14398
+    },
+    {
+      "epoch": 0.39545741671472906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3776030352211814e-05,
+      "loss": 0.9229,
+      "step": 14399
+    },
+    {
+      "epoch": 0.3954848809425723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3775206674290042e-05,
+      "loss": 0.947,
+      "step": 14400
+    },
+    {
+      "epoch": 0.39551234517041556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3774382966497375e-05,
+      "loss": 1.0174,
+      "step": 14401
+    },
+    {
+      "epoch": 0.3955398093982588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3773559228840327e-05,
+      "loss": 0.9911,
+      "step": 14402
+    },
+    {
+      "epoch": 0.395567273626102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3772735461325416e-05,
+      "loss": 0.9748,
+      "step": 14403
+    },
+    {
+      "epoch": 0.3955947378539452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.377191166395916e-05,
+      "loss": 0.9465,
+      "step": 14404
+    },
+    {
+      "epoch": 0.39562220208178844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3771087836748079e-05,
+      "loss": 0.9023,
+      "step": 14405
+    },
+    {
+      "epoch": 0.3956496663096317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.377026397969869e-05,
+      "loss": 0.8029,
+      "step": 14406
+    },
+    {
+      "epoch": 0.39567713053747494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3769440092817515e-05,
+      "loss": 0.8795,
+      "step": 14407
+    },
+    {
+      "epoch": 0.39570459476531816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3768616176111063e-05,
+      "loss": 0.8475,
+      "step": 14408
+    },
+    {
+      "epoch": 0.3957320589931614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3767792229585863e-05,
+      "loss": 0.9268,
+      "step": 14409
+    },
+    {
+      "epoch": 0.39575952322100466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3766968253248436e-05,
+      "loss": 0.9226,
+      "step": 14410
+    },
+    {
+      "epoch": 0.3957869874488479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3766144247105292e-05,
+      "loss": 0.9477,
+      "step": 14411
+    },
+    {
+      "epoch": 0.3958144516766911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3765320211162955e-05,
+      "loss": 0.9378,
+      "step": 14412
+    },
+    {
+      "epoch": 0.3958419159045343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.376449614542795e-05,
+      "loss": 0.9492,
+      "step": 14413
+    },
+    {
+      "epoch": 0.3958693801323776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3763672049906792e-05,
+      "loss": 0.9736,
+      "step": 14414
+    },
+    {
+      "epoch": 0.3958968443602208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3762847924606e-05,
+      "loss": 0.9584,
+      "step": 14415
+    },
+    {
+      "epoch": 0.39592430858806404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.37620237695321e-05,
+      "loss": 0.9388,
+      "step": 14416
+    },
+    {
+      "epoch": 0.39595177281590727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.376119958469161e-05,
+      "loss": 0.985,
+      "step": 14417
+    },
+    {
+      "epoch": 0.39597923704375054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3760375370091052e-05,
+      "loss": 0.7847,
+      "step": 14418
+    },
+    {
+      "epoch": 0.39600670127159376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3759551125736949e-05,
+      "loss": 0.9982,
+      "step": 14419
+    },
+    {
+      "epoch": 0.396034165499437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.375872685163582e-05,
+      "loss": 0.9441,
+      "step": 14420
+    },
+    {
+      "epoch": 0.3960616297272802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3757902547794186e-05,
+      "loss": 0.8874,
+      "step": 14421
+    },
+    {
+      "epoch": 0.39608909395512343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3757078214218575e-05,
+      "loss": 0.987,
+      "step": 14422
+    },
+    {
+      "epoch": 0.3961165581829667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3756253850915505e-05,
+      "loss": 0.8367,
+      "step": 14423
+    },
+    {
+      "epoch": 0.3961440224108099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3755429457891498e-05,
+      "loss": 0.9141,
+      "step": 14424
+    },
+    {
+      "epoch": 0.39617148663865315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.375460503515308e-05,
+      "loss": 0.9662,
+      "step": 14425
+    },
+    {
+      "epoch": 0.39619895086649637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3753780582706772e-05,
+      "loss": 0.8942,
+      "step": 14426
+    },
+    {
+      "epoch": 0.39622641509433965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3752956100559096e-05,
+      "loss": 0.8177,
+      "step": 14427
+    },
+    {
+      "epoch": 0.39625387932218287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3752131588716582e-05,
+      "loss": 0.8017,
+      "step": 14428
+    },
+    {
+      "epoch": 0.3962813435500261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3751307047185745e-05,
+      "loss": 1.0087,
+      "step": 14429
+    },
+    {
+      "epoch": 0.3963088077778693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3750482475973115e-05,
+      "loss": 1.0144,
+      "step": 14430
+    },
+    {
+      "epoch": 0.3963362720057126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3749657875085215e-05,
+      "loss": 0.886,
+      "step": 14431
+    },
+    {
+      "epoch": 0.3963637362335558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.374883324452857e-05,
+      "loss": 0.833,
+      "step": 14432
+    },
+    {
+      "epoch": 0.39639120046139903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3748008584309705e-05,
+      "loss": 0.9784,
+      "step": 14433
+    },
+    {
+      "epoch": 0.39641866468924225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3747183894435143e-05,
+      "loss": 0.8436,
+      "step": 14434
+    },
+    {
+      "epoch": 0.39644612891708547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.374635917491141e-05,
+      "loss": 1.0064,
+      "step": 14435
+    },
+    {
+      "epoch": 0.39647359314492875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3745534425745031e-05,
+      "loss": 0.9108,
+      "step": 14436
+    },
+    {
+      "epoch": 0.39650105737277197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3744709646942535e-05,
+      "loss": 0.8896,
+      "step": 14437
+    },
+    {
+      "epoch": 0.3965285216006152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3743884838510446e-05,
+      "loss": 0.9302,
+      "step": 14438
+    },
+    {
+      "epoch": 0.3965559858284584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3743060000455288e-05,
+      "loss": 0.9137,
+      "step": 14439
+    },
+    {
+      "epoch": 0.3965834500563017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.374223513278359e-05,
+      "loss": 0.9256,
+      "step": 14440
+    },
+    {
+      "epoch": 0.3966109142841449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3741410235501877e-05,
+      "loss": 0.9146,
+      "step": 14441
+    },
+    {
+      "epoch": 0.39663837851198813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3740585308616679e-05,
+      "loss": 0.9061,
+      "step": 14442
+    },
+    {
+      "epoch": 0.39666584273983135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3739760352134521e-05,
+      "loss": 0.9073,
+      "step": 14443
+    },
+    {
+      "epoch": 0.39669330696767463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.373893536606193e-05,
+      "loss": 0.9399,
+      "step": 14444
+    },
+    {
+      "epoch": 0.39672077119551785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3738110350405433e-05,
+      "loss": 0.8769,
+      "step": 14445
+    },
+    {
+      "epoch": 0.3967482354233611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3737285305171562e-05,
+      "loss": 0.8385,
+      "step": 14446
+    },
+    {
+      "epoch": 0.3967756996512043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3736460230366838e-05,
+      "loss": 0.928,
+      "step": 14447
+    },
+    {
+      "epoch": 0.3968031638790475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3735635125997794e-05,
+      "loss": 0.8535,
+      "step": 14448
+    },
+    {
+      "epoch": 0.3968306281068908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3734809992070959e-05,
+      "loss": 0.9644,
+      "step": 14449
+    },
+    {
+      "epoch": 0.396858092334734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3733984828592861e-05,
+      "loss": 1.0217,
+      "step": 14450
+    },
+    {
+      "epoch": 0.39688555656257724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3733159635570029e-05,
+      "loss": 0.9568,
+      "step": 14451
+    },
+    {
+      "epoch": 0.39691302079042046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.373233441300899e-05,
+      "loss": 1.0233,
+      "step": 14452
+    },
+    {
+      "epoch": 0.39694048501826373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3731509160916274e-05,
+      "loss": 0.9261,
+      "step": 14453
+    },
+    {
+      "epoch": 0.39696794924610695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3730683879298417e-05,
+      "loss": 0.948,
+      "step": 14454
+    },
+    {
+      "epoch": 0.3969954134739502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372985856816194e-05,
+      "loss": 1.0122,
+      "step": 14455
+    },
+    {
+      "epoch": 0.3970228777017934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3729033227513377e-05,
+      "loss": 0.9639,
+      "step": 14456
+    },
+    {
+      "epoch": 0.3970503419296367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372820785735926e-05,
+      "loss": 0.9703,
+      "step": 14457
+    },
+    {
+      "epoch": 0.3970778061574799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3727382457706123e-05,
+      "loss": 0.9868,
+      "step": 14458
+    },
+    {
+      "epoch": 0.3971052703853231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3726557028560485e-05,
+      "loss": 0.9393,
+      "step": 14459
+    },
+    {
+      "epoch": 0.39713273461316634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3725731569928886e-05,
+      "loss": 0.8809,
+      "step": 14460
+    },
+    {
+      "epoch": 0.39716019884100956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3724906081817858e-05,
+      "loss": 0.9431,
+      "step": 14461
+    },
+    {
+      "epoch": 0.39718766306885284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3724080564233928e-05,
+      "loss": 1.0242,
+      "step": 14462
+    },
+    {
+      "epoch": 0.39721512729669606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3723255017183632e-05,
+      "loss": 0.9124,
+      "step": 14463
+    },
+    {
+      "epoch": 0.3972425915245393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3722429440673497e-05,
+      "loss": 0.9218,
+      "step": 14464
+    },
+    {
+      "epoch": 0.3972700557523825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3721603834710057e-05,
+      "loss": 1.0039,
+      "step": 14465
+    },
+    {
+      "epoch": 0.3972975199802258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3720778199299853e-05,
+      "loss": 0.9461,
+      "step": 14466
+    },
+    {
+      "epoch": 0.397324984208069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3719952534449405e-05,
+      "loss": 0.9662,
+      "step": 14467
+    },
+    {
+      "epoch": 0.3973524484359122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3719126840165253e-05,
+      "loss": 0.9882,
+      "step": 14468
+    },
+    {
+      "epoch": 0.39737991266375544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.371830111645393e-05,
+      "loss": 0.9646,
+      "step": 14469
+    },
+    {
+      "epoch": 0.3974073768915987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3717475363321966e-05,
+      "loss": 0.9286,
+      "step": 14470
+    },
+    {
+      "epoch": 0.39743484111944194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3716649580775898e-05,
+      "loss": 0.9995,
+      "step": 14471
+    },
+    {
+      "epoch": 0.39746230534728516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3715823768822261e-05,
+      "loss": 0.9647,
+      "step": 14472
+    },
+    {
+      "epoch": 0.3974897695751284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3714997927467582e-05,
+      "loss": 0.937,
+      "step": 14473
+    },
+    {
+      "epoch": 0.3975172338029716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3714172056718405e-05,
+      "loss": 0.9554,
+      "step": 14474
+    },
+    {
+      "epoch": 0.3975446980308149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3713346156581258e-05,
+      "loss": 0.9253,
+      "step": 14475
+    },
+    {
+      "epoch": 0.3975721622586581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3712520227062678e-05,
+      "loss": 0.91,
+      "step": 14476
+    },
+    {
+      "epoch": 0.3975996264865013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3711694268169198e-05,
+      "loss": 0.9492,
+      "step": 14477
+    },
+    {
+      "epoch": 0.39762709071434454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3710868279907359e-05,
+      "loss": 0.9297,
+      "step": 14478
+    },
+    {
+      "epoch": 0.3976545549421878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3710042262283689e-05,
+      "loss": 1.0121,
+      "step": 14479
+    },
+    {
+      "epoch": 0.39768201917003104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3709216215304728e-05,
+      "loss": 0.9006,
+      "step": 14480
+    },
+    {
+      "epoch": 0.39770948339787426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3708390138977014e-05,
+      "loss": 0.9171,
+      "step": 14481
+    },
+    {
+      "epoch": 0.3977369476257175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.370756403330708e-05,
+      "loss": 0.9763,
+      "step": 14482
+    },
+    {
+      "epoch": 0.39776441185356076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3706737898301461e-05,
+      "loss": 0.8905,
+      "step": 14483
+    },
+    {
+      "epoch": 0.397791876081404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3705911733966697e-05,
+      "loss": 0.9234,
+      "step": 14484
+    },
+    {
+      "epoch": 0.3978193403092472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3705085540309325e-05,
+      "loss": 1.0549,
+      "step": 14485
+    },
+    {
+      "epoch": 0.3978468045370904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3704259317335879e-05,
+      "loss": 0.8649,
+      "step": 14486
+    },
+    {
+      "epoch": 0.39787426876493365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3703433065052903e-05,
+      "loss": 0.9653,
+      "step": 14487
+    },
+    {
+      "epoch": 0.3979017329927769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3702606783466922e-05,
+      "loss": 0.9473,
+      "step": 14488
+    },
+    {
+      "epoch": 0.39792919722062015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3701780472584488e-05,
+      "loss": 0.9807,
+      "step": 14489
+    },
+    {
+      "epoch": 0.39795666144846337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3700954132412135e-05,
+      "loss": 0.9834,
+      "step": 14490
+    },
+    {
+      "epoch": 0.3979841256763066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3700127762956394e-05,
+      "loss": 0.8724,
+      "step": 14491
+    },
+    {
+      "epoch": 0.39801158990414987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3699301364223809e-05,
+      "loss": 0.8339,
+      "step": 14492
+    },
+    {
+      "epoch": 0.3980390541319931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.369847493622092e-05,
+      "loss": 0.9578,
+      "step": 14493
+    },
+    {
+      "epoch": 0.3980665183598363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3697648478954267e-05,
+      "loss": 0.965,
+      "step": 14494
+    },
+    {
+      "epoch": 0.39809398258767953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3696821992430383e-05,
+      "loss": 0.9422,
+      "step": 14495
+    },
+    {
+      "epoch": 0.3981214468155228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3695995476655814e-05,
+      "loss": 0.8845,
+      "step": 14496
+    },
+    {
+      "epoch": 0.398148911043366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3695168931637095e-05,
+      "loss": 0.8833,
+      "step": 14497
+    },
+    {
+      "epoch": 0.39817637527120925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.369434235738077e-05,
+      "loss": 0.861,
+      "step": 14498
+    },
+    {
+      "epoch": 0.39820383949905247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3693515753893378e-05,
+      "loss": 0.9781,
+      "step": 14499
+    },
+    {
+      "epoch": 0.3982313037268957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3692689121181455e-05,
+      "loss": 0.9688,
+      "step": 14500
+    },
+    {
+      "epoch": 0.39825876795473897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3691862459251546e-05,
+      "loss": 0.9765,
+      "step": 14501
+    },
+    {
+      "epoch": 0.3982862321825822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3691035768110197e-05,
+      "loss": 1.0139,
+      "step": 14502
+    },
+    {
+      "epoch": 0.3983136964104254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3690209047763938e-05,
+      "loss": 0.9216,
+      "step": 14503
+    },
+    {
+      "epoch": 0.39834116063826863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3689382298219315e-05,
+      "loss": 0.9982,
+      "step": 14504
+    },
+    {
+      "epoch": 0.3983686248661119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.368855551948287e-05,
+      "loss": 0.899,
+      "step": 14505
+    },
+    {
+      "epoch": 0.39839608909395513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3687728711561148e-05,
+      "loss": 0.9795,
+      "step": 14506
+    },
+    {
+      "epoch": 0.39842355332179835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3686901874460688e-05,
+      "loss": 0.9304,
+      "step": 14507
+    },
+    {
+      "epoch": 0.3984510175496416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.368607500818803e-05,
+      "loss": 1.0366,
+      "step": 14508
+    },
+    {
+      "epoch": 0.39847848177748485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3685248112749716e-05,
+      "loss": 1.0119,
+      "step": 14509
+    },
+    {
+      "epoch": 0.39850594600532807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3684421188152299e-05,
+      "loss": 0.9649,
+      "step": 14510
+    },
+    {
+      "epoch": 0.3985334102331713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3683594234402308e-05,
+      "loss": 0.9061,
+      "step": 14511
+    },
+    {
+      "epoch": 0.3985608744610145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3682767251506295e-05,
+      "loss": 0.9668,
+      "step": 14512
+    },
+    {
+      "epoch": 0.39858833868885774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3681940239470801e-05,
+      "loss": 1.0217,
+      "step": 14513
+    },
+    {
+      "epoch": 0.398615802916701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3681113198302369e-05,
+      "loss": 0.9074,
+      "step": 14514
+    },
+    {
+      "epoch": 0.39864326714454423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3680286128007543e-05,
+      "loss": 0.9277,
+      "step": 14515
+    },
+    {
+      "epoch": 0.39867073137238745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3679459028592867e-05,
+      "loss": 0.9143,
+      "step": 14516
+    },
+    {
+      "epoch": 0.3986981956002307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3678631900064887e-05,
+      "loss": 0.9552,
+      "step": 14517
+    },
+    {
+      "epoch": 0.39872565982807395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3677804742430147e-05,
+      "loss": 0.9387,
+      "step": 14518
+    },
+    {
+      "epoch": 0.3987531240559172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3676977555695193e-05,
+      "loss": 0.8369,
+      "step": 14519
+    },
+    {
+      "epoch": 0.3987805882837604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3676150339866562e-05,
+      "loss": 0.8961,
+      "step": 14520
+    },
+    {
+      "epoch": 0.3988080525116036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.367532309495081e-05,
+      "loss": 0.9217,
+      "step": 14521
+    },
+    {
+      "epoch": 0.3988355167394469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.367449582095448e-05,
+      "loss": 0.9579,
+      "step": 14522
+    },
+    {
+      "epoch": 0.3988629809672901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3673668517884111e-05,
+      "loss": 0.8461,
+      "step": 14523
+    },
+    {
+      "epoch": 0.39889044519513334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3672841185746256e-05,
+      "loss": 0.903,
+      "step": 14524
+    },
+    {
+      "epoch": 0.39891790942297656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.367201382454746e-05,
+      "loss": 1.0071,
+      "step": 14525
+    },
+    {
+      "epoch": 0.39894537365081983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3671186434294265e-05,
+      "loss": 0.9222,
+      "step": 14526
+    },
+    {
+      "epoch": 0.39897283787866306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3670359014993222e-05,
+      "loss": 0.9787,
+      "step": 14527
+    },
+    {
+      "epoch": 0.3990003021065063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3669531566650878e-05,
+      "loss": 0.873,
+      "step": 14528
+    },
+    {
+      "epoch": 0.3990277663343495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3668704089273777e-05,
+      "loss": 0.9081,
+      "step": 14529
+    },
+    {
+      "epoch": 0.3990552305621927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3667876582868471e-05,
+      "loss": 0.8893,
+      "step": 14530
+    },
+    {
+      "epoch": 0.399082694790036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3667049047441503e-05,
+      "loss": 0.9231,
+      "step": 14531
+    },
+    {
+      "epoch": 0.3991101590178792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3666221482999422e-05,
+      "loss": 0.9371,
+      "step": 14532
+    },
+    {
+      "epoch": 0.39913762324572244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3665393889548776e-05,
+      "loss": 0.9424,
+      "step": 14533
+    },
+    {
+      "epoch": 0.39916508747356566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3664566267096117e-05,
+      "loss": 0.9014,
+      "step": 14534
+    },
+    {
+      "epoch": 0.39919255170140894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3663738615647986e-05,
+      "loss": 0.9131,
+      "step": 14535
+    },
+    {
+      "epoch": 0.39922001592925216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3662910935210935e-05,
+      "loss": 0.9849,
+      "step": 14536
+    },
+    {
+      "epoch": 0.3992474801570954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3662083225791519e-05,
+      "loss": 0.8291,
+      "step": 14537
+    },
+    {
+      "epoch": 0.3992749443849386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.366125548739628e-05,
+      "loss": 0.9643,
+      "step": 14538
+    },
+    {
+      "epoch": 0.3993024086127819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3660427720031767e-05,
+      "loss": 0.9859,
+      "step": 14539
+    },
+    {
+      "epoch": 0.3993298728406251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3659599923704534e-05,
+      "loss": 0.8456,
+      "step": 14540
+    },
+    {
+      "epoch": 0.3993573370684683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3658772098421128e-05,
+      "loss": 0.9181,
+      "step": 14541
+    },
+    {
+      "epoch": 0.39938480129631154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3657944244188102e-05,
+      "loss": 0.8566,
+      "step": 14542
+    },
+    {
+      "epoch": 0.39941226552415476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3657116361012002e-05,
+      "loss": 0.946,
+      "step": 14543
+    },
+    {
+      "epoch": 0.39943972975199804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.365628844889938e-05,
+      "loss": 0.8582,
+      "step": 14544
+    },
+    {
+      "epoch": 0.39946719397984126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3655460507856788e-05,
+      "loss": 0.9382,
+      "step": 14545
+    },
+    {
+      "epoch": 0.3994946582076845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3654632537890779e-05,
+      "loss": 0.8479,
+      "step": 14546
+    },
+    {
+      "epoch": 0.3995221224355277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.36538045390079e-05,
+      "loss": 0.9724,
+      "step": 14547
+    },
+    {
+      "epoch": 0.399549586663371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3652976511214701e-05,
+      "loss": 0.937,
+      "step": 14548
+    },
+    {
+      "epoch": 0.3995770508912142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3652148454517741e-05,
+      "loss": 0.9243,
+      "step": 14549
+    },
+    {
+      "epoch": 0.3996045151190574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3651320368923568e-05,
+      "loss": 1.0573,
+      "step": 14550
+    },
+    {
+      "epoch": 0.39963197934690065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3650492254438732e-05,
+      "loss": 0.9277,
+      "step": 14551
+    },
+    {
+      "epoch": 0.3996594435747439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364966411106979e-05,
+      "loss": 0.8772,
+      "step": 14552
+    },
+    {
+      "epoch": 0.39968690780258714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3648835938823287e-05,
+      "loss": 0.9868,
+      "step": 14553
+    },
+    {
+      "epoch": 0.39971437203043036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3648007737705784e-05,
+      "loss": 0.8859,
+      "step": 14554
+    },
+    {
+      "epoch": 0.3997418362582736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3647179507723828e-05,
+      "loss": 0.9141,
+      "step": 14555
+    },
+    {
+      "epoch": 0.3997693004861168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3646351248883978e-05,
+      "loss": 0.9028,
+      "step": 14556
+    },
+    {
+      "epoch": 0.3997967647139601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364552296119278e-05,
+      "loss": 0.8731,
+      "step": 14557
+    },
+    {
+      "epoch": 0.3998242289418033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.36446946446568e-05,
+      "loss": 0.9465,
+      "step": 14558
+    },
+    {
+      "epoch": 0.3998516931696465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3643866299282578e-05,
+      "loss": 1.0053,
+      "step": 14559
+    },
+    {
+      "epoch": 0.39987915739748975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3643037925076675e-05,
+      "loss": 0.9253,
+      "step": 14560
+    },
+    {
+      "epoch": 0.399906621625333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3642209522045646e-05,
+      "loss": 0.9246,
+      "step": 14561
+    },
+    {
+      "epoch": 0.39993408585317625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3641381090196047e-05,
+      "loss": 0.9071,
+      "step": 14562
+    },
+    {
+      "epoch": 0.39996155008101947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3640552629534426e-05,
+      "loss": 0.9312,
+      "step": 14563
+    },
+    {
+      "epoch": 0.3999890143088627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3639724140067343e-05,
+      "loss": 0.9294,
+      "step": 14564
+    },
+    {
+      "epoch": 0.40001647853670597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3638895621801351e-05,
+      "loss": 0.8876,
+      "step": 14565
+    },
+    {
+      "epoch": 0.4000439427645492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3638067074743013e-05,
+      "loss": 0.9543,
+      "step": 14566
+    },
+    {
+      "epoch": 0.4000714069923924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3637238498898875e-05,
+      "loss": 0.9683,
+      "step": 14567
+    },
+    {
+      "epoch": 0.40009887122023563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3636409894275495e-05,
+      "loss": 0.8966,
+      "step": 14568
+    },
+    {
+      "epoch": 0.40012633544807885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3635581260879436e-05,
+      "loss": 0.8762,
+      "step": 14569
+    },
+    {
+      "epoch": 0.40015379967592213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3634752598717249e-05,
+      "loss": 0.9526,
+      "step": 14570
+    },
+    {
+      "epoch": 0.40018126390376535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.363392390779549e-05,
+      "loss": 0.9235,
+      "step": 14571
+    },
+    {
+      "epoch": 0.40020872813160857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3633095188120718e-05,
+      "loss": 0.9523,
+      "step": 14572
+    },
+    {
+      "epoch": 0.4002361923594518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3632266439699489e-05,
+      "loss": 0.9105,
+      "step": 14573
+    },
+    {
+      "epoch": 0.40026365658729507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.363143766253836e-05,
+      "loss": 0.9194,
+      "step": 14574
+    },
+    {
+      "epoch": 0.4002911208151383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3630608856643893e-05,
+      "loss": 0.8873,
+      "step": 14575
+    },
+    {
+      "epoch": 0.4003185850429815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3629780022022638e-05,
+      "loss": 0.8984,
+      "step": 14576
+    },
+    {
+      "epoch": 0.40034604927082473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3628951158681159e-05,
+      "loss": 0.8841,
+      "step": 14577
+    },
+    {
+      "epoch": 0.400373513498668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3628122266626017e-05,
+      "loss": 1.0032,
+      "step": 14578
+    },
+    {
+      "epoch": 0.40040097772651123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.362729334586376e-05,
+      "loss": 0.9534,
+      "step": 14579
+    },
+    {
+      "epoch": 0.40042844195435445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3626464396400954e-05,
+      "loss": 0.9775,
+      "step": 14580
+    },
+    {
+      "epoch": 0.4004559061821977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3625635418244159e-05,
+      "loss": 0.9067,
+      "step": 14581
+    },
+    {
+      "epoch": 0.4004833704100409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3624806411399933e-05,
+      "loss": 1.025,
+      "step": 14582
+    },
+    {
+      "epoch": 0.40051083463788417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3623977375874834e-05,
+      "loss": 0.8681,
+      "step": 14583
+    },
+    {
+      "epoch": 0.4005382988657274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.362314831167542e-05,
+      "loss": 0.9651,
+      "step": 14584
+    },
+    {
+      "epoch": 0.4005657630935706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3622319218808253e-05,
+      "loss": 0.9362,
+      "step": 14585
+    },
+    {
+      "epoch": 0.40059322732141384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3621490097279895e-05,
+      "loss": 0.905,
+      "step": 14586
+    },
+    {
+      "epoch": 0.4006206915492571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3620660947096904e-05,
+      "loss": 0.9079,
+      "step": 14587
+    },
+    {
+      "epoch": 0.40064815577710033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3619831768265842e-05,
+      "loss": 0.9711,
+      "step": 14588
+    },
+    {
+      "epoch": 0.40067562000494356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3619002560793267e-05,
+      "loss": 0.9853,
+      "step": 14589
+    },
+    {
+      "epoch": 0.4007030842327868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3618173324685744e-05,
+      "loss": 0.9332,
+      "step": 14590
+    },
+    {
+      "epoch": 0.40073054846063005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361734405994983e-05,
+      "loss": 0.9106,
+      "step": 14591
+    },
+    {
+      "epoch": 0.4007580126884733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3616514766592087e-05,
+      "loss": 0.9225,
+      "step": 14592
+    },
+    {
+      "epoch": 0.4007854769163165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3615685444619083e-05,
+      "loss": 0.8121,
+      "step": 14593
+    },
+    {
+      "epoch": 0.4008129411441597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3614856094037375e-05,
+      "loss": 1.0622,
+      "step": 14594
+    },
+    {
+      "epoch": 0.40084040537200294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3614026714853522e-05,
+      "loss": 1.0061,
+      "step": 14595
+    },
+    {
+      "epoch": 0.4008678695998462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3613197307074091e-05,
+      "loss": 0.9074,
+      "step": 14596
+    },
+    {
+      "epoch": 0.40089533382768944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3612367870705643e-05,
+      "loss": 0.9094,
+      "step": 14597
+    },
+    {
+      "epoch": 0.40092279805553266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361153840575474e-05,
+      "loss": 1.0256,
+      "step": 14598
+    },
+    {
+      "epoch": 0.4009502622833759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361070891222795e-05,
+      "loss": 0.8719,
+      "step": 14599
+    },
+    {
+      "epoch": 0.40097772651121916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3609879390131828e-05,
+      "loss": 0.9883,
+      "step": 14600
+    },
+    {
+      "epoch": 0.4010051907390624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3609049839472943e-05,
+      "loss": 0.947,
+      "step": 14601
+    },
+    {
+      "epoch": 0.4010326549669056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.360822026025786e-05,
+      "loss": 0.8661,
+      "step": 14602
+    },
+    {
+      "epoch": 0.4010601191947488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3607390652493139e-05,
+      "loss": 0.9072,
+      "step": 14603
+    },
+    {
+      "epoch": 0.4010875834225921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3606561016185344e-05,
+      "loss": 0.8376,
+      "step": 14604
+    },
+    {
+      "epoch": 0.4011150476504353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3605731351341044e-05,
+      "loss": 0.9268,
+      "step": 14605
+    },
+    {
+      "epoch": 0.40114251187827854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3604901657966803e-05,
+      "loss": 0.8692,
+      "step": 14606
+    },
+    {
+      "epoch": 0.40116997610612176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.360407193606918e-05,
+      "loss": 0.8965,
+      "step": 14607
+    },
+    {
+      "epoch": 0.401197440333965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3603242185654743e-05,
+      "loss": 0.8382,
+      "step": 14608
+    },
+    {
+      "epoch": 0.40122490456180826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.360241240673006e-05,
+      "loss": 0.8848,
+      "step": 14609
+    },
+    {
+      "epoch": 0.4012523687896515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3601582599301693e-05,
+      "loss": 0.8503,
+      "step": 14610
+    },
+    {
+      "epoch": 0.4012798330174947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3600752763376213e-05,
+      "loss": 0.8995,
+      "step": 14611
+    },
+    {
+      "epoch": 0.4013072972453379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3599922898960177e-05,
+      "loss": 0.8988,
+      "step": 14612
+    },
+    {
+      "epoch": 0.4013347614731812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3599093006060157e-05,
+      "loss": 0.9988,
+      "step": 14613
+    },
+    {
+      "epoch": 0.4013622257010244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3598263084682726e-05,
+      "loss": 0.8737,
+      "step": 14614
+    },
+    {
+      "epoch": 0.40138968992886764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3597433134834436e-05,
+      "loss": 0.9734,
+      "step": 14615
+    },
+    {
+      "epoch": 0.40141715415671086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3596603156521866e-05,
+      "loss": 0.881,
+      "step": 14616
+    },
+    {
+      "epoch": 0.40144461838455414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3595773149751577e-05,
+      "loss": 0.8813,
+      "step": 14617
+    },
+    {
+      "epoch": 0.40147208261239736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3594943114530139e-05,
+      "loss": 1.0187,
+      "step": 14618
+    },
+    {
+      "epoch": 0.4014995468402406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3594113050864115e-05,
+      "loss": 1.0738,
+      "step": 14619
+    },
+    {
+      "epoch": 0.4015270110680838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3593282958760081e-05,
+      "loss": 0.8013,
+      "step": 14620
+    },
+    {
+      "epoch": 0.4015544752959271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3592452838224597e-05,
+      "loss": 0.9494,
+      "step": 14621
+    },
+    {
+      "epoch": 0.4015819395237703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3591622689264237e-05,
+      "loss": 0.9232,
+      "step": 14622
+    },
+    {
+      "epoch": 0.4016094037516135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3590792511885566e-05,
+      "loss": 1.0724,
+      "step": 14623
+    },
+    {
+      "epoch": 0.40163686797945675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3589962306095152e-05,
+      "loss": 0.951,
+      "step": 14624
+    },
+    {
+      "epoch": 0.40166433220729997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358913207189957e-05,
+      "loss": 0.9772,
+      "step": 14625
+    },
+    {
+      "epoch": 0.40169179643514324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358830180930538e-05,
+      "loss": 0.9257,
+      "step": 14626
+    },
+    {
+      "epoch": 0.40171926066298647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3587471518319158e-05,
+      "loss": 0.9695,
+      "step": 14627
+    },
+    {
+      "epoch": 0.4017467248908297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3586641198947472e-05,
+      "loss": 0.914,
+      "step": 14628
+    },
+    {
+      "epoch": 0.4017741891186729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358581085119689e-05,
+      "loss": 0.8054,
+      "step": 14629
+    },
+    {
+      "epoch": 0.4018016533465162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3584980475073985e-05,
+      "loss": 0.858,
+      "step": 14630
+    },
+    {
+      "epoch": 0.4018291175743594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3584150070585327e-05,
+      "loss": 0.8956,
+      "step": 14631
+    },
+    {
+      "epoch": 0.40185658180220263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3583319637737482e-05,
+      "loss": 0.989,
+      "step": 14632
+    },
+    {
+      "epoch": 0.40188404603004585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3582489176537025e-05,
+      "loss": 0.9408,
+      "step": 14633
+    },
+    {
+      "epoch": 0.4019115102578891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3581658686990528e-05,
+      "loss": 0.9422,
+      "step": 14634
+    },
+    {
+      "epoch": 0.40193897448573235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3580828169104558e-05,
+      "loss": 0.9169,
+      "step": 14635
+    },
+    {
+      "epoch": 0.40196643871357557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.357999762288569e-05,
+      "loss": 0.9061,
+      "step": 14636
+    },
+    {
+      "epoch": 0.4019939029414188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3579167048340494e-05,
+      "loss": 1.0094,
+      "step": 14637
+    },
+    {
+      "epoch": 0.402021367169262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3578336445475542e-05,
+      "loss": 0.944,
+      "step": 14638
+    },
+    {
+      "epoch": 0.4020488313971053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3577505814297403e-05,
+      "loss": 0.9514,
+      "step": 14639
+    },
+    {
+      "epoch": 0.4020762956249485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3576675154812656e-05,
+      "loss": 0.8688,
+      "step": 14640
+    },
+    {
+      "epoch": 0.40210375985279173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3575844467027865e-05,
+      "loss": 0.9274,
+      "step": 14641
+    },
+    {
+      "epoch": 0.40213122408063495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3575013750949613e-05,
+      "loss": 0.997,
+      "step": 14642
+    },
+    {
+      "epoch": 0.40215868830847823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3574183006584465e-05,
+      "loss": 0.9897,
+      "step": 14643
+    },
+    {
+      "epoch": 0.40218615253632145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3573352233938994e-05,
+      "loss": 0.9627,
+      "step": 14644
+    },
+    {
+      "epoch": 0.40221361676416467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3572521433019778e-05,
+      "loss": 0.9152,
+      "step": 14645
+    },
+    {
+      "epoch": 0.4022410809920079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3571690603833391e-05,
+      "loss": 1.0089,
+      "step": 14646
+    },
+    {
+      "epoch": 0.40226854521985117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3570859746386402e-05,
+      "loss": 0.9147,
+      "step": 14647
+    },
+    {
+      "epoch": 0.4022960094476944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3570028860685387e-05,
+      "loss": 0.9835,
+      "step": 14648
+    },
+    {
+      "epoch": 0.4023234736755376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356919794673692e-05,
+      "loss": 0.896,
+      "step": 14649
+    },
+    {
+      "epoch": 0.40235093790338083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356836700454758e-05,
+      "loss": 0.9014,
+      "step": 14650
+    },
+    {
+      "epoch": 0.40237840213122406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3567536034123934e-05,
+      "loss": 0.9242,
+      "step": 14651
+    },
+    {
+      "epoch": 0.40240586635906733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3566705035472563e-05,
+      "loss": 0.9759,
+      "step": 14652
+    },
+    {
+      "epoch": 0.40243333058691055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356587400860004e-05,
+      "loss": 0.9341,
+      "step": 14653
+    },
+    {
+      "epoch": 0.4024607948147538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3565042953512942e-05,
+      "loss": 0.8542,
+      "step": 14654
+    },
+    {
+      "epoch": 0.402488259042597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356421187021784e-05,
+      "loss": 0.8729,
+      "step": 14655
+    },
+    {
+      "epoch": 0.4025157232704403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3563380758721314e-05,
+      "loss": 0.9952,
+      "step": 14656
+    },
+    {
+      "epoch": 0.4025431874982835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3562549619029938e-05,
+      "loss": 0.8955,
+      "step": 14657
+    },
+    {
+      "epoch": 0.4025706517261267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3561718451150294e-05,
+      "loss": 0.9979,
+      "step": 14658
+    },
+    {
+      "epoch": 0.40259811595396994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356088725508895e-05,
+      "loss": 0.9643,
+      "step": 14659
+    },
+    {
+      "epoch": 0.4026255801818132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3560056030852486e-05,
+      "loss": 0.9696,
+      "step": 14660
+    },
+    {
+      "epoch": 0.40265304440965644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3559224778447485e-05,
+      "loss": 0.9863,
+      "step": 14661
+    },
+    {
+      "epoch": 0.40268050863749966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3558393497880515e-05,
+      "loss": 0.9962,
+      "step": 14662
+    },
+    {
+      "epoch": 0.4027079728653429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3557562189158156e-05,
+      "loss": 0.9272,
+      "step": 14663
+    },
+    {
+      "epoch": 0.4027354370931861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.355673085228699e-05,
+      "loss": 0.9124,
+      "step": 14664
+    },
+    {
+      "epoch": 0.4027629013210294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3555899487273588e-05,
+      "loss": 0.9456,
+      "step": 14665
+    },
+    {
+      "epoch": 0.4027903655488726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3555068094124535e-05,
+      "loss": 0.9981,
+      "step": 14666
+    },
+    {
+      "epoch": 0.4028178297767158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3554236672846405e-05,
+      "loss": 0.9622,
+      "step": 14667
+    },
+    {
+      "epoch": 0.40284529400455904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3553405223445777e-05,
+      "loss": 0.972,
+      "step": 14668
+    },
+    {
+      "epoch": 0.4028727582324023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.355257374592923e-05,
+      "loss": 0.9302,
+      "step": 14669
+    },
+    {
+      "epoch": 0.40290022246024554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3551742240303346e-05,
+      "loss": 0.9119,
+      "step": 14670
+    },
+    {
+      "epoch": 0.40292768668808876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.35509107065747e-05,
+      "loss": 0.8936,
+      "step": 14671
+    },
+    {
+      "epoch": 0.402955150915932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3550079144749872e-05,
+      "loss": 1.0442,
+      "step": 14672
+    },
+    {
+      "epoch": 0.40298261514377526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3549247554835445e-05,
+      "loss": 0.9304,
+      "step": 14673
+    },
+    {
+      "epoch": 0.4030100793716185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3548415936837998e-05,
+      "loss": 0.8588,
+      "step": 14674
+    },
+    {
+      "epoch": 0.4030375435994617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3547584290764103e-05,
+      "loss": 0.9831,
+      "step": 14675
+    },
+    {
+      "epoch": 0.4030650078273049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3546752616620354e-05,
+      "loss": 0.9267,
+      "step": 14676
+    },
+    {
+      "epoch": 0.40309247205514814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.354592091441332e-05,
+      "loss": 0.9438,
+      "step": 14677
+    },
+    {
+      "epoch": 0.4031199362829914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3545089184149588e-05,
+      "loss": 0.991,
+      "step": 14678
+    },
+    {
+      "epoch": 0.40314740051083464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3544257425835737e-05,
+      "loss": 0.9568,
+      "step": 14679
+    },
+    {
+      "epoch": 0.40317486473867786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3543425639478347e-05,
+      "loss": 0.9706,
+      "step": 14680
+    },
+    {
+      "epoch": 0.4032023289665211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3542593825084e-05,
+      "loss": 0.8414,
+      "step": 14681
+    },
+    {
+      "epoch": 0.40322979319436436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3541761982659283e-05,
+      "loss": 0.8565,
+      "step": 14682
+    },
+    {
+      "epoch": 0.4032572574222076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3540930112210769e-05,
+      "loss": 0.9142,
+      "step": 14683
+    },
+    {
+      "epoch": 0.4032847216500508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3540098213745045e-05,
+      "loss": 0.9921,
+      "step": 14684
+    },
+    {
+      "epoch": 0.403312185877894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3539266287268693e-05,
+      "loss": 0.8353,
+      "step": 14685
+    },
+    {
+      "epoch": 0.4033396501057373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3538434332788295e-05,
+      "loss": 0.907,
+      "step": 14686
+    },
+    {
+      "epoch": 0.4033671143335805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3537602350310434e-05,
+      "loss": 0.9766,
+      "step": 14687
+    },
+    {
+      "epoch": 0.40339457856142374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3536770339841694e-05,
+      "loss": 1.0555,
+      "step": 14688
+    },
+    {
+      "epoch": 0.40342204278926697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3535938301388654e-05,
+      "loss": 0.9829,
+      "step": 14689
+    },
+    {
+      "epoch": 0.4034495070171102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3535106234957905e-05,
+      "loss": 0.9168,
+      "step": 14690
+    },
+    {
+      "epoch": 0.40347697124495346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3534274140556023e-05,
+      "loss": 0.9181,
+      "step": 14691
+    },
+    {
+      "epoch": 0.4035044354727967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3533442018189595e-05,
+      "loss": 0.9914,
+      "step": 14692
+    },
+    {
+      "epoch": 0.4035318997006399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3532609867865207e-05,
+      "loss": 0.9686,
+      "step": 14693
+    },
+    {
+      "epoch": 0.40355936392848313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.353177768958944e-05,
+      "loss": 0.9195,
+      "step": 14694
+    },
+    {
+      "epoch": 0.4035868281563264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.353094548336888e-05,
+      "loss": 0.9216,
+      "step": 14695
+    },
+    {
+      "epoch": 0.4036142923841696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.353011324921011e-05,
+      "loss": 0.8743,
+      "step": 14696
+    },
+    {
+      "epoch": 0.40364175661201285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.352928098711972e-05,
+      "loss": 0.9507,
+      "step": 14697
+    },
+    {
+      "epoch": 0.40366922083985607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3528448697104289e-05,
+      "loss": 0.9605,
+      "step": 14698
+    },
+    {
+      "epoch": 0.40369668506769935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3527616379170407e-05,
+      "loss": 0.8823,
+      "step": 14699
+    },
+    {
+      "epoch": 0.40372414929554257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3526784033324653e-05,
+      "loss": 0.8286,
+      "step": 14700
+    },
+    {
+      "epoch": 0.4037516135233858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3525951659573622e-05,
+      "loss": 1.0121,
+      "step": 14701
+    },
+    {
+      "epoch": 0.403779077751229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3525119257923898e-05,
+      "loss": 0.9566,
+      "step": 14702
+    },
+    {
+      "epoch": 0.40380654197907223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3524286828382061e-05,
+      "loss": 0.9392,
+      "step": 14703
+    },
+    {
+      "epoch": 0.4038340062069155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3523454370954699e-05,
+      "loss": 0.9838,
+      "step": 14704
+    },
+    {
+      "epoch": 0.40386147043475873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3522621885648406e-05,
+      "loss": 0.9158,
+      "step": 14705
+    },
+    {
+      "epoch": 0.40388893466260195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3521789372469762e-05,
+      "loss": 0.9268,
+      "step": 14706
+    },
+    {
+      "epoch": 0.40391639889044517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3520956831425357e-05,
+      "loss": 0.9612,
+      "step": 14707
+    },
+    {
+      "epoch": 0.40394386311828845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3520124262521779e-05,
+      "loss": 0.8965,
+      "step": 14708
+    },
+    {
+      "epoch": 0.40397132734613167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3519291665765611e-05,
+      "loss": 0.9734,
+      "step": 14709
+    },
+    {
+      "epoch": 0.4039987915739749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3518459041163445e-05,
+      "loss": 1.0494,
+      "step": 14710
+    },
+    {
+      "epoch": 0.4040262558018181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3517626388721872e-05,
+      "loss": 0.9147,
+      "step": 14711
+    },
+    {
+      "epoch": 0.4040537200296614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3516793708447471e-05,
+      "loss": 0.8153,
+      "step": 14712
+    },
+    {
+      "epoch": 0.4040811842575046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3515961000346835e-05,
+      "loss": 0.9356,
+      "step": 14713
+    },
+    {
+      "epoch": 0.40410864848534783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3515128264426563e-05,
+      "loss": 0.9578,
+      "step": 14714
+    },
+    {
+      "epoch": 0.40413611271319105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3514295500693227e-05,
+      "loss": 0.8931,
+      "step": 14715
+    },
+    {
+      "epoch": 0.40416357694103433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3513462709153422e-05,
+      "loss": 0.9159,
+      "step": 14716
+    },
+    {
+      "epoch": 0.40419104116887755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3512629889813745e-05,
+      "loss": 0.8843,
+      "step": 14717
+    },
+    {
+      "epoch": 0.4042185053967208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3511797042680774e-05,
+      "loss": 0.899,
+      "step": 14718
+    },
+    {
+      "epoch": 0.404245969624564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3510964167761107e-05,
+      "loss": 0.8536,
+      "step": 14719
+    },
+    {
+      "epoch": 0.4042734338524072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3510131265061332e-05,
+      "loss": 0.9225,
+      "step": 14720
+    },
+    {
+      "epoch": 0.4043008980802505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3509298334588037e-05,
+      "loss": 0.987,
+      "step": 14721
+    },
+    {
+      "epoch": 0.4043283623080937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3508465376347816e-05,
+      "loss": 0.9138,
+      "step": 14722
+    },
+    {
+      "epoch": 0.40435582653593694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350763239034726e-05,
+      "loss": 0.9426,
+      "step": 14723
+    },
+    {
+      "epoch": 0.40438329076378016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3506799376592954e-05,
+      "loss": 0.9473,
+      "step": 14724
+    },
+    {
+      "epoch": 0.40441075499162343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3505966335091493e-05,
+      "loss": 0.8757,
+      "step": 14725
+    },
+    {
+      "epoch": 0.40443821921946665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3505133265849472e-05,
+      "loss": 0.9469,
+      "step": 14726
+    },
+    {
+      "epoch": 0.4044656834473099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3504300168873474e-05,
+      "loss": 0.9228,
+      "step": 14727
+    },
+    {
+      "epoch": 0.4044931476751531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3503467044170098e-05,
+      "loss": 1.0013,
+      "step": 14728
+    },
+    {
+      "epoch": 0.4045206119029964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3502633891745935e-05,
+      "loss": 0.9107,
+      "step": 14729
+    },
+    {
+      "epoch": 0.4045480761308396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3501800711607575e-05,
+      "loss": 0.9139,
+      "step": 14730
+    },
+    {
+      "epoch": 0.4045755403586828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350096750376161e-05,
+      "loss": 0.9595,
+      "step": 14731
+    },
+    {
+      "epoch": 0.40460300458652604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3500134268214636e-05,
+      "loss": 0.9128,
+      "step": 14732
+    },
+    {
+      "epoch": 0.40463046881436926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349930100497324e-05,
+      "loss": 0.9752,
+      "step": 14733
+    },
+    {
+      "epoch": 0.40465793304221254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3498467714044024e-05,
+      "loss": 0.9741,
+      "step": 14734
+    },
+    {
+      "epoch": 0.40468539727005576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3497634395433575e-05,
+      "loss": 0.9946,
+      "step": 14735
+    },
+    {
+      "epoch": 0.404712861497899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3496801049148484e-05,
+      "loss": 0.8318,
+      "step": 14736
+    },
+    {
+      "epoch": 0.4047403257257422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349596767519535e-05,
+      "loss": 0.914,
+      "step": 14737
+    },
+    {
+      "epoch": 0.4047677899535855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349513427358077e-05,
+      "loss": 0.9305,
+      "step": 14738
+    },
+    {
+      "epoch": 0.4047952541814287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3494300844311328e-05,
+      "loss": 0.8064,
+      "step": 14739
+    },
+    {
+      "epoch": 0.4048227184092719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3493467387393626e-05,
+      "loss": 0.9443,
+      "step": 14740
+    },
+    {
+      "epoch": 0.40485018263711514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3492633902834259e-05,
+      "loss": 0.8201,
+      "step": 14741
+    },
+    {
+      "epoch": 0.4048776468649584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349180039063982e-05,
+      "loss": 1.013,
+      "step": 14742
+    },
+    {
+      "epoch": 0.40490511109280164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34909668508169e-05,
+      "loss": 0.8427,
+      "step": 14743
+    },
+    {
+      "epoch": 0.40493257532064486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34901332833721e-05,
+      "loss": 0.9433,
+      "step": 14744
+    },
+    {
+      "epoch": 0.4049600395484881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3489299688312013e-05,
+      "loss": 0.8449,
+      "step": 14745
+    },
+    {
+      "epoch": 0.4049875037763313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3488466065643236e-05,
+      "loss": 1.0219,
+      "step": 14746
+    },
+    {
+      "epoch": 0.4050149680041746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3487632415372364e-05,
+      "loss": 0.8079,
+      "step": 14747
+    },
+    {
+      "epoch": 0.4050424322320178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3486798737505992e-05,
+      "loss": 0.798,
+      "step": 14748
+    },
+    {
+      "epoch": 0.405069896459861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3485965032050719e-05,
+      "loss": 0.8872,
+      "step": 14749
+    },
+    {
+      "epoch": 0.40509736068770424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3485131299013141e-05,
+      "loss": 0.8389,
+      "step": 14750
+    },
+    {
+      "epoch": 0.4051248249155475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3484297538399851e-05,
+      "loss": 0.9835,
+      "step": 14751
+    },
+    {
+      "epoch": 0.40515228914339074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3483463750217453e-05,
+      "loss": 0.9404,
+      "step": 14752
+    },
+    {
+      "epoch": 0.40517975337123396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3482629934472537e-05,
+      "loss": 0.803,
+      "step": 14753
+    },
+    {
+      "epoch": 0.4052072175990772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3481796091171709e-05,
+      "loss": 0.9686,
+      "step": 14754
+    },
+    {
+      "epoch": 0.40523468182692046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3480962220321558e-05,
+      "loss": 0.8142,
+      "step": 14755
+    },
+    {
+      "epoch": 0.4052621460547637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3480128321928685e-05,
+      "loss": 0.8621,
+      "step": 14756
+    },
+    {
+      "epoch": 0.4052896102826069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3479294395999689e-05,
+      "loss": 0.9014,
+      "step": 14757
+    },
+    {
+      "epoch": 0.4053170745104501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3478460442541168e-05,
+      "loss": 0.9399,
+      "step": 14758
+    },
+    {
+      "epoch": 0.40534453873829335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3477626461559722e-05,
+      "loss": 0.8934,
+      "step": 14759
+    },
+    {
+      "epoch": 0.4053720029661366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3476792453061946e-05,
+      "loss": 0.8863,
+      "step": 14760
+    },
+    {
+      "epoch": 0.40539946719397985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3475958417054445e-05,
+      "loss": 0.9594,
+      "step": 14761
+    },
+    {
+      "epoch": 0.40542693142182307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3475124353543813e-05,
+      "loss": 0.9328,
+      "step": 14762
+    },
+    {
+      "epoch": 0.4054543956496663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3474290262536649e-05,
+      "loss": 0.9666,
+      "step": 14763
+    },
+    {
+      "epoch": 0.40548185987750957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3473456144039556e-05,
+      "loss": 0.9543,
+      "step": 14764
+    },
+    {
+      "epoch": 0.4055093241053528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3472621998059133e-05,
+      "loss": 0.9217,
+      "step": 14765
+    },
+    {
+      "epoch": 0.405536788333196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3471787824601978e-05,
+      "loss": 0.9253,
+      "step": 14766
+    },
+    {
+      "epoch": 0.40556425256103923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3470953623674696e-05,
+      "loss": 0.9559,
+      "step": 14767
+    },
+    {
+      "epoch": 0.4055917167888825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.347011939528388e-05,
+      "loss": 0.9265,
+      "step": 14768
+    },
+    {
+      "epoch": 0.4056191810167257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3469285139436137e-05,
+      "loss": 0.9011,
+      "step": 14769
+    },
+    {
+      "epoch": 0.40564664524456895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3468450856138068e-05,
+      "loss": 1.0051,
+      "step": 14770
+    },
+    {
+      "epoch": 0.40567410947241217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3467616545396271e-05,
+      "loss": 0.9444,
+      "step": 14771
+    },
+    {
+      "epoch": 0.4057015737002554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3466782207217348e-05,
+      "loss": 0.8771,
+      "step": 14772
+    },
+    {
+      "epoch": 0.40572903792809867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3465947841607902e-05,
+      "loss": 0.8987,
+      "step": 14773
+    },
+    {
+      "epoch": 0.4057565021559419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3465113448574534e-05,
+      "loss": 0.9783,
+      "step": 14774
+    },
+    {
+      "epoch": 0.4057839663837851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3464279028123845e-05,
+      "loss": 0.9919,
+      "step": 14775
+    },
+    {
+      "epoch": 0.40581143061162833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3463444580262442e-05,
+      "loss": 0.9819,
+      "step": 14776
+    },
+    {
+      "epoch": 0.4058388948394716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3462610104996921e-05,
+      "loss": 1.0641,
+      "step": 14777
+    },
+    {
+      "epoch": 0.40586635906731483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3461775602333886e-05,
+      "loss": 1.027,
+      "step": 14778
+    },
+    {
+      "epoch": 0.40589382329515805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3460941072279944e-05,
+      "loss": 0.9332,
+      "step": 14779
+    },
+    {
+      "epoch": 0.4059212875230013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3460106514841696e-05,
+      "loss": 1.0189,
+      "step": 14780
+    },
+    {
+      "epoch": 0.40594875175084455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3459271930025741e-05,
+      "loss": 0.894,
+      "step": 14781
+    },
+    {
+      "epoch": 0.40597621597868777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3458437317838693e-05,
+      "loss": 0.9458,
+      "step": 14782
+    },
+    {
+      "epoch": 0.406003680206531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3457602678287145e-05,
+      "loss": 0.9101,
+      "step": 14783
+    },
+    {
+      "epoch": 0.4060311444343742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3456768011377706e-05,
+      "loss": 1.0184,
+      "step": 14784
+    },
+    {
+      "epoch": 0.40605860866221744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3455933317116978e-05,
+      "loss": 0.8928,
+      "step": 14785
+    },
+    {
+      "epoch": 0.4060860728900607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.345509859551157e-05,
+      "loss": 0.869,
+      "step": 14786
+    },
+    {
+      "epoch": 0.40611353711790393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.345426384656808e-05,
+      "loss": 0.9793,
+      "step": 14787
+    },
+    {
+      "epoch": 0.40614100134574715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3453429070293118e-05,
+      "loss": 0.9316,
+      "step": 14788
+    },
+    {
+      "epoch": 0.4061684655735904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3452594266693288e-05,
+      "loss": 0.8089,
+      "step": 14789
+    },
+    {
+      "epoch": 0.40619592980143365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3451759435775195e-05,
+      "loss": 0.7872,
+      "step": 14790
+    },
+    {
+      "epoch": 0.4062233940292769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3450924577545443e-05,
+      "loss": 0.8942,
+      "step": 14791
+    },
+    {
+      "epoch": 0.4062508582571201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.345008969201064e-05,
+      "loss": 0.8418,
+      "step": 14792
+    },
+    {
+      "epoch": 0.4062783224849633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344925477917739e-05,
+      "loss": 0.9551,
+      "step": 14793
+    },
+    {
+      "epoch": 0.4063057867128066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3448419839052303e-05,
+      "loss": 1.0002,
+      "step": 14794
+    },
+    {
+      "epoch": 0.4063332509406498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344758487164198e-05,
+      "loss": 0.8857,
+      "step": 14795
+    },
+    {
+      "epoch": 0.40636071516849304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3446749876953028e-05,
+      "loss": 0.8898,
+      "step": 14796
+    },
+    {
+      "epoch": 0.40638817939633626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344591485499206e-05,
+      "loss": 0.8986,
+      "step": 14797
+    },
+    {
+      "epoch": 0.4064156436241795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3445079805765677e-05,
+      "loss": 0.9714,
+      "step": 14798
+    },
+    {
+      "epoch": 0.40644310785202276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3444244729280485e-05,
+      "loss": 1.0332,
+      "step": 14799
+    },
+    {
+      "epoch": 0.406470572079866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34434096255431e-05,
+      "loss": 0.9939,
+      "step": 14800
+    },
+    {
+      "epoch": 0.4064980363077092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3442574494560118e-05,
+      "loss": 0.8486,
+      "step": 14801
+    },
+    {
+      "epoch": 0.4065255005355524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344173933633816e-05,
+      "loss": 0.8978,
+      "step": 14802
+    },
+    {
+      "epoch": 0.4065529647633957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3440904150883825e-05,
+      "loss": 0.9402,
+      "step": 14803
+    },
+    {
+      "epoch": 0.4065804289912389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3440068938203719e-05,
+      "loss": 0.9524,
+      "step": 14804
+    },
+    {
+      "epoch": 0.40660789321908214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3439233698304462e-05,
+      "loss": 0.8537,
+      "step": 14805
+    },
+    {
+      "epoch": 0.40663535744692536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3438398431192652e-05,
+      "loss": 0.9352,
+      "step": 14806
+    },
+    {
+      "epoch": 0.40666282167476864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3437563136874899e-05,
+      "loss": 0.9079,
+      "step": 14807
+    },
+    {
+      "epoch": 0.40669028590261186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343672781535782e-05,
+      "loss": 0.8872,
+      "step": 14808
+    },
+    {
+      "epoch": 0.4067177501304551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3435892466648016e-05,
+      "loss": 0.9064,
+      "step": 14809
+    },
+    {
+      "epoch": 0.4067452143582983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3435057090752101e-05,
+      "loss": 0.8955,
+      "step": 14810
+    },
+    {
+      "epoch": 0.4067726785861416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3434221687676683e-05,
+      "loss": 0.916,
+      "step": 14811
+    },
+    {
+      "epoch": 0.4068001428139848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3433386257428374e-05,
+      "loss": 1.0967,
+      "step": 14812
+    },
+    {
+      "epoch": 0.406827607041828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343255080001378e-05,
+      "loss": 0.9727,
+      "step": 14813
+    },
+    {
+      "epoch": 0.40685507126967124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3431715315439518e-05,
+      "loss": 0.9999,
+      "step": 14814
+    },
+    {
+      "epoch": 0.40688253549751446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3430879803712194e-05,
+      "loss": 0.8996,
+      "step": 14815
+    },
+    {
+      "epoch": 0.40690999972535774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3430044264838418e-05,
+      "loss": 1.0466,
+      "step": 14816
+    },
+    {
+      "epoch": 0.40693746395320096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3429208698824808e-05,
+      "loss": 0.9245,
+      "step": 14817
+    },
+    {
+      "epoch": 0.4069649281810442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3428373105677965e-05,
+      "loss": 0.9047,
+      "step": 14818
+    },
+    {
+      "epoch": 0.4069923924088874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3427537485404507e-05,
+      "loss": 1.0291,
+      "step": 14819
+    },
+    {
+      "epoch": 0.4070198566367307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3426701838011048e-05,
+      "loss": 0.9196,
+      "step": 14820
+    },
+    {
+      "epoch": 0.4070473208645739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3425866163504193e-05,
+      "loss": 0.8086,
+      "step": 14821
+    },
+    {
+      "epoch": 0.4070747850924171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.342503046189056e-05,
+      "loss": 1.0887,
+      "step": 14822
+    },
+    {
+      "epoch": 0.40710224932026035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3424194733176758e-05,
+      "loss": 0.9992,
+      "step": 14823
+    },
+    {
+      "epoch": 0.4071297135481036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3423358977369403e-05,
+      "loss": 0.9428,
+      "step": 14824
+    },
+    {
+      "epoch": 0.40715717777594684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34225231944751e-05,
+      "loss": 0.8782,
+      "step": 14825
+    },
+    {
+      "epoch": 0.40718464200379006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3421687384500474e-05,
+      "loss": 0.8347,
+      "step": 14826
+    },
+    {
+      "epoch": 0.4072121062316333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3420851547452132e-05,
+      "loss": 1.021,
+      "step": 14827
+    },
+    {
+      "epoch": 0.4072395704594765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3420015683336684e-05,
+      "loss": 0.9031,
+      "step": 14828
+    },
+    {
+      "epoch": 0.4072670346873198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341917979216075e-05,
+      "loss": 0.9074,
+      "step": 14829
+    },
+    {
+      "epoch": 0.407294498915163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3418343873930942e-05,
+      "loss": 0.952,
+      "step": 14830
+    },
+    {
+      "epoch": 0.4073219631430062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3417507928653868e-05,
+      "loss": 0.9105,
+      "step": 14831
+    },
+    {
+      "epoch": 0.40734942737084945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341667195633615e-05,
+      "loss": 0.9679,
+      "step": 14832
+    },
+    {
+      "epoch": 0.4073768915986927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34158359569844e-05,
+      "loss": 0.8947,
+      "step": 14833
+    },
+    {
+      "epoch": 0.40740435582653595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3414999930605233e-05,
+      "loss": 0.9265,
+      "step": 14834
+    },
+    {
+      "epoch": 0.40743182005437917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3414163877205264e-05,
+      "loss": 0.9404,
+      "step": 14835
+    },
+    {
+      "epoch": 0.4074592842822224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3413327796791106e-05,
+      "loss": 0.8759,
+      "step": 14836
+    },
+    {
+      "epoch": 0.40748674851006567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3412491689369378e-05,
+      "loss": 1.0662,
+      "step": 14837
+    },
+    {
+      "epoch": 0.4075142127379089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3411655554946694e-05,
+      "loss": 0.9341,
+      "step": 14838
+    },
+    {
+      "epoch": 0.4075416769657521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341081939352967e-05,
+      "loss": 0.8808,
+      "step": 14839
+    },
+    {
+      "epoch": 0.40756914119359533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3409983205124919e-05,
+      "loss": 0.8708,
+      "step": 14840
+    },
+    {
+      "epoch": 0.40759660542143855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3409146989739064e-05,
+      "loss": 0.9516,
+      "step": 14841
+    },
+    {
+      "epoch": 0.40762406964928183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3408310747378717e-05,
+      "loss": 0.9561,
+      "step": 14842
+    },
+    {
+      "epoch": 0.40765153387712505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340747447805049e-05,
+      "loss": 0.9897,
+      "step": 14843
+    },
+    {
+      "epoch": 0.40767899810496827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340663818176101e-05,
+      "loss": 0.8426,
+      "step": 14844
+    },
+    {
+      "epoch": 0.4077064623328115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3405801858516887e-05,
+      "loss": 0.8798,
+      "step": 14845
+    },
+    {
+      "epoch": 0.40773392656065477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3404965508324743e-05,
+      "loss": 0.9819,
+      "step": 14846
+    },
+    {
+      "epoch": 0.407761390788498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340412913119119e-05,
+      "loss": 0.841,
+      "step": 14847
+    },
+    {
+      "epoch": 0.4077888550163412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3403292727122849e-05,
+      "loss": 0.8908,
+      "step": 14848
+    },
+    {
+      "epoch": 0.40781631924418443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3402456296126339e-05,
+      "loss": 0.9213,
+      "step": 14849
+    },
+    {
+      "epoch": 0.4078437834720277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3401619838208276e-05,
+      "loss": 0.9843,
+      "step": 14850
+    },
+    {
+      "epoch": 0.40787124769987093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340078335337528e-05,
+      "loss": 0.8968,
+      "step": 14851
+    },
+    {
+      "epoch": 0.40789871192771415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3399946841633965e-05,
+      "loss": 0.8823,
+      "step": 14852
+    },
+    {
+      "epoch": 0.4079261761555574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3399110302990959e-05,
+      "loss": 1.0624,
+      "step": 14853
+    },
+    {
+      "epoch": 0.4079536403834006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3398273737452874e-05,
+      "loss": 0.9711,
+      "step": 14854
+    },
+    {
+      "epoch": 0.40798110461124387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3397437145026328e-05,
+      "loss": 0.9524,
+      "step": 14855
+    },
+    {
+      "epoch": 0.4080085688390871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3396600525717946e-05,
+      "loss": 0.9739,
+      "step": 14856
+    },
+    {
+      "epoch": 0.4080360330669303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3395763879534345e-05,
+      "loss": 0.9282,
+      "step": 14857
+    },
+    {
+      "epoch": 0.40806349729477354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3394927206482145e-05,
+      "loss": 0.9975,
+      "step": 14858
+    },
+    {
+      "epoch": 0.4080909615226168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3394090506567965e-05,
+      "loss": 1.0131,
+      "step": 14859
+    },
+    {
+      "epoch": 0.40811842575046003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3393253779798425e-05,
+      "loss": 1.0502,
+      "step": 14860
+    },
+    {
+      "epoch": 0.40814588997830326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3392417026180147e-05,
+      "loss": 0.7176,
+      "step": 14861
+    },
+    {
+      "epoch": 0.4081733542061465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3391580245719754e-05,
+      "loss": 0.9135,
+      "step": 14862
+    },
+    {
+      "epoch": 0.40820081843398975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3390743438423864e-05,
+      "loss": 0.899,
+      "step": 14863
+    },
+    {
+      "epoch": 0.408228282661833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3389906604299095e-05,
+      "loss": 0.9382,
+      "step": 14864
+    },
+    {
+      "epoch": 0.4082557468896762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3389069743352075e-05,
+      "loss": 0.9296,
+      "step": 14865
+    },
+    {
+      "epoch": 0.4082832111175194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3388232855589421e-05,
+      "loss": 0.8291,
+      "step": 14866
+    },
+    {
+      "epoch": 0.40831067534536264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3387395941017758e-05,
+      "loss": 0.9152,
+      "step": 14867
+    },
+    {
+      "epoch": 0.4083381395732059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3386558999643702e-05,
+      "loss": 0.9155,
+      "step": 14868
+    },
+    {
+      "epoch": 0.40836560380104914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338572203147388e-05,
+      "loss": 0.9482,
+      "step": 14869
+    },
+    {
+      "epoch": 0.40839306802889236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3384885036514916e-05,
+      "loss": 0.964,
+      "step": 14870
+    },
+    {
+      "epoch": 0.4084205322567356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3384048014773433e-05,
+      "loss": 0.8801,
+      "step": 14871
+    },
+    {
+      "epoch": 0.40844799648457886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3383210966256046e-05,
+      "loss": 0.9064,
+      "step": 14872
+    },
+    {
+      "epoch": 0.4084754607124221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3382373890969387e-05,
+      "loss": 1.058,
+      "step": 14873
+    },
+    {
+      "epoch": 0.4085029249402653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3381536788920076e-05,
+      "loss": 0.9442,
+      "step": 14874
+    },
+    {
+      "epoch": 0.4085303891681085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338069966011473e-05,
+      "loss": 1.0043,
+      "step": 14875
+    },
+    {
+      "epoch": 0.4085578533959518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3379862504559986e-05,
+      "loss": 0.9801,
+      "step": 14876
+    },
+    {
+      "epoch": 0.408585317623795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3379025322262455e-05,
+      "loss": 1.0022,
+      "step": 14877
+    },
+    {
+      "epoch": 0.40861278185163824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3378188113228771e-05,
+      "loss": 0.998,
+      "step": 14878
+    },
+    {
+      "epoch": 0.40864024607948146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3377350877465553e-05,
+      "loss": 0.8922,
+      "step": 14879
+    },
+    {
+      "epoch": 0.4086677103073247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3376513614979425e-05,
+      "loss": 0.8821,
+      "step": 14880
+    },
+    {
+      "epoch": 0.40869517453516796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3375676325777014e-05,
+      "loss": 0.9456,
+      "step": 14881
+    },
+    {
+      "epoch": 0.4087226387630112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3374839009864943e-05,
+      "loss": 0.9184,
+      "step": 14882
+    },
+    {
+      "epoch": 0.4087501029908544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3374001667249842e-05,
+      "loss": 0.858,
+      "step": 14883
+    },
+    {
+      "epoch": 0.4087775672186976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3373164297938331e-05,
+      "loss": 0.9219,
+      "step": 14884
+    },
+    {
+      "epoch": 0.4088050314465409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3372326901937039e-05,
+      "loss": 0.999,
+      "step": 14885
+    },
+    {
+      "epoch": 0.4088324956743841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3371489479252589e-05,
+      "loss": 0.8233,
+      "step": 14886
+    },
+    {
+      "epoch": 0.40885995990222734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3370652029891607e-05,
+      "loss": 0.866,
+      "step": 14887
+    },
+    {
+      "epoch": 0.40888742413007056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3369814553860721e-05,
+      "loss": 0.8646,
+      "step": 14888
+    },
+    {
+      "epoch": 0.40891488835791384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3368977051166556e-05,
+      "loss": 0.9321,
+      "step": 14889
+    },
+    {
+      "epoch": 0.40894235258575706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3368139521815742e-05,
+      "loss": 0.9536,
+      "step": 14890
+    },
+    {
+      "epoch": 0.4089698168136003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3367301965814903e-05,
+      "loss": 0.8663,
+      "step": 14891
+    },
+    {
+      "epoch": 0.4089972810414435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3366464383170663e-05,
+      "loss": 0.9026,
+      "step": 14892
+    },
+    {
+      "epoch": 0.4090247452692867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3365626773889656e-05,
+      "loss": 0.8895,
+      "step": 14893
+    },
+    {
+      "epoch": 0.40905220949713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3364789137978507e-05,
+      "loss": 1.0015,
+      "step": 14894
+    },
+    {
+      "epoch": 0.4090796737249732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.336395147544384e-05,
+      "loss": 1.0124,
+      "step": 14895
+    },
+    {
+      "epoch": 0.40910713795281645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3363113786292287e-05,
+      "loss": 1.0582,
+      "step": 14896
+    },
+    {
+      "epoch": 0.40913460218065967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3362276070530477e-05,
+      "loss": 1.0287,
+      "step": 14897
+    },
+    {
+      "epoch": 0.40916206640850294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3361438328165036e-05,
+      "loss": 0.8999,
+      "step": 14898
+    },
+    {
+      "epoch": 0.40918953063634617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.336060055920259e-05,
+      "loss": 0.8785,
+      "step": 14899
+    },
+    {
+      "epoch": 0.4092169948641894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3359762763649773e-05,
+      "loss": 0.8723,
+      "step": 14900
+    },
+    {
+      "epoch": 0.4092444590920326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3358924941513209e-05,
+      "loss": 0.8851,
+      "step": 14901
+    },
+    {
+      "epoch": 0.4092719233198759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3358087092799532e-05,
+      "loss": 0.904,
+      "step": 14902
+    },
+    {
+      "epoch": 0.4092993875477191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3357249217515371e-05,
+      "loss": 0.9415,
+      "step": 14903
+    },
+    {
+      "epoch": 0.40932685177556233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335641131566735e-05,
+      "loss": 0.9435,
+      "step": 14904
+    },
+    {
+      "epoch": 0.40935431600340555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3355573387262104e-05,
+      "loss": 1.0909,
+      "step": 14905
+    },
+    {
+      "epoch": 0.40938178023124877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335473543230626e-05,
+      "loss": 0.8811,
+      "step": 14906
+    },
+    {
+      "epoch": 0.40940924445909205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3353897450806455e-05,
+      "loss": 0.8143,
+      "step": 14907
+    },
+    {
+      "epoch": 0.40943670868693527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3353059442769308e-05,
+      "loss": 0.8688,
+      "step": 14908
+    },
+    {
+      "epoch": 0.4094641729147785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3352221408201457e-05,
+      "loss": 0.9001,
+      "step": 14909
+    },
+    {
+      "epoch": 0.4094916371426217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3351383347109534e-05,
+      "loss": 0.8661,
+      "step": 14910
+    },
+    {
+      "epoch": 0.409519101370465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3350545259500166e-05,
+      "loss": 0.9187,
+      "step": 14911
+    },
+    {
+      "epoch": 0.4095465655983082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334970714537999e-05,
+      "loss": 1.0014,
+      "step": 14912
+    },
+    {
+      "epoch": 0.40957402982615143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3348869004755626e-05,
+      "loss": 0.9055,
+      "step": 14913
+    },
+    {
+      "epoch": 0.40960149405399465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334803083763372e-05,
+      "loss": 0.9317,
+      "step": 14914
+    },
+    {
+      "epoch": 0.40962895828183793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3347192644020894e-05,
+      "loss": 0.9224,
+      "step": 14915
+    },
+    {
+      "epoch": 0.40965642250968115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3346354423923784e-05,
+      "loss": 0.886,
+      "step": 14916
+    },
+    {
+      "epoch": 0.40968388673752437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3345516177349021e-05,
+      "loss": 0.8917,
+      "step": 14917
+    },
+    {
+      "epoch": 0.4097113509653676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3344677904303238e-05,
+      "loss": 0.9322,
+      "step": 14918
+    },
+    {
+      "epoch": 0.40973881519321087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334383960479307e-05,
+      "loss": 0.8477,
+      "step": 14919
+    },
+    {
+      "epoch": 0.4097662794210541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3343001278825144e-05,
+      "loss": 0.9115,
+      "step": 14920
+    },
+    {
+      "epoch": 0.4097937436488973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.33421629264061e-05,
+      "loss": 0.9493,
+      "step": 14921
+    },
+    {
+      "epoch": 0.40982120787674053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3341324547542566e-05,
+      "loss": 0.9894,
+      "step": 14922
+    },
+    {
+      "epoch": 0.40984867210458376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334048614224118e-05,
+      "loss": 0.8006,
+      "step": 14923
+    },
+    {
+      "epoch": 0.40987613633242703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3339647710508574e-05,
+      "loss": 0.8965,
+      "step": 14924
+    },
+    {
+      "epoch": 0.40990360056027025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3338809252351379e-05,
+      "loss": 0.811,
+      "step": 14925
+    },
+    {
+      "epoch": 0.4099310647881135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3337970767776233e-05,
+      "loss": 0.9699,
+      "step": 14926
+    },
+    {
+      "epoch": 0.4099585290159567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3337132256789772e-05,
+      "loss": 0.9411,
+      "step": 14927
+    },
+    {
+      "epoch": 0.4099859932438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3336293719398623e-05,
+      "loss": 0.9154,
+      "step": 14928
+    },
+    {
+      "epoch": 0.4100134574716432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3335455155609429e-05,
+      "loss": 0.9577,
+      "step": 14929
+    },
+    {
+      "epoch": 0.4100409216994864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3334616565428819e-05,
+      "loss": 0.903,
+      "step": 14930
+    },
+    {
+      "epoch": 0.41006838592732964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3333777948863436e-05,
+      "loss": 0.8911,
+      "step": 14931
+    },
+    {
+      "epoch": 0.4100958501551729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3332939305919907e-05,
+      "loss": 0.8881,
+      "step": 14932
+    },
+    {
+      "epoch": 0.41012331438301614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.333210063660487e-05,
+      "loss": 0.9622,
+      "step": 14933
+    },
+    {
+      "epoch": 0.41015077861085936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3331261940924964e-05,
+      "loss": 0.8771,
+      "step": 14934
+    },
+    {
+      "epoch": 0.4101782428387026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3330423218886824e-05,
+      "loss": 0.9106,
+      "step": 14935
+    },
+    {
+      "epoch": 0.4102057070665458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3329584470497083e-05,
+      "loss": 1.0072,
+      "step": 14936
+    },
+    {
+      "epoch": 0.4102331712943891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332874569576238e-05,
+      "loss": 1.0069,
+      "step": 14937
+    },
+    {
+      "epoch": 0.4102606355222323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3327906894689353e-05,
+      "loss": 0.923,
+      "step": 14938
+    },
+    {
+      "epoch": 0.4102880997500755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3327068067284639e-05,
+      "loss": 0.9383,
+      "step": 14939
+    },
+    {
+      "epoch": 0.41031556397791874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332622921355487e-05,
+      "loss": 0.9664,
+      "step": 14940
+    },
+    {
+      "epoch": 0.410343028205762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332539033350669e-05,
+      "loss": 0.9294,
+      "step": 14941
+    },
+    {
+      "epoch": 0.41037049243360524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3324551427146735e-05,
+      "loss": 0.9704,
+      "step": 14942
+    },
+    {
+      "epoch": 0.41039795666144846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3323712494481638e-05,
+      "loss": 1.009,
+      "step": 14943
+    },
+    {
+      "epoch": 0.4104254208892917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332287353551804e-05,
+      "loss": 0.9226,
+      "step": 14944
+    },
+    {
+      "epoch": 0.41045288511713496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332203455026258e-05,
+      "loss": 0.9972,
+      "step": 14945
+    },
+    {
+      "epoch": 0.4104803493449782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3321195538721896e-05,
+      "loss": 0.8953,
+      "step": 14946
+    },
+    {
+      "epoch": 0.4105078135728214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3320356500902629e-05,
+      "loss": 0.8988,
+      "step": 14947
+    },
+    {
+      "epoch": 0.4105352778006646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3319517436811411e-05,
+      "loss": 0.8977,
+      "step": 14948
+    },
+    {
+      "epoch": 0.41056274202850784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3318678346454885e-05,
+      "loss": 0.866,
+      "step": 14949
+    },
+    {
+      "epoch": 0.4105902062563511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3317839229839693e-05,
+      "loss": 0.8855,
+      "step": 14950
+    },
+    {
+      "epoch": 0.41061767048419434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3317000086972474e-05,
+      "loss": 0.9424,
+      "step": 14951
+    },
+    {
+      "epoch": 0.41064513471203756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3316160917859859e-05,
+      "loss": 0.8756,
+      "step": 14952
+    },
+    {
+      "epoch": 0.4106725989398808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3315321722508501e-05,
+      "loss": 0.9398,
+      "step": 14953
+    },
+    {
+      "epoch": 0.41070006316772406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.331448250092503e-05,
+      "loss": 1.0333,
+      "step": 14954
+    },
+    {
+      "epoch": 0.4107275273955673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.331364325311609e-05,
+      "loss": 0.933,
+      "step": 14955
+    },
+    {
+      "epoch": 0.4107549916234105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.331280397908832e-05,
+      "loss": 1.0325,
+      "step": 14956
+    },
+    {
+      "epoch": 0.4107824558512537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3311964678848361e-05,
+      "loss": 0.9154,
+      "step": 14957
+    },
+    {
+      "epoch": 0.410809920079097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3311125352402856e-05,
+      "loss": 1.0187,
+      "step": 14958
+    },
+    {
+      "epoch": 0.4108373843069402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3310285999758445e-05,
+      "loss": 0.9726,
+      "step": 14959
+    },
+    {
+      "epoch": 0.41086484853478344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3309446620921768e-05,
+      "loss": 0.9458,
+      "step": 14960
+    },
+    {
+      "epoch": 0.41089231276262667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3308607215899468e-05,
+      "loss": 0.9438,
+      "step": 14961
+    },
+    {
+      "epoch": 0.4109197769904699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3307767784698186e-05,
+      "loss": 0.9034,
+      "step": 14962
+    },
+    {
+      "epoch": 0.41094724121831316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3306928327324564e-05,
+      "loss": 0.8417,
+      "step": 14963
+    },
+    {
+      "epoch": 0.4109747054461564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3306088843785243e-05,
+      "loss": 0.9551,
+      "step": 14964
+    },
+    {
+      "epoch": 0.4110021696739996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3305249334086869e-05,
+      "loss": 0.8757,
+      "step": 14965
+    },
+    {
+      "epoch": 0.41102963390184283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330440979823608e-05,
+      "loss": 0.8528,
+      "step": 14966
+    },
+    {
+      "epoch": 0.4110570981296861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330357023623952e-05,
+      "loss": 0.8326,
+      "step": 14967
+    },
+    {
+      "epoch": 0.4110845623575293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3302730648103833e-05,
+      "loss": 0.9677,
+      "step": 14968
+    },
+    {
+      "epoch": 0.41111202658537255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3301891033835664e-05,
+      "loss": 0.8954,
+      "step": 14969
+    },
+    {
+      "epoch": 0.41113949081321577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3301051393441654e-05,
+      "loss": 1.0616,
+      "step": 14970
+    },
+    {
+      "epoch": 0.41116695504105905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3300211726928444e-05,
+      "loss": 0.9339,
+      "step": 14971
+    },
+    {
+      "epoch": 0.41119441926890227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.329937203430268e-05,
+      "loss": 1.0719,
+      "step": 14972
+    },
+    {
+      "epoch": 0.4112218834967455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3298532315571007e-05,
+      "loss": 0.9288,
+      "step": 14973
+    },
+    {
+      "epoch": 0.4112493477245887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3297692570740071e-05,
+      "loss": 0.9844,
+      "step": 14974
+    },
+    {
+      "epoch": 0.41127681195243193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3296852799816514e-05,
+      "loss": 0.9562,
+      "step": 14975
+    },
+    {
+      "epoch": 0.4113042761802752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3296013002806977e-05,
+      "loss": 0.8223,
+      "step": 14976
+    },
+    {
+      "epoch": 0.41133174040811843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3295173179718113e-05,
+      "loss": 0.9339,
+      "step": 14977
+    },
+    {
+      "epoch": 0.41135920463596165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3294333330556559e-05,
+      "loss": 1.0271,
+      "step": 14978
+    },
+    {
+      "epoch": 0.41138666886380487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3293493455328963e-05,
+      "loss": 0.8423,
+      "step": 14979
+    },
+    {
+      "epoch": 0.41141413309164815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3292653554041974e-05,
+      "loss": 0.9801,
+      "step": 14980
+    },
+    {
+      "epoch": 0.41144159731949137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3291813626702233e-05,
+      "loss": 0.9975,
+      "step": 14981
+    },
+    {
+      "epoch": 0.4114690615473346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3290973673316388e-05,
+      "loss": 0.9761,
+      "step": 14982
+    },
+    {
+      "epoch": 0.4114965257751778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3290133693891085e-05,
+      "loss": 0.8988,
+      "step": 14983
+    },
+    {
+      "epoch": 0.4115239900030211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3289293688432968e-05,
+      "loss": 1.024,
+      "step": 14984
+    },
+    {
+      "epoch": 0.4115514542308643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3288453656948684e-05,
+      "loss": 0.9651,
+      "step": 14985
+    },
+    {
+      "epoch": 0.41157891845870753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3287613599444882e-05,
+      "loss": 0.937,
+      "step": 14986
+    },
+    {
+      "epoch": 0.41160638268655075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3286773515928211e-05,
+      "loss": 0.8674,
+      "step": 14987
+    },
+    {
+      "epoch": 0.411633846914394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3285933406405312e-05,
+      "loss": 0.983,
+      "step": 14988
+    },
+    {
+      "epoch": 0.41166131114223725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3285093270882834e-05,
+      "loss": 0.9779,
+      "step": 14989
+    },
+    {
+      "epoch": 0.4116887753700805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.328425310936743e-05,
+      "loss": 0.8614,
+      "step": 14990
+    },
+    {
+      "epoch": 0.4117162395979237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.328341292186574e-05,
+      "loss": 0.9473,
+      "step": 14991
+    },
+    {
+      "epoch": 0.4117437038257669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3282572708384413e-05,
+      "loss": 0.8802,
+      "step": 14992
+    },
+    {
+      "epoch": 0.4117711680536102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.32817324689301e-05,
+      "loss": 0.9775,
+      "step": 14993
+    },
+    {
+      "epoch": 0.4117986322814534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.328089220350945e-05,
+      "loss": 0.8925,
+      "step": 14994
+    },
+    {
+      "epoch": 0.41182609650929664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3280051912129111e-05,
+      "loss": 1.0312,
+      "step": 14995
+    },
+    {
+      "epoch": 0.41185356073713986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.327921159479573e-05,
+      "loss": 0.8744,
+      "step": 14996
+    },
+    {
+      "epoch": 0.41188102496498313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3278371251515956e-05,
+      "loss": 0.9893,
+      "step": 14997
+    },
+    {
+      "epoch": 0.41190848919282635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3277530882296436e-05,
+      "loss": 0.9649,
+      "step": 14998
+    },
+    {
+      "epoch": 0.4119359534206696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3276690487143826e-05,
+      "loss": 1.0175,
+      "step": 14999
+    },
+    {
+      "epoch": 0.4119634176485128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3275850066064773e-05,
+      "loss": 1.0497,
+      "step": 15000
+    },
+    {
+      "epoch": 0.411990881876356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3275009619065919e-05,
+      "loss": 1.0136,
+      "step": 15001
+    },
+    {
+      "epoch": 0.4120183461041993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3274169146153928e-05,
+      "loss": 0.8961,
+      "step": 15002
+    },
+    {
+      "epoch": 0.4120458103320425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3273328647335437e-05,
+      "loss": 0.9338,
+      "step": 15003
+    },
+    {
+      "epoch": 0.41207327455988574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3272488122617104e-05,
+      "loss": 0.8731,
+      "step": 15004
+    },
+    {
+      "epoch": 0.41210073878772896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3271647572005575e-05,
+      "loss": 1.0043,
+      "step": 15005
+    },
+    {
+      "epoch": 0.41212820301557224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3270806995507506e-05,
+      "loss": 0.9487,
+      "step": 15006
+    },
+    {
+      "epoch": 0.41215566724341546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3269966393129543e-05,
+      "loss": 1.0889,
+      "step": 15007
+    },
+    {
+      "epoch": 0.4121831314712587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3269125764878339e-05,
+      "loss": 0.9297,
+      "step": 15008
+    },
+    {
+      "epoch": 0.4122105956991019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3268285110760548e-05,
+      "loss": 0.9362,
+      "step": 15009
+    },
+    {
+      "epoch": 0.4122380599269452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3267444430782818e-05,
+      "loss": 0.9074,
+      "step": 15010
+    },
+    {
+      "epoch": 0.4122655241547884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.32666037249518e-05,
+      "loss": 0.9737,
+      "step": 15011
+    },
+    {
+      "epoch": 0.4122929883826316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3265762993274152e-05,
+      "loss": 0.8969,
+      "step": 15012
+    },
+    {
+      "epoch": 0.41232045261047484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3264922235756519e-05,
+      "loss": 0.9134,
+      "step": 15013
+    },
+    {
+      "epoch": 0.4123479168383181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3264081452405558e-05,
+      "loss": 0.9827,
+      "step": 15014
+    },
+    {
+      "epoch": 0.41237538106616134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3263240643227921e-05,
+      "loss": 0.9451,
+      "step": 15015
+    },
+    {
+      "epoch": 0.41240284529400456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3262399808230257e-05,
+      "loss": 0.8289,
+      "step": 15016
+    },
+    {
+      "epoch": 0.4124303095218478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3261558947419222e-05,
+      "loss": 0.8837,
+      "step": 15017
+    },
+    {
+      "epoch": 0.412457773749691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3260718060801473e-05,
+      "loss": 0.9142,
+      "step": 15018
+    },
+    {
+      "epoch": 0.4124852379775343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3259877148383657e-05,
+      "loss": 0.9452,
+      "step": 15019
+    },
+    {
+      "epoch": 0.4125127022053775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3259036210172428e-05,
+      "loss": 0.8693,
+      "step": 15020
+    },
+    {
+      "epoch": 0.4125401664332207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3258195246174448e-05,
+      "loss": 1.036,
+      "step": 15021
+    },
+    {
+      "epoch": 0.41256763066106394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3257354256396362e-05,
+      "loss": 0.9476,
+      "step": 15022
+    },
+    {
+      "epoch": 0.4125950948889072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3256513240844826e-05,
+      "loss": 0.9556,
+      "step": 15023
+    },
+    {
+      "epoch": 0.41262255911675044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3255672199526498e-05,
+      "loss": 0.9507,
+      "step": 15024
+    },
+    {
+      "epoch": 0.41265002334459366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3254831132448027e-05,
+      "loss": 0.8429,
+      "step": 15025
+    },
+    {
+      "epoch": 0.4126774875724369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3253990039616076e-05,
+      "loss": 0.9029,
+      "step": 15026
+    },
+    {
+      "epoch": 0.41270495180028016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3253148921037292e-05,
+      "loss": 0.9732,
+      "step": 15027
+    },
+    {
+      "epoch": 0.4127324160281234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3252307776718336e-05,
+      "loss": 0.9563,
+      "step": 15028
+    },
+    {
+      "epoch": 0.4127598802559666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3251466606665856e-05,
+      "loss": 0.957,
+      "step": 15029
+    },
+    {
+      "epoch": 0.4127873444838098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3250625410886518e-05,
+      "loss": 0.9116,
+      "step": 15030
+    },
+    {
+      "epoch": 0.41281480871165305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3249784189386971e-05,
+      "loss": 0.9209,
+      "step": 15031
+    },
+    {
+      "epoch": 0.4128422729394963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3248942942173873e-05,
+      "loss": 0.9702,
+      "step": 15032
+    },
+    {
+      "epoch": 0.41286973716733955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3248101669253881e-05,
+      "loss": 0.9412,
+      "step": 15033
+    },
+    {
+      "epoch": 0.41289720139518277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324726037063365e-05,
+      "loss": 0.927,
+      "step": 15034
+    },
+    {
+      "epoch": 0.412924665623026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3246419046319833e-05,
+      "loss": 1.0356,
+      "step": 15035
+    },
+    {
+      "epoch": 0.41295212985086927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3245577696319094e-05,
+      "loss": 1.0829,
+      "step": 15036
+    },
+    {
+      "epoch": 0.4129795940787125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3244736320638087e-05,
+      "loss": 1.0586,
+      "step": 15037
+    },
+    {
+      "epoch": 0.4130070583065557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324389491928347e-05,
+      "loss": 0.9369,
+      "step": 15038
+    },
+    {
+      "epoch": 0.41303452253439893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3243053492261901e-05,
+      "loss": 0.9175,
+      "step": 15039
+    },
+    {
+      "epoch": 0.4130619867622422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3242212039580033e-05,
+      "loss": 0.8788,
+      "step": 15040
+    },
+    {
+      "epoch": 0.4130894509900854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3241370561244527e-05,
+      "loss": 0.9572,
+      "step": 15041
+    },
+    {
+      "epoch": 0.41311691521792865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3240529057262045e-05,
+      "loss": 0.954,
+      "step": 15042
+    },
+    {
+      "epoch": 0.41314437944577187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3239687527639243e-05,
+      "loss": 0.9514,
+      "step": 15043
+    },
+    {
+      "epoch": 0.4131718436736151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3238845972382772e-05,
+      "loss": 0.8759,
+      "step": 15044
+    },
+    {
+      "epoch": 0.41319930790145837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3238004391499304e-05,
+      "loss": 0.9012,
+      "step": 15045
+    },
+    {
+      "epoch": 0.4132267721293016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3237162784995487e-05,
+      "loss": 0.8333,
+      "step": 15046
+    },
+    {
+      "epoch": 0.4132542363571448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3236321152877986e-05,
+      "loss": 0.9456,
+      "step": 15047
+    },
+    {
+      "epoch": 0.41328170058498803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3235479495153457e-05,
+      "loss": 0.9421,
+      "step": 15048
+    },
+    {
+      "epoch": 0.4133091648128313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3234637811828562e-05,
+      "loss": 0.9258,
+      "step": 15049
+    },
+    {
+      "epoch": 0.41333662904067453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3233796102909957e-05,
+      "loss": 0.9527,
+      "step": 15050
+    },
+    {
+      "epoch": 0.41336409326851775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.323295436840431e-05,
+      "loss": 0.9764,
+      "step": 15051
+    },
+    {
+      "epoch": 0.413391557496361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.323211260831827e-05,
+      "loss": 0.9762,
+      "step": 15052
+    },
+    {
+      "epoch": 0.41341902172420425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3231270822658506e-05,
+      "loss": 0.9872,
+      "step": 15053
+    },
+    {
+      "epoch": 0.41344648595204747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3230429011431674e-05,
+      "loss": 1.0633,
+      "step": 15054
+    },
+    {
+      "epoch": 0.4134739501798907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3229587174644439e-05,
+      "loss": 0.9695,
+      "step": 15055
+    },
+    {
+      "epoch": 0.4135014144077339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3228745312303457e-05,
+      "loss": 0.7478,
+      "step": 15056
+    },
+    {
+      "epoch": 0.41352887863557714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3227903424415391e-05,
+      "loss": 0.961,
+      "step": 15057
+    },
+    {
+      "epoch": 0.4135563428634204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3227061510986902e-05,
+      "loss": 0.7562,
+      "step": 15058
+    },
+    {
+      "epoch": 0.41358380709126363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3226219572024657e-05,
+      "loss": 0.9427,
+      "step": 15059
+    },
+    {
+      "epoch": 0.41361127131910685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3225377607535307e-05,
+      "loss": 0.9497,
+      "step": 15060
+    },
+    {
+      "epoch": 0.4136387355469501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3224535617525522e-05,
+      "loss": 1.0093,
+      "step": 15061
+    },
+    {
+      "epoch": 0.41366619977479335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3223693602001966e-05,
+      "loss": 0.8589,
+      "step": 15062
+    },
+    {
+      "epoch": 0.4136936640026366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3222851560971294e-05,
+      "loss": 0.8709,
+      "step": 15063
+    },
+    {
+      "epoch": 0.4137211282304798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3222009494440171e-05,
+      "loss": 0.8356,
+      "step": 15064
+    },
+    {
+      "epoch": 0.413748592458323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3221167402415263e-05,
+      "loss": 0.9791,
+      "step": 15065
+    },
+    {
+      "epoch": 0.4137760566861663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3220325284903228e-05,
+      "loss": 0.9362,
+      "step": 15066
+    },
+    {
+      "epoch": 0.4138035209140095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3219483141910736e-05,
+      "loss": 0.915,
+      "step": 15067
+    },
+    {
+      "epoch": 0.41383098514185274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3218640973444446e-05,
+      "loss": 0.8639,
+      "step": 15068
+    },
+    {
+      "epoch": 0.41385844936969596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3217798779511018e-05,
+      "loss": 0.9429,
+      "step": 15069
+    },
+    {
+      "epoch": 0.4138859135975392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3216956560117122e-05,
+      "loss": 0.8845,
+      "step": 15070
+    },
+    {
+      "epoch": 0.41391337782538246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321611431526942e-05,
+      "loss": 0.9425,
+      "step": 15071
+    },
+    {
+      "epoch": 0.4139408420532257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3215272044974573e-05,
+      "loss": 0.9409,
+      "step": 15072
+    },
+    {
+      "epoch": 0.4139683062810689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321442974923925e-05,
+      "loss": 0.9797,
+      "step": 15073
+    },
+    {
+      "epoch": 0.4139957705089121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3213587428070114e-05,
+      "loss": 0.9619,
+      "step": 15074
+    },
+    {
+      "epoch": 0.4140232347367554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321274508147383e-05,
+      "loss": 0.8754,
+      "step": 15075
+    },
+    {
+      "epoch": 0.4140506989645986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321190270945706e-05,
+      "loss": 0.9413,
+      "step": 15076
+    },
+    {
+      "epoch": 0.41407816319244184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3211060312026475e-05,
+      "loss": 0.9622,
+      "step": 15077
+    },
+    {
+      "epoch": 0.41410562742028506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3210217889188734e-05,
+      "loss": 1.0282,
+      "step": 15078
+    },
+    {
+      "epoch": 0.41413309164812834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3209375440950506e-05,
+      "loss": 1.0549,
+      "step": 15079
+    },
+    {
+      "epoch": 0.41416055587597156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.320853296731846e-05,
+      "loss": 0.9639,
+      "step": 15080
+    },
+    {
+      "epoch": 0.4141880201038148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3207690468299253e-05,
+      "loss": 0.8495,
+      "step": 15081
+    },
+    {
+      "epoch": 0.414215484331658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3206847943899561e-05,
+      "loss": 0.8891,
+      "step": 15082
+    },
+    {
+      "epoch": 0.4142429485595012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3206005394126042e-05,
+      "loss": 0.9774,
+      "step": 15083
+    },
+    {
+      "epoch": 0.4142704127873445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3205162818985368e-05,
+      "loss": 1.0245,
+      "step": 15084
+    },
+    {
+      "epoch": 0.4142978770151877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3204320218484203e-05,
+      "loss": 0.949,
+      "step": 15085
+    },
+    {
+      "epoch": 0.41432534124303094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.320347759262922e-05,
+      "loss": 0.9886,
+      "step": 15086
+    },
+    {
+      "epoch": 0.41435280547087416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3202634941427077e-05,
+      "loss": 1.0925,
+      "step": 15087
+    },
+    {
+      "epoch": 0.41438026969871744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3201792264884446e-05,
+      "loss": 0.9013,
+      "step": 15088
+    },
+    {
+      "epoch": 0.41440773392656066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3200949563007994e-05,
+      "loss": 0.8591,
+      "step": 15089
+    },
+    {
+      "epoch": 0.4144351981544039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3200106835804391e-05,
+      "loss": 0.9801,
+      "step": 15090
+    },
+    {
+      "epoch": 0.4144626623822471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3199264083280304e-05,
+      "loss": 0.917,
+      "step": 15091
+    },
+    {
+      "epoch": 0.4144901266100904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3198421305442398e-05,
+      "loss": 0.8456,
+      "step": 15092
+    },
+    {
+      "epoch": 0.4145175908379336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3197578502297343e-05,
+      "loss": 0.8979,
+      "step": 15093
+    },
+    {
+      "epoch": 0.4145450550657768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.319673567385181e-05,
+      "loss": 1.0048,
+      "step": 15094
+    },
+    {
+      "epoch": 0.41457251929362005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3195892820112466e-05,
+      "loss": 0.9307,
+      "step": 15095
+    },
+    {
+      "epoch": 0.41459998352146327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3195049941085979e-05,
+      "loss": 0.9021,
+      "step": 15096
+    },
+    {
+      "epoch": 0.41462744774930654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3194207036779018e-05,
+      "loss": 0.9121,
+      "step": 15097
+    },
+    {
+      "epoch": 0.41465491197714976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3193364107198257e-05,
+      "loss": 0.8789,
+      "step": 15098
+    },
+    {
+      "epoch": 0.414682376204993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.319252115235036e-05,
+      "loss": 1.0448,
+      "step": 15099
+    },
+    {
+      "epoch": 0.4147098404328362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3191678172242e-05,
+      "loss": 0.9641,
+      "step": 15100
+    },
+    {
+      "epoch": 0.4147373046606795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3190835166879844e-05,
+      "loss": 0.8871,
+      "step": 15101
+    },
+    {
+      "epoch": 0.4147647688885227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3189992136270568e-05,
+      "loss": 0.7857,
+      "step": 15102
+    },
+    {
+      "epoch": 0.4147922331163659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3189149080420833e-05,
+      "loss": 0.8634,
+      "step": 15103
+    },
+    {
+      "epoch": 0.41481969734420915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.318830599933732e-05,
+      "loss": 0.9401,
+      "step": 15104
+    },
+    {
+      "epoch": 0.4148471615720524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3187462893026692e-05,
+      "loss": 0.9661,
+      "step": 15105
+    },
+    {
+      "epoch": 0.41487462579989565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3186619761495627e-05,
+      "loss": 0.8945,
+      "step": 15106
+    },
+    {
+      "epoch": 0.41490209002773887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3185776604750788e-05,
+      "loss": 0.8731,
+      "step": 15107
+    },
+    {
+      "epoch": 0.4149295542555821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.318493342279885e-05,
+      "loss": 0.8808,
+      "step": 15108
+    },
+    {
+      "epoch": 0.41495701848342537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3184090215646488e-05,
+      "loss": 0.9332,
+      "step": 15109
+    },
+    {
+      "epoch": 0.4149844827112686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3183246983300369e-05,
+      "loss": 0.8728,
+      "step": 15110
+    },
+    {
+      "epoch": 0.4150119469391118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3182403725767168e-05,
+      "loss": 0.9274,
+      "step": 15111
+    },
+    {
+      "epoch": 0.41503941116695503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3181560443053556e-05,
+      "loss": 0.9302,
+      "step": 15112
+    },
+    {
+      "epoch": 0.41506687539479825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3180717135166203e-05,
+      "loss": 0.9785,
+      "step": 15113
+    },
+    {
+      "epoch": 0.41509433962264153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3179873802111787e-05,
+      "loss": 0.9174,
+      "step": 15114
+    },
+    {
+      "epoch": 0.41512180385048475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3179030443896976e-05,
+      "loss": 1.0139,
+      "step": 15115
+    },
+    {
+      "epoch": 0.41514926807832797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3178187060528444e-05,
+      "loss": 0.9426,
+      "step": 15116
+    },
+    {
+      "epoch": 0.4151767323061712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3177343652012863e-05,
+      "loss": 0.9289,
+      "step": 15117
+    },
+    {
+      "epoch": 0.41520419653401447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3176500218356913e-05,
+      "loss": 0.8446,
+      "step": 15118
+    },
+    {
+      "epoch": 0.4152316607618577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3175656759567264e-05,
+      "loss": 1.0008,
+      "step": 15119
+    },
+    {
+      "epoch": 0.4152591249897009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3174813275650583e-05,
+      "loss": 0.9191,
+      "step": 15120
+    },
+    {
+      "epoch": 0.41528658921754413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3173969766613552e-05,
+      "loss": 0.8888,
+      "step": 15121
+    },
+    {
+      "epoch": 0.4153140534453874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3173126232462843e-05,
+      "loss": 1.0058,
+      "step": 15122
+    },
+    {
+      "epoch": 0.41534151767323063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.317228267320513e-05,
+      "loss": 0.9769,
+      "step": 15123
+    },
+    {
+      "epoch": 0.41536898190107385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3171439088847089e-05,
+      "loss": 0.9772,
+      "step": 15124
+    },
+    {
+      "epoch": 0.4153964461289171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.317059547939539e-05,
+      "loss": 1.0425,
+      "step": 15125
+    },
+    {
+      "epoch": 0.4154239103567603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3169751844856714e-05,
+      "loss": 0.93,
+      "step": 15126
+    },
+    {
+      "epoch": 0.41545137458460357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3168908185237734e-05,
+      "loss": 1.0203,
+      "step": 15127
+    },
+    {
+      "epoch": 0.4154788388124468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3168064500545121e-05,
+      "loss": 0.9417,
+      "step": 15128
+    },
+    {
+      "epoch": 0.41550630304029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3167220790785558e-05,
+      "loss": 0.9671,
+      "step": 15129
+    },
+    {
+      "epoch": 0.41553376726813324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3166377055965719e-05,
+      "loss": 0.9534,
+      "step": 15130
+    },
+    {
+      "epoch": 0.4155612314959765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3165533296092274e-05,
+      "loss": 0.9211,
+      "step": 15131
+    },
+    {
+      "epoch": 0.41558869572381973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3164689511171905e-05,
+      "loss": 0.8276,
+      "step": 15132
+    },
+    {
+      "epoch": 0.41561615995166296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3163845701211286e-05,
+      "loss": 0.9791,
+      "step": 15133
+    },
+    {
+      "epoch": 0.4156436241795062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3163001866217095e-05,
+      "loss": 0.9476,
+      "step": 15134
+    },
+    {
+      "epoch": 0.41567108840734945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3162158006196008e-05,
+      "loss": 0.8818,
+      "step": 15135
+    },
+    {
+      "epoch": 0.4156985526351927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3161314121154702e-05,
+      "loss": 0.8912,
+      "step": 15136
+    },
+    {
+      "epoch": 0.4157260168630359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3160470211099853e-05,
+      "loss": 0.8802,
+      "step": 15137
+    },
+    {
+      "epoch": 0.4157534810908791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3159626276038142e-05,
+      "loss": 0.9321,
+      "step": 15138
+    },
+    {
+      "epoch": 0.41578094531872234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3158782315976243e-05,
+      "loss": 0.8459,
+      "step": 15139
+    },
+    {
+      "epoch": 0.4158084095465656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3157938330920835e-05,
+      "loss": 0.9267,
+      "step": 15140
+    },
+    {
+      "epoch": 0.41583587377440884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3157094320878595e-05,
+      "loss": 0.9692,
+      "step": 15141
+    },
+    {
+      "epoch": 0.41586333800225206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3156250285856201e-05,
+      "loss": 1.0008,
+      "step": 15142
+    },
+    {
+      "epoch": 0.4158908022300953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3155406225860335e-05,
+      "loss": 1.0193,
+      "step": 15143
+    },
+    {
+      "epoch": 0.41591826645793856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3154562140897667e-05,
+      "loss": 0.9389,
+      "step": 15144
+    },
+    {
+      "epoch": 0.4159457306857818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3153718030974885e-05,
+      "loss": 0.9381,
+      "step": 15145
+    },
+    {
+      "epoch": 0.415973194913625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3152873896098664e-05,
+      "loss": 0.8923,
+      "step": 15146
+    },
+    {
+      "epoch": 0.4160006591414682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3152029736275685e-05,
+      "loss": 1.0433,
+      "step": 15147
+    },
+    {
+      "epoch": 0.4160281233693115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3151185551512625e-05,
+      "loss": 0.8772,
+      "step": 15148
+    },
+    {
+      "epoch": 0.4160555875971547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3150341341816162e-05,
+      "loss": 0.9983,
+      "step": 15149
+    },
+    {
+      "epoch": 0.41608305182499794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3149497107192982e-05,
+      "loss": 0.9839,
+      "step": 15150
+    },
+    {
+      "epoch": 0.41611051605284116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3148652847649759e-05,
+      "loss": 0.9533,
+      "step": 15151
+    },
+    {
+      "epoch": 0.4161379802806844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3147808563193172e-05,
+      "loss": 0.9787,
+      "step": 15152
+    },
+    {
+      "epoch": 0.41616544450852766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3146964253829905e-05,
+      "loss": 0.8852,
+      "step": 15153
+    },
+    {
+      "epoch": 0.4161929087363709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.314611991956664e-05,
+      "loss": 1.0045,
+      "step": 15154
+    },
+    {
+      "epoch": 0.4162203729642141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3145275560410055e-05,
+      "loss": 0.7978,
+      "step": 15155
+    },
+    {
+      "epoch": 0.4162478371920573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3144431176366833e-05,
+      "loss": 0.843,
+      "step": 15156
+    },
+    {
+      "epoch": 0.4162753014199006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3143586767443652e-05,
+      "loss": 0.8568,
+      "step": 15157
+    },
+    {
+      "epoch": 0.4163027656477438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3142742333647193e-05,
+      "loss": 0.7642,
+      "step": 15158
+    },
+    {
+      "epoch": 0.41633022987558704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3141897874984142e-05,
+      "loss": 0.8657,
+      "step": 15159
+    },
+    {
+      "epoch": 0.41635769410343026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3141053391461178e-05,
+      "loss": 1.0258,
+      "step": 15160
+    },
+    {
+      "epoch": 0.41638515833127354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3140208883084982e-05,
+      "loss": 0.949,
+      "step": 15161
+    },
+    {
+      "epoch": 0.41641262255911676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3139364349862236e-05,
+      "loss": 0.9384,
+      "step": 15162
+    },
+    {
+      "epoch": 0.41644008678696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3138519791799624e-05,
+      "loss": 0.8953,
+      "step": 15163
+    },
+    {
+      "epoch": 0.4164675510148032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3137675208903828e-05,
+      "loss": 0.9933,
+      "step": 15164
+    },
+    {
+      "epoch": 0.4164950152426464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3136830601181527e-05,
+      "loss": 0.9382,
+      "step": 15165
+    },
+    {
+      "epoch": 0.4165224794704897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313598596863941e-05,
+      "loss": 0.8542,
+      "step": 15166
+    },
+    {
+      "epoch": 0.4165499436983329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3135141311284159e-05,
+      "loss": 0.918,
+      "step": 15167
+    },
+    {
+      "epoch": 0.41657740792617615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3134296629122456e-05,
+      "loss": 0.9455,
+      "step": 15168
+    },
+    {
+      "epoch": 0.41660487215401937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313345192216098e-05,
+      "loss": 0.9432,
+      "step": 15169
+    },
+    {
+      "epoch": 0.41663233638186264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3132607190406421e-05,
+      "loss": 0.9417,
+      "step": 15170
+    },
+    {
+      "epoch": 0.41665980060970587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313176243386546e-05,
+      "loss": 0.9904,
+      "step": 15171
+    },
+    {
+      "epoch": 0.4166872648375491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3130917652544782e-05,
+      "loss": 0.9289,
+      "step": 15172
+    },
+    {
+      "epoch": 0.4167147290653923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313007284645107e-05,
+      "loss": 0.8739,
+      "step": 15173
+    },
+    {
+      "epoch": 0.4167421932932356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.312922801559101e-05,
+      "loss": 0.9997,
+      "step": 15174
+    },
+    {
+      "epoch": 0.4167696575210788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3128383159971282e-05,
+      "loss": 1.0038,
+      "step": 15175
+    },
+    {
+      "epoch": 0.41679712174892203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.312753827959858e-05,
+      "loss": 0.9185,
+      "step": 15176
+    },
+    {
+      "epoch": 0.41682458597676525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3126693374479582e-05,
+      "loss": 0.907,
+      "step": 15177
+    },
+    {
+      "epoch": 0.41685205020460847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3125848444620972e-05,
+      "loss": 0.8833,
+      "step": 15178
+    },
+    {
+      "epoch": 0.41687951443245175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.312500349002944e-05,
+      "loss": 0.9719,
+      "step": 15179
+    },
+    {
+      "epoch": 0.41690697866029497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3124158510711674e-05,
+      "loss": 0.9499,
+      "step": 15180
+    },
+    {
+      "epoch": 0.4169344428881382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3123313506674349e-05,
+      "loss": 0.9273,
+      "step": 15181
+    },
+    {
+      "epoch": 0.4169619071159814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3122468477924162e-05,
+      "loss": 0.9459,
+      "step": 15182
+    },
+    {
+      "epoch": 0.4169893713438247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3121623424467793e-05,
+      "loss": 0.9873,
+      "step": 15183
+    },
+    {
+      "epoch": 0.4170168355716679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.312077834631193e-05,
+      "loss": 0.9267,
+      "step": 15184
+    },
+    {
+      "epoch": 0.41704429979951113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3119933243463261e-05,
+      "loss": 0.8386,
+      "step": 15185
+    },
+    {
+      "epoch": 0.41707176402735435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3119088115928471e-05,
+      "loss": 1.0363,
+      "step": 15186
+    },
+    {
+      "epoch": 0.41709922825519763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3118242963714249e-05,
+      "loss": 0.9112,
+      "step": 15187
+    },
+    {
+      "epoch": 0.41712669248304085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.311739778682728e-05,
+      "loss": 0.9959,
+      "step": 15188
+    },
+    {
+      "epoch": 0.41715415671088407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3116552585274252e-05,
+      "loss": 0.9296,
+      "step": 15189
+    },
+    {
+      "epoch": 0.4171816209387273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3115707359061852e-05,
+      "loss": 0.8947,
+      "step": 15190
+    },
+    {
+      "epoch": 0.4172090851665705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3114862108196768e-05,
+      "loss": 0.8705,
+      "step": 15191
+    },
+    {
+      "epoch": 0.4172365493944138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.311401683268569e-05,
+      "loss": 1.0276,
+      "step": 15192
+    },
+    {
+      "epoch": 0.417264013622257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3113171532535303e-05,
+      "loss": 0.9223,
+      "step": 15193
+    },
+    {
+      "epoch": 0.41729147785010023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3112326207752299e-05,
+      "loss": 0.8457,
+      "step": 15194
+    },
+    {
+      "epoch": 0.41731894207794346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3111480858343365e-05,
+      "loss": 0.942,
+      "step": 15195
+    },
+    {
+      "epoch": 0.41734640630578673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3110635484315186e-05,
+      "loss": 0.9782,
+      "step": 15196
+    },
+    {
+      "epoch": 0.41737387053362995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3109790085674454e-05,
+      "loss": 0.9672,
+      "step": 15197
+    },
+    {
+      "epoch": 0.4174013347614732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310894466242786e-05,
+      "loss": 0.9082,
+      "step": 15198
+    },
+    {
+      "epoch": 0.4174287989893164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3108099214582094e-05,
+      "loss": 0.8832,
+      "step": 15199
+    },
+    {
+      "epoch": 0.4174562632171597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3107253742143838e-05,
+      "loss": 0.9207,
+      "step": 15200
+    },
+    {
+      "epoch": 0.4174837274450029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310640824511979e-05,
+      "loss": 0.9835,
+      "step": 15201
+    },
+    {
+      "epoch": 0.4175111916728461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3105562723516636e-05,
+      "loss": 1.0043,
+      "step": 15202
+    },
+    {
+      "epoch": 0.41753865590068934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3104717177341067e-05,
+      "loss": 0.986,
+      "step": 15203
+    },
+    {
+      "epoch": 0.41756612012853256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3103871606599774e-05,
+      "loss": 0.9352,
+      "step": 15204
+    },
+    {
+      "epoch": 0.41759358435637584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3103026011299442e-05,
+      "loss": 0.9442,
+      "step": 15205
+    },
+    {
+      "epoch": 0.41762104858421906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3102180391446771e-05,
+      "loss": 0.9627,
+      "step": 15206
+    },
+    {
+      "epoch": 0.4176485128120623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3101334747048447e-05,
+      "loss": 0.9221,
+      "step": 15207
+    },
+    {
+      "epoch": 0.4176759770399055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3100489078111158e-05,
+      "loss": 0.7866,
+      "step": 15208
+    },
+    {
+      "epoch": 0.4177034412677488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3099643384641598e-05,
+      "loss": 0.989,
+      "step": 15209
+    },
+    {
+      "epoch": 0.417730905495592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3098797666646462e-05,
+      "loss": 0.9872,
+      "step": 15210
+    },
+    {
+      "epoch": 0.4177583697234352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3097951924132439e-05,
+      "loss": 0.9302,
+      "step": 15211
+    },
+    {
+      "epoch": 0.41778583395127844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3097106157106216e-05,
+      "loss": 0.8619,
+      "step": 15212
+    },
+    {
+      "epoch": 0.4178132981791217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3096260365574493e-05,
+      "loss": 1.0025,
+      "step": 15213
+    },
+    {
+      "epoch": 0.41784076240696494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3095414549543956e-05,
+      "loss": 0.9316,
+      "step": 15214
+    },
+    {
+      "epoch": 0.41786822663480816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3094568709021304e-05,
+      "loss": 0.9246,
+      "step": 15215
+    },
+    {
+      "epoch": 0.4178956908626514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3093722844013224e-05,
+      "loss": 1.0203,
+      "step": 15216
+    },
+    {
+      "epoch": 0.41792315509049466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3092876954526408e-05,
+      "loss": 1.0168,
+      "step": 15217
+    },
+    {
+      "epoch": 0.4179506193183379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3092031040567555e-05,
+      "loss": 0.9599,
+      "step": 15218
+    },
+    {
+      "epoch": 0.4179780835461811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3091185102143354e-05,
+      "loss": 0.8457,
+      "step": 15219
+    },
+    {
+      "epoch": 0.4180055477740243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090339139260494e-05,
+      "loss": 0.7811,
+      "step": 15220
+    },
+    {
+      "epoch": 0.41803301200186754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3089493151925679e-05,
+      "loss": 0.9435,
+      "step": 15221
+    },
+    {
+      "epoch": 0.4180604762297108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3088647140145596e-05,
+      "loss": 0.9392,
+      "step": 15222
+    },
+    {
+      "epoch": 0.41808794045755404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3087801103926943e-05,
+      "loss": 0.955,
+      "step": 15223
+    },
+    {
+      "epoch": 0.41811540468539726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3086955043276408e-05,
+      "loss": 0.9463,
+      "step": 15224
+    },
+    {
+      "epoch": 0.4181428689132405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3086108958200689e-05,
+      "loss": 0.918,
+      "step": 15225
+    },
+    {
+      "epoch": 0.41817033314108376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3085262848706482e-05,
+      "loss": 0.9966,
+      "step": 15226
+    },
+    {
+      "epoch": 0.418197797368927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.308441671480048e-05,
+      "loss": 0.9889,
+      "step": 15227
+    },
+    {
+      "epoch": 0.4182252615967702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3083570556489378e-05,
+      "loss": 0.9818,
+      "step": 15228
+    },
+    {
+      "epoch": 0.4182527258246134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.308272437377987e-05,
+      "loss": 0.894,
+      "step": 15229
+    },
+    {
+      "epoch": 0.4182801900524567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3081878166678654e-05,
+      "loss": 0.8781,
+      "step": 15230
+    },
+    {
+      "epoch": 0.4183076542802999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3081031935192425e-05,
+      "loss": 0.9741,
+      "step": 15231
+    },
+    {
+      "epoch": 0.41833511850814314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3080185679327877e-05,
+      "loss": 0.9573,
+      "step": 15232
+    },
+    {
+      "epoch": 0.41836258273598637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3079339399091706e-05,
+      "loss": 0.9202,
+      "step": 15233
+    },
+    {
+      "epoch": 0.4183900469638296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3078493094490607e-05,
+      "loss": 1.007,
+      "step": 15234
+    },
+    {
+      "epoch": 0.41841751119167286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3077646765531282e-05,
+      "loss": 1.0452,
+      "step": 15235
+    },
+    {
+      "epoch": 0.4184449754195161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3076800412220423e-05,
+      "loss": 0.9026,
+      "step": 15236
+    },
+    {
+      "epoch": 0.4184724396473593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3075954034564727e-05,
+      "loss": 0.9047,
+      "step": 15237
+    },
+    {
+      "epoch": 0.41849990387520253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3075107632570888e-05,
+      "loss": 0.8707,
+      "step": 15238
+    },
+    {
+      "epoch": 0.4185273681030458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3074261206245614e-05,
+      "loss": 0.9405,
+      "step": 15239
+    },
+    {
+      "epoch": 0.418554832330889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3073414755595586e-05,
+      "loss": 0.9916,
+      "step": 15240
+    },
+    {
+      "epoch": 0.41858229655873225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3072568280627511e-05,
+      "loss": 0.9161,
+      "step": 15241
+    },
+    {
+      "epoch": 0.41860976078657547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.307172178134809e-05,
+      "loss": 0.8488,
+      "step": 15242
+    },
+    {
+      "epoch": 0.41863722501441875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3070875257764015e-05,
+      "loss": 0.9344,
+      "step": 15243
+    },
+    {
+      "epoch": 0.41866468924226197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3070028709881984e-05,
+      "loss": 0.9037,
+      "step": 15244
+    },
+    {
+      "epoch": 0.4186921534701052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3069182137708698e-05,
+      "loss": 0.9352,
+      "step": 15245
+    },
+    {
+      "epoch": 0.4187196176979484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3068335541250852e-05,
+      "loss": 0.9383,
+      "step": 15246
+    },
+    {
+      "epoch": 0.41874708192579163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3067488920515148e-05,
+      "loss": 1.0104,
+      "step": 15247
+    },
+    {
+      "epoch": 0.4187745461536349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3066642275508286e-05,
+      "loss": 0.9014,
+      "step": 15248
+    },
+    {
+      "epoch": 0.41880201038147813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3065795606236956e-05,
+      "loss": 0.9386,
+      "step": 15249
+    },
+    {
+      "epoch": 0.41882947460932135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.306494891270787e-05,
+      "loss": 1.0362,
+      "step": 15250
+    },
+    {
+      "epoch": 0.41885693883716457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3064102194927719e-05,
+      "loss": 0.942,
+      "step": 15251
+    },
+    {
+      "epoch": 0.41888440306500785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3063255452903204e-05,
+      "loss": 0.9087,
+      "step": 15252
+    },
+    {
+      "epoch": 0.41891186729285107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3062408686641022e-05,
+      "loss": 0.8332,
+      "step": 15253
+    },
+    {
+      "epoch": 0.4189393315206943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.306156189614788e-05,
+      "loss": 0.8935,
+      "step": 15254
+    },
+    {
+      "epoch": 0.4189667957485375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3060715081430476e-05,
+      "loss": 0.9121,
+      "step": 15255
+    },
+    {
+      "epoch": 0.4189942599763808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3059868242495503e-05,
+      "loss": 0.8964,
+      "step": 15256
+    },
+    {
+      "epoch": 0.419021724204224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3059021379349673e-05,
+      "loss": 1.0483,
+      "step": 15257
+    },
+    {
+      "epoch": 0.41904918843206723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3058174491999676e-05,
+      "loss": 1.0017,
+      "step": 15258
+    },
+    {
+      "epoch": 0.41907665265991045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.305732758045222e-05,
+      "loss": 0.9131,
+      "step": 15259
+    },
+    {
+      "epoch": 0.4191041168877537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3056480644714006e-05,
+      "loss": 0.9059,
+      "step": 15260
+    },
+    {
+      "epoch": 0.41913158111559695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.305563368479173e-05,
+      "loss": 0.9687,
+      "step": 15261
+    },
+    {
+      "epoch": 0.4191590453434402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3054786700692099e-05,
+      "loss": 0.9742,
+      "step": 15262
+    },
+    {
+      "epoch": 0.4191865095712834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3053939692421813e-05,
+      "loss": 0.8365,
+      "step": 15263
+    },
+    {
+      "epoch": 0.4192139737991266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.305309265998757e-05,
+      "loss": 1.0085,
+      "step": 15264
+    },
+    {
+      "epoch": 0.4192414380269699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3052245603396075e-05,
+      "loss": 0.9056,
+      "step": 15265
+    },
+    {
+      "epoch": 0.4192689022548131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3051398522654033e-05,
+      "loss": 1.0379,
+      "step": 15266
+    },
+    {
+      "epoch": 0.41929636648265634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3050551417768143e-05,
+      "loss": 0.9302,
+      "step": 15267
+    },
+    {
+      "epoch": 0.41932383071049956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3049704288745106e-05,
+      "loss": 1.0334,
+      "step": 15268
+    },
+    {
+      "epoch": 0.41935129493834283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3048857135591632e-05,
+      "loss": 1.0393,
+      "step": 15269
+    },
+    {
+      "epoch": 0.41937875916618605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3048009958314415e-05,
+      "loss": 0.9044,
+      "step": 15270
+    },
+    {
+      "epoch": 0.4194062233940293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3047162756920167e-05,
+      "loss": 0.8569,
+      "step": 15271
+    },
+    {
+      "epoch": 0.4194336876218725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3046315531415586e-05,
+      "loss": 0.9517,
+      "step": 15272
+    },
+    {
+      "epoch": 0.4194611518497157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3045468281807372e-05,
+      "loss": 0.8839,
+      "step": 15273
+    },
+    {
+      "epoch": 0.419488616077559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3044621008102239e-05,
+      "loss": 0.8885,
+      "step": 15274
+    },
+    {
+      "epoch": 0.4195160803054022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3043773710306881e-05,
+      "loss": 0.9753,
+      "step": 15275
+    },
+    {
+      "epoch": 0.41954354453324544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3042926388428008e-05,
+      "loss": 0.9411,
+      "step": 15276
+    },
+    {
+      "epoch": 0.41957100876108866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3042079042472322e-05,
+      "loss": 0.8685,
+      "step": 15277
+    },
+    {
+      "epoch": 0.41959847298893194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3041231672446529e-05,
+      "loss": 0.9316,
+      "step": 15278
+    },
+    {
+      "epoch": 0.41962593721677516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3040384278357335e-05,
+      "loss": 0.9492,
+      "step": 15279
+    },
+    {
+      "epoch": 0.4196534014446184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.303953686021144e-05,
+      "loss": 0.9311,
+      "step": 15280
+    },
+    {
+      "epoch": 0.4196808656724616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3038689418015551e-05,
+      "loss": 0.8717,
+      "step": 15281
+    },
+    {
+      "epoch": 0.4197083299003049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3037841951776375e-05,
+      "loss": 0.981,
+      "step": 15282
+    },
+    {
+      "epoch": 0.4197357941281481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3036994461500618e-05,
+      "loss": 0.8563,
+      "step": 15283
+    },
+    {
+      "epoch": 0.4197632583559913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3036146947194986e-05,
+      "loss": 0.9442,
+      "step": 15284
+    },
+    {
+      "epoch": 0.41979072258383454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3035299408866178e-05,
+      "loss": 0.9461,
+      "step": 15285
+    },
+    {
+      "epoch": 0.41981818681167776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.303445184652091e-05,
+      "loss": 0.8461,
+      "step": 15286
+    },
+    {
+      "epoch": 0.41984565103952104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3033604260165884e-05,
+      "loss": 0.8757,
+      "step": 15287
+    },
+    {
+      "epoch": 0.41987311526736426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30327566498078e-05,
+      "loss": 0.9291,
+      "step": 15288
+    },
+    {
+      "epoch": 0.4199005794952075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3031909015453377e-05,
+      "loss": 0.8661,
+      "step": 15289
+    },
+    {
+      "epoch": 0.4199280437230507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3031061357109312e-05,
+      "loss": 0.9323,
+      "step": 15290
+    },
+    {
+      "epoch": 0.419955507950894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3030213674782318e-05,
+      "loss": 0.9597,
+      "step": 15291
+    },
+    {
+      "epoch": 0.4199829721787372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3029365968479097e-05,
+      "loss": 0.9326,
+      "step": 15292
+    },
+    {
+      "epoch": 0.4200104364065804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.302851823820636e-05,
+      "loss": 0.9093,
+      "step": 15293
+    },
+    {
+      "epoch": 0.42003790063442364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3027670483970813e-05,
+      "loss": 0.8645,
+      "step": 15294
+    },
+    {
+      "epoch": 0.4200653648622669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3026822705779167e-05,
+      "loss": 0.9832,
+      "step": 15295
+    },
+    {
+      "epoch": 0.42009282909011014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3025974903638125e-05,
+      "loss": 0.9093,
+      "step": 15296
+    },
+    {
+      "epoch": 0.42012029331795336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3025127077554397e-05,
+      "loss": 0.9508,
+      "step": 15297
+    },
+    {
+      "epoch": 0.4201477575457966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3024279227534692e-05,
+      "loss": 0.9247,
+      "step": 15298
+    },
+    {
+      "epoch": 0.4201752217736398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3023431353585721e-05,
+      "loss": 0.936,
+      "step": 15299
+    },
+    {
+      "epoch": 0.4202026860014831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3022583455714186e-05,
+      "loss": 0.9512,
+      "step": 15300
+    },
+    {
+      "epoch": 0.4202301502293263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3021735533926803e-05,
+      "loss": 0.974,
+      "step": 15301
+    },
+    {
+      "epoch": 0.4202576144571695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3020887588230276e-05,
+      "loss": 0.9576,
+      "step": 15302
+    },
+    {
+      "epoch": 0.42028507868501275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3020039618631315e-05,
+      "loss": 0.965,
+      "step": 15303
+    },
+    {
+      "epoch": 0.420312542912856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3019191625136633e-05,
+      "loss": 0.9507,
+      "step": 15304
+    },
+    {
+      "epoch": 0.42034000714069925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3018343607752938e-05,
+      "loss": 0.9553,
+      "step": 15305
+    },
+    {
+      "epoch": 0.42036747136854247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3017495566486934e-05,
+      "loss": 0.9034,
+      "step": 15306
+    },
+    {
+      "epoch": 0.4203949355963857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3016647501345342e-05,
+      "loss": 1.039,
+      "step": 15307
+    },
+    {
+      "epoch": 0.42042239982422897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3015799412334867e-05,
+      "loss": 0.8532,
+      "step": 15308
+    },
+    {
+      "epoch": 0.4204498640520722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3014951299462213e-05,
+      "loss": 0.8885,
+      "step": 15309
+    },
+    {
+      "epoch": 0.4204773282799154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3014103162734102e-05,
+      "loss": 0.9235,
+      "step": 15310
+    },
+    {
+      "epoch": 0.42050479250775863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3013255002157237e-05,
+      "loss": 0.9296,
+      "step": 15311
+    },
+    {
+      "epoch": 0.4205322567356019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.301240681773833e-05,
+      "loss": 0.9466,
+      "step": 15312
+    },
+    {
+      "epoch": 0.4205597209634451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3011558609484096e-05,
+      "loss": 0.9019,
+      "step": 15313
+    },
+    {
+      "epoch": 0.42058718519128835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3010710377401242e-05,
+      "loss": 0.8994,
+      "step": 15314
+    },
+    {
+      "epoch": 0.42061464941913157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3009862121496481e-05,
+      "loss": 0.8942,
+      "step": 15315
+    },
+    {
+      "epoch": 0.4206421136469748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3009013841776526e-05,
+      "loss": 0.841,
+      "step": 15316
+    },
+    {
+      "epoch": 0.42066957787481807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3008165538248085e-05,
+      "loss": 0.9005,
+      "step": 15317
+    },
+    {
+      "epoch": 0.4206970421026613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3007317210917877e-05,
+      "loss": 0.8733,
+      "step": 15318
+    },
+    {
+      "epoch": 0.4207245063305045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3006468859792612e-05,
+      "loss": 0.9521,
+      "step": 15319
+    },
+    {
+      "epoch": 0.42075197055834773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3005620484878994e-05,
+      "loss": 0.9333,
+      "step": 15320
+    },
+    {
+      "epoch": 0.420779434786191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3004772086183745e-05,
+      "loss": 0.9946,
+      "step": 15321
+    },
+    {
+      "epoch": 0.42080689901403423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3003923663713578e-05,
+      "loss": 0.9642,
+      "step": 15322
+    },
+    {
+      "epoch": 0.42083436324187745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3003075217475204e-05,
+      "loss": 0.9431,
+      "step": 15323
+    },
+    {
+      "epoch": 0.4208618274697207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.300222674747533e-05,
+      "loss": 1.005,
+      "step": 15324
+    },
+    {
+      "epoch": 0.42088929169756395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.300137825372068e-05,
+      "loss": 0.9163,
+      "step": 15325
+    },
+    {
+      "epoch": 0.42091675592540717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3000529736217959e-05,
+      "loss": 0.8545,
+      "step": 15326
+    },
+    {
+      "epoch": 0.4209442201532504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2999681194973888e-05,
+      "loss": 1.0609,
+      "step": 15327
+    },
+    {
+      "epoch": 0.4209716843810936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2998832629995177e-05,
+      "loss": 0.9214,
+      "step": 15328
+    },
+    {
+      "epoch": 0.42099914860893684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2997984041288536e-05,
+      "loss": 0.9756,
+      "step": 15329
+    },
+    {
+      "epoch": 0.4210266128367801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.299713542886069e-05,
+      "loss": 0.9382,
+      "step": 15330
+    },
+    {
+      "epoch": 0.42105407706462333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2996286792718345e-05,
+      "loss": 0.9561,
+      "step": 15331
+    },
+    {
+      "epoch": 0.42108154129246655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2995438132868217e-05,
+      "loss": 0.92,
+      "step": 15332
+    },
+    {
+      "epoch": 0.4211090055203098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2994589449317023e-05,
+      "loss": 0.9253,
+      "step": 15333
+    },
+    {
+      "epoch": 0.42113646974815305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2993740742071477e-05,
+      "loss": 0.9857,
+      "step": 15334
+    },
+    {
+      "epoch": 0.4211639339759963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2992892011138296e-05,
+      "loss": 0.8738,
+      "step": 15335
+    },
+    {
+      "epoch": 0.4211913982038395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2992043256524192e-05,
+      "loss": 0.8885,
+      "step": 15336
+    },
+    {
+      "epoch": 0.4212188624316827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2991194478235885e-05,
+      "loss": 0.8577,
+      "step": 15337
+    },
+    {
+      "epoch": 0.421246326659526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2990345676280085e-05,
+      "loss": 0.9446,
+      "step": 15338
+    },
+    {
+      "epoch": 0.4212737908873692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2989496850663517e-05,
+      "loss": 0.9663,
+      "step": 15339
+    },
+    {
+      "epoch": 0.42130125511521244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2988648001392889e-05,
+      "loss": 0.9566,
+      "step": 15340
+    },
+    {
+      "epoch": 0.42132871934305566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298779912847492e-05,
+      "loss": 0.9848,
+      "step": 15341
+    },
+    {
+      "epoch": 0.4213561835708989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298695023191633e-05,
+      "loss": 0.983,
+      "step": 15342
+    },
+    {
+      "epoch": 0.42138364779874216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2986101311723829e-05,
+      "loss": 0.8864,
+      "step": 15343
+    },
+    {
+      "epoch": 0.4214111120265854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2985252367904137e-05,
+      "loss": 0.9225,
+      "step": 15344
+    },
+    {
+      "epoch": 0.4214385762544286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2984403400463973e-05,
+      "loss": 0.9456,
+      "step": 15345
+    },
+    {
+      "epoch": 0.4214660404822718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2983554409410054e-05,
+      "loss": 0.9677,
+      "step": 15346
+    },
+    {
+      "epoch": 0.4214935047101151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2982705394749097e-05,
+      "loss": 0.9999,
+      "step": 15347
+    },
+    {
+      "epoch": 0.4215209689379583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298185635648782e-05,
+      "loss": 0.9984,
+      "step": 15348
+    },
+    {
+      "epoch": 0.42154843316580154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298100729463294e-05,
+      "loss": 0.943,
+      "step": 15349
+    },
+    {
+      "epoch": 0.42157589739364476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2980158209191173e-05,
+      "loss": 0.9431,
+      "step": 15350
+    },
+    {
+      "epoch": 0.42160336162148804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2979309100169246e-05,
+      "loss": 0.9162,
+      "step": 15351
+    },
+    {
+      "epoch": 0.42163082584933126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2978459967573866e-05,
+      "loss": 0.9504,
+      "step": 15352
+    },
+    {
+      "epoch": 0.4216582900771745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2977610811411758e-05,
+      "loss": 0.9974,
+      "step": 15353
+    },
+    {
+      "epoch": 0.4216857543050177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2976761631689641e-05,
+      "loss": 0.9111,
+      "step": 15354
+    },
+    {
+      "epoch": 0.4217132185328609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2975912428414232e-05,
+      "loss": 0.9258,
+      "step": 15355
+    },
+    {
+      "epoch": 0.4217406827607042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.297506320159225e-05,
+      "loss": 0.9554,
+      "step": 15356
+    },
+    {
+      "epoch": 0.4217681469885474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2974213951230418e-05,
+      "loss": 0.9768,
+      "step": 15357
+    },
+    {
+      "epoch": 0.42179561121639064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2973364677335451e-05,
+      "loss": 0.9986,
+      "step": 15358
+    },
+    {
+      "epoch": 0.42182307544423386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2972515379914073e-05,
+      "loss": 1.0325,
+      "step": 15359
+    },
+    {
+      "epoch": 0.42185053967207714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2971666058973002e-05,
+      "loss": 0.8731,
+      "step": 15360
+    },
+    {
+      "epoch": 0.42187800389992036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2970816714518954e-05,
+      "loss": 0.9602,
+      "step": 15361
+    },
+    {
+      "epoch": 0.4219054681277636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2969967346558656e-05,
+      "loss": 0.9842,
+      "step": 15362
+    },
+    {
+      "epoch": 0.4219329323556068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296911795509883e-05,
+      "loss": 0.9418,
+      "step": 15363
+    },
+    {
+      "epoch": 0.4219603965834501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2968268540146185e-05,
+      "loss": 0.9767,
+      "step": 15364
+    },
+    {
+      "epoch": 0.4219878608112933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2967419101707453e-05,
+      "loss": 1.0207,
+      "step": 15365
+    },
+    {
+      "epoch": 0.4220153250391365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2966569639789353e-05,
+      "loss": 0.9518,
+      "step": 15366
+    },
+    {
+      "epoch": 0.42204278926697975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2965720154398604e-05,
+      "loss": 0.9614,
+      "step": 15367
+    },
+    {
+      "epoch": 0.42207025349482297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2964870645541927e-05,
+      "loss": 0.887,
+      "step": 15368
+    },
+    {
+      "epoch": 0.42209771772266624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2964021113226046e-05,
+      "loss": 0.9695,
+      "step": 15369
+    },
+    {
+      "epoch": 0.42212518195050946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296317155745768e-05,
+      "loss": 0.9779,
+      "step": 15370
+    },
+    {
+      "epoch": 0.4221526461783527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2962321978243555e-05,
+      "loss": 0.8701,
+      "step": 15371
+    },
+    {
+      "epoch": 0.4221801104061959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2961472375590392e-05,
+      "loss": 0.9358,
+      "step": 15372
+    },
+    {
+      "epoch": 0.4222075746340392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2960622749504909e-05,
+      "loss": 0.8825,
+      "step": 15373
+    },
+    {
+      "epoch": 0.4222350388618824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2959773099993834e-05,
+      "loss": 0.92,
+      "step": 15374
+    },
+    {
+      "epoch": 0.4222625030897256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.295892342706389e-05,
+      "loss": 0.8698,
+      "step": 15375
+    },
+    {
+      "epoch": 0.42228996731756885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2958073730721795e-05,
+      "loss": 0.8224,
+      "step": 15376
+    },
+    {
+      "epoch": 0.4223174315454121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2957224010974275e-05,
+      "loss": 0.9877,
+      "step": 15377
+    },
+    {
+      "epoch": 0.42234489577325535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2956374267828053e-05,
+      "loss": 0.9025,
+      "step": 15378
+    },
+    {
+      "epoch": 0.42237236000109857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2955524501289854e-05,
+      "loss": 0.939,
+      "step": 15379
+    },
+    {
+      "epoch": 0.4223998242289418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2954674711366398e-05,
+      "loss": 1.0573,
+      "step": 15380
+    },
+    {
+      "epoch": 0.422427288456785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2953824898064412e-05,
+      "loss": 0.8255,
+      "step": 15381
+    },
+    {
+      "epoch": 0.4224547526846283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2952975061390618e-05,
+      "loss": 0.9309,
+      "step": 15382
+    },
+    {
+      "epoch": 0.4224822169124715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2952125201351743e-05,
+      "loss": 0.921,
+      "step": 15383
+    },
+    {
+      "epoch": 0.42250968114031473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2951275317954512e-05,
+      "loss": 0.8984,
+      "step": 15384
+    },
+    {
+      "epoch": 0.42253714536815795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2950425411205643e-05,
+      "loss": 0.9406,
+      "step": 15385
+    },
+    {
+      "epoch": 0.42256460959600123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2949575481111867e-05,
+      "loss": 0.9262,
+      "step": 15386
+    },
+    {
+      "epoch": 0.42259207382384445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294872552767991e-05,
+      "loss": 1.0212,
+      "step": 15387
+    },
+    {
+      "epoch": 0.42261953805168767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294787555091649e-05,
+      "loss": 1.0118,
+      "step": 15388
+    },
+    {
+      "epoch": 0.4226470022795309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2947025550828338e-05,
+      "loss": 0.9336,
+      "step": 15389
+    },
+    {
+      "epoch": 0.42267446650737417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294617552742218e-05,
+      "loss": 0.9917,
+      "step": 15390
+    },
+    {
+      "epoch": 0.4227019307352174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294532548070474e-05,
+      "loss": 0.9673,
+      "step": 15391
+    },
+    {
+      "epoch": 0.4227293949630606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2944475410682741e-05,
+      "loss": 0.8928,
+      "step": 15392
+    },
+    {
+      "epoch": 0.42275685919090383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2943625317362916e-05,
+      "loss": 0.9139,
+      "step": 15393
+    },
+    {
+      "epoch": 0.42278432341874705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2942775200751985e-05,
+      "loss": 0.8619,
+      "step": 15394
+    },
+    {
+      "epoch": 0.42281178764659033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2941925060856677e-05,
+      "loss": 0.8795,
+      "step": 15395
+    },
+    {
+      "epoch": 0.42283925187443355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294107489768372e-05,
+      "loss": 0.8764,
+      "step": 15396
+    },
+    {
+      "epoch": 0.4228667161022768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2940224711239835e-05,
+      "loss": 0.8939,
+      "step": 15397
+    },
+    {
+      "epoch": 0.42289418033012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2939374501531754e-05,
+      "loss": 0.9625,
+      "step": 15398
+    },
+    {
+      "epoch": 0.42292164455796327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2938524268566208e-05,
+      "loss": 0.9336,
+      "step": 15399
+    },
+    {
+      "epoch": 0.4229491087858065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2937674012349915e-05,
+      "loss": 0.8324,
+      "step": 15400
+    },
+    {
+      "epoch": 0.4229765730136497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2936823732889607e-05,
+      "loss": 0.9444,
+      "step": 15401
+    },
+    {
+      "epoch": 0.42300403724149294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2935973430192017e-05,
+      "loss": 0.9235,
+      "step": 15402
+    },
+    {
+      "epoch": 0.4230315014693362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2935123104263864e-05,
+      "loss": 0.9355,
+      "step": 15403
+    },
+    {
+      "epoch": 0.42305896569717943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.293427275511188e-05,
+      "loss": 0.8994,
+      "step": 15404
+    },
+    {
+      "epoch": 0.42308642992502266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2933422382742795e-05,
+      "loss": 0.7981,
+      "step": 15405
+    },
+    {
+      "epoch": 0.4231138941528659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2932571987163333e-05,
+      "loss": 0.8582,
+      "step": 15406
+    },
+    {
+      "epoch": 0.42314135838070915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2931721568380229e-05,
+      "loss": 1.0881,
+      "step": 15407
+    },
+    {
+      "epoch": 0.4231688226085524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2930871126400208e-05,
+      "loss": 0.9375,
+      "step": 15408
+    },
+    {
+      "epoch": 0.4231962868363956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2930020661229997e-05,
+      "loss": 0.9417,
+      "step": 15409
+    },
+    {
+      "epoch": 0.4232237510642388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2929170172876329e-05,
+      "loss": 1.0349,
+      "step": 15410
+    },
+    {
+      "epoch": 0.42325121529208204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2928319661345932e-05,
+      "loss": 0.9703,
+      "step": 15411
+    },
+    {
+      "epoch": 0.4232786795199253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2927469126645533e-05,
+      "loss": 0.8507,
+      "step": 15412
+    },
+    {
+      "epoch": 0.42330614374776854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2926618568781869e-05,
+      "loss": 0.8997,
+      "step": 15413
+    },
+    {
+      "epoch": 0.42333360797561176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2925767987761661e-05,
+      "loss": 0.9918,
+      "step": 15414
+    },
+    {
+      "epoch": 0.423361072203455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2924917383591647e-05,
+      "loss": 0.8692,
+      "step": 15415
+    },
+    {
+      "epoch": 0.42338853643129826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.292406675627855e-05,
+      "loss": 0.9023,
+      "step": 15416
+    },
+    {
+      "epoch": 0.4234160006591415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2923216105829103e-05,
+      "loss": 0.9948,
+      "step": 15417
+    },
+    {
+      "epoch": 0.4234434648869847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2922365432250042e-05,
+      "loss": 0.9531,
+      "step": 15418
+    },
+    {
+      "epoch": 0.4234709291148279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2921514735548095e-05,
+      "loss": 1.0306,
+      "step": 15419
+    },
+    {
+      "epoch": 0.4234983933426712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2920664015729986e-05,
+      "loss": 0.9807,
+      "step": 15420
+    },
+    {
+      "epoch": 0.4235258575705144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2919813272802455e-05,
+      "loss": 0.9341,
+      "step": 15421
+    },
+    {
+      "epoch": 0.42355332179835764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291896250677223e-05,
+      "loss": 0.931,
+      "step": 15422
+    },
+    {
+      "epoch": 0.42358078602620086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2918111717646045e-05,
+      "loss": 1.0122,
+      "step": 15423
+    },
+    {
+      "epoch": 0.4236082502540441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2917260905430626e-05,
+      "loss": 0.8891,
+      "step": 15424
+    },
+    {
+      "epoch": 0.42363571448188736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2916410070132711e-05,
+      "loss": 0.9296,
+      "step": 15425
+    },
+    {
+      "epoch": 0.4236631787097306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291555921175903e-05,
+      "loss": 0.8835,
+      "step": 15426
+    },
+    {
+      "epoch": 0.4236906429375738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2914708330316314e-05,
+      "loss": 0.9319,
+      "step": 15427
+    },
+    {
+      "epoch": 0.423718107165417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.29138574258113e-05,
+      "loss": 1.0527,
+      "step": 15428
+    },
+    {
+      "epoch": 0.4237455713932603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2913006498250714e-05,
+      "loss": 0.9429,
+      "step": 15429
+    },
+    {
+      "epoch": 0.4237730356211035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291215554764129e-05,
+      "loss": 0.8184,
+      "step": 15430
+    },
+    {
+      "epoch": 0.42380049984894674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291130457398977e-05,
+      "loss": 0.8704,
+      "step": 15431
+    },
+    {
+      "epoch": 0.42382796407678996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2910453577302874e-05,
+      "loss": 0.9619,
+      "step": 15432
+    },
+    {
+      "epoch": 0.42385542830463324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2909602557587344e-05,
+      "loss": 0.9079,
+      "step": 15433
+    },
+    {
+      "epoch": 0.42388289253247646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.290875151484991e-05,
+      "loss": 0.9094,
+      "step": 15434
+    },
+    {
+      "epoch": 0.4239103567603197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.290790044909731e-05,
+      "loss": 0.9175,
+      "step": 15435
+    },
+    {
+      "epoch": 0.4239378209881629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2907049360336275e-05,
+      "loss": 0.9348,
+      "step": 15436
+    },
+    {
+      "epoch": 0.4239652852160061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2906198248573541e-05,
+      "loss": 0.9466,
+      "step": 15437
+    },
+    {
+      "epoch": 0.4239927494438494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2905347113815836e-05,
+      "loss": 0.7922,
+      "step": 15438
+    },
+    {
+      "epoch": 0.4240202136716926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2904495956069902e-05,
+      "loss": 0.8981,
+      "step": 15439
+    },
+    {
+      "epoch": 0.42404767789953585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2903644775342472e-05,
+      "loss": 0.9453,
+      "step": 15440
+    },
+    {
+      "epoch": 0.42407514212737907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2902793571640278e-05,
+      "loss": 0.9532,
+      "step": 15441
+    },
+    {
+      "epoch": 0.42410260635522234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2901942344970056e-05,
+      "loss": 0.8486,
+      "step": 15442
+    },
+    {
+      "epoch": 0.42413007058306557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2901091095338545e-05,
+      "loss": 0.8739,
+      "step": 15443
+    },
+    {
+      "epoch": 0.4241575348109088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2900239822752474e-05,
+      "loss": 0.805,
+      "step": 15444
+    },
+    {
+      "epoch": 0.424184999038752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2899388527218583e-05,
+      "loss": 0.8275,
+      "step": 15445
+    },
+    {
+      "epoch": 0.4242124632665953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2898537208743607e-05,
+      "loss": 0.8729,
+      "step": 15446
+    },
+    {
+      "epoch": 0.4242399274944385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2897685867334285e-05,
+      "loss": 0.9632,
+      "step": 15447
+    },
+    {
+      "epoch": 0.42426739172228173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2896834502997346e-05,
+      "loss": 0.9598,
+      "step": 15448
+    },
+    {
+      "epoch": 0.42429485595012495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2895983115739532e-05,
+      "loss": 0.9325,
+      "step": 15449
+    },
+    {
+      "epoch": 0.42432232017796817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2895131705567577e-05,
+      "loss": 0.882,
+      "step": 15450
+    },
+    {
+      "epoch": 0.42434978440581145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.289428027248822e-05,
+      "loss": 0.9799,
+      "step": 15451
+    },
+    {
+      "epoch": 0.42437724863365467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2893428816508195e-05,
+      "loss": 1.0514,
+      "step": 15452
+    },
+    {
+      "epoch": 0.4244047128614979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.289257733763424e-05,
+      "loss": 0.9554,
+      "step": 15453
+    },
+    {
+      "epoch": 0.4244321770893411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2891725835873092e-05,
+      "loss": 0.951,
+      "step": 15454
+    },
+    {
+      "epoch": 0.4244596413171844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2890874311231492e-05,
+      "loss": 0.9088,
+      "step": 15455
+    },
+    {
+      "epoch": 0.4244871055450276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2890022763716173e-05,
+      "loss": 0.9384,
+      "step": 15456
+    },
+    {
+      "epoch": 0.42451456977287083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2889171193333873e-05,
+      "loss": 1.0066,
+      "step": 15457
+    },
+    {
+      "epoch": 0.42454203400071405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2888319600091332e-05,
+      "loss": 0.9732,
+      "step": 15458
+    },
+    {
+      "epoch": 0.42456949822855733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.288746798399529e-05,
+      "loss": 0.9151,
+      "step": 15459
+    },
+    {
+      "epoch": 0.42459696245640055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2886616345052481e-05,
+      "loss": 0.9458,
+      "step": 15460
+    },
+    {
+      "epoch": 0.42462442668424377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2885764683269646e-05,
+      "loss": 0.9504,
+      "step": 15461
+    },
+    {
+      "epoch": 0.424651890912087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.288491299865352e-05,
+      "loss": 0.9247,
+      "step": 15462
+    },
+    {
+      "epoch": 0.4246793551399302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2884061291210848e-05,
+      "loss": 0.8283,
+      "step": 15463
+    },
+    {
+      "epoch": 0.4247068193677735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2883209560948366e-05,
+      "loss": 0.8922,
+      "step": 15464
+    },
+    {
+      "epoch": 0.4247342835956167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2882357807872811e-05,
+      "loss": 0.9682,
+      "step": 15465
+    },
+    {
+      "epoch": 0.42476174782345993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2881506031990925e-05,
+      "loss": 0.8859,
+      "step": 15466
+    },
+    {
+      "epoch": 0.42478921205130316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2880654233309451e-05,
+      "loss": 0.9362,
+      "step": 15467
+    },
+    {
+      "epoch": 0.42481667627914643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.287980241183512e-05,
+      "loss": 0.9549,
+      "step": 15468
+    },
+    {
+      "epoch": 0.42484414050698965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.287895056757468e-05,
+      "loss": 0.9698,
+      "step": 15469
+    },
+    {
+      "epoch": 0.4248716047348329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2878098700534864e-05,
+      "loss": 1.0673,
+      "step": 15470
+    },
+    {
+      "epoch": 0.4248990689626761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2877246810722422e-05,
+      "loss": 0.9308,
+      "step": 15471
+    },
+    {
+      "epoch": 0.4249265331905194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2876394898144084e-05,
+      "loss": 0.8633,
+      "step": 15472
+    },
+    {
+      "epoch": 0.4249539974183626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2875542962806595e-05,
+      "loss": 0.9633,
+      "step": 15473
+    },
+    {
+      "epoch": 0.4249814616462058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2874691004716697e-05,
+      "loss": 0.9808,
+      "step": 15474
+    },
+    {
+      "epoch": 0.42500892587404904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2873839023881134e-05,
+      "loss": 1.0092,
+      "step": 15475
+    },
+    {
+      "epoch": 0.42503639010189226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.287298702030664e-05,
+      "loss": 0.8525,
+      "step": 15476
+    },
+    {
+      "epoch": 0.42506385432973554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2872134993999958e-05,
+      "loss": 1.0364,
+      "step": 15477
+    },
+    {
+      "epoch": 0.42509131855757876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2871282944967836e-05,
+      "loss": 0.961,
+      "step": 15478
+    },
+    {
+      "epoch": 0.425118782785422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2870430873217007e-05,
+      "loss": 0.9294,
+      "step": 15479
+    },
+    {
+      "epoch": 0.4251462470132652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2869578778754219e-05,
+      "loss": 0.7975,
+      "step": 15480
+    },
+    {
+      "epoch": 0.4251737112411085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2868726661586212e-05,
+      "loss": 0.9528,
+      "step": 15481
+    },
+    {
+      "epoch": 0.4252011754689517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2867874521719728e-05,
+      "loss": 0.8062,
+      "step": 15482
+    },
+    {
+      "epoch": 0.4252286396967949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2867022359161512e-05,
+      "loss": 0.853,
+      "step": 15483
+    },
+    {
+      "epoch": 0.42525610392463814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2866170173918302e-05,
+      "loss": 0.9318,
+      "step": 15484
+    },
+    {
+      "epoch": 0.4252835681524814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2865317965996843e-05,
+      "loss": 0.9598,
+      "step": 15485
+    },
+    {
+      "epoch": 0.42531103238032464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2864465735403879e-05,
+      "loss": 0.8828,
+      "step": 15486
+    },
+    {
+      "epoch": 0.42533849660816786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2863613482146155e-05,
+      "loss": 0.9141,
+      "step": 15487
+    },
+    {
+      "epoch": 0.4253659608360111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.286276120623041e-05,
+      "loss": 0.8225,
+      "step": 15488
+    },
+    {
+      "epoch": 0.4253934250638543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2861908907663386e-05,
+      "loss": 0.883,
+      "step": 15489
+    },
+    {
+      "epoch": 0.4254208892916976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2861056586451834e-05,
+      "loss": 0.9027,
+      "step": 15490
+    },
+    {
+      "epoch": 0.4254483535195408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2860204242602495e-05,
+      "loss": 0.895,
+      "step": 15491
+    },
+    {
+      "epoch": 0.425475817747384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2859351876122109e-05,
+      "loss": 0.9837,
+      "step": 15492
+    },
+    {
+      "epoch": 0.42550328197522724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2858499487017427e-05,
+      "loss": 0.9816,
+      "step": 15493
+    },
+    {
+      "epoch": 0.4255307462030705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2857647075295185e-05,
+      "loss": 0.9653,
+      "step": 15494
+    },
+    {
+      "epoch": 0.42555821043091374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2856794640962136e-05,
+      "loss": 0.9399,
+      "step": 15495
+    },
+    {
+      "epoch": 0.42558567465875696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2855942184025021e-05,
+      "loss": 0.892,
+      "step": 15496
+    },
+    {
+      "epoch": 0.4256131388866002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2855089704490582e-05,
+      "loss": 0.9734,
+      "step": 15497
+    },
+    {
+      "epoch": 0.42564060311444346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2854237202365569e-05,
+      "loss": 0.9664,
+      "step": 15498
+    },
+    {
+      "epoch": 0.4256680673422867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2853384677656728e-05,
+      "loss": 0.9298,
+      "step": 15499
+    },
+    {
+      "epoch": 0.4256955315701299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28525321303708e-05,
+      "loss": 0.9784,
+      "step": 15500
+    },
+    {
+      "epoch": 0.4257229957979731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2851679560514532e-05,
+      "loss": 0.9236,
+      "step": 15501
+    },
+    {
+      "epoch": 0.4257504600258164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2850826968094672e-05,
+      "loss": 0.9685,
+      "step": 15502
+    },
+    {
+      "epoch": 0.4257779242536596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2849974353117967e-05,
+      "loss": 0.9609,
+      "step": 15503
+    },
+    {
+      "epoch": 0.42580538848150284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2849121715591157e-05,
+      "loss": 0.9569,
+      "step": 15504
+    },
+    {
+      "epoch": 0.42583285270934607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2848269055520994e-05,
+      "loss": 0.9048,
+      "step": 15505
+    },
+    {
+      "epoch": 0.4258603169371893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2847416372914221e-05,
+      "loss": 0.8941,
+      "step": 15506
+    },
+    {
+      "epoch": 0.42588778116503256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284656366777759e-05,
+      "loss": 1.0406,
+      "step": 15507
+    },
+    {
+      "epoch": 0.4259152453928758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2845710940117843e-05,
+      "loss": 0.886,
+      "step": 15508
+    },
+    {
+      "epoch": 0.425942709620719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2844858189941729e-05,
+      "loss": 0.9577,
+      "step": 15509
+    },
+    {
+      "epoch": 0.42597017384856223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2844005417255994e-05,
+      "loss": 0.9194,
+      "step": 15510
+    },
+    {
+      "epoch": 0.4259976380764055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284315262206739e-05,
+      "loss": 0.91,
+      "step": 15511
+    },
+    {
+      "epoch": 0.4260251023042487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2842299804382656e-05,
+      "loss": 0.9463,
+      "step": 15512
+    },
+    {
+      "epoch": 0.42605256653209195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2841446964208546e-05,
+      "loss": 0.9069,
+      "step": 15513
+    },
+    {
+      "epoch": 0.42608003075993517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2840594101551809e-05,
+      "loss": 0.9534,
+      "step": 15514
+    },
+    {
+      "epoch": 0.42610749498777845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2839741216419192e-05,
+      "loss": 0.9307,
+      "step": 15515
+    },
+    {
+      "epoch": 0.42613495921562167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2838888308817439e-05,
+      "loss": 0.8314,
+      "step": 15516
+    },
+    {
+      "epoch": 0.4261624234434649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2838035378753304e-05,
+      "loss": 0.9476,
+      "step": 15517
+    },
+    {
+      "epoch": 0.4261898876713081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2837182426233534e-05,
+      "loss": 0.9753,
+      "step": 15518
+    },
+    {
+      "epoch": 0.42621735189915133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2836329451264877e-05,
+      "loss": 0.8428,
+      "step": 15519
+    },
+    {
+      "epoch": 0.4262448161269946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2835476453854084e-05,
+      "loss": 0.9221,
+      "step": 15520
+    },
+    {
+      "epoch": 0.42627228035483783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2834623434007902e-05,
+      "loss": 0.9606,
+      "step": 15521
+    },
+    {
+      "epoch": 0.42629974458268105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283377039173308e-05,
+      "loss": 0.9027,
+      "step": 15522
+    },
+    {
+      "epoch": 0.42632720881052427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2832917327036372e-05,
+      "loss": 0.9219,
+      "step": 15523
+    },
+    {
+      "epoch": 0.42635467303836755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2832064239924522e-05,
+      "loss": 0.9393,
+      "step": 15524
+    },
+    {
+      "epoch": 0.42638213726621077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2831211130404282e-05,
+      "loss": 0.9,
+      "step": 15525
+    },
+    {
+      "epoch": 0.426409601494054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2830357998482405e-05,
+      "loss": 0.8275,
+      "step": 15526
+    },
+    {
+      "epoch": 0.4264370657218972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2829504844165641e-05,
+      "loss": 1.0213,
+      "step": 15527
+    },
+    {
+      "epoch": 0.4264645299497405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2828651667460737e-05,
+      "loss": 0.7364,
+      "step": 15528
+    },
+    {
+      "epoch": 0.4264919941775837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2827798468374445e-05,
+      "loss": 0.8298,
+      "step": 15529
+    },
+    {
+      "epoch": 0.42651945840542693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2826945246913511e-05,
+      "loss": 0.9601,
+      "step": 15530
+    },
+    {
+      "epoch": 0.42654692263327015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28260920030847e-05,
+      "loss": 0.902,
+      "step": 15531
+    },
+    {
+      "epoch": 0.4265743868611134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.282523873689475e-05,
+      "loss": 0.9644,
+      "step": 15532
+    },
+    {
+      "epoch": 0.42660185108895665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2824385448350413e-05,
+      "loss": 0.9622,
+      "step": 15533
+    },
+    {
+      "epoch": 0.4266293153167999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2823532137458449e-05,
+      "loss": 0.8873,
+      "step": 15534
+    },
+    {
+      "epoch": 0.4266567795446431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2822678804225608e-05,
+      "loss": 0.9415,
+      "step": 15535
+    },
+    {
+      "epoch": 0.4266842437724863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2821825448658634e-05,
+      "loss": 0.8481,
+      "step": 15536
+    },
+    {
+      "epoch": 0.4267117080003296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2820972070764283e-05,
+      "loss": 0.8862,
+      "step": 15537
+    },
+    {
+      "epoch": 0.4267391722281728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2820118670549312e-05,
+      "loss": 0.9886,
+      "step": 15538
+    },
+    {
+      "epoch": 0.42676663645601604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2819265248020466e-05,
+      "loss": 0.9304,
+      "step": 15539
+    },
+    {
+      "epoch": 0.42679410068385926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2818411803184505e-05,
+      "loss": 0.937,
+      "step": 15540
+    },
+    {
+      "epoch": 0.42682156491170253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2817558336048175e-05,
+      "loss": 0.8511,
+      "step": 15541
+    },
+    {
+      "epoch": 0.42684902913954575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2816704846618233e-05,
+      "loss": 0.9106,
+      "step": 15542
+    },
+    {
+      "epoch": 0.426876493367389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2815851334901435e-05,
+      "loss": 0.8755,
+      "step": 15543
+    },
+    {
+      "epoch": 0.4269039575952322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2814997800904524e-05,
+      "loss": 0.8875,
+      "step": 15544
+    },
+    {
+      "epoch": 0.4269314218230754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.281414424463426e-05,
+      "loss": 0.8964,
+      "step": 15545
+    },
+    {
+      "epoch": 0.4269588860509187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2813290666097402e-05,
+      "loss": 0.9288,
+      "step": 15546
+    },
+    {
+      "epoch": 0.4269863502787619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2812437065300698e-05,
+      "loss": 0.9938,
+      "step": 15547
+    },
+    {
+      "epoch": 0.42701381450660514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2811583442250899e-05,
+      "loss": 0.8912,
+      "step": 15548
+    },
+    {
+      "epoch": 0.42704127873444836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2810729796954764e-05,
+      "loss": 0.8976,
+      "step": 15549
+    },
+    {
+      "epoch": 0.42706874296229164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2809876129419045e-05,
+      "loss": 0.8926,
+      "step": 15550
+    },
+    {
+      "epoch": 0.42709620719013486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28090224396505e-05,
+      "loss": 0.9176,
+      "step": 15551
+    },
+    {
+      "epoch": 0.4271236714179781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2808168727655882e-05,
+      "loss": 0.9513,
+      "step": 15552
+    },
+    {
+      "epoch": 0.4271511356458213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2807314993441944e-05,
+      "loss": 0.7719,
+      "step": 15553
+    },
+    {
+      "epoch": 0.4271785998736646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.280646123701544e-05,
+      "loss": 0.8917,
+      "step": 15554
+    },
+    {
+      "epoch": 0.4272060641015078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2805607458383132e-05,
+      "loss": 0.8659,
+      "step": 15555
+    },
+    {
+      "epoch": 0.427233528329351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2804753657551768e-05,
+      "loss": 0.8883,
+      "step": 15556
+    },
+    {
+      "epoch": 0.42726099255719424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2803899834528103e-05,
+      "loss": 0.8802,
+      "step": 15557
+    },
+    {
+      "epoch": 0.42728845678503746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2803045989318903e-05,
+      "loss": 0.8833,
+      "step": 15558
+    },
+    {
+      "epoch": 0.42731592101288074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2802192121930917e-05,
+      "loss": 0.9234,
+      "step": 15559
+    },
+    {
+      "epoch": 0.42734338524072396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2801338232370897e-05,
+      "loss": 0.9769,
+      "step": 15560
+    },
+    {
+      "epoch": 0.4273708494685672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2800484320645605e-05,
+      "loss": 0.8415,
+      "step": 15561
+    },
+    {
+      "epoch": 0.4273983136964104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2799630386761797e-05,
+      "loss": 1.008,
+      "step": 15562
+    },
+    {
+      "epoch": 0.4274257779242537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2798776430726231e-05,
+      "loss": 0.7906,
+      "step": 15563
+    },
+    {
+      "epoch": 0.4274532421520969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2797922452545662e-05,
+      "loss": 0.9203,
+      "step": 15564
+    },
+    {
+      "epoch": 0.4274807063799401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2797068452226842e-05,
+      "loss": 0.9179,
+      "step": 15565
+    },
+    {
+      "epoch": 0.42750817060778334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2796214429776535e-05,
+      "loss": 0.8366,
+      "step": 15566
+    },
+    {
+      "epoch": 0.4275356348356266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2795360385201499e-05,
+      "loss": 0.9323,
+      "step": 15567
+    },
+    {
+      "epoch": 0.42756309906346984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2794506318508487e-05,
+      "loss": 0.9212,
+      "step": 15568
+    },
+    {
+      "epoch": 0.42759056329131306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2793652229704256e-05,
+      "loss": 0.9058,
+      "step": 15569
+    },
+    {
+      "epoch": 0.4276180275191563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2792798118795568e-05,
+      "loss": 0.9064,
+      "step": 15570
+    },
+    {
+      "epoch": 0.4276454917469995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2791943985789182e-05,
+      "loss": 0.8901,
+      "step": 15571
+    },
+    {
+      "epoch": 0.4276729559748428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2791089830691849e-05,
+      "loss": 0.8456,
+      "step": 15572
+    },
+    {
+      "epoch": 0.427700420202686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2790235653510336e-05,
+      "loss": 0.8373,
+      "step": 15573
+    },
+    {
+      "epoch": 0.4277278844305292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2789381454251397e-05,
+      "loss": 0.9063,
+      "step": 15574
+    },
+    {
+      "epoch": 0.42775534865837245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.278852723292179e-05,
+      "loss": 0.9401,
+      "step": 15575
+    },
+    {
+      "epoch": 0.4277828128862157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2787672989528276e-05,
+      "loss": 0.9006,
+      "step": 15576
+    },
+    {
+      "epoch": 0.42781027711405895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2786818724077614e-05,
+      "loss": 0.8509,
+      "step": 15577
+    },
+    {
+      "epoch": 0.42783774134190217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2785964436576562e-05,
+      "loss": 0.8154,
+      "step": 15578
+    },
+    {
+      "epoch": 0.4278652055697454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2785110127031884e-05,
+      "loss": 0.902,
+      "step": 15579
+    },
+    {
+      "epoch": 0.42789266979758867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.278425579545033e-05,
+      "loss": 0.7917,
+      "step": 15580
+    },
+    {
+      "epoch": 0.4279201340254319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2783401441838668e-05,
+      "loss": 0.9733,
+      "step": 15581
+    },
+    {
+      "epoch": 0.4279475982532751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2782547066203657e-05,
+      "loss": 1.0198,
+      "step": 15582
+    },
+    {
+      "epoch": 0.42797506248111833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2781692668552055e-05,
+      "loss": 0.8856,
+      "step": 15583
+    },
+    {
+      "epoch": 0.42800252670896155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2780838248890622e-05,
+      "loss": 0.8777,
+      "step": 15584
+    },
+    {
+      "epoch": 0.4280299909368048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277998380722612e-05,
+      "loss": 0.9313,
+      "step": 15585
+    },
+    {
+      "epoch": 0.42805745516464805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2779129343565309e-05,
+      "loss": 0.9321,
+      "step": 15586
+    },
+    {
+      "epoch": 0.42808491939249127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2778274857914949e-05,
+      "loss": 0.814,
+      "step": 15587
+    },
+    {
+      "epoch": 0.4281123836203345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2777420350281806e-05,
+      "loss": 0.9015,
+      "step": 15588
+    },
+    {
+      "epoch": 0.42813984784817777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2776565820672635e-05,
+      "loss": 0.9684,
+      "step": 15589
+    },
+    {
+      "epoch": 0.428167312076021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2775711269094199e-05,
+      "loss": 0.8995,
+      "step": 15590
+    },
+    {
+      "epoch": 0.4281947763038642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2774856695553265e-05,
+      "loss": 0.9696,
+      "step": 15591
+    },
+    {
+      "epoch": 0.42822224053170743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2774002100056582e-05,
+      "loss": 1.0036,
+      "step": 15592
+    },
+    {
+      "epoch": 0.4282497047595507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2773147482610923e-05,
+      "loss": 0.9561,
+      "step": 15593
+    },
+    {
+      "epoch": 0.42827716898739393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2772292843223048e-05,
+      "loss": 1.009,
+      "step": 15594
+    },
+    {
+      "epoch": 0.42830463321523715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2771438181899717e-05,
+      "loss": 1.0023,
+      "step": 15595
+    },
+    {
+      "epoch": 0.4283320974430804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2770583498647698e-05,
+      "loss": 0.8964,
+      "step": 15596
+    },
+    {
+      "epoch": 0.4283595616709236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2769728793473743e-05,
+      "loss": 0.8363,
+      "step": 15597
+    },
+    {
+      "epoch": 0.42838702589876687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2768874066384623e-05,
+      "loss": 0.8149,
+      "step": 15598
+    },
+    {
+      "epoch": 0.4284144901266101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2768019317387102e-05,
+      "loss": 0.9997,
+      "step": 15599
+    },
+    {
+      "epoch": 0.4284419543544533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2767164546487936e-05,
+      "loss": 0.8782,
+      "step": 15600
+    },
+    {
+      "epoch": 0.42846941858229654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276630975369389e-05,
+      "loss": 0.9996,
+      "step": 15601
+    },
+    {
+      "epoch": 0.4284968828101398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2765454939011734e-05,
+      "loss": 0.8893,
+      "step": 15602
+    },
+    {
+      "epoch": 0.42852434703798303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2764600102448228e-05,
+      "loss": 0.8936,
+      "step": 15603
+    },
+    {
+      "epoch": 0.42855181126582625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2763745244010134e-05,
+      "loss": 0.8794,
+      "step": 15604
+    },
+    {
+      "epoch": 0.4285792754936695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2762890363704217e-05,
+      "loss": 0.9611,
+      "step": 15605
+    },
+    {
+      "epoch": 0.42860673972151275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2762035461537238e-05,
+      "loss": 0.9227,
+      "step": 15606
+    },
+    {
+      "epoch": 0.428634203949356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276118053751597e-05,
+      "loss": 0.9045,
+      "step": 15607
+    },
+    {
+      "epoch": 0.4286616681771992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2760325591647168e-05,
+      "loss": 0.9386,
+      "step": 15608
+    },
+    {
+      "epoch": 0.4286891324050424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.27594706239376e-05,
+      "loss": 0.9501,
+      "step": 15609
+    },
+    {
+      "epoch": 0.4287165966328857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2758615634394031e-05,
+      "loss": 0.9446,
+      "step": 15610
+    },
+    {
+      "epoch": 0.4287440608607289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2757760623023235e-05,
+      "loss": 0.8157,
+      "step": 15611
+    },
+    {
+      "epoch": 0.42877152508857214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2756905589831959e-05,
+      "loss": 0.9055,
+      "step": 15612
+    },
+    {
+      "epoch": 0.42879898931641536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.275605053482698e-05,
+      "loss": 0.9697,
+      "step": 15613
+    },
+    {
+      "epoch": 0.4288264535442586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2755195458015064e-05,
+      "loss": 0.846,
+      "step": 15614
+    },
+    {
+      "epoch": 0.42885391777210186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2754340359402975e-05,
+      "loss": 0.9712,
+      "step": 15615
+    },
+    {
+      "epoch": 0.4288813819999451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2753485238997476e-05,
+      "loss": 0.8664,
+      "step": 15616
+    },
+    {
+      "epoch": 0.4289088462277883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2752630096805336e-05,
+      "loss": 0.8097,
+      "step": 15617
+    },
+    {
+      "epoch": 0.4289363104556315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2751774932833318e-05,
+      "loss": 0.956,
+      "step": 15618
+    },
+    {
+      "epoch": 0.4289637746834748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2750919747088195e-05,
+      "loss": 0.9526,
+      "step": 15619
+    },
+    {
+      "epoch": 0.428991238911318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2750064539576728e-05,
+      "loss": 0.9937,
+      "step": 15620
+    },
+    {
+      "epoch": 0.42901870313916124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2749209310305683e-05,
+      "loss": 1.0081,
+      "step": 15621
+    },
+    {
+      "epoch": 0.42904616736700446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2748354059281829e-05,
+      "loss": 1.0453,
+      "step": 15622
+    },
+    {
+      "epoch": 0.42907363159484774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2747498786511937e-05,
+      "loss": 1.0142,
+      "step": 15623
+    },
+    {
+      "epoch": 0.42910109582269096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2746643492002769e-05,
+      "loss": 0.9449,
+      "step": 15624
+    },
+    {
+      "epoch": 0.4291285600505342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.274578817576109e-05,
+      "loss": 0.8412,
+      "step": 15625
+    },
+    {
+      "epoch": 0.4291560242783774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2744932837793673e-05,
+      "loss": 0.8396,
+      "step": 15626
+    },
+    {
+      "epoch": 0.4291834885062206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2744077478107286e-05,
+      "loss": 0.9655,
+      "step": 15627
+    },
+    {
+      "epoch": 0.4292109527340639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2743222096708695e-05,
+      "loss": 0.9146,
+      "step": 15628
+    },
+    {
+      "epoch": 0.4292384169619071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2742366693604666e-05,
+      "loss": 0.9382,
+      "step": 15629
+    },
+    {
+      "epoch": 0.42926588118975034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2741511268801969e-05,
+      "loss": 1.0281,
+      "step": 15630
+    },
+    {
+      "epoch": 0.42929334541759356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2740655822307378e-05,
+      "loss": 0.9781,
+      "step": 15631
+    },
+    {
+      "epoch": 0.42932080964543684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2739800354127651e-05,
+      "loss": 0.8085,
+      "step": 15632
+    },
+    {
+      "epoch": 0.42934827387328006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2738944864269564e-05,
+      "loss": 0.9331,
+      "step": 15633
+    },
+    {
+      "epoch": 0.4293757381011233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2738089352739884e-05,
+      "loss": 0.8912,
+      "step": 15634
+    },
+    {
+      "epoch": 0.4294032023289665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2737233819545382e-05,
+      "loss": 0.9532,
+      "step": 15635
+    },
+    {
+      "epoch": 0.4294306665568098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2736378264692826e-05,
+      "loss": 0.9602,
+      "step": 15636
+    },
+    {
+      "epoch": 0.429458130784653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2735522688188981e-05,
+      "loss": 0.8782,
+      "step": 15637
+    },
+    {
+      "epoch": 0.4294855950124962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2734667090040624e-05,
+      "loss": 0.8723,
+      "step": 15638
+    },
+    {
+      "epoch": 0.42951305924033945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2733811470254525e-05,
+      "loss": 0.8153,
+      "step": 15639
+    },
+    {
+      "epoch": 0.42954052346818267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2732955828837445e-05,
+      "loss": 1.0307,
+      "step": 15640
+    },
+    {
+      "epoch": 0.42956798769602594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2732100165796162e-05,
+      "loss": 0.83,
+      "step": 15641
+    },
+    {
+      "epoch": 0.42959545192386916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2731244481137443e-05,
+      "loss": 0.8508,
+      "step": 15642
+    },
+    {
+      "epoch": 0.4296229161517124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.273038877486806e-05,
+      "loss": 1.0187,
+      "step": 15643
+    },
+    {
+      "epoch": 0.4296503803795556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2729533046994786e-05,
+      "loss": 0.9585,
+      "step": 15644
+    },
+    {
+      "epoch": 0.4296778446073989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2728677297524388e-05,
+      "loss": 0.8777,
+      "step": 15645
+    },
+    {
+      "epoch": 0.4297053088352421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2727821526463634e-05,
+      "loss": 0.9151,
+      "step": 15646
+    },
+    {
+      "epoch": 0.4297327730630853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2726965733819307e-05,
+      "loss": 0.854,
+      "step": 15647
+    },
+    {
+      "epoch": 0.42976023729092855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2726109919598165e-05,
+      "loss": 0.8465,
+      "step": 15648
+    },
+    {
+      "epoch": 0.4297877015187718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2725254083806989e-05,
+      "loss": 0.9202,
+      "step": 15649
+    },
+    {
+      "epoch": 0.42981516574661505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2724398226452543e-05,
+      "loss": 0.9716,
+      "step": 15650
+    },
+    {
+      "epoch": 0.42984262997445827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2723542347541607e-05,
+      "loss": 0.962,
+      "step": 15651
+    },
+    {
+      "epoch": 0.4298700942023015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2722686447080948e-05,
+      "loss": 0.9319,
+      "step": 15652
+    },
+    {
+      "epoch": 0.4298975584301447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2721830525077336e-05,
+      "loss": 0.866,
+      "step": 15653
+    },
+    {
+      "epoch": 0.429925022657988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2720974581537546e-05,
+      "loss": 0.89,
+      "step": 15654
+    },
+    {
+      "epoch": 0.4299524868858312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2720118616468356e-05,
+      "loss": 0.8627,
+      "step": 15655
+    },
+    {
+      "epoch": 0.42997995111367443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2719262629876532e-05,
+      "loss": 0.9405,
+      "step": 15656
+    },
+    {
+      "epoch": 0.43000741534151765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2718406621768846e-05,
+      "loss": 0.9082,
+      "step": 15657
+    },
+    {
+      "epoch": 0.43003487956936093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271755059215208e-05,
+      "loss": 0.9888,
+      "step": 15658
+    },
+    {
+      "epoch": 0.43006234379720415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2716694541032996e-05,
+      "loss": 0.8934,
+      "step": 15659
+    },
+    {
+      "epoch": 0.43008980802504737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2715838468418373e-05,
+      "loss": 0.8013,
+      "step": 15660
+    },
+    {
+      "epoch": 0.4301172722528906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2714982374314986e-05,
+      "loss": 0.9699,
+      "step": 15661
+    },
+    {
+      "epoch": 0.43014473648073387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2714126258729604e-05,
+      "loss": 0.9539,
+      "step": 15662
+    },
+    {
+      "epoch": 0.4301722007085771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2713270121669006e-05,
+      "loss": 0.8508,
+      "step": 15663
+    },
+    {
+      "epoch": 0.4301996649364203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2712413963139962e-05,
+      "loss": 0.8482,
+      "step": 15664
+    },
+    {
+      "epoch": 0.43022712916426353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271155778314925e-05,
+      "loss": 1.0069,
+      "step": 15665
+    },
+    {
+      "epoch": 0.43025459339210675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271070158170364e-05,
+      "loss": 0.8946,
+      "step": 15666
+    },
+    {
+      "epoch": 0.43028205761995003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2709845358809912e-05,
+      "loss": 0.8308,
+      "step": 15667
+    },
+    {
+      "epoch": 0.43030952184779325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2708989114474837e-05,
+      "loss": 0.8994,
+      "step": 15668
+    },
+    {
+      "epoch": 0.4303369860756365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2708132848705189e-05,
+      "loss": 0.9664,
+      "step": 15669
+    },
+    {
+      "epoch": 0.4303644503034797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2707276561507747e-05,
+      "loss": 0.9928,
+      "step": 15670
+    },
+    {
+      "epoch": 0.43039191453132297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2706420252889281e-05,
+      "loss": 0.9487,
+      "step": 15671
+    },
+    {
+      "epoch": 0.4304193787591662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2705563922856572e-05,
+      "loss": 0.9108,
+      "step": 15672
+    },
+    {
+      "epoch": 0.4304468429870094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2704707571416393e-05,
+      "loss": 0.9599,
+      "step": 15673
+    },
+    {
+      "epoch": 0.43047430721485264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2703851198575519e-05,
+      "loss": 0.9789,
+      "step": 15674
+    },
+    {
+      "epoch": 0.4305017714426959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2702994804340729e-05,
+      "loss": 0.9431,
+      "step": 15675
+    },
+    {
+      "epoch": 0.43052923567053913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2702138388718797e-05,
+      "loss": 0.9229,
+      "step": 15676
+    },
+    {
+      "epoch": 0.43055669989838236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2701281951716497e-05,
+      "loss": 0.9038,
+      "step": 15677
+    },
+    {
+      "epoch": 0.4305841641262256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2700425493340608e-05,
+      "loss": 0.8642,
+      "step": 15678
+    },
+    {
+      "epoch": 0.4306116283540688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.269956901359791e-05,
+      "loss": 0.8251,
+      "step": 15679
+    },
+    {
+      "epoch": 0.4306390925819121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2698712512495173e-05,
+      "loss": 0.8976,
+      "step": 15680
+    },
+    {
+      "epoch": 0.4306665568097553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2697855990039176e-05,
+      "loss": 0.8807,
+      "step": 15681
+    },
+    {
+      "epoch": 0.4306940210375985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.26969994462367e-05,
+      "loss": 0.9367,
+      "step": 15682
+    },
+    {
+      "epoch": 0.43072148526544174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2696142881094519e-05,
+      "loss": 0.8838,
+      "step": 15683
+    },
+    {
+      "epoch": 0.430748949493285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2695286294619411e-05,
+      "loss": 0.8911,
+      "step": 15684
+    },
+    {
+      "epoch": 0.43077641372112824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2694429686818155e-05,
+      "loss": 0.9982,
+      "step": 15685
+    },
+    {
+      "epoch": 0.43080387794897146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2693573057697526e-05,
+      "loss": 0.8547,
+      "step": 15686
+    },
+    {
+      "epoch": 0.4308313421768147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2692716407264305e-05,
+      "loss": 0.9164,
+      "step": 15687
+    },
+    {
+      "epoch": 0.43085880640465796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2691859735525267e-05,
+      "loss": 0.9563,
+      "step": 15688
+    },
+    {
+      "epoch": 0.4308862706325012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.269100304248719e-05,
+      "loss": 0.8228,
+      "step": 15689
+    },
+    {
+      "epoch": 0.4309137348603444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2690146328156859e-05,
+      "loss": 0.86,
+      "step": 15690
+    },
+    {
+      "epoch": 0.4309411990881876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.268928959254105e-05,
+      "loss": 0.9146,
+      "step": 15691
+    },
+    {
+      "epoch": 0.43096866331603084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2688432835646534e-05,
+      "loss": 0.9387,
+      "step": 15692
+    },
+    {
+      "epoch": 0.4309961275438741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2687576057480098e-05,
+      "loss": 0.9368,
+      "step": 15693
+    },
+    {
+      "epoch": 0.43102359177171734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2686719258048518e-05,
+      "loss": 0.9174,
+      "step": 15694
+    },
+    {
+      "epoch": 0.43105105599956056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2685862437358579e-05,
+      "loss": 0.9603,
+      "step": 15695
+    },
+    {
+      "epoch": 0.4310785202274038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2685005595417053e-05,
+      "loss": 0.9447,
+      "step": 15696
+    },
+    {
+      "epoch": 0.43110598445524706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2684148732230724e-05,
+      "loss": 0.9464,
+      "step": 15697
+    },
+    {
+      "epoch": 0.4311334486830903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2683291847806368e-05,
+      "loss": 0.8931,
+      "step": 15698
+    },
+    {
+      "epoch": 0.4311609129109335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.268243494215077e-05,
+      "loss": 0.9227,
+      "step": 15699
+    },
+    {
+      "epoch": 0.4311883771387767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2681578015270705e-05,
+      "loss": 0.9224,
+      "step": 15700
+    },
+    {
+      "epoch": 0.43121584136662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2680721067172956e-05,
+      "loss": 0.8438,
+      "step": 15701
+    },
+    {
+      "epoch": 0.4312433055944632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2679864097864305e-05,
+      "loss": 0.9522,
+      "step": 15702
+    },
+    {
+      "epoch": 0.43127076982230644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2679007107351535e-05,
+      "loss": 0.9538,
+      "step": 15703
+    },
+    {
+      "epoch": 0.43129823405014966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2678150095641417e-05,
+      "loss": 0.897,
+      "step": 15704
+    },
+    {
+      "epoch": 0.43132569827799294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2677293062740739e-05,
+      "loss": 0.9094,
+      "step": 15705
+    },
+    {
+      "epoch": 0.43135316250583616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2676436008656283e-05,
+      "loss": 0.8945,
+      "step": 15706
+    },
+    {
+      "epoch": 0.4313806267336794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.267557893339483e-05,
+      "loss": 0.8813,
+      "step": 15707
+    },
+    {
+      "epoch": 0.4314080909615226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2674721836963157e-05,
+      "loss": 0.8242,
+      "step": 15708
+    },
+    {
+      "epoch": 0.4314355551893658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.267386471936805e-05,
+      "loss": 0.9738,
+      "step": 15709
+    },
+    {
+      "epoch": 0.4314630194172091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2673007580616287e-05,
+      "loss": 0.9619,
+      "step": 15710
+    },
+    {
+      "epoch": 0.4314904836450523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2672150420714657e-05,
+      "loss": 0.9397,
+      "step": 15711
+    },
+    {
+      "epoch": 0.43151794787289555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2671293239669937e-05,
+      "loss": 0.9309,
+      "step": 15712
+    },
+    {
+      "epoch": 0.43154541210073877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2670436037488907e-05,
+      "loss": 0.9803,
+      "step": 15713
+    },
+    {
+      "epoch": 0.43157287632858204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2669578814178357e-05,
+      "loss": 0.9606,
+      "step": 15714
+    },
+    {
+      "epoch": 0.43160034055642527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2668721569745062e-05,
+      "loss": 0.769,
+      "step": 15715
+    },
+    {
+      "epoch": 0.4316278047842685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2667864304195809e-05,
+      "loss": 0.883,
+      "step": 15716
+    },
+    {
+      "epoch": 0.4316552690121117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.266700701753738e-05,
+      "loss": 0.9083,
+      "step": 15717
+    },
+    {
+      "epoch": 0.431682733239955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2666149709776558e-05,
+      "loss": 0.9033,
+      "step": 15718
+    },
+    {
+      "epoch": 0.4317101974677982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2665292380920129e-05,
+      "loss": 0.9535,
+      "step": 15719
+    },
+    {
+      "epoch": 0.43173766169564143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2664435030974873e-05,
+      "loss": 0.9662,
+      "step": 15720
+    },
+    {
+      "epoch": 0.43176512592348465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2663577659947573e-05,
+      "loss": 0.9265,
+      "step": 15721
+    },
+    {
+      "epoch": 0.43179259015132787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2662720267845014e-05,
+      "loss": 0.9312,
+      "step": 15722
+    },
+    {
+      "epoch": 0.43182005437917115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2661862854673985e-05,
+      "loss": 0.8687,
+      "step": 15723
+    },
+    {
+      "epoch": 0.43184751860701437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2661005420441266e-05,
+      "loss": 0.8107,
+      "step": 15724
+    },
+    {
+      "epoch": 0.4318749828348576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2660147965153638e-05,
+      "loss": 0.926,
+      "step": 15725
+    },
+    {
+      "epoch": 0.4319024470627008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2659290488817891e-05,
+      "loss": 0.8595,
+      "step": 15726
+    },
+    {
+      "epoch": 0.4319299112905441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2658432991440808e-05,
+      "loss": 0.908,
+      "step": 15727
+    },
+    {
+      "epoch": 0.4319573755183873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2657575473029172e-05,
+      "loss": 0.8945,
+      "step": 15728
+    },
+    {
+      "epoch": 0.43198483974623053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2656717933589769e-05,
+      "loss": 0.7991,
+      "step": 15729
+    },
+    {
+      "epoch": 0.43201230397407375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2655860373129385e-05,
+      "loss": 0.8747,
+      "step": 15730
+    },
+    {
+      "epoch": 0.43203976820191703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2655002791654804e-05,
+      "loss": 1.0175,
+      "step": 15731
+    },
+    {
+      "epoch": 0.43206723242976025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2654145189172816e-05,
+      "loss": 0.8757,
+      "step": 15732
+    },
+    {
+      "epoch": 0.43209469665760347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2653287565690197e-05,
+      "loss": 0.891,
+      "step": 15733
+    },
+    {
+      "epoch": 0.4321221608854467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2652429921213742e-05,
+      "loss": 0.975,
+      "step": 15734
+    },
+    {
+      "epoch": 0.4321496251132899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2651572255750237e-05,
+      "loss": 1.0201,
+      "step": 15735
+    },
+    {
+      "epoch": 0.4321770893411332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2650714569306464e-05,
+      "loss": 0.9103,
+      "step": 15736
+    },
+    {
+      "epoch": 0.4322045535689764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2649856861889207e-05,
+      "loss": 0.7989,
+      "step": 15737
+    },
+    {
+      "epoch": 0.43223201779681963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2648999133505262e-05,
+      "loss": 0.9439,
+      "step": 15738
+    },
+    {
+      "epoch": 0.43225948202466286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2648141384161406e-05,
+      "loss": 0.8917,
+      "step": 15739
+    },
+    {
+      "epoch": 0.43228694625250613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.264728361386443e-05,
+      "loss": 0.9229,
+      "step": 15740
+    },
+    {
+      "epoch": 0.43231441048034935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.264642582262112e-05,
+      "loss": 0.9765,
+      "step": 15741
+    },
+    {
+      "epoch": 0.4323418747081926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2645568010438262e-05,
+      "loss": 0.9645,
+      "step": 15742
+    },
+    {
+      "epoch": 0.4323693389360358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.264471017732265e-05,
+      "loss": 0.9669,
+      "step": 15743
+    },
+    {
+      "epoch": 0.4323968031638791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2643852323281064e-05,
+      "loss": 0.9396,
+      "step": 15744
+    },
+    {
+      "epoch": 0.4324242673917223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2642994448320295e-05,
+      "loss": 0.9586,
+      "step": 15745
+    },
+    {
+      "epoch": 0.4324517316195655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2642136552447127e-05,
+      "loss": 0.9509,
+      "step": 15746
+    },
+    {
+      "epoch": 0.43247919584740874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2641278635668352e-05,
+      "loss": 0.9275,
+      "step": 15747
+    },
+    {
+      "epoch": 0.43250666007525196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2640420697990761e-05,
+      "loss": 0.8759,
+      "step": 15748
+    },
+    {
+      "epoch": 0.43253412430309524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2639562739421136e-05,
+      "loss": 0.9232,
+      "step": 15749
+    },
+    {
+      "epoch": 0.43256158853093846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2638704759966269e-05,
+      "loss": 0.9972,
+      "step": 15750
+    },
+    {
+      "epoch": 0.4325890527587817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2637846759632947e-05,
+      "loss": 1.003,
+      "step": 15751
+    },
+    {
+      "epoch": 0.4326165169866249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.263698873842796e-05,
+      "loss": 0.9409,
+      "step": 15752
+    },
+    {
+      "epoch": 0.4326439812144682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2636130696358098e-05,
+      "loss": 0.8251,
+      "step": 15753
+    },
+    {
+      "epoch": 0.4326714454423114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2635272633430145e-05,
+      "loss": 1.0114,
+      "step": 15754
+    },
+    {
+      "epoch": 0.4326989096701546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2634414549650898e-05,
+      "loss": 0.8607,
+      "step": 15755
+    },
+    {
+      "epoch": 0.43272637389799784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2633556445027141e-05,
+      "loss": 0.9983,
+      "step": 15756
+    },
+    {
+      "epoch": 0.4327538381258411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2632698319565666e-05,
+      "loss": 0.848,
+      "step": 15757
+    },
+    {
+      "epoch": 0.43278130235368434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.263184017327326e-05,
+      "loss": 0.9984,
+      "step": 15758
+    },
+    {
+      "epoch": 0.43280876658152756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.263098200615672e-05,
+      "loss": 0.8792,
+      "step": 15759
+    },
+    {
+      "epoch": 0.4328362308093708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2630123818222827e-05,
+      "loss": 0.9855,
+      "step": 15760
+    },
+    {
+      "epoch": 0.432863695037214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2629265609478373e-05,
+      "loss": 0.7518,
+      "step": 15761
+    },
+    {
+      "epoch": 0.4328911592650573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2628407379930157e-05,
+      "loss": 0.9634,
+      "step": 15762
+    },
+    {
+      "epoch": 0.4329186234929005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2627549129584961e-05,
+      "loss": 0.8496,
+      "step": 15763
+    },
+    {
+      "epoch": 0.4329460877207437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2626690858449577e-05,
+      "loss": 0.8924,
+      "step": 15764
+    },
+    {
+      "epoch": 0.43297355194858694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.26258325665308e-05,
+      "loss": 1.0136,
+      "step": 15765
+    },
+    {
+      "epoch": 0.4330010161764302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2624974253835416e-05,
+      "loss": 0.8785,
+      "step": 15766
+    },
+    {
+      "epoch": 0.43302848040427344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2624115920370221e-05,
+      "loss": 0.97,
+      "step": 15767
+    },
+    {
+      "epoch": 0.43305594463211666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2623257566142005e-05,
+      "loss": 0.8583,
+      "step": 15768
+    },
+    {
+      "epoch": 0.4330834088599599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2622399191157557e-05,
+      "loss": 0.837,
+      "step": 15769
+    },
+    {
+      "epoch": 0.43311087308780316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2621540795423672e-05,
+      "loss": 0.929,
+      "step": 15770
+    },
+    {
+      "epoch": 0.4331383373156464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2620682378947142e-05,
+      "loss": 0.9407,
+      "step": 15771
+    },
+    {
+      "epoch": 0.4331658015434896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2619823941734755e-05,
+      "loss": 0.851,
+      "step": 15772
+    },
+    {
+      "epoch": 0.4331932657713328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2618965483793307e-05,
+      "loss": 1.0128,
+      "step": 15773
+    },
+    {
+      "epoch": 0.43322072999917605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261810700512959e-05,
+      "loss": 0.8661,
+      "step": 15774
+    },
+    {
+      "epoch": 0.4332481942270193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2617248505750398e-05,
+      "loss": 0.955,
+      "step": 15775
+    },
+    {
+      "epoch": 0.43327565845486254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2616389985662518e-05,
+      "loss": 0.8211,
+      "step": 15776
+    },
+    {
+      "epoch": 0.43330312268270577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2615531444872749e-05,
+      "loss": 0.8667,
+      "step": 15777
+    },
+    {
+      "epoch": 0.433330586910549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2614672883387882e-05,
+      "loss": 0.9091,
+      "step": 15778
+    },
+    {
+      "epoch": 0.43335805113839226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261381430121471e-05,
+      "loss": 0.8396,
+      "step": 15779
+    },
+    {
+      "epoch": 0.4333855153662355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2612955698360028e-05,
+      "loss": 0.8923,
+      "step": 15780
+    },
+    {
+      "epoch": 0.4334129795940787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2612097074830627e-05,
+      "loss": 0.8949,
+      "step": 15781
+    },
+    {
+      "epoch": 0.43344044382192193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2611238430633303e-05,
+      "loss": 0.8884,
+      "step": 15782
+    },
+    {
+      "epoch": 0.4334679080497652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261037976577485e-05,
+      "loss": 0.9126,
+      "step": 15783
+    },
+    {
+      "epoch": 0.4334953722776084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2609521080262056e-05,
+      "loss": 0.9693,
+      "step": 15784
+    },
+    {
+      "epoch": 0.43352283650545165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2608662374101726e-05,
+      "loss": 0.8957,
+      "step": 15785
+    },
+    {
+      "epoch": 0.43355030073329487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2607803647300647e-05,
+      "loss": 0.9478,
+      "step": 15786
+    },
+    {
+      "epoch": 0.4335777649611381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606944899865616e-05,
+      "loss": 0.9638,
+      "step": 15787
+    },
+    {
+      "epoch": 0.43360522918898137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606086131803427e-05,
+      "loss": 1.0291,
+      "step": 15788
+    },
+    {
+      "epoch": 0.4336326934168246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2605227343120873e-05,
+      "loss": 0.9675,
+      "step": 15789
+    },
+    {
+      "epoch": 0.4336601576446678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2604368533824751e-05,
+      "loss": 1.0021,
+      "step": 15790
+    },
+    {
+      "epoch": 0.43368762187251103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2603509703921859e-05,
+      "loss": 0.8594,
+      "step": 15791
+    },
+    {
+      "epoch": 0.4337150861003543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.260265085341899e-05,
+      "loss": 0.9092,
+      "step": 15792
+    },
+    {
+      "epoch": 0.43374255032819753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2601791982322937e-05,
+      "loss": 0.8911,
+      "step": 15793
+    },
+    {
+      "epoch": 0.43377001455604075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.26009330906405e-05,
+      "loss": 0.9188,
+      "step": 15794
+    },
+    {
+      "epoch": 0.43379747878388397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2600074178378471e-05,
+      "loss": 0.9057,
+      "step": 15795
+    },
+    {
+      "epoch": 0.43382494301172725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2599215245543647e-05,
+      "loss": 0.9049,
+      "step": 15796
+    },
+    {
+      "epoch": 0.43385240723957047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2598356292142828e-05,
+      "loss": 0.8737,
+      "step": 15797
+    },
+    {
+      "epoch": 0.4338798714674137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2597497318182805e-05,
+      "loss": 0.942,
+      "step": 15798
+    },
+    {
+      "epoch": 0.4339073356952569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2596638323670381e-05,
+      "loss": 0.8444,
+      "step": 15799
+    },
+    {
+      "epoch": 0.4339347999231002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2595779308612346e-05,
+      "loss": 0.8237,
+      "step": 15800
+    },
+    {
+      "epoch": 0.4339622641509434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2594920273015499e-05,
+      "loss": 0.8312,
+      "step": 15801
+    },
+    {
+      "epoch": 0.43398972837878663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2594061216886637e-05,
+      "loss": 0.9913,
+      "step": 15802
+    },
+    {
+      "epoch": 0.43401719260662985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2593202140232561e-05,
+      "loss": 0.9778,
+      "step": 15803
+    },
+    {
+      "epoch": 0.4340446568344731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2592343043060064e-05,
+      "loss": 0.7689,
+      "step": 15804
+    },
+    {
+      "epoch": 0.43407212106231635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2591483925375943e-05,
+      "loss": 0.8622,
+      "step": 15805
+    },
+    {
+      "epoch": 0.4340995852901596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2590624787186998e-05,
+      "loss": 0.8729,
+      "step": 15806
+    },
+    {
+      "epoch": 0.4341270495180028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2589765628500028e-05,
+      "loss": 0.9576,
+      "step": 15807
+    },
+    {
+      "epoch": 0.434154513745846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2588906449321827e-05,
+      "loss": 0.9588,
+      "step": 15808
+    },
+    {
+      "epoch": 0.4341819779736893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2588047249659196e-05,
+      "loss": 0.9517,
+      "step": 15809
+    },
+    {
+      "epoch": 0.4342094422015325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2587188029518931e-05,
+      "loss": 0.9722,
+      "step": 15810
+    },
+    {
+      "epoch": 0.43423690642937574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2586328788907835e-05,
+      "loss": 0.7939,
+      "step": 15811
+    },
+    {
+      "epoch": 0.43426437065721896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2585469527832703e-05,
+      "loss": 0.9033,
+      "step": 15812
+    },
+    {
+      "epoch": 0.43429183488506223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2584610246300332e-05,
+      "loss": 0.9816,
+      "step": 15813
+    },
+    {
+      "epoch": 0.43431929911290545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2583750944317525e-05,
+      "loss": 0.957,
+      "step": 15814
+    },
+    {
+      "epoch": 0.4343467633407487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2582891621891083e-05,
+      "loss": 0.9216,
+      "step": 15815
+    },
+    {
+      "epoch": 0.4343742275685919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2582032279027801e-05,
+      "loss": 0.998,
+      "step": 15816
+    },
+    {
+      "epoch": 0.4344016917964351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2581172915734476e-05,
+      "loss": 0.8653,
+      "step": 15817
+    },
+    {
+      "epoch": 0.4344291560242784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2580313532017914e-05,
+      "loss": 0.8446,
+      "step": 15818
+    },
+    {
+      "epoch": 0.4344566202521216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2579454127884909e-05,
+      "loss": 0.8521,
+      "step": 15819
+    },
+    {
+      "epoch": 0.43448408447996484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2578594703342267e-05,
+      "loss": 0.8952,
+      "step": 15820
+    },
+    {
+      "epoch": 0.43451154870780806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2577735258396785e-05,
+      "loss": 0.9163,
+      "step": 15821
+    },
+    {
+      "epoch": 0.43453901293565134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2576875793055256e-05,
+      "loss": 0.9033,
+      "step": 15822
+    },
+    {
+      "epoch": 0.43456647716349456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2576016307324497e-05,
+      "loss": 0.9287,
+      "step": 15823
+    },
+    {
+      "epoch": 0.4345939413913378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2575156801211295e-05,
+      "loss": 0.9419,
+      "step": 15824
+    },
+    {
+      "epoch": 0.434621405619181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2574297274722454e-05,
+      "loss": 0.9426,
+      "step": 15825
+    },
+    {
+      "epoch": 0.4346488698470243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2573437727864775e-05,
+      "loss": 0.8838,
+      "step": 15826
+    },
+    {
+      "epoch": 0.4346763340748675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.257257816064506e-05,
+      "loss": 0.907,
+      "step": 15827
+    },
+    {
+      "epoch": 0.4347037983027107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2571718573070113e-05,
+      "loss": 1.0154,
+      "step": 15828
+    },
+    {
+      "epoch": 0.43473126253055394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2570858965146734e-05,
+      "loss": 0.8028,
+      "step": 15829
+    },
+    {
+      "epoch": 0.43475872675839716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2569999336881717e-05,
+      "loss": 0.9601,
+      "step": 15830
+    },
+    {
+      "epoch": 0.43478619098624044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2569139688281874e-05,
+      "loss": 0.8707,
+      "step": 15831
+    },
+    {
+      "epoch": 0.43481365521408366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2568280019354001e-05,
+      "loss": 0.8871,
+      "step": 15832
+    },
+    {
+      "epoch": 0.4348411194419269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2567420330104902e-05,
+      "loss": 0.8886,
+      "step": 15833
+    },
+    {
+      "epoch": 0.4348685836697701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2566560620541379e-05,
+      "loss": 0.8444,
+      "step": 15834
+    },
+    {
+      "epoch": 0.4348960478976134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2565700890670234e-05,
+      "loss": 0.8853,
+      "step": 15835
+    },
+    {
+      "epoch": 0.4349235121254566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2564841140498269e-05,
+      "loss": 0.8333,
+      "step": 15836
+    },
+    {
+      "epoch": 0.4349509763532998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2563981370032287e-05,
+      "loss": 1.0096,
+      "step": 15837
+    },
+    {
+      "epoch": 0.43497844058114304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2563121579279094e-05,
+      "loss": 0.8884,
+      "step": 15838
+    },
+    {
+      "epoch": 0.4350059048089863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.256226176824549e-05,
+      "loss": 1.02,
+      "step": 15839
+    },
+    {
+      "epoch": 0.43503336903682954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2561401936938275e-05,
+      "loss": 0.9838,
+      "step": 15840
+    },
+    {
+      "epoch": 0.43506083326467276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2560542085364258e-05,
+      "loss": 0.9526,
+      "step": 15841
+    },
+    {
+      "epoch": 0.435088297492516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255968221353024e-05,
+      "loss": 0.906,
+      "step": 15842
+    },
+    {
+      "epoch": 0.4351157617203592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2558822321443024e-05,
+      "loss": 0.8721,
+      "step": 15843
+    },
+    {
+      "epoch": 0.4351432259482025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2557962409109417e-05,
+      "loss": 0.8465,
+      "step": 15844
+    },
+    {
+      "epoch": 0.4351706901760457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2557102476536219e-05,
+      "loss": 0.9488,
+      "step": 15845
+    },
+    {
+      "epoch": 0.4351981544038889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2556242523730231e-05,
+      "loss": 0.9641,
+      "step": 15846
+    },
+    {
+      "epoch": 0.43522561863173215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2555382550698267e-05,
+      "loss": 0.8373,
+      "step": 15847
+    },
+    {
+      "epoch": 0.4352530828595754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2554522557447128e-05,
+      "loss": 0.8378,
+      "step": 15848
+    },
+    {
+      "epoch": 0.43528054708741865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2553662543983613e-05,
+      "loss": 0.8372,
+      "step": 15849
+    },
+    {
+      "epoch": 0.43530801131526187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2552802510314535e-05,
+      "loss": 0.9502,
+      "step": 15850
+    },
+    {
+      "epoch": 0.4353354755431051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2551942456446691e-05,
+      "loss": 0.9659,
+      "step": 15851
+    },
+    {
+      "epoch": 0.43536293977094837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255108238238689e-05,
+      "loss": 0.8852,
+      "step": 15852
+    },
+    {
+      "epoch": 0.4353904039987916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2550222288141937e-05,
+      "loss": 0.9251,
+      "step": 15853
+    },
+    {
+      "epoch": 0.4354178682266348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.254936217371864e-05,
+      "loss": 0.8663,
+      "step": 15854
+    },
+    {
+      "epoch": 0.43544533245447803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25485020391238e-05,
+      "loss": 0.9617,
+      "step": 15855
+    },
+    {
+      "epoch": 0.43547279668232125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2547641884364226e-05,
+      "loss": 0.8922,
+      "step": 15856
+    },
+    {
+      "epoch": 0.4355002609101645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.254678170944672e-05,
+      "loss": 0.9732,
+      "step": 15857
+    },
+    {
+      "epoch": 0.43552772513800775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2545921514378091e-05,
+      "loss": 0.8117,
+      "step": 15858
+    },
+    {
+      "epoch": 0.43555518936585097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2545061299165147e-05,
+      "loss": 0.889,
+      "step": 15859
+    },
+    {
+      "epoch": 0.4355826535936942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2544201063814692e-05,
+      "loss": 0.9789,
+      "step": 15860
+    },
+    {
+      "epoch": 0.43561011782153747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.254334080833353e-05,
+      "loss": 1.0031,
+      "step": 15861
+    },
+    {
+      "epoch": 0.4356375820493807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2542480532728471e-05,
+      "loss": 0.9314,
+      "step": 15862
+    },
+    {
+      "epoch": 0.4356650462772239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2541620237006324e-05,
+      "loss": 0.9607,
+      "step": 15863
+    },
+    {
+      "epoch": 0.43569251050506713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.254075992117389e-05,
+      "loss": 0.8899,
+      "step": 15864
+    },
+    {
+      "epoch": 0.4357199747329104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2539899585237983e-05,
+      "loss": 0.9379,
+      "step": 15865
+    },
+    {
+      "epoch": 0.43574743896075363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2539039229205403e-05,
+      "loss": 0.8374,
+      "step": 15866
+    },
+    {
+      "epoch": 0.43577490318859685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2538178853082963e-05,
+      "loss": 0.8626,
+      "step": 15867
+    },
+    {
+      "epoch": 0.4358023674164401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2537318456877468e-05,
+      "loss": 0.9722,
+      "step": 15868
+    },
+    {
+      "epoch": 0.4358298316442833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2536458040595725e-05,
+      "loss": 0.8079,
+      "step": 15869
+    },
+    {
+      "epoch": 0.43585729587212657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2535597604244542e-05,
+      "loss": 1.0115,
+      "step": 15870
+    },
+    {
+      "epoch": 0.4358847600999698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2534737147830734e-05,
+      "loss": 0.9734,
+      "step": 15871
+    },
+    {
+      "epoch": 0.435912224327813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2533876671361102e-05,
+      "loss": 0.8863,
+      "step": 15872
+    },
+    {
+      "epoch": 0.43593968855565624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2533016174842454e-05,
+      "loss": 0.9027,
+      "step": 15873
+    },
+    {
+      "epoch": 0.4359671527834995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2532155658281605e-05,
+      "loss": 0.8808,
+      "step": 15874
+    },
+    {
+      "epoch": 0.43599461701134273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2531295121685356e-05,
+      "loss": 0.9943,
+      "step": 15875
+    },
+    {
+      "epoch": 0.43602208123918595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2530434565060522e-05,
+      "loss": 1.0129,
+      "step": 15876
+    },
+    {
+      "epoch": 0.4360495454670292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2529573988413906e-05,
+      "loss": 0.9182,
+      "step": 15877
+    },
+    {
+      "epoch": 0.43607700969487245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2528713391752323e-05,
+      "loss": 0.925,
+      "step": 15878
+    },
+    {
+      "epoch": 0.4361044739227157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.252785277508258e-05,
+      "loss": 0.9236,
+      "step": 15879
+    },
+    {
+      "epoch": 0.4361319381505589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2526992138411487e-05,
+      "loss": 0.8472,
+      "step": 15880
+    },
+    {
+      "epoch": 0.4361594023784021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2526131481745852e-05,
+      "loss": 0.8326,
+      "step": 15881
+    },
+    {
+      "epoch": 0.43618686660624534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2525270805092485e-05,
+      "loss": 0.981,
+      "step": 15882
+    },
+    {
+      "epoch": 0.4362143308340886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25244101084582e-05,
+      "loss": 0.9168,
+      "step": 15883
+    },
+    {
+      "epoch": 0.43624179506193184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2523549391849805e-05,
+      "loss": 0.8926,
+      "step": 15884
+    },
+    {
+      "epoch": 0.43626925928977506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2522688655274107e-05,
+      "loss": 0.8546,
+      "step": 15885
+    },
+    {
+      "epoch": 0.4362967235176183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2521827898737922e-05,
+      "loss": 0.922,
+      "step": 15886
+    },
+    {
+      "epoch": 0.43632418774546156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2520967122248056e-05,
+      "loss": 0.9607,
+      "step": 15887
+    },
+    {
+      "epoch": 0.4363516519733048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.252010632581132e-05,
+      "loss": 1.0444,
+      "step": 15888
+    },
+    {
+      "epoch": 0.436379116201148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.251924550943453e-05,
+      "loss": 0.8135,
+      "step": 15889
+    },
+    {
+      "epoch": 0.4364065804289912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2518384673124491e-05,
+      "loss": 0.905,
+      "step": 15890
+    },
+    {
+      "epoch": 0.4364340446568345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2517523816888018e-05,
+      "loss": 0.9181,
+      "step": 15891
+    },
+    {
+      "epoch": 0.4364615088846777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2516662940731923e-05,
+      "loss": 0.8193,
+      "step": 15892
+    },
+    {
+      "epoch": 0.43648897311252094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2515802044663012e-05,
+      "loss": 0.9722,
+      "step": 15893
+    },
+    {
+      "epoch": 0.43651643734036416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2514941128688105e-05,
+      "loss": 0.9792,
+      "step": 15894
+    },
+    {
+      "epoch": 0.43654390156820744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2514080192814006e-05,
+      "loss": 0.9366,
+      "step": 15895
+    },
+    {
+      "epoch": 0.43657136579605066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2513219237047531e-05,
+      "loss": 0.8951,
+      "step": 15896
+    },
+    {
+      "epoch": 0.4365988300238939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2512358261395497e-05,
+      "loss": 0.948,
+      "step": 15897
+    },
+    {
+      "epoch": 0.4366262942517371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2511497265864705e-05,
+      "loss": 0.9851,
+      "step": 15898
+    },
+    {
+      "epoch": 0.4366537584795803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2510636250461976e-05,
+      "loss": 0.9289,
+      "step": 15899
+    },
+    {
+      "epoch": 0.4366812227074236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2509775215194123e-05,
+      "loss": 0.923,
+      "step": 15900
+    },
+    {
+      "epoch": 0.4367086869352668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2508914160067951e-05,
+      "loss": 0.9003,
+      "step": 15901
+    },
+    {
+      "epoch": 0.43673615116311004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250805308509028e-05,
+      "loss": 0.9405,
+      "step": 15902
+    },
+    {
+      "epoch": 0.43676361539095326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2507191990267924e-05,
+      "loss": 0.9387,
+      "step": 15903
+    },
+    {
+      "epoch": 0.43679107961879654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2506330875607693e-05,
+      "loss": 0.9585,
+      "step": 15904
+    },
+    {
+      "epoch": 0.43681854384663976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2505469741116398e-05,
+      "loss": 0.9179,
+      "step": 15905
+    },
+    {
+      "epoch": 0.436846008074483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2504608586800862e-05,
+      "loss": 1.0005,
+      "step": 15906
+    },
+    {
+      "epoch": 0.4368734723023262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2503747412667888e-05,
+      "loss": 0.9246,
+      "step": 15907
+    },
+    {
+      "epoch": 0.4369009365301695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2502886218724295e-05,
+      "loss": 0.8819,
+      "step": 15908
+    },
+    {
+      "epoch": 0.4369284007580127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25020250049769e-05,
+      "loss": 0.9376,
+      "step": 15909
+    },
+    {
+      "epoch": 0.4369558649858559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250116377143251e-05,
+      "loss": 0.82,
+      "step": 15910
+    },
+    {
+      "epoch": 0.43698332921369915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2500302518097945e-05,
+      "loss": 0.8662,
+      "step": 15911
+    },
+    {
+      "epoch": 0.43701079344154237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2499441244980019e-05,
+      "loss": 0.934,
+      "step": 15912
+    },
+    {
+      "epoch": 0.43703825766938564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2498579952085543e-05,
+      "loss": 0.8759,
+      "step": 15913
+    },
+    {
+      "epoch": 0.43706572189722886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2497718639421335e-05,
+      "loss": 1.0195,
+      "step": 15914
+    },
+    {
+      "epoch": 0.4370931861250721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2496857306994213e-05,
+      "loss": 0.8828,
+      "step": 15915
+    },
+    {
+      "epoch": 0.4371206503529153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2495995954810986e-05,
+      "loss": 0.8824,
+      "step": 15916
+    },
+    {
+      "epoch": 0.4371481145807586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2495134582878472e-05,
+      "loss": 0.9434,
+      "step": 15917
+    },
+    {
+      "epoch": 0.4371755788086018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2494273191203488e-05,
+      "loss": 0.8972,
+      "step": 15918
+    },
+    {
+      "epoch": 0.437203043036445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2493411779792848e-05,
+      "loss": 0.9071,
+      "step": 15919
+    },
+    {
+      "epoch": 0.43723050726428825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2492550348653367e-05,
+      "loss": 0.9734,
+      "step": 15920
+    },
+    {
+      "epoch": 0.4372579714921315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2491688897791862e-05,
+      "loss": 0.9944,
+      "step": 15921
+    },
+    {
+      "epoch": 0.43728543571997475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2490827427215151e-05,
+      "loss": 0.87,
+      "step": 15922
+    },
+    {
+      "epoch": 0.43731289994781797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2489965936930046e-05,
+      "loss": 0.9692,
+      "step": 15923
+    },
+    {
+      "epoch": 0.4373403641756612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248910442694337e-05,
+      "loss": 0.9474,
+      "step": 15924
+    },
+    {
+      "epoch": 0.4373678284035044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248824289726193e-05,
+      "loss": 0.9386,
+      "step": 15925
+    },
+    {
+      "epoch": 0.4373952926313477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2487381347892554e-05,
+      "loss": 0.8751,
+      "step": 15926
+    },
+    {
+      "epoch": 0.4374227568591909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248651977884205e-05,
+      "loss": 0.9156,
+      "step": 15927
+    },
+    {
+      "epoch": 0.43745022108703413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248565819011724e-05,
+      "loss": 0.9899,
+      "step": 15928
+    },
+    {
+      "epoch": 0.43747768531487735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2484796581724935e-05,
+      "loss": 1.0157,
+      "step": 15929
+    },
+    {
+      "epoch": 0.43750514954272063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2483934953671963e-05,
+      "loss": 0.8754,
+      "step": 15930
+    },
+    {
+      "epoch": 0.43753261377056385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2483073305965135e-05,
+      "loss": 0.9043,
+      "step": 15931
+    },
+    {
+      "epoch": 0.43756007799840707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2482211638611265e-05,
+      "loss": 0.8794,
+      "step": 15932
+    },
+    {
+      "epoch": 0.4375875422262503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2481349951617179e-05,
+      "loss": 0.9027,
+      "step": 15933
+    },
+    {
+      "epoch": 0.43761500645409357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2480488244989688e-05,
+      "loss": 0.8879,
+      "step": 15934
+    },
+    {
+      "epoch": 0.4376424706819368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2479626518735614e-05,
+      "loss": 0.947,
+      "step": 15935
+    },
+    {
+      "epoch": 0.43766993490978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2478764772861775e-05,
+      "loss": 0.89,
+      "step": 15936
+    },
+    {
+      "epoch": 0.43769739913762323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.247790300737499e-05,
+      "loss": 0.9858,
+      "step": 15937
+    },
+    {
+      "epoch": 0.43772486336546645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2477041222282074e-05,
+      "loss": 0.9241,
+      "step": 15938
+    },
+    {
+      "epoch": 0.43775232759330973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.247617941758985e-05,
+      "loss": 0.9201,
+      "step": 15939
+    },
+    {
+      "epoch": 0.43777979182115295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2475317593305136e-05,
+      "loss": 0.9059,
+      "step": 15940
+    },
+    {
+      "epoch": 0.4378072560489962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2474455749434749e-05,
+      "loss": 0.8826,
+      "step": 15941
+    },
+    {
+      "epoch": 0.4378347202768394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2473593885985508e-05,
+      "loss": 0.8628,
+      "step": 15942
+    },
+    {
+      "epoch": 0.43786218450468267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2472732002964238e-05,
+      "loss": 0.9025,
+      "step": 15943
+    },
+    {
+      "epoch": 0.4378896487325259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2471870100377752e-05,
+      "loss": 1.0117,
+      "step": 15944
+    },
+    {
+      "epoch": 0.4379171129603691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2471008178232873e-05,
+      "loss": 0.9047,
+      "step": 15945
+    },
+    {
+      "epoch": 0.43794457718821234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2470146236536417e-05,
+      "loss": 0.8854,
+      "step": 15946
+    },
+    {
+      "epoch": 0.4379720414160556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2469284275295212e-05,
+      "loss": 0.8092,
+      "step": 15947
+    },
+    {
+      "epoch": 0.43799950564389883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2468422294516072e-05,
+      "loss": 0.8219,
+      "step": 15948
+    },
+    {
+      "epoch": 0.43802696987174206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2467560294205816e-05,
+      "loss": 0.8533,
+      "step": 15949
+    },
+    {
+      "epoch": 0.4380544340995853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2466698274371271e-05,
+      "loss": 0.9647,
+      "step": 15950
+    },
+    {
+      "epoch": 0.4380818983274285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2465836235019251e-05,
+      "loss": 0.8737,
+      "step": 15951
+    },
+    {
+      "epoch": 0.4381093625552718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.246497417615658e-05,
+      "loss": 0.917,
+      "step": 15952
+    },
+    {
+      "epoch": 0.438136826783115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.246411209779008e-05,
+      "loss": 0.8246,
+      "step": 15953
+    },
+    {
+      "epoch": 0.4381642910109582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2463249999926567e-05,
+      "loss": 0.9235,
+      "step": 15954
+    },
+    {
+      "epoch": 0.43819175523880144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.246238788257287e-05,
+      "loss": 0.9387,
+      "step": 15955
+    },
+    {
+      "epoch": 0.4382192194666447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2461525745735805e-05,
+      "loss": 0.8192,
+      "step": 15956
+    },
+    {
+      "epoch": 0.43824668369448794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2460663589422192e-05,
+      "loss": 0.8795,
+      "step": 15957
+    },
+    {
+      "epoch": 0.43827414792233116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2459801413638858e-05,
+      "loss": 0.8564,
+      "step": 15958
+    },
+    {
+      "epoch": 0.4383016121501744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2458939218392623e-05,
+      "loss": 0.7669,
+      "step": 15959
+    },
+    {
+      "epoch": 0.43832907637801766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2458077003690308e-05,
+      "loss": 1.0093,
+      "step": 15960
+    },
+    {
+      "epoch": 0.4383565406058609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2457214769538735e-05,
+      "loss": 0.9132,
+      "step": 15961
+    },
+    {
+      "epoch": 0.4383840048337041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2456352515944728e-05,
+      "loss": 0.9707,
+      "step": 15962
+    },
+    {
+      "epoch": 0.4384114690615473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2455490242915104e-05,
+      "loss": 0.9886,
+      "step": 15963
+    },
+    {
+      "epoch": 0.43843893328939054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2454627950456695e-05,
+      "loss": 0.9884,
+      "step": 15964
+    },
+    {
+      "epoch": 0.4384663975172338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2453765638576318e-05,
+      "loss": 0.8892,
+      "step": 15965
+    },
+    {
+      "epoch": 0.43849386174507704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2452903307280794e-05,
+      "loss": 0.9835,
+      "step": 15966
+    },
+    {
+      "epoch": 0.43852132597292026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2452040956576953e-05,
+      "loss": 1.0114,
+      "step": 15967
+    },
+    {
+      "epoch": 0.4385487902007635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2451178586471613e-05,
+      "loss": 0.8376,
+      "step": 15968
+    },
+    {
+      "epoch": 0.43857625442860676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2450316196971597e-05,
+      "loss": 0.8875,
+      "step": 15969
+    },
+    {
+      "epoch": 0.43860371865645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2449453788083728e-05,
+      "loss": 0.8813,
+      "step": 15970
+    },
+    {
+      "epoch": 0.4386311828842932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2448591359814837e-05,
+      "loss": 0.8328,
+      "step": 15971
+    },
+    {
+      "epoch": 0.4386586471121364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2447728912171739e-05,
+      "loss": 0.953,
+      "step": 15972
+    },
+    {
+      "epoch": 0.4386861113399797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2446866445161264e-05,
+      "loss": 0.944,
+      "step": 15973
+    },
+    {
+      "epoch": 0.4387135755678229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2446003958790231e-05,
+      "loss": 0.9694,
+      "step": 15974
+    },
+    {
+      "epoch": 0.43874103979566614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2445141453065468e-05,
+      "loss": 0.9943,
+      "step": 15975
+    },
+    {
+      "epoch": 0.43876850402350936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.24442789279938e-05,
+      "loss": 0.9599,
+      "step": 15976
+    },
+    {
+      "epoch": 0.4387959682513526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2443416383582052e-05,
+      "loss": 0.8797,
+      "step": 15977
+    },
+    {
+      "epoch": 0.43882343247919586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2442553819837042e-05,
+      "loss": 0.874,
+      "step": 15978
+    },
+    {
+      "epoch": 0.4388508967070391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2441691236765606e-05,
+      "loss": 0.9759,
+      "step": 15979
+    },
+    {
+      "epoch": 0.4388783609348823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.244082863437456e-05,
+      "loss": 0.9663,
+      "step": 15980
+    },
+    {
+      "epoch": 0.4389058251627255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.243996601267073e-05,
+      "loss": 0.8017,
+      "step": 15981
+    },
+    {
+      "epoch": 0.4389332893905688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2439103371660946e-05,
+      "loss": 0.9505,
+      "step": 15982
+    },
+    {
+      "epoch": 0.438960753618412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2438240711352032e-05,
+      "loss": 0.8923,
+      "step": 15983
+    },
+    {
+      "epoch": 0.43898821784625525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2437378031750814e-05,
+      "loss": 0.9868,
+      "step": 15984
+    },
+    {
+      "epoch": 0.43901568207409847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2436515332864114e-05,
+      "loss": 0.8967,
+      "step": 15985
+    },
+    {
+      "epoch": 0.43904314630194174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2435652614698763e-05,
+      "loss": 0.9179,
+      "step": 15986
+    },
+    {
+      "epoch": 0.43907061052978497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2434789877261584e-05,
+      "loss": 0.8954,
+      "step": 15987
+    },
+    {
+      "epoch": 0.4390980747576282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2433927120559404e-05,
+      "loss": 0.8896,
+      "step": 15988
+    },
+    {
+      "epoch": 0.4391255389854714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.243306434459905e-05,
+      "loss": 0.8281,
+      "step": 15989
+    },
+    {
+      "epoch": 0.43915300321331463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2432201549387349e-05,
+      "loss": 0.8825,
+      "step": 15990
+    },
+    {
+      "epoch": 0.4391804674411579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2431338734931126e-05,
+      "loss": 0.894,
+      "step": 15991
+    },
+    {
+      "epoch": 0.43920793166900113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2430475901237212e-05,
+      "loss": 0.8683,
+      "step": 15992
+    },
+    {
+      "epoch": 0.43923539589684435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2429613048312428e-05,
+      "loss": 0.9811,
+      "step": 15993
+    },
+    {
+      "epoch": 0.43926286012468757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2428750176163604e-05,
+      "loss": 0.9186,
+      "step": 15994
+    },
+    {
+      "epoch": 0.43929032435253085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242788728479757e-05,
+      "loss": 0.875,
+      "step": 15995
+    },
+    {
+      "epoch": 0.43931778858037407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2427024374221152e-05,
+      "loss": 0.886,
+      "step": 15996
+    },
+    {
+      "epoch": 0.4393452528082173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2426161444441175e-05,
+      "loss": 0.8853,
+      "step": 15997
+    },
+    {
+      "epoch": 0.4393727170360605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2425298495464471e-05,
+      "loss": 0.9043,
+      "step": 15998
+    },
+    {
+      "epoch": 0.4394001812639038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2424435527297864e-05,
+      "loss": 0.9319,
+      "step": 15999
+    },
+    {
+      "epoch": 0.439427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2423572539948183e-05,
+      "loss": 0.9312,
+      "step": 16000
+    },
+    {
+      "epoch": 0.43945510971959023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2422709533422261e-05,
+      "loss": 0.989,
+      "step": 16001
+    },
+    {
+      "epoch": 0.43948257394743345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2421846507726918e-05,
+      "loss": 0.9495,
+      "step": 16002
+    },
+    {
+      "epoch": 0.43951003817527673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2420983462868991e-05,
+      "loss": 0.9075,
+      "step": 16003
+    },
+    {
+      "epoch": 0.43953750240311995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2420120398855304e-05,
+      "loss": 0.8677,
+      "step": 16004
+    },
+    {
+      "epoch": 0.43956496663096317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2419257315692685e-05,
+      "loss": 0.8061,
+      "step": 16005
+    },
+    {
+      "epoch": 0.4395924308588064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2418394213387967e-05,
+      "loss": 0.8248,
+      "step": 16006
+    },
+    {
+      "epoch": 0.4396198950866496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2417531091947978e-05,
+      "loss": 0.9457,
+      "step": 16007
+    },
+    {
+      "epoch": 0.4396473593144929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2416667951379547e-05,
+      "loss": 0.9157,
+      "step": 16008
+    },
+    {
+      "epoch": 0.4396748235423361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2415804791689498e-05,
+      "loss": 0.9249,
+      "step": 16009
+    },
+    {
+      "epoch": 0.43970228777017933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2414941612884673e-05,
+      "loss": 0.8604,
+      "step": 16010
+    },
+    {
+      "epoch": 0.43972975199802256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.241407841497189e-05,
+      "loss": 0.9852,
+      "step": 16011
+    },
+    {
+      "epoch": 0.43975721622586583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2413215197957986e-05,
+      "loss": 0.8811,
+      "step": 16012
+    },
+    {
+      "epoch": 0.43978468045370905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2412351961849784e-05,
+      "loss": 1.0165,
+      "step": 16013
+    },
+    {
+      "epoch": 0.4398121446815523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.241148870665412e-05,
+      "loss": 0.8979,
+      "step": 16014
+    },
+    {
+      "epoch": 0.4398396089093955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2410625432377827e-05,
+      "loss": 0.8965,
+      "step": 16015
+    },
+    {
+      "epoch": 0.4398670731372388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240976213902773e-05,
+      "loss": 1.0875,
+      "step": 16016
+    },
+    {
+      "epoch": 0.439894537365082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240889882661066e-05,
+      "loss": 0.9399,
+      "step": 16017
+    },
+    {
+      "epoch": 0.4399220015929252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2408035495133451e-05,
+      "loss": 0.9619,
+      "step": 16018
+    },
+    {
+      "epoch": 0.43994946582076844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2407172144602931e-05,
+      "loss": 0.9271,
+      "step": 16019
+    },
+    {
+      "epoch": 0.43997693004861166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2406308775025933e-05,
+      "loss": 0.9255,
+      "step": 16020
+    },
+    {
+      "epoch": 0.44000439427645494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240544538640929e-05,
+      "loss": 0.8218,
+      "step": 16021
+    },
+    {
+      "epoch": 0.44003185850429816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240458197875983e-05,
+      "loss": 0.9222,
+      "step": 16022
+    },
+    {
+      "epoch": 0.4400593227321414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2403718552084385e-05,
+      "loss": 0.8429,
+      "step": 16023
+    },
+    {
+      "epoch": 0.4400867869599846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240285510638979e-05,
+      "loss": 0.9025,
+      "step": 16024
+    },
+    {
+      "epoch": 0.4401142511878279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2401991641682868e-05,
+      "loss": 0.9625,
+      "step": 16025
+    },
+    {
+      "epoch": 0.4401417154156711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2401128157970463e-05,
+      "loss": 0.9109,
+      "step": 16026
+    },
+    {
+      "epoch": 0.4401691796435143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2400264655259401e-05,
+      "loss": 0.9407,
+      "step": 16027
+    },
+    {
+      "epoch": 0.44019664387135754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2399401133556518e-05,
+      "loss": 0.8827,
+      "step": 16028
+    },
+    {
+      "epoch": 0.4402241080992008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.239853759286864e-05,
+      "loss": 0.9004,
+      "step": 16029
+    },
+    {
+      "epoch": 0.44025157232704404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2397674033202604e-05,
+      "loss": 0.9024,
+      "step": 16030
+    },
+    {
+      "epoch": 0.44027903655488726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2396810454565244e-05,
+      "loss": 0.8505,
+      "step": 16031
+    },
+    {
+      "epoch": 0.4403065007827305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.239594685696339e-05,
+      "loss": 0.9359,
+      "step": 16032
+    },
+    {
+      "epoch": 0.4403339650105737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2395083240403877e-05,
+      "loss": 0.913,
+      "step": 16033
+    },
+    {
+      "epoch": 0.440361429238417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2394219604893539e-05,
+      "loss": 1.11,
+      "step": 16034
+    },
+    {
+      "epoch": 0.4403888934662602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2393355950439205e-05,
+      "loss": 0.9088,
+      "step": 16035
+    },
+    {
+      "epoch": 0.4404163576941034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2392492277047714e-05,
+      "loss": 0.8906,
+      "step": 16036
+    },
+    {
+      "epoch": 0.44044382192194664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2391628584725897e-05,
+      "loss": 0.8915,
+      "step": 16037
+    },
+    {
+      "epoch": 0.4404712861497899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2390764873480585e-05,
+      "loss": 0.9723,
+      "step": 16038
+    },
+    {
+      "epoch": 0.44049875037763314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.238990114331862e-05,
+      "loss": 0.8836,
+      "step": 16039
+    },
+    {
+      "epoch": 0.44052621460547636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.238903739424683e-05,
+      "loss": 0.8703,
+      "step": 16040
+    },
+    {
+      "epoch": 0.4405536788333196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2388173626272048e-05,
+      "loss": 0.8349,
+      "step": 16041
+    },
+    {
+      "epoch": 0.44058114306116286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2387309839401113e-05,
+      "loss": 0.899,
+      "step": 16042
+    },
+    {
+      "epoch": 0.4406086072890061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2386446033640857e-05,
+      "loss": 0.8854,
+      "step": 16043
+    },
+    {
+      "epoch": 0.4406360715168493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2385582208998117e-05,
+      "loss": 0.9677,
+      "step": 16044
+    },
+    {
+      "epoch": 0.4406635357446925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2384718365479729e-05,
+      "loss": 0.9921,
+      "step": 16045
+    },
+    {
+      "epoch": 0.44069099997253575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2383854503092521e-05,
+      "loss": 0.9823,
+      "step": 16046
+    },
+    {
+      "epoch": 0.440718464200379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2382990621843334e-05,
+      "loss": 0.9442,
+      "step": 16047
+    },
+    {
+      "epoch": 0.44074592842822224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2382126721739004e-05,
+      "loss": 0.9081,
+      "step": 16048
+    },
+    {
+      "epoch": 0.44077339265606547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2381262802786362e-05,
+      "loss": 0.7965,
+      "step": 16049
+    },
+    {
+      "epoch": 0.4408008568839087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2380398864992245e-05,
+      "loss": 0.9644,
+      "step": 16050
+    },
+    {
+      "epoch": 0.44082832111175196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2379534908363493e-05,
+      "loss": 0.8704,
+      "step": 16051
+    },
+    {
+      "epoch": 0.4408557853395952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.237867093290694e-05,
+      "loss": 0.9243,
+      "step": 16052
+    },
+    {
+      "epoch": 0.4408832495674384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2377806938629418e-05,
+      "loss": 0.8981,
+      "step": 16053
+    },
+    {
+      "epoch": 0.44091071379528163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.237694292553777e-05,
+      "loss": 0.8061,
+      "step": 16054
+    },
+    {
+      "epoch": 0.4409381780231249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2376078893638826e-05,
+      "loss": 0.868,
+      "step": 16055
+    },
+    {
+      "epoch": 0.4409656422509681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2375214842939426e-05,
+      "loss": 0.9104,
+      "step": 16056
+    },
+    {
+      "epoch": 0.44099310647881135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2374350773446409e-05,
+      "loss": 1.0168,
+      "step": 16057
+    },
+    {
+      "epoch": 0.44102057070665457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2373486685166604e-05,
+      "loss": 0.9457,
+      "step": 16058
+    },
+    {
+      "epoch": 0.4410480349344978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2372622578106857e-05,
+      "loss": 0.9275,
+      "step": 16059
+    },
+    {
+      "epoch": 0.44107549916234107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2371758452273999e-05,
+      "loss": 0.9216,
+      "step": 16060
+    },
+    {
+      "epoch": 0.4411029633901843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2370894307674868e-05,
+      "loss": 0.9375,
+      "step": 16061
+    },
+    {
+      "epoch": 0.4411304276180275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2370030144316305e-05,
+      "loss": 0.8722,
+      "step": 16062
+    },
+    {
+      "epoch": 0.44115789184587073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2369165962205145e-05,
+      "loss": 1.0568,
+      "step": 16063
+    },
+    {
+      "epoch": 0.441185356073714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2368301761348227e-05,
+      "loss": 0.8916,
+      "step": 16064
+    },
+    {
+      "epoch": 0.44121282030155723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2367437541752387e-05,
+      "loss": 0.8681,
+      "step": 16065
+    },
+    {
+      "epoch": 0.44124028452940045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2366573303424465e-05,
+      "loss": 0.9002,
+      "step": 16066
+    },
+    {
+      "epoch": 0.44126774875724367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2365709046371299e-05,
+      "loss": 0.8829,
+      "step": 16067
+    },
+    {
+      "epoch": 0.44129521298508695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2364844770599724e-05,
+      "loss": 0.9335,
+      "step": 16068
+    },
+    {
+      "epoch": 0.44132267721293017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2363980476116582e-05,
+      "loss": 0.8935,
+      "step": 16069
+    },
+    {
+      "epoch": 0.4413501414407734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236311616292871e-05,
+      "loss": 0.9711,
+      "step": 16070
+    },
+    {
+      "epoch": 0.4413776056686166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236225183104295e-05,
+      "loss": 1.0247,
+      "step": 16071
+    },
+    {
+      "epoch": 0.44140506989645983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2361387480466139e-05,
+      "loss": 0.9261,
+      "step": 16072
+    },
+    {
+      "epoch": 0.4414325341243031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2360523111205112e-05,
+      "loss": 0.9445,
+      "step": 16073
+    },
+    {
+      "epoch": 0.44145999835214633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2359658723266714e-05,
+      "loss": 0.8966,
+      "step": 16074
+    },
+    {
+      "epoch": 0.44148746257998955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.235879431665778e-05,
+      "loss": 0.8484,
+      "step": 16075
+    },
+    {
+      "epoch": 0.4415149268078328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2357929891385154e-05,
+      "loss": 0.9872,
+      "step": 16076
+    },
+    {
+      "epoch": 0.44154239103567605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2357065447455674e-05,
+      "loss": 0.9756,
+      "step": 16077
+    },
+    {
+      "epoch": 0.4415698552635193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2356200984876177e-05,
+      "loss": 0.9222,
+      "step": 16078
+    },
+    {
+      "epoch": 0.4415973194913625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2355336503653507e-05,
+      "loss": 0.8674,
+      "step": 16079
+    },
+    {
+      "epoch": 0.4416247837192057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2354472003794503e-05,
+      "loss": 0.9154,
+      "step": 16080
+    },
+    {
+      "epoch": 0.441652247947049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2353607485306001e-05,
+      "loss": 0.9728,
+      "step": 16081
+    },
+    {
+      "epoch": 0.4416797121748922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2352742948194845e-05,
+      "loss": 0.8844,
+      "step": 16082
+    },
+    {
+      "epoch": 0.44170717640273544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2351878392467877e-05,
+      "loss": 0.9649,
+      "step": 16083
+    },
+    {
+      "epoch": 0.44173464063057866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2351013818131939e-05,
+      "loss": 0.8684,
+      "step": 16084
+    },
+    {
+      "epoch": 0.4417621048584219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2350149225193865e-05,
+      "loss": 0.9064,
+      "step": 16085
+    },
+    {
+      "epoch": 0.44178956908626515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.23492846136605e-05,
+      "loss": 0.9094,
+      "step": 16086
+    },
+    {
+      "epoch": 0.4418170333141084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2348419983538684e-05,
+      "loss": 0.9518,
+      "step": 16087
+    },
+    {
+      "epoch": 0.4418444975419516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2347555334835262e-05,
+      "loss": 0.8069,
+      "step": 16088
+    },
+    {
+      "epoch": 0.4418719617697948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2346690667557075e-05,
+      "loss": 0.9164,
+      "step": 16089
+    },
+    {
+      "epoch": 0.4418994259976381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2345825981710957e-05,
+      "loss": 0.8772,
+      "step": 16090
+    },
+    {
+      "epoch": 0.4419268902254813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.234496127730376e-05,
+      "loss": 0.9354,
+      "step": 16091
+    },
+    {
+      "epoch": 0.44195435445332454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.234409655434232e-05,
+      "loss": 0.9005,
+      "step": 16092
+    },
+    {
+      "epoch": 0.44198181868116776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2343231812833476e-05,
+      "loss": 0.9836,
+      "step": 16093
+    },
+    {
+      "epoch": 0.44200928290901104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2342367052784076e-05,
+      "loss": 1.0088,
+      "step": 16094
+    },
+    {
+      "epoch": 0.44203674713685426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2341502274200963e-05,
+      "loss": 1.0236,
+      "step": 16095
+    },
+    {
+      "epoch": 0.4420642113646975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2340637477090977e-05,
+      "loss": 0.8309,
+      "step": 16096
+    },
+    {
+      "epoch": 0.4420916755925407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2339772661460958e-05,
+      "loss": 0.8952,
+      "step": 16097
+    },
+    {
+      "epoch": 0.442119139820384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2338907827317755e-05,
+      "loss": 0.9469,
+      "step": 16098
+    },
+    {
+      "epoch": 0.4421466040482272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2338042974668203e-05,
+      "loss": 0.9086,
+      "step": 16099
+    },
+    {
+      "epoch": 0.4421740682760704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2337178103519152e-05,
+      "loss": 0.8792,
+      "step": 16100
+    },
+    {
+      "epoch": 0.44220153250391364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2336313213877442e-05,
+      "loss": 0.9834,
+      "step": 16101
+    },
+    {
+      "epoch": 0.44222899673175686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2335448305749915e-05,
+      "loss": 0.8823,
+      "step": 16102
+    },
+    {
+      "epoch": 0.44225646095960014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2334583379143418e-05,
+      "loss": 0.9801,
+      "step": 16103
+    },
+    {
+      "epoch": 0.44228392518744336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2333718434064795e-05,
+      "loss": 0.9183,
+      "step": 16104
+    },
+    {
+      "epoch": 0.4423113894152866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2332853470520885e-05,
+      "loss": 0.8325,
+      "step": 16105
+    },
+    {
+      "epoch": 0.4423388536431298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2331988488518535e-05,
+      "loss": 0.8659,
+      "step": 16106
+    },
+    {
+      "epoch": 0.4423663178709731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2331123488064586e-05,
+      "loss": 0.8641,
+      "step": 16107
+    },
+    {
+      "epoch": 0.4423937820988163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233025846916589e-05,
+      "loss": 0.8952,
+      "step": 16108
+    },
+    {
+      "epoch": 0.4424212463266595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2329393431829284e-05,
+      "loss": 0.8494,
+      "step": 16109
+    },
+    {
+      "epoch": 0.44244871055450274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2328528376061618e-05,
+      "loss": 0.8457,
+      "step": 16110
+    },
+    {
+      "epoch": 0.442476174782346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2327663301869728e-05,
+      "loss": 0.9757,
+      "step": 16111
+    },
+    {
+      "epoch": 0.44250363901018924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2326798209260469e-05,
+      "loss": 0.9222,
+      "step": 16112
+    },
+    {
+      "epoch": 0.44253110323803246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2325933098240679e-05,
+      "loss": 0.8817,
+      "step": 16113
+    },
+    {
+      "epoch": 0.4425585674658757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2325067968817204e-05,
+      "loss": 0.8953,
+      "step": 16114
+    },
+    {
+      "epoch": 0.4425860316937189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2324202820996893e-05,
+      "loss": 0.8662,
+      "step": 16115
+    },
+    {
+      "epoch": 0.4426134959215622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232333765478659e-05,
+      "loss": 0.848,
+      "step": 16116
+    },
+    {
+      "epoch": 0.4426409601494054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2322472470193137e-05,
+      "loss": 0.8922,
+      "step": 16117
+    },
+    {
+      "epoch": 0.4426684243772486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232160726722338e-05,
+      "loss": 0.9478,
+      "step": 16118
+    },
+    {
+      "epoch": 0.44269588860509185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2320742045884172e-05,
+      "loss": 0.9277,
+      "step": 16119
+    },
+    {
+      "epoch": 0.4427233528329351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2319876806182352e-05,
+      "loss": 1.0034,
+      "step": 16120
+    },
+    {
+      "epoch": 0.44275081706077835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2319011548124766e-05,
+      "loss": 0.9383,
+      "step": 16121
+    },
+    {
+      "epoch": 0.44277828128862157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2318146271718265e-05,
+      "loss": 0.9475,
+      "step": 16122
+    },
+    {
+      "epoch": 0.4428057455164648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2317280976969692e-05,
+      "loss": 0.9152,
+      "step": 16123
+    },
+    {
+      "epoch": 0.44283320974430807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2316415663885895e-05,
+      "loss": 1.0223,
+      "step": 16124
+    },
+    {
+      "epoch": 0.4428606739721513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231555033247372e-05,
+      "loss": 0.9675,
+      "step": 16125
+    },
+    {
+      "epoch": 0.4428881381999945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2314684982740013e-05,
+      "loss": 0.9192,
+      "step": 16126
+    },
+    {
+      "epoch": 0.44291560242783773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2313819614691623e-05,
+      "loss": 0.8886,
+      "step": 16127
+    },
+    {
+      "epoch": 0.44294306665568095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2312954228335395e-05,
+      "loss": 0.9301,
+      "step": 16128
+    },
+    {
+      "epoch": 0.4429705308835242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2312088823678174e-05,
+      "loss": 0.8914,
+      "step": 16129
+    },
+    {
+      "epoch": 0.44299799511136745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2311223400726813e-05,
+      "loss": 0.9697,
+      "step": 16130
+    },
+    {
+      "epoch": 0.44302545933921067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2310357959488159e-05,
+      "loss": 0.8356,
+      "step": 16131
+    },
+    {
+      "epoch": 0.4430529235670539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2309492499969059e-05,
+      "loss": 0.9943,
+      "step": 16132
+    },
+    {
+      "epoch": 0.44308038779489717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2308627022176356e-05,
+      "loss": 0.8691,
+      "step": 16133
+    },
+    {
+      "epoch": 0.4431078520227404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2307761526116901e-05,
+      "loss": 0.8991,
+      "step": 16134
+    },
+    {
+      "epoch": 0.4431353162505836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2306896011797544e-05,
+      "loss": 0.8568,
+      "step": 16135
+    },
+    {
+      "epoch": 0.44316278047842683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2306030479225137e-05,
+      "loss": 0.9797,
+      "step": 16136
+    },
+    {
+      "epoch": 0.4431902447062701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2305164928406518e-05,
+      "loss": 0.9056,
+      "step": 16137
+    },
+    {
+      "epoch": 0.44321770893411333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2304299359348541e-05,
+      "loss": 0.9506,
+      "step": 16138
+    },
+    {
+      "epoch": 0.44324517316195655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2303433772058058e-05,
+      "loss": 0.8064,
+      "step": 16139
+    },
+    {
+      "epoch": 0.4432726373897998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2302568166541914e-05,
+      "loss": 0.9077,
+      "step": 16140
+    },
+    {
+      "epoch": 0.443300101617643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2301702542806958e-05,
+      "loss": 0.9297,
+      "step": 16141
+    },
+    {
+      "epoch": 0.44332756584548627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.230083690086004e-05,
+      "loss": 0.9666,
+      "step": 16142
+    },
+    {
+      "epoch": 0.4433550300733295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2299971240708008e-05,
+      "loss": 0.9371,
+      "step": 16143
+    },
+    {
+      "epoch": 0.4433824943011727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2299105562357716e-05,
+      "loss": 0.8954,
+      "step": 16144
+    },
+    {
+      "epoch": 0.44340995852901594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2298239865816008e-05,
+      "loss": 1.0323,
+      "step": 16145
+    },
+    {
+      "epoch": 0.4434374227568592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2297374151089733e-05,
+      "loss": 0.8165,
+      "step": 16146
+    },
+    {
+      "epoch": 0.44346488698470243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2296508418185748e-05,
+      "loss": 0.9552,
+      "step": 16147
+    },
+    {
+      "epoch": 0.44349235121254565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22956426671109e-05,
+      "loss": 1.0396,
+      "step": 16148
+    },
+    {
+      "epoch": 0.4435198154403889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2294776897872033e-05,
+      "loss": 0.928,
+      "step": 16149
+    },
+    {
+      "epoch": 0.44354727966823215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2293911110476005e-05,
+      "loss": 0.9369,
+      "step": 16150
+    },
+    {
+      "epoch": 0.4435747438960754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2293045304929664e-05,
+      "loss": 0.9428,
+      "step": 16151
+    },
+    {
+      "epoch": 0.4436022081239186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2292179481239861e-05,
+      "loss": 0.8864,
+      "step": 16152
+    },
+    {
+      "epoch": 0.4436296723517618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2291313639413443e-05,
+      "loss": 0.8963,
+      "step": 16153
+    },
+    {
+      "epoch": 0.44365713657960504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2290447779457266e-05,
+      "loss": 0.9078,
+      "step": 16154
+    },
+    {
+      "epoch": 0.4436846008074483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2289581901378178e-05,
+      "loss": 0.9367,
+      "step": 16155
+    },
+    {
+      "epoch": 0.44371206503529154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2288716005183032e-05,
+      "loss": 0.9459,
+      "step": 16156
+    },
+    {
+      "epoch": 0.44373952926313476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2287850090878681e-05,
+      "loss": 0.9532,
+      "step": 16157
+    },
+    {
+      "epoch": 0.443766993490978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2286984158471969e-05,
+      "loss": 0.8804,
+      "step": 16158
+    },
+    {
+      "epoch": 0.44379445771882126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2286118207969755e-05,
+      "loss": 0.906,
+      "step": 16159
+    },
+    {
+      "epoch": 0.4438219219466645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2285252239378888e-05,
+      "loss": 1.0349,
+      "step": 16160
+    },
+    {
+      "epoch": 0.4438493861745077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2284386252706219e-05,
+      "loss": 0.8918,
+      "step": 16161
+    },
+    {
+      "epoch": 0.4438768504023509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2283520247958602e-05,
+      "loss": 0.9134,
+      "step": 16162
+    },
+    {
+      "epoch": 0.4439043146301942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.228265422514289e-05,
+      "loss": 0.82,
+      "step": 16163
+    },
+    {
+      "epoch": 0.4439317788580374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.228178818426593e-05,
+      "loss": 0.9453,
+      "step": 16164
+    },
+    {
+      "epoch": 0.44395924308588064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.228092212533458e-05,
+      "loss": 0.7881,
+      "step": 16165
+    },
+    {
+      "epoch": 0.44398670731372386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.228005604835569e-05,
+      "loss": 0.9243,
+      "step": 16166
+    },
+    {
+      "epoch": 0.4440141715415671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2279189953336113e-05,
+      "loss": 0.9599,
+      "step": 16167
+    },
+    {
+      "epoch": 0.44404163576941036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2278323840282705e-05,
+      "loss": 0.8201,
+      "step": 16168
+    },
+    {
+      "epoch": 0.4440690999972536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2277457709202314e-05,
+      "loss": 0.8098,
+      "step": 16169
+    },
+    {
+      "epoch": 0.4440965642250968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2276591560101794e-05,
+      "loss": 1.0435,
+      "step": 16170
+    },
+    {
+      "epoch": 0.44412402845294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2275725392988002e-05,
+      "loss": 1.0264,
+      "step": 16171
+    },
+    {
+      "epoch": 0.4441514926807833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2274859207867789e-05,
+      "loss": 1.014,
+      "step": 16172
+    },
+    {
+      "epoch": 0.4441789569086265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2273993004748006e-05,
+      "loss": 0.9534,
+      "step": 16173
+    },
+    {
+      "epoch": 0.44420642113646974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2273126783635512e-05,
+      "loss": 0.907,
+      "step": 16174
+    },
+    {
+      "epoch": 0.44423388536431296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2272260544537158e-05,
+      "loss": 0.8437,
+      "step": 16175
+    },
+    {
+      "epoch": 0.44426134959215624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2271394287459798e-05,
+      "loss": 0.9733,
+      "step": 16176
+    },
+    {
+      "epoch": 0.44428881381999946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2270528012410285e-05,
+      "loss": 0.9425,
+      "step": 16177
+    },
+    {
+      "epoch": 0.4443162780478427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2269661719395477e-05,
+      "loss": 1.0081,
+      "step": 16178
+    },
+    {
+      "epoch": 0.4443437422756859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2268795408422228e-05,
+      "loss": 0.9397,
+      "step": 16179
+    },
+    {
+      "epoch": 0.4443712065035291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2267929079497389e-05,
+      "loss": 0.9322,
+      "step": 16180
+    },
+    {
+      "epoch": 0.4443986707313724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2267062732627816e-05,
+      "loss": 1.0125,
+      "step": 16181
+    },
+    {
+      "epoch": 0.4444261349592156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2266196367820365e-05,
+      "loss": 0.8647,
+      "step": 16182
+    },
+    {
+      "epoch": 0.44445359918705885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2265329985081891e-05,
+      "loss": 0.9434,
+      "step": 16183
+    },
+    {
+      "epoch": 0.44448106341490207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2264463584419248e-05,
+      "loss": 0.8898,
+      "step": 16184
+    },
+    {
+      "epoch": 0.44450852764274534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2263597165839292e-05,
+      "loss": 0.8196,
+      "step": 16185
+    },
+    {
+      "epoch": 0.44453599187058856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2262730729348877e-05,
+      "loss": 0.8507,
+      "step": 16186
+    },
+    {
+      "epoch": 0.4445634560984318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2261864274954862e-05,
+      "loss": 0.8537,
+      "step": 16187
+    },
+    {
+      "epoch": 0.444590920326275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2260997802664101e-05,
+      "loss": 0.9337,
+      "step": 16188
+    },
+    {
+      "epoch": 0.4446183845541183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2260131312483447e-05,
+      "loss": 0.9664,
+      "step": 16189
+    },
+    {
+      "epoch": 0.4446458487819615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2259264804419761e-05,
+      "loss": 0.9191,
+      "step": 16190
+    },
+    {
+      "epoch": 0.4446733130098047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2258398278479895e-05,
+      "loss": 0.8823,
+      "step": 16191
+    },
+    {
+      "epoch": 0.44470077723764795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.225753173467071e-05,
+      "loss": 0.9343,
+      "step": 16192
+    },
+    {
+      "epoch": 0.4447282414654912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2256665172999058e-05,
+      "loss": 0.8881,
+      "step": 16193
+    },
+    {
+      "epoch": 0.44475570569333445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2255798593471795e-05,
+      "loss": 0.9262,
+      "step": 16194
+    },
+    {
+      "epoch": 0.44478316992117767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2254931996095782e-05,
+      "loss": 1.0027,
+      "step": 16195
+    },
+    {
+      "epoch": 0.4448106341490209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2254065380877873e-05,
+      "loss": 1.0551,
+      "step": 16196
+    },
+    {
+      "epoch": 0.4448380983768641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2253198747824924e-05,
+      "loss": 0.9274,
+      "step": 16197
+    },
+    {
+      "epoch": 0.4448655626047074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2252332096943795e-05,
+      "loss": 0.9957,
+      "step": 16198
+    },
+    {
+      "epoch": 0.4448930268325506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2251465428241342e-05,
+      "loss": 0.8482,
+      "step": 16199
+    },
+    {
+      "epoch": 0.44492049106039383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2250598741724423e-05,
+      "loss": 0.8492,
+      "step": 16200
+    },
+    {
+      "epoch": 0.44494795528823705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2249732037399896e-05,
+      "loss": 0.9548,
+      "step": 16201
+    },
+    {
+      "epoch": 0.44497541951608033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2248865315274614e-05,
+      "loss": 0.9374,
+      "step": 16202
+    },
+    {
+      "epoch": 0.44500288374392355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.224799857535544e-05,
+      "loss": 0.8448,
+      "step": 16203
+    },
+    {
+      "epoch": 0.44503034797176677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2247131817649232e-05,
+      "loss": 0.9933,
+      "step": 16204
+    },
+    {
+      "epoch": 0.44505781219961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2246265042162846e-05,
+      "loss": 0.8252,
+      "step": 16205
+    },
+    {
+      "epoch": 0.44508527642745327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2245398248903138e-05,
+      "loss": 1.0396,
+      "step": 16206
+    },
+    {
+      "epoch": 0.4451127406552965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2244531437876971e-05,
+      "loss": 0.8441,
+      "step": 16207
+    },
+    {
+      "epoch": 0.4451402048831397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2243664609091205e-05,
+      "loss": 1.009,
+      "step": 16208
+    },
+    {
+      "epoch": 0.44516766911098293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2242797762552692e-05,
+      "loss": 0.9211,
+      "step": 16209
+    },
+    {
+      "epoch": 0.44519513333882615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2241930898268296e-05,
+      "loss": 0.9035,
+      "step": 16210
+    },
+    {
+      "epoch": 0.44522259756666943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2241064016244872e-05,
+      "loss": 0.8422,
+      "step": 16211
+    },
+    {
+      "epoch": 0.44525006179451265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2240197116489284e-05,
+      "loss": 1.0296,
+      "step": 16212
+    },
+    {
+      "epoch": 0.4452775260223559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2239330199008387e-05,
+      "loss": 0.856,
+      "step": 16213
+    },
+    {
+      "epoch": 0.4453049902501991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2238463263809042e-05,
+      "loss": 1.0214,
+      "step": 16214
+    },
+    {
+      "epoch": 0.44533245447804237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.223759631089811e-05,
+      "loss": 0.9054,
+      "step": 16215
+    },
+    {
+      "epoch": 0.4453599187058856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.223672934028245e-05,
+      "loss": 0.9249,
+      "step": 16216
+    },
+    {
+      "epoch": 0.4453873829337288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2235862351968919e-05,
+      "loss": 0.8704,
+      "step": 16217
+    },
+    {
+      "epoch": 0.44541484716157204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2234995345964378e-05,
+      "loss": 0.865,
+      "step": 16218
+    },
+    {
+      "epoch": 0.4454423113894153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.223412832227569e-05,
+      "loss": 0.8747,
+      "step": 16219
+    },
+    {
+      "epoch": 0.44546977561725853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2233261280909714e-05,
+      "loss": 0.8199,
+      "step": 16220
+    },
+    {
+      "epoch": 0.44549723984510176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2232394221873308e-05,
+      "loss": 0.9531,
+      "step": 16221
+    },
+    {
+      "epoch": 0.445524704072945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2231527145173335e-05,
+      "loss": 0.8495,
+      "step": 16222
+    },
+    {
+      "epoch": 0.4455521683007882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2230660050816651e-05,
+      "loss": 0.9659,
+      "step": 16223
+    },
+    {
+      "epoch": 0.4455796325286315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2229792938810125e-05,
+      "loss": 0.916,
+      "step": 16224
+    },
+    {
+      "epoch": 0.4456070967564747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2228925809160614e-05,
+      "loss": 0.983,
+      "step": 16225
+    },
+    {
+      "epoch": 0.4456345609843179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2228058661874978e-05,
+      "loss": 0.9146,
+      "step": 16226
+    },
+    {
+      "epoch": 0.44566202521216114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2227191496960077e-05,
+      "loss": 0.843,
+      "step": 16227
+    },
+    {
+      "epoch": 0.4456894894400044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2226324314422778e-05,
+      "loss": 0.945,
+      "step": 16228
+    },
+    {
+      "epoch": 0.44571695366784764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2225457114269934e-05,
+      "loss": 0.9738,
+      "step": 16229
+    },
+    {
+      "epoch": 0.44574441789569086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2224589896508412e-05,
+      "loss": 0.8086,
+      "step": 16230
+    },
+    {
+      "epoch": 0.4457718821235341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2223722661145075e-05,
+      "loss": 0.9513,
+      "step": 16231
+    },
+    {
+      "epoch": 0.44579934635137736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2222855408186783e-05,
+      "loss": 0.8812,
+      "step": 16232
+    },
+    {
+      "epoch": 0.4458268105792206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2221988137640396e-05,
+      "loss": 0.9547,
+      "step": 16233
+    },
+    {
+      "epoch": 0.4458542748070638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2221120849512777e-05,
+      "loss": 0.9792,
+      "step": 16234
+    },
+    {
+      "epoch": 0.445881739034907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2220253543810792e-05,
+      "loss": 0.9914,
+      "step": 16235
+    },
+    {
+      "epoch": 0.44590920326275024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2219386220541301e-05,
+      "loss": 0.9257,
+      "step": 16236
+    },
+    {
+      "epoch": 0.4459366674905935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2218518879711168e-05,
+      "loss": 0.8944,
+      "step": 16237
+    },
+    {
+      "epoch": 0.44596413171843674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.221765152132725e-05,
+      "loss": 0.9462,
+      "step": 16238
+    },
+    {
+      "epoch": 0.44599159594627996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2216784145396417e-05,
+      "loss": 0.9527,
+      "step": 16239
+    },
+    {
+      "epoch": 0.4460190601741232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.221591675192553e-05,
+      "loss": 0.9045,
+      "step": 16240
+    },
+    {
+      "epoch": 0.44604652440196646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2215049340921447e-05,
+      "loss": 0.8903,
+      "step": 16241
+    },
+    {
+      "epoch": 0.4460739886298097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2214181912391038e-05,
+      "loss": 0.93,
+      "step": 16242
+    },
+    {
+      "epoch": 0.4461014528576529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2213314466341166e-05,
+      "loss": 1.0321,
+      "step": 16243
+    },
+    {
+      "epoch": 0.4461289170854961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2212447002778692e-05,
+      "loss": 0.8432,
+      "step": 16244
+    },
+    {
+      "epoch": 0.4461563813133394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2211579521710477e-05,
+      "loss": 0.9609,
+      "step": 16245
+    },
+    {
+      "epoch": 0.4461838455411826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2210712023143392e-05,
+      "loss": 0.8531,
+      "step": 16246
+    },
+    {
+      "epoch": 0.44621130976902584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2209844507084293e-05,
+      "loss": 0.916,
+      "step": 16247
+    },
+    {
+      "epoch": 0.44623877399686906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2208976973540053e-05,
+      "loss": 1.0002,
+      "step": 16248
+    },
+    {
+      "epoch": 0.4462662382247123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220810942251753e-05,
+      "loss": 1.0044,
+      "step": 16249
+    },
+    {
+      "epoch": 0.44629370245255556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2207241854023586e-05,
+      "loss": 0.8486,
+      "step": 16250
+    },
+    {
+      "epoch": 0.4463211666803988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2206374268065095e-05,
+      "loss": 1.0294,
+      "step": 16251
+    },
+    {
+      "epoch": 0.446348630908242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2205506664648914e-05,
+      "loss": 1.0087,
+      "step": 16252
+    },
+    {
+      "epoch": 0.4463760951360852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220463904378191e-05,
+      "loss": 1.095,
+      "step": 16253
+    },
+    {
+      "epoch": 0.4464035593639285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2203771405470947e-05,
+      "loss": 1.059,
+      "step": 16254
+    },
+    {
+      "epoch": 0.4464310235917717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220290374972289e-05,
+      "loss": 0.8575,
+      "step": 16255
+    },
+    {
+      "epoch": 0.44645848781961495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2202036076544609e-05,
+      "loss": 0.835,
+      "step": 16256
+    },
+    {
+      "epoch": 0.44648595204745817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2201168385942963e-05,
+      "loss": 0.9539,
+      "step": 16257
+    },
+    {
+      "epoch": 0.44651341627530144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2200300677924819e-05,
+      "loss": 0.8937,
+      "step": 16258
+    },
+    {
+      "epoch": 0.44654088050314467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2199432952497043e-05,
+      "loss": 0.9224,
+      "step": 16259
+    },
+    {
+      "epoch": 0.4465683447309879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2198565209666509e-05,
+      "loss": 0.9199,
+      "step": 16260
+    },
+    {
+      "epoch": 0.4465958089588311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.219769744944007e-05,
+      "loss": 0.9285,
+      "step": 16261
+    },
+    {
+      "epoch": 0.44662327318667433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2196829671824595e-05,
+      "loss": 1.0131,
+      "step": 16262
+    },
+    {
+      "epoch": 0.4466507374145176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2195961876826957e-05,
+      "loss": 0.9111,
+      "step": 16263
+    },
+    {
+      "epoch": 0.44667820164236083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2195094064454016e-05,
+      "loss": 0.9203,
+      "step": 16264
+    },
+    {
+      "epoch": 0.44670566587020405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2194226234712639e-05,
+      "loss": 0.932,
+      "step": 16265
+    },
+    {
+      "epoch": 0.44673313009804727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2193358387609698e-05,
+      "loss": 0.9741,
+      "step": 16266
+    },
+    {
+      "epoch": 0.44676059432589055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.219249052315205e-05,
+      "loss": 0.8415,
+      "step": 16267
+    },
+    {
+      "epoch": 0.44678805855373377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2191622641346572e-05,
+      "loss": 0.9413,
+      "step": 16268
+    },
+    {
+      "epoch": 0.446815522781577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2190754742200127e-05,
+      "loss": 0.8792,
+      "step": 16269
+    },
+    {
+      "epoch": 0.4468429870094202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2189886825719579e-05,
+      "loss": 0.8816,
+      "step": 16270
+    },
+    {
+      "epoch": 0.4468704512372635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.21890188919118e-05,
+      "loss": 1.0076,
+      "step": 16271
+    },
+    {
+      "epoch": 0.4468979154651067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2188150940783658e-05,
+      "loss": 0.9584,
+      "step": 16272
+    },
+    {
+      "epoch": 0.44692537969294993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2187282972342014e-05,
+      "loss": 0.9321,
+      "step": 16273
+    },
+    {
+      "epoch": 0.44695284392079315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2186414986593742e-05,
+      "loss": 0.8717,
+      "step": 16274
+    },
+    {
+      "epoch": 0.4469803081486364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2185546983545706e-05,
+      "loss": 0.9321,
+      "step": 16275
+    },
+    {
+      "epoch": 0.44700777237647965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2184678963204778e-05,
+      "loss": 0.9982,
+      "step": 16276
+    },
+    {
+      "epoch": 0.44703523660432287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2183810925577822e-05,
+      "loss": 0.8544,
+      "step": 16277
+    },
+    {
+      "epoch": 0.4470627008321661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.218294287067171e-05,
+      "loss": 0.9335,
+      "step": 16278
+    },
+    {
+      "epoch": 0.4470901650600093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2182074798493306e-05,
+      "loss": 1.0117,
+      "step": 16279
+    },
+    {
+      "epoch": 0.4471176292878526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2181206709049482e-05,
+      "loss": 0.8681,
+      "step": 16280
+    },
+    {
+      "epoch": 0.4471450935156958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2180338602347107e-05,
+      "loss": 0.9816,
+      "step": 16281
+    },
+    {
+      "epoch": 0.44717255774353903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2179470478393045e-05,
+      "loss": 0.858,
+      "step": 16282
+    },
+    {
+      "epoch": 0.44720002197138226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217860233719417e-05,
+      "loss": 0.9031,
+      "step": 16283
+    },
+    {
+      "epoch": 0.44722748619922553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2177734178757352e-05,
+      "loss": 0.9456,
+      "step": 16284
+    },
+    {
+      "epoch": 0.44725495042706875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2176866003089453e-05,
+      "loss": 0.9082,
+      "step": 16285
+    },
+    {
+      "epoch": 0.447282414654912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2175997810197351e-05,
+      "loss": 0.9617,
+      "step": 16286
+    },
+    {
+      "epoch": 0.4473098788827552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2175129600087909e-05,
+      "loss": 0.9226,
+      "step": 16287
+    },
+    {
+      "epoch": 0.4473373431105985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2174261372768002e-05,
+      "loss": 0.8896,
+      "step": 16288
+    },
+    {
+      "epoch": 0.4473648073384417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2173393128244493e-05,
+      "loss": 0.8766,
+      "step": 16289
+    },
+    {
+      "epoch": 0.4473922715662849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2172524866524258e-05,
+      "loss": 0.8212,
+      "step": 16290
+    },
+    {
+      "epoch": 0.44741973579412814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2171656587614162e-05,
+      "loss": 0.9174,
+      "step": 16291
+    },
+    {
+      "epoch": 0.44744720002197136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217078829152108e-05,
+      "loss": 0.8941,
+      "step": 16292
+    },
+    {
+      "epoch": 0.44747466424981464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2169919978251883e-05,
+      "loss": 0.839,
+      "step": 16293
+    },
+    {
+      "epoch": 0.44750212847765786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2169051647813432e-05,
+      "loss": 0.9085,
+      "step": 16294
+    },
+    {
+      "epoch": 0.4475295927055011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2168183300212608e-05,
+      "loss": 1.0731,
+      "step": 16295
+    },
+    {
+      "epoch": 0.4475570569333443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216731493545628e-05,
+      "loss": 0.8878,
+      "step": 16296
+    },
+    {
+      "epoch": 0.4475845211611876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2166446553551312e-05,
+      "loss": 0.961,
+      "step": 16297
+    },
+    {
+      "epoch": 0.4476119853890308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2165578154504582e-05,
+      "loss": 0.8456,
+      "step": 16298
+    },
+    {
+      "epoch": 0.447639449616874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216470973832296e-05,
+      "loss": 1.0363,
+      "step": 16299
+    },
+    {
+      "epoch": 0.44766691384471724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2163841305013316e-05,
+      "loss": 0.8256,
+      "step": 16300
+    },
+    {
+      "epoch": 0.4476943780725605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2162972854582518e-05,
+      "loss": 0.9909,
+      "step": 16301
+    },
+    {
+      "epoch": 0.44772184230040374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2162104387037444e-05,
+      "loss": 0.9568,
+      "step": 16302
+    },
+    {
+      "epoch": 0.44774930652824696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216123590238496e-05,
+      "loss": 0.8829,
+      "step": 16303
+    },
+    {
+      "epoch": 0.4477767707560902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2160367400631943e-05,
+      "loss": 0.912,
+      "step": 16304
+    },
+    {
+      "epoch": 0.4478042349839334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2159498881785261e-05,
+      "loss": 0.881,
+      "step": 16305
+    },
+    {
+      "epoch": 0.4478316992117767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2158630345851787e-05,
+      "loss": 0.8579,
+      "step": 16306
+    },
+    {
+      "epoch": 0.4478591634396199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2157761792838396e-05,
+      "loss": 1.0277,
+      "step": 16307
+    },
+    {
+      "epoch": 0.4478866276674631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2156893222751958e-05,
+      "loss": 0.8652,
+      "step": 16308
+    },
+    {
+      "epoch": 0.44791409189530634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2156024635599342e-05,
+      "loss": 1.0274,
+      "step": 16309
+    },
+    {
+      "epoch": 0.4479415561231496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2155156031387422e-05,
+      "loss": 0.9347,
+      "step": 16310
+    },
+    {
+      "epoch": 0.44796902035099284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2154287410123078e-05,
+      "loss": 0.8894,
+      "step": 16311
+    },
+    {
+      "epoch": 0.44799648457883606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2153418771813174e-05,
+      "loss": 0.8981,
+      "step": 16312
+    },
+    {
+      "epoch": 0.4480239488066793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.215255011646459e-05,
+      "loss": 1.0174,
+      "step": 16313
+    },
+    {
+      "epoch": 0.44805141303452256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2151681444084194e-05,
+      "loss": 0.9599,
+      "step": 16314
+    },
+    {
+      "epoch": 0.4480788772623658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2150812754678858e-05,
+      "loss": 0.8615,
+      "step": 16315
+    },
+    {
+      "epoch": 0.448106341490209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2149944048255464e-05,
+      "loss": 0.9982,
+      "step": 16316
+    },
+    {
+      "epoch": 0.4481338057180522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2149075324820876e-05,
+      "loss": 0.8994,
+      "step": 16317
+    },
+    {
+      "epoch": 0.44816126994589545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2148206584381969e-05,
+      "loss": 0.951,
+      "step": 16318
+    },
+    {
+      "epoch": 0.4481887341737387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2147337826945625e-05,
+      "loss": 1.0015,
+      "step": 16319
+    },
+    {
+      "epoch": 0.44821619840158194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.214646905251871e-05,
+      "loss": 0.9819,
+      "step": 16320
+    },
+    {
+      "epoch": 0.44824366262942517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2145600261108099e-05,
+      "loss": 0.9491,
+      "step": 16321
+    },
+    {
+      "epoch": 0.4482711268572684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2144731452720671e-05,
+      "loss": 0.9539,
+      "step": 16322
+    },
+    {
+      "epoch": 0.44829859108511166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2143862627363294e-05,
+      "loss": 0.9337,
+      "step": 16323
+    },
+    {
+      "epoch": 0.4483260553129549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2142993785042847e-05,
+      "loss": 0.8864,
+      "step": 16324
+    },
+    {
+      "epoch": 0.4483535195407981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2142124925766204e-05,
+      "loss": 0.9862,
+      "step": 16325
+    },
+    {
+      "epoch": 0.44838098376864133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2141256049540236e-05,
+      "loss": 0.9006,
+      "step": 16326
+    },
+    {
+      "epoch": 0.4484084479964846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.214038715637182e-05,
+      "loss": 0.9903,
+      "step": 16327
+    },
+    {
+      "epoch": 0.4484359122243278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2139518246267837e-05,
+      "loss": 0.9889,
+      "step": 16328
+    },
+    {
+      "epoch": 0.44846337645217105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2138649319235152e-05,
+      "loss": 0.9923,
+      "step": 16329
+    },
+    {
+      "epoch": 0.44849084068001427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2137780375280645e-05,
+      "loss": 0.8587,
+      "step": 16330
+    },
+    {
+      "epoch": 0.4485183049078575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2136911414411194e-05,
+      "loss": 0.9367,
+      "step": 16331
+    },
+    {
+      "epoch": 0.44854576913570077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2136042436633672e-05,
+      "loss": 0.8946,
+      "step": 16332
+    },
+    {
+      "epoch": 0.448573233363544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2135173441954953e-05,
+      "loss": 0.9952,
+      "step": 16333
+    },
+    {
+      "epoch": 0.4486006975913872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2134304430381916e-05,
+      "loss": 1.0253,
+      "step": 16334
+    },
+    {
+      "epoch": 0.44862816181923043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2133435401921434e-05,
+      "loss": 0.9946,
+      "step": 16335
+    },
+    {
+      "epoch": 0.4486556260470737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2132566356580388e-05,
+      "loss": 0.9833,
+      "step": 16336
+    },
+    {
+      "epoch": 0.44868309027491693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.213169729436565e-05,
+      "loss": 0.8883,
+      "step": 16337
+    },
+    {
+      "epoch": 0.44871055450276015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2130828215284094e-05,
+      "loss": 0.9293,
+      "step": 16338
+    },
+    {
+      "epoch": 0.44873801873060337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.21299591193426e-05,
+      "loss": 0.8118,
+      "step": 16339
+    },
+    {
+      "epoch": 0.44876548295844665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.212909000654805e-05,
+      "loss": 1.0325,
+      "step": 16340
+    },
+    {
+      "epoch": 0.44879294718628987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2128220876907311e-05,
+      "loss": 1.0172,
+      "step": 16341
+    },
+    {
+      "epoch": 0.4488204114141331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2127351730427264e-05,
+      "loss": 0.9485,
+      "step": 16342
+    },
+    {
+      "epoch": 0.4488478756419763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2126482567114788e-05,
+      "loss": 0.9584,
+      "step": 16343
+    },
+    {
+      "epoch": 0.44887533986981953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2125613386976758e-05,
+      "loss": 0.8652,
+      "step": 16344
+    },
+    {
+      "epoch": 0.4489028040976628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2124744190020047e-05,
+      "loss": 0.879,
+      "step": 16345
+    },
+    {
+      "epoch": 0.44893026832550603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2123874976251542e-05,
+      "loss": 0.845,
+      "step": 16346
+    },
+    {
+      "epoch": 0.44895773255334925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.212300574567811e-05,
+      "loss": 1.0684,
+      "step": 16347
+    },
+    {
+      "epoch": 0.4489851967811925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.212213649830664e-05,
+      "loss": 0.9957,
+      "step": 16348
+    },
+    {
+      "epoch": 0.44901266100903575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2121267234144003e-05,
+      "loss": 0.9366,
+      "step": 16349
+    },
+    {
+      "epoch": 0.449040125236879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2120397953197075e-05,
+      "loss": 0.9169,
+      "step": 16350
+    },
+    {
+      "epoch": 0.4490675894647222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211952865547274e-05,
+      "loss": 0.986,
+      "step": 16351
+    },
+    {
+      "epoch": 0.4490950536925654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2118659340977874e-05,
+      "loss": 0.974,
+      "step": 16352
+    },
+    {
+      "epoch": 0.4491225179204087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2117790009719353e-05,
+      "loss": 0.9551,
+      "step": 16353
+    },
+    {
+      "epoch": 0.4491499821482519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2116920661704055e-05,
+      "loss": 0.8376,
+      "step": 16354
+    },
+    {
+      "epoch": 0.44917744637609514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2116051296938862e-05,
+      "loss": 1.0543,
+      "step": 16355
+    },
+    {
+      "epoch": 0.44920491060393836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2115181915430655e-05,
+      "loss": 0.9379,
+      "step": 16356
+    },
+    {
+      "epoch": 0.4492323748317816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2114312517186305e-05,
+      "loss": 0.8287,
+      "step": 16357
+    },
+    {
+      "epoch": 0.44925983905962485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2113443102212697e-05,
+      "loss": 0.9691,
+      "step": 16358
+    },
+    {
+      "epoch": 0.4492873032874681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2112573670516705e-05,
+      "loss": 0.969,
+      "step": 16359
+    },
+    {
+      "epoch": 0.4493147675153113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2111704222105216e-05,
+      "loss": 0.8883,
+      "step": 16360
+    },
+    {
+      "epoch": 0.4493422317431545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2110834756985106e-05,
+      "loss": 0.9006,
+      "step": 16361
+    },
+    {
+      "epoch": 0.4493696959709978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210996527516325e-05,
+      "loss": 0.8031,
+      "step": 16362
+    },
+    {
+      "epoch": 0.449397160198841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2109095776646534e-05,
+      "loss": 0.8545,
+      "step": 16363
+    },
+    {
+      "epoch": 0.44942462442668424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2108226261441835e-05,
+      "loss": 0.9714,
+      "step": 16364
+    },
+    {
+      "epoch": 0.44945208865452746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2107356729556032e-05,
+      "loss": 0.956,
+      "step": 16365
+    },
+    {
+      "epoch": 0.44947955288237074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2106487180996007e-05,
+      "loss": 0.9088,
+      "step": 16366
+    },
+    {
+      "epoch": 0.44950701711021396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2105617615768639e-05,
+      "loss": 0.7824,
+      "step": 16367
+    },
+    {
+      "epoch": 0.4495344813380572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210474803388081e-05,
+      "loss": 0.9058,
+      "step": 16368
+    },
+    {
+      "epoch": 0.4495619455659004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2103878435339398e-05,
+      "loss": 0.9728,
+      "step": 16369
+    },
+    {
+      "epoch": 0.4495894097937436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2103008820151285e-05,
+      "loss": 0.8915,
+      "step": 16370
+    },
+    {
+      "epoch": 0.4496168740215869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2102139188323351e-05,
+      "loss": 1.0767,
+      "step": 16371
+    },
+    {
+      "epoch": 0.4496443382494301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210126953986248e-05,
+      "loss": 1.0101,
+      "step": 16372
+    },
+    {
+      "epoch": 0.44967180247727334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210039987477555e-05,
+      "loss": 0.7957,
+      "step": 16373
+    },
+    {
+      "epoch": 0.44969926670511656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.209953019306944e-05,
+      "loss": 0.9746,
+      "step": 16374
+    },
+    {
+      "epoch": 0.44972673093295984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2098660494751035e-05,
+      "loss": 0.8376,
+      "step": 16375
+    },
+    {
+      "epoch": 0.44975419516080306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2097790779827216e-05,
+      "loss": 0.8498,
+      "step": 16376
+    },
+    {
+      "epoch": 0.4497816593886463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2096921048304864e-05,
+      "loss": 0.9826,
+      "step": 16377
+    },
+    {
+      "epoch": 0.4498091236164895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2096051300190859e-05,
+      "loss": 0.8497,
+      "step": 16378
+    },
+    {
+      "epoch": 0.4498365878443328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2095181535492085e-05,
+      "loss": 0.9509,
+      "step": 16379
+    },
+    {
+      "epoch": 0.449864052072176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2094311754215421e-05,
+      "loss": 0.934,
+      "step": 16380
+    },
+    {
+      "epoch": 0.4498915163000192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2093441956367756e-05,
+      "loss": 0.9272,
+      "step": 16381
+    },
+    {
+      "epoch": 0.44991898052786244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2092572141955963e-05,
+      "loss": 0.9025,
+      "step": 16382
+    },
+    {
+      "epoch": 0.44994644475570567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2091702310986929e-05,
+      "loss": 0.9612,
+      "step": 16383
+    },
+    {
+      "epoch": 0.44997390898354894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2090832463467539e-05,
+      "loss": 0.9232,
+      "step": 16384
+    },
+    {
+      "epoch": 0.45000137321139216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.208996259940467e-05,
+      "loss": 0.903,
+      "step": 16385
+    },
+    {
+      "epoch": 0.4500288374392354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2089092718805206e-05,
+      "loss": 0.954,
+      "step": 16386
+    },
+    {
+      "epoch": 0.4500563016670786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2088222821676034e-05,
+      "loss": 0.9962,
+      "step": 16387
+    },
+    {
+      "epoch": 0.4500837658949219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2087352908024035e-05,
+      "loss": 0.9734,
+      "step": 16388
+    },
+    {
+      "epoch": 0.4501112301227651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.208648297785609e-05,
+      "loss": 1.1248,
+      "step": 16389
+    },
+    {
+      "epoch": 0.4501386943506083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2085613031179082e-05,
+      "loss": 0.8839,
+      "step": 16390
+    },
+    {
+      "epoch": 0.45016615857845155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2084743067999895e-05,
+      "loss": 0.8559,
+      "step": 16391
+    },
+    {
+      "epoch": 0.4501936228062948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2083873088325415e-05,
+      "loss": 0.8903,
+      "step": 16392
+    },
+    {
+      "epoch": 0.45022108703413805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2083003092162527e-05,
+      "loss": 0.8856,
+      "step": 16393
+    },
+    {
+      "epoch": 0.45024855126198127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2082133079518106e-05,
+      "loss": 0.898,
+      "step": 16394
+    },
+    {
+      "epoch": 0.4502760154898245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2081263050399044e-05,
+      "loss": 0.8874,
+      "step": 16395
+    },
+    {
+      "epoch": 0.45030347971766777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2080393004812226e-05,
+      "loss": 0.8436,
+      "step": 16396
+    },
+    {
+      "epoch": 0.450330943945511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2079522942764528e-05,
+      "loss": 0.8928,
+      "step": 16397
+    },
+    {
+      "epoch": 0.4503584081733542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2078652864262838e-05,
+      "loss": 0.7699,
+      "step": 16398
+    },
+    {
+      "epoch": 0.45038587240119743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2077782769314045e-05,
+      "loss": 0.905,
+      "step": 16399
+    },
+    {
+      "epoch": 0.45041333662904065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2076912657925028e-05,
+      "loss": 0.9292,
+      "step": 16400
+    },
+    {
+      "epoch": 0.4504408008568839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2076042530102674e-05,
+      "loss": 0.857,
+      "step": 16401
+    },
+    {
+      "epoch": 0.45046826508472715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2075172385853868e-05,
+      "loss": 0.7526,
+      "step": 16402
+    },
+    {
+      "epoch": 0.45049572931257037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2074302225185492e-05,
+      "loss": 0.8714,
+      "step": 16403
+    },
+    {
+      "epoch": 0.4505231935404136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2073432048104437e-05,
+      "loss": 0.914,
+      "step": 16404
+    },
+    {
+      "epoch": 0.45055065776825687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2072561854617582e-05,
+      "loss": 0.9294,
+      "step": 16405
+    },
+    {
+      "epoch": 0.4505781219961001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2071691644731815e-05,
+      "loss": 0.9636,
+      "step": 16406
+    },
+    {
+      "epoch": 0.4506055862239433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2070821418454019e-05,
+      "loss": 0.9055,
+      "step": 16407
+    },
+    {
+      "epoch": 0.45063305045178653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2069951175791087e-05,
+      "loss": 0.9716,
+      "step": 16408
+    },
+    {
+      "epoch": 0.4506605146796298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2069080916749896e-05,
+      "loss": 0.871,
+      "step": 16409
+    },
+    {
+      "epoch": 0.45068797890747303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2068210641337335e-05,
+      "loss": 0.8895,
+      "step": 16410
+    },
+    {
+      "epoch": 0.45071544313531625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2067340349560292e-05,
+      "loss": 0.9075,
+      "step": 16411
+    },
+    {
+      "epoch": 0.4507429073631595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206647004142565e-05,
+      "loss": 0.9323,
+      "step": 16412
+    },
+    {
+      "epoch": 0.4507703715910027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2065599716940297e-05,
+      "loss": 0.8843,
+      "step": 16413
+    },
+    {
+      "epoch": 0.45079783581884597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206472937611112e-05,
+      "loss": 0.9867,
+      "step": 16414
+    },
+    {
+      "epoch": 0.4508253000466892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2063859018945e-05,
+      "loss": 0.8773,
+      "step": 16415
+    },
+    {
+      "epoch": 0.4508527642745324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2062988645448835e-05,
+      "loss": 0.9072,
+      "step": 16416
+    },
+    {
+      "epoch": 0.45088022850237564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2062118255629502e-05,
+      "loss": 0.9289,
+      "step": 16417
+    },
+    {
+      "epoch": 0.4509076927302189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2061247849493887e-05,
+      "loss": 0.8851,
+      "step": 16418
+    },
+    {
+      "epoch": 0.45093515695806213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2060377427048882e-05,
+      "loss": 0.8889,
+      "step": 16419
+    },
+    {
+      "epoch": 0.45096262118590535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2059506988301377e-05,
+      "loss": 0.8952,
+      "step": 16420
+    },
+    {
+      "epoch": 0.4509900854137486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.205863653325825e-05,
+      "loss": 0.8576,
+      "step": 16421
+    },
+    {
+      "epoch": 0.45101754964159185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2057766061926394e-05,
+      "loss": 0.8999,
+      "step": 16422
+    },
+    {
+      "epoch": 0.4510450138694351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.20568955743127e-05,
+      "loss": 1.039,
+      "step": 16423
+    },
+    {
+      "epoch": 0.4510724780972783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2056025070424048e-05,
+      "loss": 0.8427,
+      "step": 16424
+    },
+    {
+      "epoch": 0.4510999423251215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.205515455026733e-05,
+      "loss": 1.0288,
+      "step": 16425
+    },
+    {
+      "epoch": 0.45112740655296474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2054284013849431e-05,
+      "loss": 0.888,
+      "step": 16426
+    },
+    {
+      "epoch": 0.451154870780808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2053413461177243e-05,
+      "loss": 0.9365,
+      "step": 16427
+    },
+    {
+      "epoch": 0.45118233500865124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2052542892257653e-05,
+      "loss": 0.9993,
+      "step": 16428
+    },
+    {
+      "epoch": 0.45120979923649446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2051672307097548e-05,
+      "loss": 0.9796,
+      "step": 16429
+    },
+    {
+      "epoch": 0.4512372634643377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2050801705703817e-05,
+      "loss": 0.9841,
+      "step": 16430
+    },
+    {
+      "epoch": 0.45126472769218096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2049931088083344e-05,
+      "loss": 0.9932,
+      "step": 16431
+    },
+    {
+      "epoch": 0.4512921919200242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2049060454243031e-05,
+      "loss": 0.8671,
+      "step": 16432
+    },
+    {
+      "epoch": 0.4513196561478674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2048189804189752e-05,
+      "loss": 0.9311,
+      "step": 16433
+    },
+    {
+      "epoch": 0.4513471203757106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2047319137930404e-05,
+      "loss": 0.9738,
+      "step": 16434
+    },
+    {
+      "epoch": 0.4513745846035539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.204644845547187e-05,
+      "loss": 0.8194,
+      "step": 16435
+    },
+    {
+      "epoch": 0.4514020488313971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2045577756821047e-05,
+      "loss": 0.903,
+      "step": 16436
+    },
+    {
+      "epoch": 0.45142951305924034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2044707041984822e-05,
+      "loss": 0.8744,
+      "step": 16437
+    },
+    {
+      "epoch": 0.45145697728708356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2043836310970078e-05,
+      "loss": 0.8241,
+      "step": 16438
+    },
+    {
+      "epoch": 0.4514844415149268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2042965563783712e-05,
+      "loss": 0.9018,
+      "step": 16439
+    },
+    {
+      "epoch": 0.45151190574277006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2042094800432613e-05,
+      "loss": 0.9138,
+      "step": 16440
+    },
+    {
+      "epoch": 0.4515393699706133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2041224020923665e-05,
+      "loss": 0.9648,
+      "step": 16441
+    },
+    {
+      "epoch": 0.4515668341984565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2040353225263763e-05,
+      "loss": 0.8834,
+      "step": 16442
+    },
+    {
+      "epoch": 0.4515942984262997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2039482413459798e-05,
+      "loss": 0.9159,
+      "step": 16443
+    },
+    {
+      "epoch": 0.451621762654143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2038611585518659e-05,
+      "loss": 0.7799,
+      "step": 16444
+    },
+    {
+      "epoch": 0.4516492268819862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203774074144723e-05,
+      "loss": 0.9337,
+      "step": 16445
+    },
+    {
+      "epoch": 0.45167669110982944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2036869881252411e-05,
+      "loss": 0.98,
+      "step": 16446
+    },
+    {
+      "epoch": 0.45170415533767266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2035999004941087e-05,
+      "loss": 0.8323,
+      "step": 16447
+    },
+    {
+      "epoch": 0.45173161956551594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203512811252015e-05,
+      "loss": 1.0031,
+      "step": 16448
+    },
+    {
+      "epoch": 0.45175908379335916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2034257203996493e-05,
+      "loss": 0.9584,
+      "step": 16449
+    },
+    {
+      "epoch": 0.4517865480212024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2033386279377005e-05,
+      "loss": 0.8495,
+      "step": 16450
+    },
+    {
+      "epoch": 0.4518140122490456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2032515338668573e-05,
+      "loss": 0.822,
+      "step": 16451
+    },
+    {
+      "epoch": 0.4518414764768888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2031644381878098e-05,
+      "loss": 0.8509,
+      "step": 16452
+    },
+    {
+      "epoch": 0.4518689407047321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2030773409012462e-05,
+      "loss": 0.9231,
+      "step": 16453
+    },
+    {
+      "epoch": 0.4518964049325753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202990242007856e-05,
+      "loss": 0.931,
+      "step": 16454
+    },
+    {
+      "epoch": 0.45192386916041855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2029031415083286e-05,
+      "loss": 0.9267,
+      "step": 16455
+    },
+    {
+      "epoch": 0.45195133338826177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202816039403353e-05,
+      "loss": 0.8689,
+      "step": 16456
+    },
+    {
+      "epoch": 0.45197879761610504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2027289356936178e-05,
+      "loss": 0.8212,
+      "step": 16457
+    },
+    {
+      "epoch": 0.45200626184394826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2026418303798133e-05,
+      "loss": 0.9065,
+      "step": 16458
+    },
+    {
+      "epoch": 0.4520337260717915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2025547234626279e-05,
+      "loss": 0.862,
+      "step": 16459
+    },
+    {
+      "epoch": 0.4520611902996347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202467614942751e-05,
+      "loss": 0.9462,
+      "step": 16460
+    },
+    {
+      "epoch": 0.452088654527478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2023805048208721e-05,
+      "loss": 0.9719,
+      "step": 16461
+    },
+    {
+      "epoch": 0.4521161187553212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.20229339309768e-05,
+      "loss": 0.9839,
+      "step": 16462
+    },
+    {
+      "epoch": 0.4521435829831644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2022062797738642e-05,
+      "loss": 0.8804,
+      "step": 16463
+    },
+    {
+      "epoch": 0.45217104721100765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2021191648501147e-05,
+      "loss": 0.9315,
+      "step": 16464
+    },
+    {
+      "epoch": 0.45219851143885087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2020320483271193e-05,
+      "loss": 1.0624,
+      "step": 16465
+    },
+    {
+      "epoch": 0.45222597566669415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.201944930205568e-05,
+      "loss": 0.8303,
+      "step": 16466
+    },
+    {
+      "epoch": 0.45225343989453737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2018578104861507e-05,
+      "loss": 0.8955,
+      "step": 16467
+    },
+    {
+      "epoch": 0.4522809041223806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2017706891695561e-05,
+      "loss": 0.8748,
+      "step": 16468
+    },
+    {
+      "epoch": 0.4523083683502238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2016835662564734e-05,
+      "loss": 0.9072,
+      "step": 16469
+    },
+    {
+      "epoch": 0.4523358325780671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2015964417475926e-05,
+      "loss": 0.8309,
+      "step": 16470
+    },
+    {
+      "epoch": 0.4523632968059103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2015093156436022e-05,
+      "loss": 0.8203,
+      "step": 16471
+    },
+    {
+      "epoch": 0.45239076103375353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2014221879451924e-05,
+      "loss": 0.9623,
+      "step": 16472
+    },
+    {
+      "epoch": 0.45241822526159675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2013350586530521e-05,
+      "loss": 0.984,
+      "step": 16473
+    },
+    {
+      "epoch": 0.45244568948944003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2012479277678706e-05,
+      "loss": 0.9611,
+      "step": 16474
+    },
+    {
+      "epoch": 0.45247315371728325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2011607952903376e-05,
+      "loss": 0.7738,
+      "step": 16475
+    },
+    {
+      "epoch": 0.45250061794512647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2010736612211428e-05,
+      "loss": 0.9054,
+      "step": 16476
+    },
+    {
+      "epoch": 0.4525280821729697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2009865255609752e-05,
+      "loss": 0.9698,
+      "step": 16477
+    },
+    {
+      "epoch": 0.4525555464008129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.200899388310524e-05,
+      "loss": 0.8591,
+      "step": 16478
+    },
+    {
+      "epoch": 0.4525830106286562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2008122494704795e-05,
+      "loss": 0.8626,
+      "step": 16479
+    },
+    {
+      "epoch": 0.4526104748564994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2007251090415304e-05,
+      "loss": 0.9626,
+      "step": 16480
+    },
+    {
+      "epoch": 0.45263793908434263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2006379670243664e-05,
+      "loss": 0.8042,
+      "step": 16481
+    },
+    {
+      "epoch": 0.45266540331218585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2005508234196774e-05,
+      "loss": 0.9159,
+      "step": 16482
+    },
+    {
+      "epoch": 0.45269286754002913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2004636782281522e-05,
+      "loss": 0.8896,
+      "step": 16483
+    },
+    {
+      "epoch": 0.45272033176787235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2003765314504812e-05,
+      "loss": 0.9509,
+      "step": 16484
+    },
+    {
+      "epoch": 0.4527477959957156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2002893830873531e-05,
+      "loss": 0.8899,
+      "step": 16485
+    },
+    {
+      "epoch": 0.4527752602235588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2002022331394581e-05,
+      "loss": 0.9557,
+      "step": 16486
+    },
+    {
+      "epoch": 0.45280272445140207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2001150816074851e-05,
+      "loss": 1.0066,
+      "step": 16487
+    },
+    {
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2000279284921247e-05,
+      "loss": 0.91,
+      "step": 16488
+    },
+    {
+      "epoch": 0.4528576529070885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1999407737940652e-05,
+      "loss": 0.8134,
+      "step": 16489
+    },
+    {
+      "epoch": 0.45288511713493174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199853617513997e-05,
+      "loss": 0.9648,
+      "step": 16490
+    },
+    {
+      "epoch": 0.452912581362775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1997664596526097e-05,
+      "loss": 1.0048,
+      "step": 16491
+    },
+    {
+      "epoch": 0.45294004559061823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1996793002105928e-05,
+      "loss": 0.9418,
+      "step": 16492
+    },
+    {
+      "epoch": 0.45296750981846146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199592139188636e-05,
+      "loss": 0.9315,
+      "step": 16493
+    },
+    {
+      "epoch": 0.4529949740463047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1995049765874284e-05,
+      "loss": 0.9611,
+      "step": 16494
+    },
+    {
+      "epoch": 0.4530224382741479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1994178124076606e-05,
+      "loss": 0.919,
+      "step": 16495
+    },
+    {
+      "epoch": 0.4530499025019912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1993306466500218e-05,
+      "loss": 0.9361,
+      "step": 16496
+    },
+    {
+      "epoch": 0.4530773667298344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1992434793152016e-05,
+      "loss": 0.8657,
+      "step": 16497
+    },
+    {
+      "epoch": 0.4531048309576776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1991563104038896e-05,
+      "loss": 0.8828,
+      "step": 16498
+    },
+    {
+      "epoch": 0.45313229518552084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1990691399167762e-05,
+      "loss": 0.9437,
+      "step": 16499
+    },
+    {
+      "epoch": 0.4531597594133641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1989819678545504e-05,
+      "loss": 1.0597,
+      "step": 16500
+    },
+    {
+      "epoch": 0.45318722364120734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1988947942179021e-05,
+      "loss": 0.8374,
+      "step": 16501
+    },
+    {
+      "epoch": 0.45321468786905056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1988076190075212e-05,
+      "loss": 1.0021,
+      "step": 16502
+    },
+    {
+      "epoch": 0.4532421520968938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1987204422240974e-05,
+      "loss": 0.9915,
+      "step": 16503
+    },
+    {
+      "epoch": 0.45326961632473706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1986332638683207e-05,
+      "loss": 0.9196,
+      "step": 16504
+    },
+    {
+      "epoch": 0.4532970805525803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1985460839408807e-05,
+      "loss": 1.0084,
+      "step": 16505
+    },
+    {
+      "epoch": 0.4533245447804235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1984589024424666e-05,
+      "loss": 0.8772,
+      "step": 16506
+    },
+    {
+      "epoch": 0.4533520090082667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1983717193737692e-05,
+      "loss": 0.8322,
+      "step": 16507
+    },
+    {
+      "epoch": 0.45337947323610994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.198284534735478e-05,
+      "loss": 1.0103,
+      "step": 16508
+    },
+    {
+      "epoch": 0.4534069374639532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1981973485282827e-05,
+      "loss": 0.873,
+      "step": 16509
+    },
+    {
+      "epoch": 0.45343440169179644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.198110160752873e-05,
+      "loss": 1.0516,
+      "step": 16510
+    },
+    {
+      "epoch": 0.45346186591963966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1980229714099392e-05,
+      "loss": 0.8918,
+      "step": 16511
+    },
+    {
+      "epoch": 0.4534893301474829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197935780500171e-05,
+      "loss": 0.9894,
+      "step": 16512
+    },
+    {
+      "epoch": 0.45351679437532616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1978485880242578e-05,
+      "loss": 0.8609,
+      "step": 16513
+    },
+    {
+      "epoch": 0.4535442586031694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1977613939828904e-05,
+      "loss": 0.9112,
+      "step": 16514
+    },
+    {
+      "epoch": 0.4535717228310126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1976741983767579e-05,
+      "loss": 0.9565,
+      "step": 16515
+    },
+    {
+      "epoch": 0.4535991870588558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197587001206551e-05,
+      "loss": 0.8817,
+      "step": 16516
+    },
+    {
+      "epoch": 0.4536266512866991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1974998024729589e-05,
+      "loss": 0.946,
+      "step": 16517
+    },
+    {
+      "epoch": 0.4536541155145423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1974126021766716e-05,
+      "loss": 0.9582,
+      "step": 16518
+    },
+    {
+      "epoch": 0.45368157974238554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1973254003183795e-05,
+      "loss": 0.9104,
+      "step": 16519
+    },
+    {
+      "epoch": 0.45370904397022876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197238196898773e-05,
+      "loss": 0.8902,
+      "step": 16520
+    },
+    {
+      "epoch": 0.453736508198072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197150991918541e-05,
+      "loss": 0.9018,
+      "step": 16521
+    },
+    {
+      "epoch": 0.45376397242591526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1970637853783737e-05,
+      "loss": 0.8069,
+      "step": 16522
+    },
+    {
+      "epoch": 0.4537914366537585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196976577278962e-05,
+      "loss": 0.9303,
+      "step": 16523
+    },
+    {
+      "epoch": 0.4538189008816017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196889367620995e-05,
+      "loss": 0.9151,
+      "step": 16524
+    },
+    {
+      "epoch": 0.4538463651094449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1968021564051629e-05,
+      "loss": 1.0192,
+      "step": 16525
+    },
+    {
+      "epoch": 0.4538738293372882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1967149436321562e-05,
+      "loss": 0.9344,
+      "step": 16526
+    },
+    {
+      "epoch": 0.4539012935651314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1966277293026643e-05,
+      "loss": 0.9315,
+      "step": 16527
+    },
+    {
+      "epoch": 0.45392875779297465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1965405134173779e-05,
+      "loss": 0.8972,
+      "step": 16528
+    },
+    {
+      "epoch": 0.45395622202081787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1964532959769868e-05,
+      "loss": 0.7886,
+      "step": 16529
+    },
+    {
+      "epoch": 0.45398368624866114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196366076982181e-05,
+      "loss": 0.9828,
+      "step": 16530
+    },
+    {
+      "epoch": 0.45401115047650437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1962788564336506e-05,
+      "loss": 0.917,
+      "step": 16531
+    },
+    {
+      "epoch": 0.4540386147043476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1961916343320865e-05,
+      "loss": 0.9035,
+      "step": 16532
+    },
+    {
+      "epoch": 0.4540660789321908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196104410678178e-05,
+      "loss": 0.8529,
+      "step": 16533
+    },
+    {
+      "epoch": 0.45409354316003403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1960171854726148e-05,
+      "loss": 0.8736,
+      "step": 16534
+    },
+    {
+      "epoch": 0.4541210073878773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1959299587160882e-05,
+      "loss": 0.9659,
+      "step": 16535
+    },
+    {
+      "epoch": 0.45414847161572053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.195842730409288e-05,
+      "loss": 0.9161,
+      "step": 16536
+    },
+    {
+      "epoch": 0.45417593584356375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1957555005529037e-05,
+      "loss": 1.0103,
+      "step": 16537
+    },
+    {
+      "epoch": 0.45420340007140697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1956682691476267e-05,
+      "loss": 0.9365,
+      "step": 16538
+    },
+    {
+      "epoch": 0.45423086429925025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1955810361941462e-05,
+      "loss": 0.86,
+      "step": 16539
+    },
+    {
+      "epoch": 0.45425832852709347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1954938016931528e-05,
+      "loss": 0.8403,
+      "step": 16540
+    },
+    {
+      "epoch": 0.4542857927549367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1954065656453369e-05,
+      "loss": 0.9539,
+      "step": 16541
+    },
+    {
+      "epoch": 0.4543132569827799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1953193280513883e-05,
+      "loss": 0.9944,
+      "step": 16542
+    },
+    {
+      "epoch": 0.4543407212106232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1952320889119975e-05,
+      "loss": 0.8825,
+      "step": 16543
+    },
+    {
+      "epoch": 0.4543681854384664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1951448482278554e-05,
+      "loss": 0.921,
+      "step": 16544
+    },
+    {
+      "epoch": 0.45439564966630963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.195057605999651e-05,
+      "loss": 0.9131,
+      "step": 16545
+    },
+    {
+      "epoch": 0.45442311389415285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1949703622280752e-05,
+      "loss": 0.912,
+      "step": 16546
+    },
+    {
+      "epoch": 0.4544505781219961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1948831169138188e-05,
+      "loss": 0.9536,
+      "step": 16547
+    },
+    {
+      "epoch": 0.45447804234983935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1947958700575718e-05,
+      "loss": 0.8717,
+      "step": 16548
+    },
+    {
+      "epoch": 0.45450550657768257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.194708621660024e-05,
+      "loss": 0.9318,
+      "step": 16549
+    },
+    {
+      "epoch": 0.4545329708055258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1946213717218661e-05,
+      "loss": 0.9575,
+      "step": 16550
+    },
+    {
+      "epoch": 0.454560435033369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1945341202437887e-05,
+      "loss": 0.8288,
+      "step": 16551
+    },
+    {
+      "epoch": 0.4545878992612123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.194446867226482e-05,
+      "loss": 1.046,
+      "step": 16552
+    },
+    {
+      "epoch": 0.4546153634890555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1943596126706365e-05,
+      "loss": 0.9098,
+      "step": 16553
+    },
+    {
+      "epoch": 0.45464282771689873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1942723565769423e-05,
+      "loss": 0.8819,
+      "step": 16554
+    },
+    {
+      "epoch": 0.45467029194474196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1941850989460899e-05,
+      "loss": 1.0187,
+      "step": 16555
+    },
+    {
+      "epoch": 0.45469775617258523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1940978397787702e-05,
+      "loss": 0.8929,
+      "step": 16556
+    },
+    {
+      "epoch": 0.45472522040042845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1940105790756727e-05,
+      "loss": 0.9447,
+      "step": 16557
+    },
+    {
+      "epoch": 0.4547526846282717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1939233168374884e-05,
+      "loss": 0.8889,
+      "step": 16558
+    },
+    {
+      "epoch": 0.4547801488561149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1938360530649078e-05,
+      "loss": 0.8465,
+      "step": 16559
+    },
+    {
+      "epoch": 0.4548076130839581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1937487877586213e-05,
+      "loss": 0.9497,
+      "step": 16560
+    },
+    {
+      "epoch": 0.4548350773118014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1936615209193194e-05,
+      "loss": 0.9897,
+      "step": 16561
+    },
+    {
+      "epoch": 0.4548625415396446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1935742525476923e-05,
+      "loss": 0.8196,
+      "step": 16562
+    },
+    {
+      "epoch": 0.45489000576748784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1934869826444309e-05,
+      "loss": 0.8168,
+      "step": 16563
+    },
+    {
+      "epoch": 0.45491746999533106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1933997112102256e-05,
+      "loss": 0.8489,
+      "step": 16564
+    },
+    {
+      "epoch": 0.45494493422317434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1933124382457668e-05,
+      "loss": 0.9559,
+      "step": 16565
+    },
+    {
+      "epoch": 0.45497239845101756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1932251637517449e-05,
+      "loss": 0.9249,
+      "step": 16566
+    },
+    {
+      "epoch": 0.4549998626788608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.193137887728851e-05,
+      "loss": 1.01,
+      "step": 16567
+    },
+    {
+      "epoch": 0.455027326906704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1930506101777753e-05,
+      "loss": 0.8575,
+      "step": 16568
+    },
+    {
+      "epoch": 0.4550547911345473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1929633310992082e-05,
+      "loss": 0.9657,
+      "step": 16569
+    },
+    {
+      "epoch": 0.4550822553623905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1928760504938406e-05,
+      "loss": 0.9024,
+      "step": 16570
+    },
+    {
+      "epoch": 0.4551097195902337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1927887683623626e-05,
+      "loss": 0.8851,
+      "step": 16571
+    },
+    {
+      "epoch": 0.45513718381807694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1927014847054656e-05,
+      "loss": 0.9678,
+      "step": 16572
+    },
+    {
+      "epoch": 0.45516464804592016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1926141995238398e-05,
+      "loss": 0.9774,
+      "step": 16573
+    },
+    {
+      "epoch": 0.45519211227376344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1925269128181758e-05,
+      "loss": 0.9407,
+      "step": 16574
+    },
+    {
+      "epoch": 0.45521957650160666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1924396245891641e-05,
+      "loss": 0.8903,
+      "step": 16575
+    },
+    {
+      "epoch": 0.4552470407294499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.192352334837496e-05,
+      "loss": 0.9704,
+      "step": 16576
+    },
+    {
+      "epoch": 0.4552745049572931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1922650435638611e-05,
+      "loss": 0.9141,
+      "step": 16577
+    },
+    {
+      "epoch": 0.4553019691851364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1921777507689509e-05,
+      "loss": 0.9767,
+      "step": 16578
+    },
+    {
+      "epoch": 0.4553294334129796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1920904564534561e-05,
+      "loss": 0.8393,
+      "step": 16579
+    },
+    {
+      "epoch": 0.4553568976408228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1920031606180671e-05,
+      "loss": 0.7746,
+      "step": 16580
+    },
+    {
+      "epoch": 0.45538436186866604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1919158632634747e-05,
+      "loss": 1.002,
+      "step": 16581
+    },
+    {
+      "epoch": 0.4554118260965093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1918285643903697e-05,
+      "loss": 0.9572,
+      "step": 16582
+    },
+    {
+      "epoch": 0.45543929032435254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1917412639994425e-05,
+      "loss": 1.0022,
+      "step": 16583
+    },
+    {
+      "epoch": 0.45546675455219576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1916539620913848e-05,
+      "loss": 0.82,
+      "step": 16584
+    },
+    {
+      "epoch": 0.455494218780039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1915666586668862e-05,
+      "loss": 0.9643,
+      "step": 16585
+    },
+    {
+      "epoch": 0.45552168300788226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.191479353726638e-05,
+      "loss": 0.8676,
+      "step": 16586
+    },
+    {
+      "epoch": 0.4555491472357255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.191392047271331e-05,
+      "loss": 0.9453,
+      "step": 16587
+    },
+    {
+      "epoch": 0.4555766114635687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1913047393016563e-05,
+      "loss": 0.9474,
+      "step": 16588
+    },
+    {
+      "epoch": 0.4556040756914119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1912174298183041e-05,
+      "loss": 0.9098,
+      "step": 16589
+    },
+    {
+      "epoch": 0.45563153991925515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1911301188219655e-05,
+      "loss": 0.986,
+      "step": 16590
+    },
+    {
+      "epoch": 0.4556590041470984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1910428063133315e-05,
+      "loss": 0.8772,
+      "step": 16591
+    },
+    {
+      "epoch": 0.45568646837494164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1909554922930927e-05,
+      "loss": 0.9237,
+      "step": 16592
+    },
+    {
+      "epoch": 0.45571393260278487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1908681767619402e-05,
+      "loss": 0.8705,
+      "step": 16593
+    },
+    {
+      "epoch": 0.4557413968306281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1907808597205648e-05,
+      "loss": 0.891,
+      "step": 16594
+    },
+    {
+      "epoch": 0.45576886105847136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.190693541169657e-05,
+      "loss": 0.9053,
+      "step": 16595
+    },
+    {
+      "epoch": 0.4557963252863146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1906062211099083e-05,
+      "loss": 1.0081,
+      "step": 16596
+    },
+    {
+      "epoch": 0.4558237895141578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1905188995420094e-05,
+      "loss": 0.9462,
+      "step": 16597
+    },
+    {
+      "epoch": 0.45585125374200103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.190431576466651e-05,
+      "loss": 0.894,
+      "step": 16598
+    },
+    {
+      "epoch": 0.4558787179698443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1903442518845241e-05,
+      "loss": 0.9324,
+      "step": 16599
+    },
+    {
+      "epoch": 0.4559061821976875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1902569257963203e-05,
+      "loss": 0.7555,
+      "step": 16600
+    },
+    {
+      "epoch": 0.45593364642553075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1901695982027295e-05,
+      "loss": 0.8611,
+      "step": 16601
+    },
+    {
+      "epoch": 0.45596111065337397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1900822691044432e-05,
+      "loss": 0.9062,
+      "step": 16602
+    },
+    {
+      "epoch": 0.4559885748812172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1899949385021525e-05,
+      "loss": 0.8177,
+      "step": 16603
+    },
+    {
+      "epoch": 0.45601603910906047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1899076063965482e-05,
+      "loss": 0.9518,
+      "step": 16604
+    },
+    {
+      "epoch": 0.4560435033369037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1898202727883212e-05,
+      "loss": 0.8918,
+      "step": 16605
+    },
+    {
+      "epoch": 0.4560709675647469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189732937678163e-05,
+      "loss": 0.8416,
+      "step": 16606
+    },
+    {
+      "epoch": 0.45609843179259013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1896456010667641e-05,
+      "loss": 0.8546,
+      "step": 16607
+    },
+    {
+      "epoch": 0.4561258960204334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1895582629548157e-05,
+      "loss": 0.8555,
+      "step": 16608
+    },
+    {
+      "epoch": 0.45615336024827663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1894709233430091e-05,
+      "loss": 0.966,
+      "step": 16609
+    },
+    {
+      "epoch": 0.45618082447611985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189383582232035e-05,
+      "loss": 0.819,
+      "step": 16610
+    },
+    {
+      "epoch": 0.45620828870396307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1892962396225845e-05,
+      "loss": 0.9001,
+      "step": 16611
+    },
+    {
+      "epoch": 0.45623575293180635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1892088955153494e-05,
+      "loss": 0.9258,
+      "step": 16612
+    },
+    {
+      "epoch": 0.45626321715964957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1891215499110197e-05,
+      "loss": 0.8844,
+      "step": 16613
+    },
+    {
+      "epoch": 0.4562906813874928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189034202810287e-05,
+      "loss": 1.032,
+      "step": 16614
+    },
+    {
+      "epoch": 0.456318145615336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1889468542138425e-05,
+      "loss": 0.9689,
+      "step": 16615
+    },
+    {
+      "epoch": 0.45634560984317923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1888595041223777e-05,
+      "loss": 0.9463,
+      "step": 16616
+    },
+    {
+      "epoch": 0.4563730740710225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1887721525365832e-05,
+      "loss": 0.8945,
+      "step": 16617
+    },
+    {
+      "epoch": 0.45640053829886573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.18868479945715e-05,
+      "loss": 0.8424,
+      "step": 16618
+    },
+    {
+      "epoch": 0.45642800252670895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1885974448847695e-05,
+      "loss": 0.9056,
+      "step": 16619
+    },
+    {
+      "epoch": 0.4564554667545522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1885100888201333e-05,
+      "loss": 0.9475,
+      "step": 16620
+    },
+    {
+      "epoch": 0.45648293098239545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1884227312639323e-05,
+      "loss": 0.9929,
+      "step": 16621
+    },
+    {
+      "epoch": 0.4565103952102387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1883353722168573e-05,
+      "loss": 0.9046,
+      "step": 16622
+    },
+    {
+      "epoch": 0.4565378594380819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1882480116796003e-05,
+      "loss": 0.9718,
+      "step": 16623
+    },
+    {
+      "epoch": 0.4565653236659251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.188160649652852e-05,
+      "loss": 0.9233,
+      "step": 16624
+    },
+    {
+      "epoch": 0.4565927878937684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1880732861373034e-05,
+      "loss": 0.7908,
+      "step": 16625
+    },
+    {
+      "epoch": 0.4566202521216116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1879859211336466e-05,
+      "loss": 0.8814,
+      "step": 16626
+    },
+    {
+      "epoch": 0.45664771634945484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1878985546425721e-05,
+      "loss": 0.9484,
+      "step": 16627
+    },
+    {
+      "epoch": 0.45667518057729806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1878111866647715e-05,
+      "loss": 0.8954,
+      "step": 16628
+    },
+    {
+      "epoch": 0.4567026448051413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1877238172009362e-05,
+      "loss": 0.8956,
+      "step": 16629
+    },
+    {
+      "epoch": 0.45673010903298455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1876364462517569e-05,
+      "loss": 0.8408,
+      "step": 16630
+    },
+    {
+      "epoch": 0.4567575732608278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1875490738179257e-05,
+      "loss": 0.8249,
+      "step": 16631
+    },
+    {
+      "epoch": 0.456785037488671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1874616999001336e-05,
+      "loss": 0.9525,
+      "step": 16632
+    },
+    {
+      "epoch": 0.4568125017165142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1873743244990718e-05,
+      "loss": 0.9034,
+      "step": 16633
+    },
+    {
+      "epoch": 0.4568399659443575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1872869476154318e-05,
+      "loss": 1.009,
+      "step": 16634
+    },
+    {
+      "epoch": 0.4568674301722007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1871995692499049e-05,
+      "loss": 0.967,
+      "step": 16635
+    },
+    {
+      "epoch": 0.45689489440004394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1871121894031825e-05,
+      "loss": 0.8658,
+      "step": 16636
+    },
+    {
+      "epoch": 0.45692235862788716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1870248080759562e-05,
+      "loss": 0.9429,
+      "step": 16637
+    },
+    {
+      "epoch": 0.45694982285573044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1869374252689167e-05,
+      "loss": 0.9046,
+      "step": 16638
+    },
+    {
+      "epoch": 0.45697728708357366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186850040982756e-05,
+      "loss": 0.8952,
+      "step": 16639
+    },
+    {
+      "epoch": 0.4570047513114169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1867626552181657e-05,
+      "loss": 0.8943,
+      "step": 16640
+    },
+    {
+      "epoch": 0.4570322155392601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1866752679758369e-05,
+      "loss": 0.9192,
+      "step": 16641
+    },
+    {
+      "epoch": 0.4570596797671033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186587879256461e-05,
+      "loss": 0.8774,
+      "step": 16642
+    },
+    {
+      "epoch": 0.4570871439949466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1865004890607292e-05,
+      "loss": 0.8615,
+      "step": 16643
+    },
+    {
+      "epoch": 0.4571146082227898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1864130973893335e-05,
+      "loss": 0.8933,
+      "step": 16644
+    },
+    {
+      "epoch": 0.45714207245063304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1863257042429655e-05,
+      "loss": 0.8675,
+      "step": 16645
+    },
+    {
+      "epoch": 0.45716953667847626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1862383096223158e-05,
+      "loss": 0.9198,
+      "step": 16646
+    },
+    {
+      "epoch": 0.45719700090631954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1861509135280768e-05,
+      "loss": 0.8449,
+      "step": 16647
+    },
+    {
+      "epoch": 0.45722446513416276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1860635159609397e-05,
+      "loss": 0.9974,
+      "step": 16648
+    },
+    {
+      "epoch": 0.457251929362006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1859761169215957e-05,
+      "loss": 0.8746,
+      "step": 16649
+    },
+    {
+      "epoch": 0.4572793935898492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1858887164107369e-05,
+      "loss": 0.9196,
+      "step": 16650
+    },
+    {
+      "epoch": 0.4573068578176925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1858013144290543e-05,
+      "loss": 0.924,
+      "step": 16651
+    },
+    {
+      "epoch": 0.4573343220455357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1857139109772399e-05,
+      "loss": 0.9459,
+      "step": 16652
+    },
+    {
+      "epoch": 0.4573617862733789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1856265060559853e-05,
+      "loss": 0.8686,
+      "step": 16653
+    },
+    {
+      "epoch": 0.45738925050122214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1855390996659814e-05,
+      "loss": 0.9028,
+      "step": 16654
+    },
+    {
+      "epoch": 0.45741671472906537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1854516918079205e-05,
+      "loss": 0.876,
+      "step": 16655
+    },
+    {
+      "epoch": 0.45744417895690864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1853642824824942e-05,
+      "loss": 0.9227,
+      "step": 16656
+    },
+    {
+      "epoch": 0.45747164318475186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1852768716903938e-05,
+      "loss": 0.888,
+      "step": 16657
+    },
+    {
+      "epoch": 0.4574991074125951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1851894594323106e-05,
+      "loss": 0.8141,
+      "step": 16658
+    },
+    {
+      "epoch": 0.4575265716404383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.185102045708937e-05,
+      "loss": 0.8379,
+      "step": 16659
+    },
+    {
+      "epoch": 0.4575540358682816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1850146305209645e-05,
+      "loss": 0.8381,
+      "step": 16660
+    },
+    {
+      "epoch": 0.4575815000961248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1849272138690844e-05,
+      "loss": 0.7999,
+      "step": 16661
+    },
+    {
+      "epoch": 0.457608964323968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1848397957539886e-05,
+      "loss": 0.9712,
+      "step": 16662
+    },
+    {
+      "epoch": 0.45763642855181125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1847523761763685e-05,
+      "loss": 1.0015,
+      "step": 16663
+    },
+    {
+      "epoch": 0.4576638927796545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1846649551369163e-05,
+      "loss": 0.9353,
+      "step": 16664
+    },
+    {
+      "epoch": 0.45769135700749775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1845775326363234e-05,
+      "loss": 0.9234,
+      "step": 16665
+    },
+    {
+      "epoch": 0.45771882123534097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1844901086752813e-05,
+      "loss": 0.9479,
+      "step": 16666
+    },
+    {
+      "epoch": 0.4577462854631842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184402683254482e-05,
+      "loss": 0.9128,
+      "step": 16667
+    },
+    {
+      "epoch": 0.4577737496910274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1843152563746179e-05,
+      "loss": 0.9487,
+      "step": 16668
+    },
+    {
+      "epoch": 0.4578012139188707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1842278280363794e-05,
+      "loss": 0.9215,
+      "step": 16669
+    },
+    {
+      "epoch": 0.4578286781467139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184140398240459e-05,
+      "loss": 0.9253,
+      "step": 16670
+    },
+    {
+      "epoch": 0.45785614237455713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1840529669875487e-05,
+      "loss": 0.9488,
+      "step": 16671
+    },
+    {
+      "epoch": 0.45788360660240035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.18396553427834e-05,
+      "loss": 0.9368,
+      "step": 16672
+    },
+    {
+      "epoch": 0.4579110708302436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1838781001135248e-05,
+      "loss": 0.8876,
+      "step": 16673
+    },
+    {
+      "epoch": 0.45793853505808685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1837906644937947e-05,
+      "loss": 0.9277,
+      "step": 16674
+    },
+    {
+      "epoch": 0.45796599928593007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1837032274198417e-05,
+      "loss": 0.9158,
+      "step": 16675
+    },
+    {
+      "epoch": 0.4579934635137733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1836157888923576e-05,
+      "loss": 0.8904,
+      "step": 16676
+    },
+    {
+      "epoch": 0.45802092774161657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1835283489120344e-05,
+      "loss": 0.8889,
+      "step": 16677
+    },
+    {
+      "epoch": 0.4580483919694598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1834409074795634e-05,
+      "loss": 0.8677,
+      "step": 16678
+    },
+    {
+      "epoch": 0.458075856197303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1833534645956373e-05,
+      "loss": 0.9524,
+      "step": 16679
+    },
+    {
+      "epoch": 0.45810332042514623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1832660202609475e-05,
+      "loss": 0.9182,
+      "step": 16680
+    },
+    {
+      "epoch": 0.4581307846529895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1831785744761857e-05,
+      "loss": 0.9508,
+      "step": 16681
+    },
+    {
+      "epoch": 0.45815824888083273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1830911272420445e-05,
+      "loss": 0.912,
+      "step": 16682
+    },
+    {
+      "epoch": 0.45818571310867595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1830036785592149e-05,
+      "loss": 0.9498,
+      "step": 16683
+    },
+    {
+      "epoch": 0.4582131773365192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1829162284283894e-05,
+      "loss": 1.0535,
+      "step": 16684
+    },
+    {
+      "epoch": 0.4582406415643624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1828287768502603e-05,
+      "loss": 0.9122,
+      "step": 16685
+    },
+    {
+      "epoch": 0.45826810579220567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1827413238255186e-05,
+      "loss": 0.9492,
+      "step": 16686
+    },
+    {
+      "epoch": 0.4582955700200489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1826538693548568e-05,
+      "loss": 0.8972,
+      "step": 16687
+    },
+    {
+      "epoch": 0.4583230342478921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1825664134389669e-05,
+      "loss": 0.9472,
+      "step": 16688
+    },
+    {
+      "epoch": 0.45835049847573534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.182478956078541e-05,
+      "loss": 0.8792,
+      "step": 16689
+    },
+    {
+      "epoch": 0.4583779627035786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1823914972742707e-05,
+      "loss": 0.8317,
+      "step": 16690
+    },
+    {
+      "epoch": 0.45840542693142183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1823040370268484e-05,
+      "loss": 0.7793,
+      "step": 16691
+    },
+    {
+      "epoch": 0.45843289115926505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.182216575336966e-05,
+      "loss": 1.0403,
+      "step": 16692
+    },
+    {
+      "epoch": 0.4584603553871083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1821291122053152e-05,
+      "loss": 0.9336,
+      "step": 16693
+    },
+    {
+      "epoch": 0.45848781961495155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1820416476325884e-05,
+      "loss": 0.9156,
+      "step": 16694
+    },
+    {
+      "epoch": 0.4585152838427948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1819541816194775e-05,
+      "loss": 0.9146,
+      "step": 16695
+    },
+    {
+      "epoch": 0.458542748070638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1818667141666749e-05,
+      "loss": 0.9256,
+      "step": 16696
+    },
+    {
+      "epoch": 0.4585702122984812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1817792452748723e-05,
+      "loss": 0.8256,
+      "step": 16697
+    },
+    {
+      "epoch": 0.45859767652632444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1816917749447616e-05,
+      "loss": 0.9727,
+      "step": 16698
+    },
+    {
+      "epoch": 0.4586251407541677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1816043031770355e-05,
+      "loss": 0.92,
+      "step": 16699
+    },
+    {
+      "epoch": 0.45865260498201094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1815168299723858e-05,
+      "loss": 1.0317,
+      "step": 16700
+    },
+    {
+      "epoch": 0.45868006920985416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1814293553315046e-05,
+      "loss": 0.991,
+      "step": 16701
+    },
+    {
+      "epoch": 0.4587075334376974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.181341879255084e-05,
+      "loss": 0.9518,
+      "step": 16702
+    },
+    {
+      "epoch": 0.45873499766554066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1812544017438163e-05,
+      "loss": 0.9221,
+      "step": 16703
+    },
+    {
+      "epoch": 0.4587624618933839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1811669227983935e-05,
+      "loss": 1.0388,
+      "step": 16704
+    },
+    {
+      "epoch": 0.4587899261212271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1810794424195078e-05,
+      "loss": 0.89,
+      "step": 16705
+    },
+    {
+      "epoch": 0.4588173903490703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1809919606078515e-05,
+      "loss": 0.8861,
+      "step": 16706
+    },
+    {
+      "epoch": 0.4588448545769136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1809044773641165e-05,
+      "loss": 0.9565,
+      "step": 16707
+    },
+    {
+      "epoch": 0.4588723188047568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1808169926889955e-05,
+      "loss": 1.0093,
+      "step": 16708
+    },
+    {
+      "epoch": 0.45889978303260004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1807295065831803e-05,
+      "loss": 0.8825,
+      "step": 16709
+    },
+    {
+      "epoch": 0.45892724726044326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1806420190473631e-05,
+      "loss": 0.9587,
+      "step": 16710
+    },
+    {
+      "epoch": 0.4589547114882865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1805545300822361e-05,
+      "loss": 0.8409,
+      "step": 16711
+    },
+    {
+      "epoch": 0.45898217571612976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1804670396884921e-05,
+      "loss": 0.9661,
+      "step": 16712
+    },
+    {
+      "epoch": 0.459009639943973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1803795478668229e-05,
+      "loss": 0.8923,
+      "step": 16713
+    },
+    {
+      "epoch": 0.4590371041718162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1802920546179206e-05,
+      "loss": 0.8293,
+      "step": 16714
+    },
+    {
+      "epoch": 0.4590645683996594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.180204559942478e-05,
+      "loss": 0.8795,
+      "step": 16715
+    },
+    {
+      "epoch": 0.4590920326275027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1801170638411873e-05,
+      "loss": 0.8635,
+      "step": 16716
+    },
+    {
+      "epoch": 0.4591194968553459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1800295663147403e-05,
+      "loss": 0.8333,
+      "step": 16717
+    },
+    {
+      "epoch": 0.45914696108318914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1799420673638297e-05,
+      "loss": 0.8437,
+      "step": 16718
+    },
+    {
+      "epoch": 0.45917442531103236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1798545669891477e-05,
+      "loss": 0.8839,
+      "step": 16719
+    },
+    {
+      "epoch": 0.45920188953887564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1797670651913868e-05,
+      "loss": 0.9691,
+      "step": 16720
+    },
+    {
+      "epoch": 0.45922935376671886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1796795619712391e-05,
+      "loss": 0.8843,
+      "step": 16721
+    },
+    {
+      "epoch": 0.4592568179945621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1795920573293974e-05,
+      "loss": 0.9743,
+      "step": 16722
+    },
+    {
+      "epoch": 0.4592842822224053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1795045512665532e-05,
+      "loss": 1.0088,
+      "step": 16723
+    },
+    {
+      "epoch": 0.4593117464502485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1794170437834001e-05,
+      "loss": 0.8367,
+      "step": 16724
+    },
+    {
+      "epoch": 0.4593392106780918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1793295348806295e-05,
+      "loss": 0.9422,
+      "step": 16725
+    },
+    {
+      "epoch": 0.459366674905935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1792420245589341e-05,
+      "loss": 0.8282,
+      "step": 16726
+    },
+    {
+      "epoch": 0.45939413913377825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1791545128190066e-05,
+      "loss": 0.9219,
+      "step": 16727
+    },
+    {
+      "epoch": 0.45942160336162147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1790669996615392e-05,
+      "loss": 0.9673,
+      "step": 16728
+    },
+    {
+      "epoch": 0.45944906758946474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1789794850872243e-05,
+      "loss": 0.8831,
+      "step": 16729
+    },
+    {
+      "epoch": 0.45947653181730796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1788919690967542e-05,
+      "loss": 0.8738,
+      "step": 16730
+    },
+    {
+      "epoch": 0.4595039960451512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1788044516908217e-05,
+      "loss": 0.8963,
+      "step": 16731
+    },
+    {
+      "epoch": 0.4595314602729944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1787169328701191e-05,
+      "loss": 0.8346,
+      "step": 16732
+    },
+    {
+      "epoch": 0.4595589245008377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.178629412635339e-05,
+      "loss": 0.9177,
+      "step": 16733
+    },
+    {
+      "epoch": 0.4595863887286809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1785418909871735e-05,
+      "loss": 0.8599,
+      "step": 16734
+    },
+    {
+      "epoch": 0.4596138529565241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1784543679263156e-05,
+      "loss": 1.0161,
+      "step": 16735
+    },
+    {
+      "epoch": 0.45964131718436735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1783668434534576e-05,
+      "loss": 0.8434,
+      "step": 16736
+    },
+    {
+      "epoch": 0.45966878141221057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.178279317569292e-05,
+      "loss": 0.8197,
+      "step": 16737
+    },
+    {
+      "epoch": 0.45969624564005385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1781917902745114e-05,
+      "loss": 0.967,
+      "step": 16738
+    },
+    {
+      "epoch": 0.45972370986789707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1781042615698084e-05,
+      "loss": 0.8856,
+      "step": 16739
+    },
+    {
+      "epoch": 0.4597511740957403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1780167314558756e-05,
+      "loss": 0.9128,
+      "step": 16740
+    },
+    {
+      "epoch": 0.4597786383235835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1779291999334053e-05,
+      "loss": 0.8458,
+      "step": 16741
+    },
+    {
+      "epoch": 0.4598061025514268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1778416670030901e-05,
+      "loss": 0.8112,
+      "step": 16742
+    },
+    {
+      "epoch": 0.45983356677927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1777541326656227e-05,
+      "loss": 0.9435,
+      "step": 16743
+    },
+    {
+      "epoch": 0.45986103100711323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.177666596921696e-05,
+      "loss": 0.8417,
+      "step": 16744
+    },
+    {
+      "epoch": 0.45988849523495645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1775790597720025e-05,
+      "loss": 0.9475,
+      "step": 16745
+    },
+    {
+      "epoch": 0.45991595946279973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1774915212172342e-05,
+      "loss": 1.0187,
+      "step": 16746
+    },
+    {
+      "epoch": 0.45994342369064295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1774039812580847e-05,
+      "loss": 0.9535,
+      "step": 16747
+    },
+    {
+      "epoch": 0.45997088791848617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1773164398952461e-05,
+      "loss": 0.8426,
+      "step": 16748
+    },
+    {
+      "epoch": 0.4599983521463294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1772288971294108e-05,
+      "loss": 0.9327,
+      "step": 16749
+    },
+    {
+      "epoch": 0.4600258163741726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1771413529612722e-05,
+      "loss": 0.9355,
+      "step": 16750
+    },
+    {
+      "epoch": 0.4600532806020159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1770538073915223e-05,
+      "loss": 0.8108,
+      "step": 16751
+    },
+    {
+      "epoch": 0.4600807448298591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1769662604208543e-05,
+      "loss": 0.8791,
+      "step": 16752
+    },
+    {
+      "epoch": 0.46010820905770233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1768787120499608e-05,
+      "loss": 0.9686,
+      "step": 16753
+    },
+    {
+      "epoch": 0.46013567328554555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.176791162279534e-05,
+      "loss": 0.9138,
+      "step": 16754
+    },
+    {
+      "epoch": 0.46016313751338883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1767036111102672e-05,
+      "loss": 0.8365,
+      "step": 16755
+    },
+    {
+      "epoch": 0.46019060174123205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1766160585428532e-05,
+      "loss": 0.9839,
+      "step": 16756
+    },
+    {
+      "epoch": 0.4602180659690753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1765285045779847e-05,
+      "loss": 0.9237,
+      "step": 16757
+    },
+    {
+      "epoch": 0.4602455301969185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1764409492163537e-05,
+      "loss": 0.9201,
+      "step": 16758
+    },
+    {
+      "epoch": 0.46027299442476177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1763533924586539e-05,
+      "loss": 0.9717,
+      "step": 16759
+    },
+    {
+      "epoch": 0.460300458652605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1762658343055779e-05,
+      "loss": 0.9865,
+      "step": 16760
+    },
+    {
+      "epoch": 0.4603279228804482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1761782747578178e-05,
+      "loss": 0.8935,
+      "step": 16761
+    },
+    {
+      "epoch": 0.46035538710829144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1760907138160676e-05,
+      "loss": 0.9834,
+      "step": 16762
+    },
+    {
+      "epoch": 0.46038285133613466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.176003151481019e-05,
+      "loss": 0.8381,
+      "step": 16763
+    },
+    {
+      "epoch": 0.46041031556397793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1759155877533654e-05,
+      "loss": 0.9364,
+      "step": 16764
+    },
+    {
+      "epoch": 0.46043777979182116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1758280226337997e-05,
+      "loss": 0.823,
+      "step": 16765
+    },
+    {
+      "epoch": 0.4604652440196644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1757404561230144e-05,
+      "loss": 1.0248,
+      "step": 16766
+    },
+    {
+      "epoch": 0.4604927082475076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1756528882217024e-05,
+      "loss": 0.9365,
+      "step": 16767
+    },
+    {
+      "epoch": 0.4605201724753509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.175565318930557e-05,
+      "loss": 0.9096,
+      "step": 16768
+    },
+    {
+      "epoch": 0.4605476367031941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1754777482502706e-05,
+      "loss": 0.8518,
+      "step": 16769
+    },
+    {
+      "epoch": 0.4605751009310373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1753901761815363e-05,
+      "loss": 1.0405,
+      "step": 16770
+    },
+    {
+      "epoch": 0.46060256515888054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1753026027250471e-05,
+      "loss": 0.9376,
+      "step": 16771
+    },
+    {
+      "epoch": 0.4606300293867238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.175215027881496e-05,
+      "loss": 0.8793,
+      "step": 16772
+    },
+    {
+      "epoch": 0.46065749361456704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1751274516515752e-05,
+      "loss": 1.0171,
+      "step": 16773
+    },
+    {
+      "epoch": 0.46068495784241026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1750398740359786e-05,
+      "loss": 0.8894,
+      "step": 16774
+    },
+    {
+      "epoch": 0.4607124220702535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1749522950353984e-05,
+      "loss": 1.006,
+      "step": 16775
+    },
+    {
+      "epoch": 0.4607398862980967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1748647146505282e-05,
+      "loss": 0.9265,
+      "step": 16776
+    },
+    {
+      "epoch": 0.46076735052594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1747771328820604e-05,
+      "loss": 0.7768,
+      "step": 16777
+    },
+    {
+      "epoch": 0.4607948147537832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1746895497306883e-05,
+      "loss": 0.9711,
+      "step": 16778
+    },
+    {
+      "epoch": 0.4608222789816264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1746019651971045e-05,
+      "loss": 0.8608,
+      "step": 16779
+    },
+    {
+      "epoch": 0.46084974320946964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1745143792820028e-05,
+      "loss": 0.9589,
+      "step": 16780
+    },
+    {
+      "epoch": 0.4608772074373129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1744267919860757e-05,
+      "loss": 0.8302,
+      "step": 16781
+    },
+    {
+      "epoch": 0.46090467166515614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1743392033100159e-05,
+      "loss": 0.8364,
+      "step": 16782
+    },
+    {
+      "epoch": 0.46093213589299936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.174251613254517e-05,
+      "loss": 0.9078,
+      "step": 16783
+    },
+    {
+      "epoch": 0.4609596001208426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1741640218202722e-05,
+      "loss": 0.8606,
+      "step": 16784
+    },
+    {
+      "epoch": 0.46098706434868586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1740764290079736e-05,
+      "loss": 0.9581,
+      "step": 16785
+    },
+    {
+      "epoch": 0.4610145285765291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173988834818315e-05,
+      "loss": 0.8842,
+      "step": 16786
+    },
+    {
+      "epoch": 0.4610419928043723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1739012392519893e-05,
+      "loss": 0.9432,
+      "step": 16787
+    },
+    {
+      "epoch": 0.4610694570322155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.17381364230969e-05,
+      "loss": 0.8518,
+      "step": 16788
+    },
+    {
+      "epoch": 0.4610969212600588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1737260439921097e-05,
+      "loss": 0.9697,
+      "step": 16789
+    },
+    {
+      "epoch": 0.461124385487902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1736384442999413e-05,
+      "loss": 0.8257,
+      "step": 16790
+    },
+    {
+      "epoch": 0.46115184971574524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1735508432338784e-05,
+      "loss": 0.9239,
+      "step": 16791
+    },
+    {
+      "epoch": 0.46117931394358846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1734632407946143e-05,
+      "loss": 0.9608,
+      "step": 16792
+    },
+    {
+      "epoch": 0.4612067781714317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1733756369828418e-05,
+      "loss": 0.8922,
+      "step": 16793
+    },
+    {
+      "epoch": 0.46123424239927496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173288031799254e-05,
+      "loss": 0.8798,
+      "step": 16794
+    },
+    {
+      "epoch": 0.4612617066271182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1732004252445439e-05,
+      "loss": 0.9384,
+      "step": 16795
+    },
+    {
+      "epoch": 0.4612891708549614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1731128173194053e-05,
+      "loss": 0.8375,
+      "step": 16796
+    },
+    {
+      "epoch": 0.4613166350828046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173025208024531e-05,
+      "loss": 0.9234,
+      "step": 16797
+    },
+    {
+      "epoch": 0.4613440993106479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.172937597360614e-05,
+      "loss": 0.9032,
+      "step": 16798
+    },
+    {
+      "epoch": 0.4613715635384911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1728499853283476e-05,
+      "loss": 0.9126,
+      "step": 16799
+    },
+    {
+      "epoch": 0.46139902776633435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1727623719284255e-05,
+      "loss": 0.9532,
+      "step": 16800
+    },
+    {
+      "epoch": 0.46142649199417757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1726747571615406e-05,
+      "loss": 0.9426,
+      "step": 16801
+    },
+    {
+      "epoch": 0.46145395622202084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.172587141028386e-05,
+      "loss": 0.8771,
+      "step": 16802
+    },
+    {
+      "epoch": 0.46148142044986407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1724995235296552e-05,
+      "loss": 0.8763,
+      "step": 16803
+    },
+    {
+      "epoch": 0.4615088846777073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1724119046660415e-05,
+      "loss": 0.8629,
+      "step": 16804
+    },
+    {
+      "epoch": 0.4615363489055505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1723242844382376e-05,
+      "loss": 0.9669,
+      "step": 16805
+    },
+    {
+      "epoch": 0.46156381313339373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1722366628469375e-05,
+      "loss": 0.912,
+      "step": 16806
+    },
+    {
+      "epoch": 0.461591277361237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1721490398928341e-05,
+      "loss": 0.8957,
+      "step": 16807
+    },
+    {
+      "epoch": 0.46161874158908023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1720614155766207e-05,
+      "loss": 1.0105,
+      "step": 16808
+    },
+    {
+      "epoch": 0.46164620581692345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1719737898989911e-05,
+      "loss": 0.904,
+      "step": 16809
+    },
+    {
+      "epoch": 0.46167367004476667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.171886162860638e-05,
+      "loss": 1.0068,
+      "step": 16810
+    },
+    {
+      "epoch": 0.46170113427260995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1717985344622549e-05,
+      "loss": 0.9356,
+      "step": 16811
+    },
+    {
+      "epoch": 0.46172859850045317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1717109047045355e-05,
+      "loss": 0.9805,
+      "step": 16812
+    },
+    {
+      "epoch": 0.4617560627282964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1716232735881727e-05,
+      "loss": 0.9174,
+      "step": 16813
+    },
+    {
+      "epoch": 0.4617835269561396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1715356411138601e-05,
+      "loss": 0.9098,
+      "step": 16814
+    },
+    {
+      "epoch": 0.4618109911839829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1714480072822912e-05,
+      "loss": 0.9055,
+      "step": 16815
+    },
+    {
+      "epoch": 0.4618384554118261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1713603720941594e-05,
+      "loss": 0.93,
+      "step": 16816
+    },
+    {
+      "epoch": 0.46186591963966933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1712727355501574e-05,
+      "loss": 0.8566,
+      "step": 16817
+    },
+    {
+      "epoch": 0.46189338386751255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1711850976509797e-05,
+      "loss": 0.8969,
+      "step": 16818
+    },
+    {
+      "epoch": 0.4619208480953558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.171097458397319e-05,
+      "loss": 0.9815,
+      "step": 16819
+    },
+    {
+      "epoch": 0.46194831232319905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1710098177898687e-05,
+      "loss": 0.8979,
+      "step": 16820
+    },
+    {
+      "epoch": 0.46197577655104227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170922175829323e-05,
+      "loss": 0.9044,
+      "step": 16821
+    },
+    {
+      "epoch": 0.4620032407788855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1708345325163744e-05,
+      "loss": 0.9667,
+      "step": 16822
+    },
+    {
+      "epoch": 0.4620307050067287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1707468878517167e-05,
+      "loss": 0.9719,
+      "step": 16823
+    },
+    {
+      "epoch": 0.462058169234572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1706592418360439e-05,
+      "loss": 0.7939,
+      "step": 16824
+    },
+    {
+      "epoch": 0.4620856334624152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1705715944700491e-05,
+      "loss": 0.9724,
+      "step": 16825
+    },
+    {
+      "epoch": 0.46211309769025843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1704839457544256e-05,
+      "loss": 0.8953,
+      "step": 16826
+    },
+    {
+      "epoch": 0.46214056191810166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170396295689867e-05,
+      "loss": 1.0208,
+      "step": 16827
+    },
+    {
+      "epoch": 0.46216802614594493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170308644277067e-05,
+      "loss": 0.8274,
+      "step": 16828
+    },
+    {
+      "epoch": 0.46219549037378815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1702209915167189e-05,
+      "loss": 0.9186,
+      "step": 16829
+    },
+    {
+      "epoch": 0.4622229546016314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1701333374095165e-05,
+      "loss": 0.9448,
+      "step": 16830
+    },
+    {
+      "epoch": 0.4622504188294746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170045681956153e-05,
+      "loss": 0.9236,
+      "step": 16831
+    },
+    {
+      "epoch": 0.4622778830573178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1699580251573226e-05,
+      "loss": 0.8847,
+      "step": 16832
+    },
+    {
+      "epoch": 0.4623053472851611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1698703670137184e-05,
+      "loss": 1.0075,
+      "step": 16833
+    },
+    {
+      "epoch": 0.4623328115130043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1697827075260337e-05,
+      "loss": 0.8136,
+      "step": 16834
+    },
+    {
+      "epoch": 0.46236027574084754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1696950466949626e-05,
+      "loss": 0.8851,
+      "step": 16835
+    },
+    {
+      "epoch": 0.46238773996869076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1696073845211987e-05,
+      "loss": 0.8789,
+      "step": 16836
+    },
+    {
+      "epoch": 0.46241520419653404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1695197210054355e-05,
+      "loss": 0.9589,
+      "step": 16837
+    },
+    {
+      "epoch": 0.46244266842437726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1694320561483664e-05,
+      "loss": 0.887,
+      "step": 16838
+    },
+    {
+      "epoch": 0.4624701326522205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1693443899506853e-05,
+      "loss": 0.8804,
+      "step": 16839
+    },
+    {
+      "epoch": 0.4624975968800637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.169256722413086e-05,
+      "loss": 0.8946,
+      "step": 16840
+    },
+    {
+      "epoch": 0.462525061107907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1691690535362617e-05,
+      "loss": 1.0325,
+      "step": 16841
+    },
+    {
+      "epoch": 0.4625525253357502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1690813833209065e-05,
+      "loss": 0.9661,
+      "step": 16842
+    },
+    {
+      "epoch": 0.4625799895635934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1689937117677136e-05,
+      "loss": 0.9556,
+      "step": 16843
+    },
+    {
+      "epoch": 0.46260745379143664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1689060388773774e-05,
+      "loss": 0.8607,
+      "step": 16844
+    },
+    {
+      "epoch": 0.46263491801927986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1688183646505911e-05,
+      "loss": 0.8221,
+      "step": 16845
+    },
+    {
+      "epoch": 0.46266238224712314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168730689088048e-05,
+      "loss": 0.85,
+      "step": 16846
+    },
+    {
+      "epoch": 0.46268984647496636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1686430121904429e-05,
+      "loss": 0.9612,
+      "step": 16847
+    },
+    {
+      "epoch": 0.4627173107028096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1685553339584687e-05,
+      "loss": 0.8943,
+      "step": 16848
+    },
+    {
+      "epoch": 0.4627447749306528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1684676543928197e-05,
+      "loss": 0.7906,
+      "step": 16849
+    },
+    {
+      "epoch": 0.4627722391584961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168379973494189e-05,
+      "loss": 0.9756,
+      "step": 16850
+    },
+    {
+      "epoch": 0.4627997033863393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1682922912632708e-05,
+      "loss": 0.8979,
+      "step": 16851
+    },
+    {
+      "epoch": 0.4628271676141825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1682046077007592e-05,
+      "loss": 0.8699,
+      "step": 16852
+    },
+    {
+      "epoch": 0.46285463184202574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168116922807347e-05,
+      "loss": 0.8842,
+      "step": 16853
+    },
+    {
+      "epoch": 0.462882096069869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168029236583729e-05,
+      "loss": 0.8468,
+      "step": 16854
+    },
+    {
+      "epoch": 0.46290956029771224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1679415490305983e-05,
+      "loss": 0.9409,
+      "step": 16855
+    },
+    {
+      "epoch": 0.46293702452555546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.167853860148649e-05,
+      "loss": 0.9147,
+      "step": 16856
+    },
+    {
+      "epoch": 0.4629644887533987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1677661699385753e-05,
+      "loss": 0.9334,
+      "step": 16857
+    },
+    {
+      "epoch": 0.4629919529812419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1676784784010704e-05,
+      "loss": 0.8608,
+      "step": 16858
+    },
+    {
+      "epoch": 0.4630194172090852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1675907855368287e-05,
+      "loss": 0.8182,
+      "step": 16859
+    },
+    {
+      "epoch": 0.4630468814369284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1675030913465433e-05,
+      "loss": 0.8665,
+      "step": 16860
+    },
+    {
+      "epoch": 0.4630743456647716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1674153958309089e-05,
+      "loss": 0.971,
+      "step": 16861
+    },
+    {
+      "epoch": 0.46310180989261485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.167327698990619e-05,
+      "loss": 0.8764,
+      "step": 16862
+    },
+    {
+      "epoch": 0.4631292741204581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1672400008263676e-05,
+      "loss": 0.8826,
+      "step": 16863
+    },
+    {
+      "epoch": 0.46315673834830134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1671523013388484e-05,
+      "loss": 0.9687,
+      "step": 16864
+    },
+    {
+      "epoch": 0.46318420257614457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1670646005287556e-05,
+      "loss": 0.8715,
+      "step": 16865
+    },
+    {
+      "epoch": 0.4632116668039878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1669768983967826e-05,
+      "loss": 0.8771,
+      "step": 16866
+    },
+    {
+      "epoch": 0.46323913103183106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1668891949436239e-05,
+      "loss": 0.8468,
+      "step": 16867
+    },
+    {
+      "epoch": 0.4632665952596743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1668014901699733e-05,
+      "loss": 0.9779,
+      "step": 16868
+    },
+    {
+      "epoch": 0.4632940594875175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1667137840765247e-05,
+      "loss": 0.9051,
+      "step": 16869
+    },
+    {
+      "epoch": 0.46332152371536073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1666260766639721e-05,
+      "loss": 0.9431,
+      "step": 16870
+    },
+    {
+      "epoch": 0.46334898794320395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1665383679330094e-05,
+      "loss": 0.8598,
+      "step": 16871
+    },
+    {
+      "epoch": 0.4633764521710472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1664506578843305e-05,
+      "loss": 0.9022,
+      "step": 16872
+    },
+    {
+      "epoch": 0.46340391639889045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1663629465186297e-05,
+      "loss": 0.8933,
+      "step": 16873
+    },
+    {
+      "epoch": 0.46343138062673367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1662752338366008e-05,
+      "loss": 0.9114,
+      "step": 16874
+    },
+    {
+      "epoch": 0.4634588448545769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1661875198389376e-05,
+      "loss": 0.8485,
+      "step": 16875
+    },
+    {
+      "epoch": 0.46348630908242017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1660998045263347e-05,
+      "loss": 0.9184,
+      "step": 16876
+    },
+    {
+      "epoch": 0.4635137733102634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1660120878994858e-05,
+      "loss": 0.9885,
+      "step": 16877
+    },
+    {
+      "epoch": 0.4635412375381066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1659243699590845e-05,
+      "loss": 0.9153,
+      "step": 16878
+    },
+    {
+      "epoch": 0.46356870176594983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1658366507058256e-05,
+      "loss": 0.9397,
+      "step": 16879
+    },
+    {
+      "epoch": 0.4635961659937931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165748930140403e-05,
+      "loss": 0.946,
+      "step": 16880
+    },
+    {
+      "epoch": 0.46362363022163633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1656612082635105e-05,
+      "loss": 0.8455,
+      "step": 16881
+    },
+    {
+      "epoch": 0.46365109444947955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1655734850758424e-05,
+      "loss": 0.8671,
+      "step": 16882
+    },
+    {
+      "epoch": 0.46367855867732277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165485760578093e-05,
+      "loss": 0.9287,
+      "step": 16883
+    },
+    {
+      "epoch": 0.46370602290516605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165398034770956e-05,
+      "loss": 0.9781,
+      "step": 16884
+    },
+    {
+      "epoch": 0.46373348713300927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1653103076551253e-05,
+      "loss": 0.7823,
+      "step": 16885
+    },
+    {
+      "epoch": 0.4637609513608525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1652225792312959e-05,
+      "loss": 0.9277,
+      "step": 16886
+    },
+    {
+      "epoch": 0.4637884155886957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165134849500161e-05,
+      "loss": 0.9861,
+      "step": 16887
+    },
+    {
+      "epoch": 0.46381587981653893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1650471184624157e-05,
+      "loss": 1.0076,
+      "step": 16888
+    },
+    {
+      "epoch": 0.4638433440443822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1649593861187536e-05,
+      "loss": 0.8549,
+      "step": 16889
+    },
+    {
+      "epoch": 0.46387080827222543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1648716524698684e-05,
+      "loss": 0.8177,
+      "step": 16890
+    },
+    {
+      "epoch": 0.46389827250006865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1647839175164552e-05,
+      "loss": 0.9268,
+      "step": 16891
+    },
+    {
+      "epoch": 0.4639257367279119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.164696181259208e-05,
+      "loss": 0.856,
+      "step": 16892
+    },
+    {
+      "epoch": 0.46395320095575515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1646084436988205e-05,
+      "loss": 0.9108,
+      "step": 16893
+    },
+    {
+      "epoch": 0.4639806651835984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1645207048359873e-05,
+      "loss": 0.8618,
+      "step": 16894
+    },
+    {
+      "epoch": 0.4640081294114416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1644329646714028e-05,
+      "loss": 0.9294,
+      "step": 16895
+    },
+    {
+      "epoch": 0.4640355936392848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1643452232057608e-05,
+      "loss": 0.9729,
+      "step": 16896
+    },
+    {
+      "epoch": 0.4640630578671281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1642574804397556e-05,
+      "loss": 0.936,
+      "step": 16897
+    },
+    {
+      "epoch": 0.4640905220949713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1641697363740819e-05,
+      "loss": 1.0388,
+      "step": 16898
+    },
+    {
+      "epoch": 0.46411798632281454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1640819910094332e-05,
+      "loss": 0.7684,
+      "step": 16899
+    },
+    {
+      "epoch": 0.46414545055065776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1639942443465047e-05,
+      "loss": 0.9607,
+      "step": 16900
+    },
+    {
+      "epoch": 0.464172914778501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.16390649638599e-05,
+      "loss": 0.8153,
+      "step": 16901
+    },
+    {
+      "epoch": 0.46420037900634425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1638187471285836e-05,
+      "loss": 0.8423,
+      "step": 16902
+    },
+    {
+      "epoch": 0.4642278432341875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1637309965749796e-05,
+      "loss": 0.9613,
+      "step": 16903
+    },
+    {
+      "epoch": 0.4642553074620307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1636432447258728e-05,
+      "loss": 0.8626,
+      "step": 16904
+    },
+    {
+      "epoch": 0.4642827716898739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1635554915819573e-05,
+      "loss": 0.8989,
+      "step": 16905
+    },
+    {
+      "epoch": 0.4643102359177172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1634677371439271e-05,
+      "loss": 0.8451,
+      "step": 16906
+    },
+    {
+      "epoch": 0.4643377001455604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1633799814124773e-05,
+      "loss": 0.8671,
+      "step": 16907
+    },
+    {
+      "epoch": 0.46436516437340364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1632922243883015e-05,
+      "loss": 1.0214,
+      "step": 16908
+    },
+    {
+      "epoch": 0.46439262860124686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1632044660720944e-05,
+      "loss": 0.9335,
+      "step": 16909
+    },
+    {
+      "epoch": 0.46442009282909014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1631167064645505e-05,
+      "loss": 0.9729,
+      "step": 16910
+    },
+    {
+      "epoch": 0.46444755705693336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1630289455663637e-05,
+      "loss": 0.9453,
+      "step": 16911
+    },
+    {
+      "epoch": 0.4644750212847766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1629411833782292e-05,
+      "loss": 0.8101,
+      "step": 16912
+    },
+    {
+      "epoch": 0.4645024855126198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1628534199008406e-05,
+      "loss": 0.9615,
+      "step": 16913
+    },
+    {
+      "epoch": 0.464529949740463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1627656551348928e-05,
+      "loss": 0.9451,
+      "step": 16914
+    },
+    {
+      "epoch": 0.4645574139683063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.16267788908108e-05,
+      "loss": 0.8041,
+      "step": 16915
+    },
+    {
+      "epoch": 0.4645848781961495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1625901217400966e-05,
+      "loss": 0.8522,
+      "step": 16916
+    },
+    {
+      "epoch": 0.46461234242399274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1625023531126374e-05,
+      "loss": 0.8394,
+      "step": 16917
+    },
+    {
+      "epoch": 0.46463980665183596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1624145831993968e-05,
+      "loss": 0.8577,
+      "step": 16918
+    },
+    {
+      "epoch": 0.46466727087967924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1623268120010684e-05,
+      "loss": 1.0677,
+      "step": 16919
+    },
+    {
+      "epoch": 0.46469473510752246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.162239039518348e-05,
+      "loss": 0.8044,
+      "step": 16920
+    },
+    {
+      "epoch": 0.4647221993353657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1621512657519297e-05,
+      "loss": 0.8681,
+      "step": 16921
+    },
+    {
+      "epoch": 0.4647496635632089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.162063490702507e-05,
+      "loss": 1.0278,
+      "step": 16922
+    },
+    {
+      "epoch": 0.4647771277910522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1619757143707755e-05,
+      "loss": 0.881,
+      "step": 16923
+    },
+    {
+      "epoch": 0.4648045920188954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1618879367574297e-05,
+      "loss": 0.9108,
+      "step": 16924
+    },
+    {
+      "epoch": 0.4648320562467386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1618001578631638e-05,
+      "loss": 0.9294,
+      "step": 16925
+    },
+    {
+      "epoch": 0.46485952047458184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1617123776886719e-05,
+      "loss": 0.8837,
+      "step": 16926
+    },
+    {
+      "epoch": 0.46488698470242507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1616245962346494e-05,
+      "loss": 0.9963,
+      "step": 16927
+    },
+    {
+      "epoch": 0.46491444893026834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1615368135017902e-05,
+      "loss": 0.9278,
+      "step": 16928
+    },
+    {
+      "epoch": 0.46494191315811156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1614490294907895e-05,
+      "loss": 0.9702,
+      "step": 16929
+    },
+    {
+      "epoch": 0.4649693773859548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1613612442023415e-05,
+      "loss": 0.8231,
+      "step": 16930
+    },
+    {
+      "epoch": 0.464996841613798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1612734576371407e-05,
+      "loss": 0.9632,
+      "step": 16931
+    },
+    {
+      "epoch": 0.4650243058416413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1611856697958819e-05,
+      "loss": 0.7847,
+      "step": 16932
+    },
+    {
+      "epoch": 0.4650517700694845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1610978806792597e-05,
+      "loss": 0.9749,
+      "step": 16933
+    },
+    {
+      "epoch": 0.4650792342973277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1610100902879684e-05,
+      "loss": 0.8858,
+      "step": 16934
+    },
+    {
+      "epoch": 0.46510669852517095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.160922298622703e-05,
+      "loss": 0.9506,
+      "step": 16935
+    },
+    {
+      "epoch": 0.4651341627530142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1608345056841581e-05,
+      "loss": 0.9756,
+      "step": 16936
+    },
+    {
+      "epoch": 0.46516162698085745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1607467114730284e-05,
+      "loss": 0.8978,
+      "step": 16937
+    },
+    {
+      "epoch": 0.46518909120870067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1606589159900081e-05,
+      "loss": 0.891,
+      "step": 16938
+    },
+    {
+      "epoch": 0.4652165554365439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1605711192357926e-05,
+      "loss": 0.9302,
+      "step": 16939
+    },
+    {
+      "epoch": 0.4652440196643871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.160483321211076e-05,
+      "loss": 0.9536,
+      "step": 16940
+    },
+    {
+      "epoch": 0.4652714838922304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1603955219165534e-05,
+      "loss": 0.9295,
+      "step": 16941
+    },
+    {
+      "epoch": 0.4652989481200736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1603077213529192e-05,
+      "loss": 0.8864,
+      "step": 16942
+    },
+    {
+      "epoch": 0.46532641234791683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1602199195208681e-05,
+      "loss": 1.0353,
+      "step": 16943
+    },
+    {
+      "epoch": 0.46535387657576005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1601321164210953e-05,
+      "loss": 0.9555,
+      "step": 16944
+    },
+    {
+      "epoch": 0.4653813408036033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1600443120542949e-05,
+      "loss": 0.8096,
+      "step": 16945
+    },
+    {
+      "epoch": 0.46540880503144655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1599565064211617e-05,
+      "loss": 0.9991,
+      "step": 16946
+    },
+    {
+      "epoch": 0.46543626925928977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1598686995223907e-05,
+      "loss": 0.8941,
+      "step": 16947
+    },
+    {
+      "epoch": 0.465463733487133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1597808913586768e-05,
+      "loss": 0.888,
+      "step": 16948
+    },
+    {
+      "epoch": 0.46549119771497627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1596930819307146e-05,
+      "loss": 0.9764,
+      "step": 16949
+    },
+    {
+      "epoch": 0.4655186619428195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1596052712391988e-05,
+      "loss": 0.8848,
+      "step": 16950
+    },
+    {
+      "epoch": 0.4655461261706627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1595174592848244e-05,
+      "loss": 0.9307,
+      "step": 16951
+    },
+    {
+      "epoch": 0.46557359039850593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1594296460682859e-05,
+      "loss": 0.9369,
+      "step": 16952
+    },
+    {
+      "epoch": 0.46560105462634915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1593418315902785e-05,
+      "loss": 0.8887,
+      "step": 16953
+    },
+    {
+      "epoch": 0.46562851885419243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1592540158514966e-05,
+      "loss": 0.8016,
+      "step": 16954
+    },
+    {
+      "epoch": 0.46565598308203565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1591661988526353e-05,
+      "loss": 0.9324,
+      "step": 16955
+    },
+    {
+      "epoch": 0.4656834473098789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1590783805943895e-05,
+      "loss": 0.8832,
+      "step": 16956
+    },
+    {
+      "epoch": 0.4657109115377221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.158990561077454e-05,
+      "loss": 0.9526,
+      "step": 16957
+    },
+    {
+      "epoch": 0.46573837576556537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1589027403025233e-05,
+      "loss": 0.8986,
+      "step": 16958
+    },
+    {
+      "epoch": 0.4657658399934086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1588149182702924e-05,
+      "loss": 0.9205,
+      "step": 16959
+    },
+    {
+      "epoch": 0.4657933042212518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1587270949814569e-05,
+      "loss": 0.827,
+      "step": 16960
+    },
+    {
+      "epoch": 0.46582076844909504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.158639270436711e-05,
+      "loss": 0.8818,
+      "step": 16961
+    },
+    {
+      "epoch": 0.4658482326769383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1585514446367494e-05,
+      "loss": 0.9934,
+      "step": 16962
+    },
+    {
+      "epoch": 0.46587569690478153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1584636175822674e-05,
+      "loss": 0.8061,
+      "step": 16963
+    },
+    {
+      "epoch": 0.46590316113262475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1583757892739602e-05,
+      "loss": 0.6989,
+      "step": 16964
+    },
+    {
+      "epoch": 0.465930625360468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1582879597125222e-05,
+      "loss": 1.0012,
+      "step": 16965
+    },
+    {
+      "epoch": 0.4659580895883112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1582001288986486e-05,
+      "loss": 0.9763,
+      "step": 16966
+    },
+    {
+      "epoch": 0.4659855538161545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1581122968330342e-05,
+      "loss": 0.8988,
+      "step": 16967
+    },
+    {
+      "epoch": 0.4660130180439977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1580244635163742e-05,
+      "loss": 0.9852,
+      "step": 16968
+    },
+    {
+      "epoch": 0.4660404822718409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1579366289493637e-05,
+      "loss": 0.9483,
+      "step": 16969
+    },
+    {
+      "epoch": 0.46606794649968414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157848793132697e-05,
+      "loss": 0.8858,
+      "step": 16970
+    },
+    {
+      "epoch": 0.4660954107275274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1577609560670693e-05,
+      "loss": 0.9235,
+      "step": 16971
+    },
+    {
+      "epoch": 0.46612287495537064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1576731177531762e-05,
+      "loss": 0.8665,
+      "step": 16972
+    },
+    {
+      "epoch": 0.46615033918321386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1575852781917123e-05,
+      "loss": 0.9514,
+      "step": 16973
+    },
+    {
+      "epoch": 0.4661778034110571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1574974373833725e-05,
+      "loss": 0.9687,
+      "step": 16974
+    },
+    {
+      "epoch": 0.46620526763890036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157409595328852e-05,
+      "loss": 0.902,
+      "step": 16975
+    },
+    {
+      "epoch": 0.4662327318667436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157321752028846e-05,
+      "loss": 0.9961,
+      "step": 16976
+    },
+    {
+      "epoch": 0.4662601960945868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1572339074840493e-05,
+      "loss": 0.9065,
+      "step": 16977
+    },
+    {
+      "epoch": 0.46628766032243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1571460616951567e-05,
+      "loss": 0.9602,
+      "step": 16978
+    },
+    {
+      "epoch": 0.4663151245502733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1570582146628637e-05,
+      "loss": 0.9182,
+      "step": 16979
+    },
+    {
+      "epoch": 0.4663425887781165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1569703663878656e-05,
+      "loss": 0.9751,
+      "step": 16980
+    },
+    {
+      "epoch": 0.46637005300595974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1568825168708568e-05,
+      "loss": 0.9085,
+      "step": 16981
+    },
+    {
+      "epoch": 0.46639751723380296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1567946661125328e-05,
+      "loss": 0.8632,
+      "step": 16982
+    },
+    {
+      "epoch": 0.4664249814616462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1567068141135888e-05,
+      "loss": 0.9276,
+      "step": 16983
+    },
+    {
+      "epoch": 0.46645244568948946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1566189608747196e-05,
+      "loss": 0.8792,
+      "step": 16984
+    },
+    {
+      "epoch": 0.4664799099173327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1565311063966206e-05,
+      "loss": 0.9335,
+      "step": 16985
+    },
+    {
+      "epoch": 0.4665073741451759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1564432506799872e-05,
+      "loss": 0.9085,
+      "step": 16986
+    },
+    {
+      "epoch": 0.4665348383730191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1563553937255136e-05,
+      "loss": 0.9547,
+      "step": 16987
+    },
+    {
+      "epoch": 0.4665623026008624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1562675355338961e-05,
+      "loss": 0.9904,
+      "step": 16988
+    },
+    {
+      "epoch": 0.4665897668287056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.156179676105829e-05,
+      "loss": 0.9713,
+      "step": 16989
+    },
+    {
+      "epoch": 0.46661723105654884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1560918154420079e-05,
+      "loss": 0.8597,
+      "step": 16990
+    },
+    {
+      "epoch": 0.46664469528439206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1560039535431277e-05,
+      "loss": 0.9167,
+      "step": 16991
+    },
+    {
+      "epoch": 0.46667215951223534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1559160904098841e-05,
+      "loss": 0.8846,
+      "step": 16992
+    },
+    {
+      "epoch": 0.46669962374007856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1558282260429719e-05,
+      "loss": 0.9396,
+      "step": 16993
+    },
+    {
+      "epoch": 0.4667270879679218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1557403604430862e-05,
+      "loss": 1.0428,
+      "step": 16994
+    },
+    {
+      "epoch": 0.466754552195765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1556524936109228e-05,
+      "loss": 0.968,
+      "step": 16995
+    },
+    {
+      "epoch": 0.4667820164236082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1555646255471761e-05,
+      "loss": 0.9731,
+      "step": 16996
+    },
+    {
+      "epoch": 0.4668094806514515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1554767562525422e-05,
+      "loss": 1.0059,
+      "step": 16997
+    },
+    {
+      "epoch": 0.4668369448792947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.155388885727716e-05,
+      "loss": 0.9474,
+      "step": 16998
+    },
+    {
+      "epoch": 0.46686440910713795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1553010139733924e-05,
+      "loss": 0.8427,
+      "step": 16999
+    },
+    {
+      "epoch": 0.46689187333498117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1552131409902674e-05,
+      "loss": 0.8021,
+      "step": 17000
+    },
+    {
+      "epoch": 0.46691933756282444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.155125266779036e-05,
+      "loss": 0.8764,
+      "step": 17001
+    },
+    {
+      "epoch": 0.46694680179066766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.155037391340393e-05,
+      "loss": 0.9114,
+      "step": 17002
+    },
+    {
+      "epoch": 0.4669742660185109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.154949514675034e-05,
+      "loss": 0.777,
+      "step": 17003
+    },
+    {
+      "epoch": 0.4670017302463541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.154861636783655e-05,
+      "loss": 0.8769,
+      "step": 17004
+    },
+    {
+      "epoch": 0.4670291944741974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1547737576669505e-05,
+      "loss": 0.8605,
+      "step": 17005
+    },
+    {
+      "epoch": 0.4670566587020406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1546858773256157e-05,
+      "loss": 0.856,
+      "step": 17006
+    },
+    {
+      "epoch": 0.4670841229298838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1545979957603468e-05,
+      "loss": 0.879,
+      "step": 17007
+    },
+    {
+      "epoch": 0.46711158715772705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1545101129718383e-05,
+      "loss": 0.7822,
+      "step": 17008
+    },
+    {
+      "epoch": 0.46713905138557027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1544222289607863e-05,
+      "loss": 0.8911,
+      "step": 17009
+    },
+    {
+      "epoch": 0.46716651561341355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1543343437278856e-05,
+      "loss": 0.8976,
+      "step": 17010
+    },
+    {
+      "epoch": 0.46719397984125677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1542464572738317e-05,
+      "loss": 0.8687,
+      "step": 17011
+    },
+    {
+      "epoch": 0.4672214440691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1541585695993202e-05,
+      "loss": 0.984,
+      "step": 17012
+    },
+    {
+      "epoch": 0.4672489082969432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1540706807050466e-05,
+      "loss": 0.9377,
+      "step": 17013
+    },
+    {
+      "epoch": 0.4672763725247865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1539827905917058e-05,
+      "loss": 0.8728,
+      "step": 17014
+    },
+    {
+      "epoch": 0.4673038367526297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1538948992599934e-05,
+      "loss": 0.9428,
+      "step": 17015
+    },
+    {
+      "epoch": 0.46733130098047293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1538070067106052e-05,
+      "loss": 0.9084,
+      "step": 17016
+    },
+    {
+      "epoch": 0.46735876520831615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1537191129442364e-05,
+      "loss": 0.9799,
+      "step": 17017
+    },
+    {
+      "epoch": 0.46738622943615943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1536312179615823e-05,
+      "loss": 0.9324,
+      "step": 17018
+    },
+    {
+      "epoch": 0.46741369366400265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1535433217633384e-05,
+      "loss": 0.7937,
+      "step": 17019
+    },
+    {
+      "epoch": 0.46744115789184587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1534554243502005e-05,
+      "loss": 0.8908,
+      "step": 17020
+    },
+    {
+      "epoch": 0.4674686221196891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1533675257228638e-05,
+      "loss": 0.9659,
+      "step": 17021
+    },
+    {
+      "epoch": 0.4674960863475323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1532796258820238e-05,
+      "loss": 0.901,
+      "step": 17022
+    },
+    {
+      "epoch": 0.4675235505753756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153191724828376e-05,
+      "loss": 0.8719,
+      "step": 17023
+    },
+    {
+      "epoch": 0.4675510148032188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153103822562616e-05,
+      "loss": 0.9097,
+      "step": 17024
+    },
+    {
+      "epoch": 0.46757847903106203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153015919085439e-05,
+      "loss": 0.9243,
+      "step": 17025
+    },
+    {
+      "epoch": 0.46760594325890525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.152928014397541e-05,
+      "loss": 0.8411,
+      "step": 17026
+    },
+    {
+      "epoch": 0.46763340748674853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.152840108499617e-05,
+      "loss": 0.9399,
+      "step": 17027
+    },
+    {
+      "epoch": 0.46766087171459175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1527522013923634e-05,
+      "loss": 0.9286,
+      "step": 17028
+    },
+    {
+      "epoch": 0.467688335942435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1526642930764748e-05,
+      "loss": 0.9659,
+      "step": 17029
+    },
+    {
+      "epoch": 0.4677158001702782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1525763835526473e-05,
+      "loss": 1.0662,
+      "step": 17030
+    },
+    {
+      "epoch": 0.46774326439812147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1524884728215763e-05,
+      "loss": 0.8824,
+      "step": 17031
+    },
+    {
+      "epoch": 0.4677707286259647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1524005608839575e-05,
+      "loss": 0.951,
+      "step": 17032
+    },
+    {
+      "epoch": 0.4677981928538079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1523126477404865e-05,
+      "loss": 0.9167,
+      "step": 17033
+    },
+    {
+      "epoch": 0.46782565708165114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1522247333918589e-05,
+      "loss": 0.9603,
+      "step": 17034
+    },
+    {
+      "epoch": 0.46785312130949436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15213681783877e-05,
+      "loss": 0.82,
+      "step": 17035
+    },
+    {
+      "epoch": 0.46788058553733763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1520489010819158e-05,
+      "loss": 0.8708,
+      "step": 17036
+    },
+    {
+      "epoch": 0.46790804976518086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1519609831219918e-05,
+      "loss": 0.7912,
+      "step": 17037
+    },
+    {
+      "epoch": 0.4679355139930241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1518730639596936e-05,
+      "loss": 0.8221,
+      "step": 17038
+    },
+    {
+      "epoch": 0.4679629782208673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1517851435957168e-05,
+      "loss": 0.9449,
+      "step": 17039
+    },
+    {
+      "epoch": 0.4679904424487106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1516972220307573e-05,
+      "loss": 0.9139,
+      "step": 17040
+    },
+    {
+      "epoch": 0.4680179066765538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1516092992655105e-05,
+      "loss": 0.857,
+      "step": 17041
+    },
+    {
+      "epoch": 0.468045370904397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1515213753006723e-05,
+      "loss": 0.9278,
+      "step": 17042
+    },
+    {
+      "epoch": 0.46807283513224024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1514334501369381e-05,
+      "loss": 1.0064,
+      "step": 17043
+    },
+    {
+      "epoch": 0.4681002993600835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.151345523775004e-05,
+      "loss": 0.8968,
+      "step": 17044
+    },
+    {
+      "epoch": 0.46812776358792674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1512575962155657e-05,
+      "loss": 0.7832,
+      "step": 17045
+    },
+    {
+      "epoch": 0.46815522781576996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.151169667459318e-05,
+      "loss": 0.9377,
+      "step": 17046
+    },
+    {
+      "epoch": 0.4681826920436132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1510817375069579e-05,
+      "loss": 0.8906,
+      "step": 17047
+    },
+    {
+      "epoch": 0.4682101562714564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1509938063591804e-05,
+      "loss": 0.9443,
+      "step": 17048
+    },
+    {
+      "epoch": 0.4682376204992997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1509058740166813e-05,
+      "loss": 0.8895,
+      "step": 17049
+    },
+    {
+      "epoch": 0.4682650847271429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1508179404801567e-05,
+      "loss": 0.901,
+      "step": 17050
+    },
+    {
+      "epoch": 0.4682925489549861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1507300057503021e-05,
+      "loss": 1.002,
+      "step": 17051
+    },
+    {
+      "epoch": 0.46832001318282934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1506420698278131e-05,
+      "loss": 0.9797,
+      "step": 17052
+    },
+    {
+      "epoch": 0.4683474774106726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1505541327133857e-05,
+      "loss": 0.9368,
+      "step": 17053
+    },
+    {
+      "epoch": 0.46837494163851584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.150466194407716e-05,
+      "loss": 0.9537,
+      "step": 17054
+    },
+    {
+      "epoch": 0.46840240586635906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1503782549114992e-05,
+      "loss": 0.8868,
+      "step": 17055
+    },
+    {
+      "epoch": 0.4684298700942023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1502903142254316e-05,
+      "loss": 0.8308,
+      "step": 17056
+    },
+    {
+      "epoch": 0.46845733432204556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1502023723502086e-05,
+      "loss": 0.8673,
+      "step": 17057
+    },
+    {
+      "epoch": 0.4684847985498888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1501144292865265e-05,
+      "loss": 0.9377,
+      "step": 17058
+    },
+    {
+      "epoch": 0.468512262777732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1500264850350806e-05,
+      "loss": 0.8883,
+      "step": 17059
+    },
+    {
+      "epoch": 0.4685397270055752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1499385395965673e-05,
+      "loss": 0.9177,
+      "step": 17060
+    },
+    {
+      "epoch": 0.46856719123341845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1498505929716821e-05,
+      "loss": 0.9631,
+      "step": 17061
+    },
+    {
+      "epoch": 0.4685946554612617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1497626451611207e-05,
+      "loss": 0.9719,
+      "step": 17062
+    },
+    {
+      "epoch": 0.46862211968910494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1496746961655796e-05,
+      "loss": 0.9323,
+      "step": 17063
+    },
+    {
+      "epoch": 0.46864958391694816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1495867459857542e-05,
+      "loss": 0.8813,
+      "step": 17064
+    },
+    {
+      "epoch": 0.4686770481447914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1494987946223406e-05,
+      "loss": 0.9109,
+      "step": 17065
+    },
+    {
+      "epoch": 0.46870451237263466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1494108420760346e-05,
+      "loss": 0.9528,
+      "step": 17066
+    },
+    {
+      "epoch": 0.4687319766004779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1493228883475319e-05,
+      "loss": 0.8661,
+      "step": 17067
+    },
+    {
+      "epoch": 0.4687594408283211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.149234933437529e-05,
+      "loss": 0.9741,
+      "step": 17068
+    },
+    {
+      "epoch": 0.4687869050561643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1491469773467214e-05,
+      "loss": 0.8449,
+      "step": 17069
+    },
+    {
+      "epoch": 0.4688143692840076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.149059020075805e-05,
+      "loss": 0.9727,
+      "step": 17070
+    },
+    {
+      "epoch": 0.4688418335118508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1489710616254759e-05,
+      "loss": 0.9111,
+      "step": 17071
+    },
+    {
+      "epoch": 0.46886929773969405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1488831019964301e-05,
+      "loss": 1.0274,
+      "step": 17072
+    },
+    {
+      "epoch": 0.46889676196753727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1487951411893637e-05,
+      "loss": 0.9051,
+      "step": 17073
+    },
+    {
+      "epoch": 0.46892422619538054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1487071792049722e-05,
+      "loss": 0.8344,
+      "step": 17074
+    },
+    {
+      "epoch": 0.46895169042322377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1486192160439523e-05,
+      "loss": 0.8943,
+      "step": 17075
+    },
+    {
+      "epoch": 0.468979154651067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1485312517069992e-05,
+      "loss": 0.8385,
+      "step": 17076
+    },
+    {
+      "epoch": 0.4690066188789102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1484432861948097e-05,
+      "loss": 0.8377,
+      "step": 17077
+    },
+    {
+      "epoch": 0.46903408310675343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.148355319508079e-05,
+      "loss": 0.8058,
+      "step": 17078
+    },
+    {
+      "epoch": 0.4690615473345967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1482673516475035e-05,
+      "loss": 0.9231,
+      "step": 17079
+    },
+    {
+      "epoch": 0.46908901156243993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1481793826137796e-05,
+      "loss": 0.9568,
+      "step": 17080
+    },
+    {
+      "epoch": 0.46911647579028315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1480914124076033e-05,
+      "loss": 0.962,
+      "step": 17081
+    },
+    {
+      "epoch": 0.46914394001812637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1480034410296696e-05,
+      "loss": 0.8262,
+      "step": 17082
+    },
+    {
+      "epoch": 0.46917140424596965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1479154684806757e-05,
+      "loss": 0.8863,
+      "step": 17083
+    },
+    {
+      "epoch": 0.46919886847381287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1478274947613175e-05,
+      "loss": 0.9689,
+      "step": 17084
+    },
+    {
+      "epoch": 0.4692263327016561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1477395198722909e-05,
+      "loss": 0.971,
+      "step": 17085
+    },
+    {
+      "epoch": 0.4692537969294993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1476515438142917e-05,
+      "loss": 0.9603,
+      "step": 17086
+    },
+    {
+      "epoch": 0.4692812611573426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1475635665880165e-05,
+      "loss": 0.9389,
+      "step": 17087
+    },
+    {
+      "epoch": 0.4693087253851858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.147475588194161e-05,
+      "loss": 0.8711,
+      "step": 17088
+    },
+    {
+      "epoch": 0.46933618961302903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1473876086334217e-05,
+      "loss": 0.9779,
+      "step": 17089
+    },
+    {
+      "epoch": 0.46936365384087225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1472996279064948e-05,
+      "loss": 1.0239,
+      "step": 17090
+    },
+    {
+      "epoch": 0.4693911180687155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1472116460140757e-05,
+      "loss": 1.1042,
+      "step": 17091
+    },
+    {
+      "epoch": 0.46941858229655875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1471236629568613e-05,
+      "loss": 0.8511,
+      "step": 17092
+    },
+    {
+      "epoch": 0.46944604652440197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1470356787355474e-05,
+      "loss": 0.9966,
+      "step": 17093
+    },
+    {
+      "epoch": 0.4694735107522452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1469476933508303e-05,
+      "loss": 0.8909,
+      "step": 17094
+    },
+    {
+      "epoch": 0.4695009749800884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1468597068034061e-05,
+      "loss": 0.8724,
+      "step": 17095
+    },
+    {
+      "epoch": 0.4695284392079317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1467717190939709e-05,
+      "loss": 0.9372,
+      "step": 17096
+    },
+    {
+      "epoch": 0.4695559034357749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1466837302232212e-05,
+      "loss": 0.7937,
+      "step": 17097
+    },
+    {
+      "epoch": 0.46958336766361813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.146595740191853e-05,
+      "loss": 0.8408,
+      "step": 17098
+    },
+    {
+      "epoch": 0.46961083189146136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1465077490005623e-05,
+      "loss": 0.8651,
+      "step": 17099
+    },
+    {
+      "epoch": 0.46963829611930463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1464197566500456e-05,
+      "loss": 0.9249,
+      "step": 17100
+    },
+    {
+      "epoch": 0.46966576034714785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1463317631409996e-05,
+      "loss": 0.8709,
+      "step": 17101
+    },
+    {
+      "epoch": 0.4696932245749911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1462437684741192e-05,
+      "loss": 0.9107,
+      "step": 17102
+    },
+    {
+      "epoch": 0.4697206888028343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1461557726501017e-05,
+      "loss": 0.8752,
+      "step": 17103
+    },
+    {
+      "epoch": 0.4697481530306775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1460677756696435e-05,
+      "loss": 0.9123,
+      "step": 17104
+    },
+    {
+      "epoch": 0.4697756172585208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1459797775334402e-05,
+      "loss": 0.8478,
+      "step": 17105
+    },
+    {
+      "epoch": 0.469803081486364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1458917782421883e-05,
+      "loss": 0.9824,
+      "step": 17106
+    },
+    {
+      "epoch": 0.46983054571420724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1458037777965842e-05,
+      "loss": 0.9784,
+      "step": 17107
+    },
+    {
+      "epoch": 0.46985800994205046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1457157761973241e-05,
+      "loss": 0.9844,
+      "step": 17108
+    },
+    {
+      "epoch": 0.46988547416989374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1456277734451041e-05,
+      "loss": 0.8732,
+      "step": 17109
+    },
+    {
+      "epoch": 0.46991293839773696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1455397695406214e-05,
+      "loss": 0.9022,
+      "step": 17110
+    },
+    {
+      "epoch": 0.4699404026255802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.145451764484571e-05,
+      "loss": 1.0014,
+      "step": 17111
+    },
+    {
+      "epoch": 0.4699678668534234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1453637582776502e-05,
+      "loss": 0.8549,
+      "step": 17112
+    },
+    {
+      "epoch": 0.4699953310812667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1452757509205553e-05,
+      "loss": 0.9425,
+      "step": 17113
+    },
+    {
+      "epoch": 0.4700227953091099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1451877424139819e-05,
+      "loss": 0.9538,
+      "step": 17114
+    },
+    {
+      "epoch": 0.4700502595369531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1450997327586268e-05,
+      "loss": 0.9871,
+      "step": 17115
+    },
+    {
+      "epoch": 0.47007772376479634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1450117219551869e-05,
+      "loss": 0.8475,
+      "step": 17116
+    },
+    {
+      "epoch": 0.47010518799263956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1449237100043578e-05,
+      "loss": 0.7905,
+      "step": 17117
+    },
+    {
+      "epoch": 0.47013265222048284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1448356969068362e-05,
+      "loss": 0.886,
+      "step": 17118
+    },
+    {
+      "epoch": 0.47016011644832606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1447476826633183e-05,
+      "loss": 0.8059,
+      "step": 17119
+    },
+    {
+      "epoch": 0.4701875806761693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1446596672745007e-05,
+      "loss": 0.8541,
+      "step": 17120
+    },
+    {
+      "epoch": 0.4702150449040125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.14457165074108e-05,
+      "loss": 0.8741,
+      "step": 17121
+    },
+    {
+      "epoch": 0.4702425091318558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1444836330637525e-05,
+      "loss": 0.8914,
+      "step": 17122
+    },
+    {
+      "epoch": 0.470269973359699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1443956142432141e-05,
+      "loss": 0.9289,
+      "step": 17123
+    },
+    {
+      "epoch": 0.4702974375875422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.144307594280162e-05,
+      "loss": 0.9045,
+      "step": 17124
+    },
+    {
+      "epoch": 0.47032490181538544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1442195731752924e-05,
+      "loss": 0.9888,
+      "step": 17125
+    },
+    {
+      "epoch": 0.4703523660432287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1441315509293011e-05,
+      "loss": 0.9213,
+      "step": 17126
+    },
+    {
+      "epoch": 0.47037983027107194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1440435275428857e-05,
+      "loss": 0.9226,
+      "step": 17127
+    },
+    {
+      "epoch": 0.47040729449891516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1439555030167417e-05,
+      "loss": 0.9434,
+      "step": 17128
+    },
+    {
+      "epoch": 0.4704347587267584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1438674773515666e-05,
+      "loss": 0.827,
+      "step": 17129
+    },
+    {
+      "epoch": 0.4704622229546016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1437794505480558e-05,
+      "loss": 1.0624,
+      "step": 17130
+    },
+    {
+      "epoch": 0.4704896871824449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1436914226069062e-05,
+      "loss": 0.9754,
+      "step": 17131
+    },
+    {
+      "epoch": 0.4705171514102881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1436033935288146e-05,
+      "loss": 0.8453,
+      "step": 17132
+    },
+    {
+      "epoch": 0.4705446156381313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1435153633144773e-05,
+      "loss": 0.9366,
+      "step": 17133
+    },
+    {
+      "epoch": 0.47057207986597455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.143427331964591e-05,
+      "loss": 0.9162,
+      "step": 17134
+    },
+    {
+      "epoch": 0.4705995440938178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1433392994798519e-05,
+      "loss": 0.9412,
+      "step": 17135
+    },
+    {
+      "epoch": 0.47062700832166104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1432512658609569e-05,
+      "loss": 0.8253,
+      "step": 17136
+    },
+    {
+      "epoch": 0.47065447254950427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1431632311086025e-05,
+      "loss": 0.9826,
+      "step": 17137
+    },
+    {
+      "epoch": 0.4706819367773475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1430751952234846e-05,
+      "loss": 0.9345,
+      "step": 17138
+    },
+    {
+      "epoch": 0.47070940100519076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1429871582063007e-05,
+      "loss": 0.9291,
+      "step": 17139
+    },
+    {
+      "epoch": 0.470736865233034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.142899120057747e-05,
+      "loss": 0.9706,
+      "step": 17140
+    },
+    {
+      "epoch": 0.4707643294608772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1428110807785202e-05,
+      "loss": 0.8918,
+      "step": 17141
+    },
+    {
+      "epoch": 0.47079179368872043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1427230403693165e-05,
+      "loss": 0.8708,
+      "step": 17142
+    },
+    {
+      "epoch": 0.47081925791656365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1426349988308334e-05,
+      "loss": 0.9318,
+      "step": 17143
+    },
+    {
+      "epoch": 0.4708467221444069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1425469561637663e-05,
+      "loss": 0.9177,
+      "step": 17144
+    },
+    {
+      "epoch": 0.47087418637225015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.142458912368813e-05,
+      "loss": 0.8506,
+      "step": 17145
+    },
+    {
+      "epoch": 0.47090165060009337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1423708674466693e-05,
+      "loss": 0.9001,
+      "step": 17146
+    },
+    {
+      "epoch": 0.4709291148279366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1422828213980321e-05,
+      "loss": 0.9141,
+      "step": 17147
+    },
+    {
+      "epoch": 0.47095657905577987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1421947742235983e-05,
+      "loss": 0.9256,
+      "step": 17148
+    },
+    {
+      "epoch": 0.4709840432836231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1421067259240643e-05,
+      "loss": 0.9886,
+      "step": 17149
+    },
+    {
+      "epoch": 0.4710115075114663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1420186765001269e-05,
+      "loss": 0.9276,
+      "step": 17150
+    },
+    {
+      "epoch": 0.47103897173930953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1419306259524822e-05,
+      "loss": 0.9386,
+      "step": 17151
+    },
+    {
+      "epoch": 0.4710664359671528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.141842574281828e-05,
+      "loss": 0.9064,
+      "step": 17152
+    },
+    {
+      "epoch": 0.47109390019499603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1417545214888603e-05,
+      "loss": 0.8948,
+      "step": 17153
+    },
+    {
+      "epoch": 0.47112136442283925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1416664675742758e-05,
+      "loss": 0.8647,
+      "step": 17154
+    },
+    {
+      "epoch": 0.47114882865068247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1415784125387717e-05,
+      "loss": 0.9666,
+      "step": 17155
+    },
+    {
+      "epoch": 0.4711762928785257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1414903563830437e-05,
+      "loss": 0.9078,
+      "step": 17156
+    },
+    {
+      "epoch": 0.47120375710636897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.14140229910779e-05,
+      "loss": 0.9072,
+      "step": 17157
+    },
+    {
+      "epoch": 0.4712312213342122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.141314240713706e-05,
+      "loss": 0.9543,
+      "step": 17158
+    },
+    {
+      "epoch": 0.4712586855620554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1412261812014888e-05,
+      "loss": 0.854,
+      "step": 17159
+    },
+    {
+      "epoch": 0.47128614978989863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1411381205718357e-05,
+      "loss": 0.9096,
+      "step": 17160
+    },
+    {
+      "epoch": 0.4713136140177419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1410500588254431e-05,
+      "loss": 0.9327,
+      "step": 17161
+    },
+    {
+      "epoch": 0.47134107824558513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1409619959630076e-05,
+      "loss": 0.8795,
+      "step": 17162
+    },
+    {
+      "epoch": 0.47136854247342835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1408739319852265e-05,
+      "loss": 0.9309,
+      "step": 17163
+    },
+    {
+      "epoch": 0.4713960067012716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1407858668927958e-05,
+      "loss": 0.9951,
+      "step": 17164
+    },
+    {
+      "epoch": 0.47142347092911485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.140697800686413e-05,
+      "loss": 0.8819,
+      "step": 17165
+    },
+    {
+      "epoch": 0.4714509351569581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1406097333667751e-05,
+      "loss": 0.8607,
+      "step": 17166
+    },
+    {
+      "epoch": 0.4714783993848013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.140521664934578e-05,
+      "loss": 0.8657,
+      "step": 17167
+    },
+    {
+      "epoch": 0.4715058636126445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1404335953905192e-05,
+      "loss": 0.9008,
+      "step": 17168
+    },
+    {
+      "epoch": 0.47153332784048774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1403455247352957e-05,
+      "loss": 0.9147,
+      "step": 17169
+    },
+    {
+      "epoch": 0.471560792068331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1402574529696038e-05,
+      "loss": 0.8467,
+      "step": 17170
+    },
+    {
+      "epoch": 0.47158825629617424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1401693800941404e-05,
+      "loss": 0.9292,
+      "step": 17171
+    },
+    {
+      "epoch": 0.47161572052401746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1400813061096027e-05,
+      "loss": 0.8499,
+      "step": 17172
+    },
+    {
+      "epoch": 0.4716431847518607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1399932310166877e-05,
+      "loss": 0.9757,
+      "step": 17173
+    },
+    {
+      "epoch": 0.47167064897970395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1399051548160917e-05,
+      "loss": 0.8325,
+      "step": 17174
+    },
+    {
+      "epoch": 0.4716981132075472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1398170775085122e-05,
+      "loss": 1.0323,
+      "step": 17175
+    },
+    {
+      "epoch": 0.4717255774353904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1397289990946456e-05,
+      "loss": 0.9908,
+      "step": 17176
+    },
+    {
+      "epoch": 0.4717530416632336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1396409195751892e-05,
+      "loss": 0.8351,
+      "step": 17177
+    },
+    {
+      "epoch": 0.4717805058910769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1395528389508397e-05,
+      "loss": 1.0558,
+      "step": 17178
+    },
+    {
+      "epoch": 0.4718079701189201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.139464757222294e-05,
+      "loss": 0.9301,
+      "step": 17179
+    },
+    {
+      "epoch": 0.47183543434676334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1393766743902491e-05,
+      "loss": 0.8967,
+      "step": 17180
+    },
+    {
+      "epoch": 0.47186289857460656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1392885904554021e-05,
+      "loss": 0.9458,
+      "step": 17181
+    },
+    {
+      "epoch": 0.47189036280244984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1392005054184497e-05,
+      "loss": 0.8745,
+      "step": 17182
+    },
+    {
+      "epoch": 0.47191782703029306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1391124192800888e-05,
+      "loss": 0.8771,
+      "step": 17183
+    },
+    {
+      "epoch": 0.4719452912581363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1390243320410171e-05,
+      "loss": 0.9622,
+      "step": 17184
+    },
+    {
+      "epoch": 0.4719727554859795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1389362437019306e-05,
+      "loss": 1.0222,
+      "step": 17185
+    },
+    {
+      "epoch": 0.4720002197138227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1388481542635268e-05,
+      "loss": 0.991,
+      "step": 17186
+    },
+    {
+      "epoch": 0.472027683941666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1387600637265025e-05,
+      "loss": 0.9287,
+      "step": 17187
+    },
+    {
+      "epoch": 0.4720551481695092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.138671972091555e-05,
+      "loss": 0.8012,
+      "step": 17188
+    },
+    {
+      "epoch": 0.47208261239735244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.138583879359381e-05,
+      "loss": 0.8502,
+      "step": 17189
+    },
+    {
+      "epoch": 0.47211007662519566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.138495785530678e-05,
+      "loss": 0.9312,
+      "step": 17190
+    },
+    {
+      "epoch": 0.47213754085303894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1384076906061422e-05,
+      "loss": 1.0004,
+      "step": 17191
+    },
+    {
+      "epoch": 0.47216500508088216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1383195945864715e-05,
+      "loss": 0.9475,
+      "step": 17192
+    },
+    {
+      "epoch": 0.4721924693087254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1382314974723628e-05,
+      "loss": 0.9476,
+      "step": 17193
+    },
+    {
+      "epoch": 0.4722199335365686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1381433992645125e-05,
+      "loss": 0.9119,
+      "step": 17194
+    },
+    {
+      "epoch": 0.4722473977644119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1380552999636181e-05,
+      "loss": 0.9328,
+      "step": 17195
+    },
+    {
+      "epoch": 0.4722748619922551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.137967199570377e-05,
+      "loss": 0.8412,
+      "step": 17196
+    },
+    {
+      "epoch": 0.4723023262200983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1378790980854859e-05,
+      "loss": 0.9131,
+      "step": 17197
+    },
+    {
+      "epoch": 0.47232979044794154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1377909955096417e-05,
+      "loss": 0.8312,
+      "step": 17198
+    },
+    {
+      "epoch": 0.47235725467578477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.137702891843542e-05,
+      "loss": 1.0102,
+      "step": 17199
+    },
+    {
+      "epoch": 0.47238471890362804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1376147870878835e-05,
+      "loss": 0.9228,
+      "step": 17200
+    },
+    {
+      "epoch": 0.47241218313147126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1375266812433638e-05,
+      "loss": 1.0112,
+      "step": 17201
+    },
+    {
+      "epoch": 0.4724396473593145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1374385743106797e-05,
+      "loss": 0.9063,
+      "step": 17202
+    },
+    {
+      "epoch": 0.4724671115871577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.137350466290528e-05,
+      "loss": 0.9089,
+      "step": 17203
+    },
+    {
+      "epoch": 0.472494575815001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1372623571836066e-05,
+      "loss": 0.9333,
+      "step": 17204
+    },
+    {
+      "epoch": 0.4725220400428442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1371742469906122e-05,
+      "loss": 1.0145,
+      "step": 17205
+    },
+    {
+      "epoch": 0.4725495042706874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1370861357122416e-05,
+      "loss": 0.7898,
+      "step": 17206
+    },
+    {
+      "epoch": 0.47257696849853065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1369980233491926e-05,
+      "loss": 0.9986,
+      "step": 17207
+    },
+    {
+      "epoch": 0.4726044327263739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1369099099021623e-05,
+      "loss": 0.9199,
+      "step": 17208
+    },
+    {
+      "epoch": 0.47263189695421715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1368217953718479e-05,
+      "loss": 0.9348,
+      "step": 17209
+    },
+    {
+      "epoch": 0.47265936118206037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1367336797589461e-05,
+      "loss": 0.8171,
+      "step": 17210
+    },
+    {
+      "epoch": 0.4726868254099036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136645563064155e-05,
+      "loss": 0.9895,
+      "step": 17211
+    },
+    {
+      "epoch": 0.4727142896377468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1365574452881707e-05,
+      "loss": 0.9207,
+      "step": 17212
+    },
+    {
+      "epoch": 0.4727417538655901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1364693264316913e-05,
+      "loss": 0.9304,
+      "step": 17213
+    },
+    {
+      "epoch": 0.4727692180934333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1363812064954136e-05,
+      "loss": 0.9182,
+      "step": 17214
+    },
+    {
+      "epoch": 0.47279668232127653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1362930854800349e-05,
+      "loss": 0.8755,
+      "step": 17215
+    },
+    {
+      "epoch": 0.47282414654911975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1362049633862528e-05,
+      "loss": 0.9118,
+      "step": 17216
+    },
+    {
+      "epoch": 0.472851610776963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1361168402147641e-05,
+      "loss": 0.8797,
+      "step": 17217
+    },
+    {
+      "epoch": 0.47287907500480625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1360287159662661e-05,
+      "loss": 0.8689,
+      "step": 17218
+    },
+    {
+      "epoch": 0.47290653923264947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1359405906414565e-05,
+      "loss": 0.9017,
+      "step": 17219
+    },
+    {
+      "epoch": 0.4729340034604927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1358524642410318e-05,
+      "loss": 0.9377,
+      "step": 17220
+    },
+    {
+      "epoch": 0.47296146768833597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1357643367656903e-05,
+      "loss": 0.8004,
+      "step": 17221
+    },
+    {
+      "epoch": 0.4729889319161792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1356762082161285e-05,
+      "loss": 0.968,
+      "step": 17222
+    },
+    {
+      "epoch": 0.4730163961440224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135588078593044e-05,
+      "loss": 0.9132,
+      "step": 17223
+    },
+    {
+      "epoch": 0.47304386037186563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135499947897134e-05,
+      "loss": 0.884,
+      "step": 17224
+    },
+    {
+      "epoch": 0.47307132459970885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1354118161290966e-05,
+      "loss": 0.9217,
+      "step": 17225
+    },
+    {
+      "epoch": 0.47309878882755213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1353236832896278e-05,
+      "loss": 0.917,
+      "step": 17226
+    },
+    {
+      "epoch": 0.47312625305539535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1352355493794255e-05,
+      "loss": 0.9034,
+      "step": 17227
+    },
+    {
+      "epoch": 0.4731537172832386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1351474143991874e-05,
+      "loss": 0.7986,
+      "step": 17228
+    },
+    {
+      "epoch": 0.4731811815110818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1350592783496109e-05,
+      "loss": 0.9247,
+      "step": 17229
+    },
+    {
+      "epoch": 0.47320864573892507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1349711412313923e-05,
+      "loss": 0.9328,
+      "step": 17230
+    },
+    {
+      "epoch": 0.4732361099667683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1348830030452304e-05,
+      "loss": 0.8576,
+      "step": 17231
+    },
+    {
+      "epoch": 0.4732635741946115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1347948637918215e-05,
+      "loss": 0.8906,
+      "step": 17232
+    },
+    {
+      "epoch": 0.47329103842245474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1347067234718639e-05,
+      "loss": 0.8351,
+      "step": 17233
+    },
+    {
+      "epoch": 0.473318502650298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1346185820860545e-05,
+      "loss": 0.8268,
+      "step": 17234
+    },
+    {
+      "epoch": 0.47334596687814123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1345304396350903e-05,
+      "loss": 0.9384,
+      "step": 17235
+    },
+    {
+      "epoch": 0.47337343110598445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1344422961196693e-05,
+      "loss": 0.8813,
+      "step": 17236
+    },
+    {
+      "epoch": 0.4734008953338277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1343541515404893e-05,
+      "loss": 0.8983,
+      "step": 17237
+    },
+    {
+      "epoch": 0.4734283595616709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1342660058982465e-05,
+      "loss": 0.8364,
+      "step": 17238
+    },
+    {
+      "epoch": 0.4734558237895142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1341778591936392e-05,
+      "loss": 0.9553,
+      "step": 17239
+    },
+    {
+      "epoch": 0.4734832880173574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1340897114273647e-05,
+      "loss": 0.7573,
+      "step": 17240
+    },
+    {
+      "epoch": 0.4735107522452006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1340015626001208e-05,
+      "loss": 0.8762,
+      "step": 17241
+    },
+    {
+      "epoch": 0.47353821647304384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1339134127126043e-05,
+      "loss": 0.8845,
+      "step": 17242
+    },
+    {
+      "epoch": 0.4735656807008871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1338252617655132e-05,
+      "loss": 0.78,
+      "step": 17243
+    },
+    {
+      "epoch": 0.47359314492873034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1337371097595444e-05,
+      "loss": 0.9848,
+      "step": 17244
+    },
+    {
+      "epoch": 0.47362060915657356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1336489566953964e-05,
+      "loss": 0.9117,
+      "step": 17245
+    },
+    {
+      "epoch": 0.4736480733844168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1335608025737657e-05,
+      "loss": 0.9047,
+      "step": 17246
+    },
+    {
+      "epoch": 0.47367553761226006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13347264739535e-05,
+      "loss": 0.975,
+      "step": 17247
+    },
+    {
+      "epoch": 0.4737030018401033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1333844911608474e-05,
+      "loss": 0.9311,
+      "step": 17248
+    },
+    {
+      "epoch": 0.4737304660679465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1332963338709551e-05,
+      "loss": 0.965,
+      "step": 17249
+    },
+    {
+      "epoch": 0.4737579302957897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1332081755263701e-05,
+      "loss": 0.8792,
+      "step": 17250
+    },
+    {
+      "epoch": 0.47378539452363294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1331200161277907e-05,
+      "loss": 0.8892,
+      "step": 17251
+    },
+    {
+      "epoch": 0.4738128587514762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1330318556759144e-05,
+      "loss": 0.906,
+      "step": 17252
+    },
+    {
+      "epoch": 0.47384032297931944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1329436941714382e-05,
+      "loss": 0.7922,
+      "step": 17253
+    },
+    {
+      "epoch": 0.47386778720716266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13285553161506e-05,
+      "loss": 0.8605,
+      "step": 17254
+    },
+    {
+      "epoch": 0.4738952514350059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1327673680074775e-05,
+      "loss": 0.9806,
+      "step": 17255
+    },
+    {
+      "epoch": 0.47392271566284916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1326792033493881e-05,
+      "loss": 1.0204,
+      "step": 17256
+    },
+    {
+      "epoch": 0.4739501798906924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1325910376414894e-05,
+      "loss": 0.9185,
+      "step": 17257
+    },
+    {
+      "epoch": 0.4739776441185356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1325028708844791e-05,
+      "loss": 0.8935,
+      "step": 17258
+    },
+    {
+      "epoch": 0.4740051083463788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1324147030790548e-05,
+      "loss": 0.9055,
+      "step": 17259
+    },
+    {
+      "epoch": 0.4740325725742221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1323265342259139e-05,
+      "loss": 0.8174,
+      "step": 17260
+    },
+    {
+      "epoch": 0.4740600368020653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1322383643257546e-05,
+      "loss": 0.9135,
+      "step": 17261
+    },
+    {
+      "epoch": 0.47408750102990854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1321501933792737e-05,
+      "loss": 0.9143,
+      "step": 17262
+    },
+    {
+      "epoch": 0.47411496525775176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1320620213871693e-05,
+      "loss": 0.903,
+      "step": 17263
+    },
+    {
+      "epoch": 0.474142429485595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1319738483501395e-05,
+      "loss": 0.9355,
+      "step": 17264
+    },
+    {
+      "epoch": 0.47416989371343826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1318856742688812e-05,
+      "loss": 0.7848,
+      "step": 17265
+    },
+    {
+      "epoch": 0.4741973579412815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131797499144092e-05,
+      "loss": 0.8899,
+      "step": 17266
+    },
+    {
+      "epoch": 0.4742248221691247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1317093229764707e-05,
+      "loss": 0.8456,
+      "step": 17267
+    },
+    {
+      "epoch": 0.4742522863969679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1316211457667135e-05,
+      "loss": 0.8547,
+      "step": 17268
+    },
+    {
+      "epoch": 0.4742797506248112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1315329675155191e-05,
+      "loss": 0.9908,
+      "step": 17269
+    },
+    {
+      "epoch": 0.4743072148526544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131444788223585e-05,
+      "loss": 0.9271,
+      "step": 17270
+    },
+    {
+      "epoch": 0.47433467908049765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1313566078916088e-05,
+      "loss": 0.8513,
+      "step": 17271
+    },
+    {
+      "epoch": 0.47436214330834087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131268426520288e-05,
+      "loss": 0.9398,
+      "step": 17272
+    },
+    {
+      "epoch": 0.47438960753618414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131180244110321e-05,
+      "loss": 0.8404,
+      "step": 17273
+    },
+    {
+      "epoch": 0.47441707176402736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1310920606624048e-05,
+      "loss": 0.9555,
+      "step": 17274
+    },
+    {
+      "epoch": 0.4744445359918706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131003876177237e-05,
+      "loss": 0.955,
+      "step": 17275
+    },
+    {
+      "epoch": 0.4744720002197138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1309156906555165e-05,
+      "loss": 0.8152,
+      "step": 17276
+    },
+    {
+      "epoch": 0.4744994644475571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1308275040979401e-05,
+      "loss": 0.8545,
+      "step": 17277
+    },
+    {
+      "epoch": 0.4745269286754003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1307393165052058e-05,
+      "loss": 0.8499,
+      "step": 17278
+    },
+    {
+      "epoch": 0.4745543929032435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1306511278780115e-05,
+      "loss": 0.9858,
+      "step": 17279
+    },
+    {
+      "epoch": 0.47458185713108675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1305629382170544e-05,
+      "loss": 0.9242,
+      "step": 17280
+    },
+    {
+      "epoch": 0.47460932135892997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1304747475230336e-05,
+      "loss": 0.8749,
+      "step": 17281
+    },
+    {
+      "epoch": 0.47463678558677325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1303865557966453e-05,
+      "loss": 0.9567,
+      "step": 17282
+    },
+    {
+      "epoch": 0.47466424981461647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1302983630385882e-05,
+      "loss": 0.8926,
+      "step": 17283
+    },
+    {
+      "epoch": 0.4746917140424597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1302101692495602e-05,
+      "loss": 0.9336,
+      "step": 17284
+    },
+    {
+      "epoch": 0.4747191782703029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1301219744302589e-05,
+      "loss": 0.9294,
+      "step": 17285
+    },
+    {
+      "epoch": 0.4747466424981462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.130033778581382e-05,
+      "loss": 0.8731,
+      "step": 17286
+    },
+    {
+      "epoch": 0.4747741067259894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1299455817036276e-05,
+      "loss": 0.957,
+      "step": 17287
+    },
+    {
+      "epoch": 0.47480157095383263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1298573837976932e-05,
+      "loss": 0.9157,
+      "step": 17288
+    },
+    {
+      "epoch": 0.47482903518167585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.129769184864277e-05,
+      "loss": 0.8547,
+      "step": 17289
+    },
+    {
+      "epoch": 0.47485649940951913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1296809849040769e-05,
+      "loss": 0.9261,
+      "step": 17290
+    },
+    {
+      "epoch": 0.47488396363736235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1295927839177903e-05,
+      "loss": 0.9414,
+      "step": 17291
+    },
+    {
+      "epoch": 0.47491142786520557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1295045819061154e-05,
+      "loss": 0.8425,
+      "step": 17292
+    },
+    {
+      "epoch": 0.4749388920930488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1294163788697505e-05,
+      "loss": 0.941,
+      "step": 17293
+    },
+    {
+      "epoch": 0.474966356320892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1293281748093928e-05,
+      "loss": 0.8978,
+      "step": 17294
+    },
+    {
+      "epoch": 0.4749938205487353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1292399697257403e-05,
+      "loss": 0.9695,
+      "step": 17295
+    },
+    {
+      "epoch": 0.4750212847765785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1291517636194915e-05,
+      "loss": 0.7721,
+      "step": 17296
+    },
+    {
+      "epoch": 0.47504874900442173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1290635564913438e-05,
+      "loss": 0.9234,
+      "step": 17297
+    },
+    {
+      "epoch": 0.47507621323226495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128975348341995e-05,
+      "loss": 1.0773,
+      "step": 17298
+    },
+    {
+      "epoch": 0.47510367746010823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1288871391721434e-05,
+      "loss": 0.9668,
+      "step": 17299
+    },
+    {
+      "epoch": 0.47513114168795145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1287989289824868e-05,
+      "loss": 0.8264,
+      "step": 17300
+    },
+    {
+      "epoch": 0.4751586059157947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1287107177737233e-05,
+      "loss": 0.8448,
+      "step": 17301
+    },
+    {
+      "epoch": 0.4751860701436379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128622505546551e-05,
+      "loss": 0.8429,
+      "step": 17302
+    },
+    {
+      "epoch": 0.47521353437148117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1285342923016671e-05,
+      "loss": 0.927,
+      "step": 17303
+    },
+    {
+      "epoch": 0.4752409985993244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1284460780397701e-05,
+      "loss": 0.9998,
+      "step": 17304
+    },
+    {
+      "epoch": 0.4752684628271676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1283578627615586e-05,
+      "loss": 0.9321,
+      "step": 17305
+    },
+    {
+      "epoch": 0.47529592705501084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1282696464677295e-05,
+      "loss": 0.9027,
+      "step": 17306
+    },
+    {
+      "epoch": 0.47532339128285406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128181429158981e-05,
+      "loss": 0.8651,
+      "step": 17307
+    },
+    {
+      "epoch": 0.47535085551069733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1280932108360118e-05,
+      "loss": 0.9663,
+      "step": 17308
+    },
+    {
+      "epoch": 0.47537831973854056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1280049914995194e-05,
+      "loss": 0.9726,
+      "step": 17309
+    },
+    {
+      "epoch": 0.4754057839663838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.127916771150202e-05,
+      "loss": 0.9463,
+      "step": 17310
+    },
+    {
+      "epoch": 0.475433248194227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1278285497887575e-05,
+      "loss": 0.8485,
+      "step": 17311
+    },
+    {
+      "epoch": 0.4754607124220703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1277403274158839e-05,
+      "loss": 0.901,
+      "step": 17312
+    },
+    {
+      "epoch": 0.4754881766499135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1276521040322794e-05,
+      "loss": 0.8946,
+      "step": 17313
+    },
+    {
+      "epoch": 0.4755156408777567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1275638796386421e-05,
+      "loss": 0.8258,
+      "step": 17314
+    },
+    {
+      "epoch": 0.47554310510559994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1274756542356698e-05,
+      "loss": 0.9194,
+      "step": 17315
+    },
+    {
+      "epoch": 0.4755705693334432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1273874278240607e-05,
+      "loss": 0.8456,
+      "step": 17316
+    },
+    {
+      "epoch": 0.47559803356128644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1272992004045133e-05,
+      "loss": 0.9753,
+      "step": 17317
+    },
+    {
+      "epoch": 0.47562549778912966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1272109719777251e-05,
+      "loss": 0.9789,
+      "step": 17318
+    },
+    {
+      "epoch": 0.4756529620169729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1271227425443943e-05,
+      "loss": 0.9373,
+      "step": 17319
+    },
+    {
+      "epoch": 0.4756804262448161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1270345121052193e-05,
+      "loss": 1.0045,
+      "step": 17320
+    },
+    {
+      "epoch": 0.4757078904726594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.126946280660898e-05,
+      "loss": 0.9545,
+      "step": 17321
+    },
+    {
+      "epoch": 0.4757353547005026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1268580482121283e-05,
+      "loss": 0.8763,
+      "step": 17322
+    },
+    {
+      "epoch": 0.4757628189283458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1267698147596089e-05,
+      "loss": 0.8892,
+      "step": 17323
+    },
+    {
+      "epoch": 0.47579028315618904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1266815803040375e-05,
+      "loss": 0.9646,
+      "step": 17324
+    },
+    {
+      "epoch": 0.4758177473840323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1265933448461123e-05,
+      "loss": 0.894,
+      "step": 17325
+    },
+    {
+      "epoch": 0.47584521161187554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1265051083865317e-05,
+      "loss": 0.912,
+      "step": 17326
+    },
+    {
+      "epoch": 0.47587267583971876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1264168709259935e-05,
+      "loss": 0.8933,
+      "step": 17327
+    },
+    {
+      "epoch": 0.475900140067562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1263286324651958e-05,
+      "loss": 0.9175,
+      "step": 17328
+    },
+    {
+      "epoch": 0.47592760429540526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1262403930048377e-05,
+      "loss": 0.8408,
+      "step": 17329
+    },
+    {
+      "epoch": 0.4759550685232485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1261521525456163e-05,
+      "loss": 1.0057,
+      "step": 17330
+    },
+    {
+      "epoch": 0.4759825327510917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1260639110882302e-05,
+      "loss": 0.8922,
+      "step": 17331
+    },
+    {
+      "epoch": 0.4760099969789349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1259756686333777e-05,
+      "loss": 0.9752,
+      "step": 17332
+    },
+    {
+      "epoch": 0.47603746120677815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1258874251817568e-05,
+      "loss": 0.9077,
+      "step": 17333
+    },
+    {
+      "epoch": 0.4760649254346214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1257991807340658e-05,
+      "loss": 0.8347,
+      "step": 17334
+    },
+    {
+      "epoch": 0.47609238966246464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1257109352910032e-05,
+      "loss": 0.9555,
+      "step": 17335
+    },
+    {
+      "epoch": 0.47611985389030786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1256226888532666e-05,
+      "loss": 0.9298,
+      "step": 17336
+    },
+    {
+      "epoch": 0.4761473181181511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1255344414215553e-05,
+      "loss": 1.004,
+      "step": 17337
+    },
+    {
+      "epoch": 0.47617478234599436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.125446192996566e-05,
+      "loss": 0.9666,
+      "step": 17338
+    },
+    {
+      "epoch": 0.4762022465738376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1253579435789984e-05,
+      "loss": 0.8246,
+      "step": 17339
+    },
+    {
+      "epoch": 0.4762297108016808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1252696931695502e-05,
+      "loss": 0.9041,
+      "step": 17340
+    },
+    {
+      "epoch": 0.476257175029524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1251814417689195e-05,
+      "loss": 0.8827,
+      "step": 17341
+    },
+    {
+      "epoch": 0.4762846392573673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1250931893778045e-05,
+      "loss": 0.9224,
+      "step": 17342
+    },
+    {
+      "epoch": 0.4763121034852105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1250049359969041e-05,
+      "loss": 0.9693,
+      "step": 17343
+    },
+    {
+      "epoch": 0.47633956771305375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1249166816269159e-05,
+      "loss": 0.9663,
+      "step": 17344
+    },
+    {
+      "epoch": 0.47636703194089697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.124828426268539e-05,
+      "loss": 0.8367,
+      "step": 17345
+    },
+    {
+      "epoch": 0.4763944961687402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.124740169922471e-05,
+      "loss": 0.9429,
+      "step": 17346
+    },
+    {
+      "epoch": 0.47642196039658347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1246519125894102e-05,
+      "loss": 0.8396,
+      "step": 17347
+    },
+    {
+      "epoch": 0.4764494246244267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1245636542700553e-05,
+      "loss": 0.9354,
+      "step": 17348
+    },
+    {
+      "epoch": 0.4764768888522699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.124475394965105e-05,
+      "loss": 0.9505,
+      "step": 17349
+    },
+    {
+      "epoch": 0.47650435308011313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1243871346752566e-05,
+      "loss": 0.9026,
+      "step": 17350
+    },
+    {
+      "epoch": 0.4765318173079564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.124298873401209e-05,
+      "loss": 0.9175,
+      "step": 17351
+    },
+    {
+      "epoch": 0.47655928153579963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1242106111436608e-05,
+      "loss": 1.0255,
+      "step": 17352
+    },
+    {
+      "epoch": 0.47658674576364285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1241223479033103e-05,
+      "loss": 0.8767,
+      "step": 17353
+    },
+    {
+      "epoch": 0.47661420999148607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1240340836808554e-05,
+      "loss": 0.9109,
+      "step": 17354
+    },
+    {
+      "epoch": 0.47664167421932935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123945818476995e-05,
+      "loss": 0.9277,
+      "step": 17355
+    },
+    {
+      "epoch": 0.47666913844717257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1238575522924268e-05,
+      "loss": 0.9407,
+      "step": 17356
+    },
+    {
+      "epoch": 0.4766966026750158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.12376928512785e-05,
+      "loss": 0.9789,
+      "step": 17357
+    },
+    {
+      "epoch": 0.476724066902859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123681016983963e-05,
+      "loss": 0.8887,
+      "step": 17358
+    },
+    {
+      "epoch": 0.47675153113070223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1235927478614634e-05,
+      "loss": 0.8964,
+      "step": 17359
+    },
+    {
+      "epoch": 0.4767789953585455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1235044777610502e-05,
+      "loss": 0.8864,
+      "step": 17360
+    },
+    {
+      "epoch": 0.47680645958638873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123416206683422e-05,
+      "loss": 0.9078,
+      "step": 17361
+    },
+    {
+      "epoch": 0.47683392381423195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1233279346292767e-05,
+      "loss": 0.9822,
+      "step": 17362
+    },
+    {
+      "epoch": 0.4768613880420752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1232396615993129e-05,
+      "loss": 0.8604,
+      "step": 17363
+    },
+    {
+      "epoch": 0.47688885226991845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1231513875942295e-05,
+      "loss": 0.8554,
+      "step": 17364
+    },
+    {
+      "epoch": 0.47691631649776167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1230631126147245e-05,
+      "loss": 0.961,
+      "step": 17365
+    },
+    {
+      "epoch": 0.4769437807256049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1229748366614963e-05,
+      "loss": 0.918,
+      "step": 17366
+    },
+    {
+      "epoch": 0.4769712449534481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1228865597352438e-05,
+      "loss": 0.9642,
+      "step": 17367
+    },
+    {
+      "epoch": 0.4769987091812914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.122798281836665e-05,
+      "loss": 0.8961,
+      "step": 17368
+    },
+    {
+      "epoch": 0.4770261734091346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1227100029664588e-05,
+      "loss": 0.8766,
+      "step": 17369
+    },
+    {
+      "epoch": 0.47705363763697783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1226217231253236e-05,
+      "loss": 0.8582,
+      "step": 17370
+    },
+    {
+      "epoch": 0.47708110186482106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1225334423139575e-05,
+      "loss": 1.0295,
+      "step": 17371
+    },
+    {
+      "epoch": 0.47710856609266433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1224451605330595e-05,
+      "loss": 0.8847,
+      "step": 17372
+    },
+    {
+      "epoch": 0.47713603032050755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1223568777833282e-05,
+      "loss": 0.8778,
+      "step": 17373
+    },
+    {
+      "epoch": 0.4771634945483508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1222685940654617e-05,
+      "loss": 0.9427,
+      "step": 17374
+    },
+    {
+      "epoch": 0.477190958776194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1221803093801584e-05,
+      "loss": 0.8903,
+      "step": 17375
+    },
+    {
+      "epoch": 0.4772184230040372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1220920237281177e-05,
+      "loss": 0.9134,
+      "step": 17376
+    },
+    {
+      "epoch": 0.4772458872318805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1220037371100375e-05,
+      "loss": 0.7348,
+      "step": 17377
+    },
+    {
+      "epoch": 0.4772733514597237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.121915449526616e-05,
+      "loss": 0.9302,
+      "step": 17378
+    },
+    {
+      "epoch": 0.47730081568756694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1218271609785529e-05,
+      "loss": 0.8155,
+      "step": 17379
+    },
+    {
+      "epoch": 0.47732827991541016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1217388714665456e-05,
+      "loss": 0.9653,
+      "step": 17380
+    },
+    {
+      "epoch": 0.47735574414325344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1216505809912935e-05,
+      "loss": 0.9435,
+      "step": 17381
+    },
+    {
+      "epoch": 0.47738320837109666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1215622895534948e-05,
+      "loss": 0.9761,
+      "step": 17382
+    },
+    {
+      "epoch": 0.4774106725989399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1214739971538479e-05,
+      "loss": 0.8979,
+      "step": 17383
+    },
+    {
+      "epoch": 0.4774381368267831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.121385703793052e-05,
+      "loss": 0.8696,
+      "step": 17384
+    },
+    {
+      "epoch": 0.4774656010546264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1212974094718056e-05,
+      "loss": 0.9343,
+      "step": 17385
+    },
+    {
+      "epoch": 0.4774930652824696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1212091141908067e-05,
+      "loss": 0.9019,
+      "step": 17386
+    },
+    {
+      "epoch": 0.4775205295103128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1211208179507544e-05,
+      "loss": 0.8098,
+      "step": 17387
+    },
+    {
+      "epoch": 0.47754799373815604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1210325207523474e-05,
+      "loss": 0.9423,
+      "step": 17388
+    },
+    {
+      "epoch": 0.47757545796599926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1209442225962843e-05,
+      "loss": 0.9518,
+      "step": 17389
+    },
+    {
+      "epoch": 0.47760292219384254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1208559234832635e-05,
+      "loss": 0.8613,
+      "step": 17390
+    },
+    {
+      "epoch": 0.47763038642168576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1207676234139838e-05,
+      "loss": 0.8531,
+      "step": 17391
+    },
+    {
+      "epoch": 0.477657850649529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1206793223891437e-05,
+      "loss": 0.9287,
+      "step": 17392
+    },
+    {
+      "epoch": 0.4776853148773722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1205910204094423e-05,
+      "loss": 0.9809,
+      "step": 17393
+    },
+    {
+      "epoch": 0.4777127791052155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1205027174755783e-05,
+      "loss": 0.8722,
+      "step": 17394
+    },
+    {
+      "epoch": 0.4777402433330587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1204144135882496e-05,
+      "loss": 0.8044,
+      "step": 17395
+    },
+    {
+      "epoch": 0.4777677075609019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1203261087481558e-05,
+      "loss": 0.9184,
+      "step": 17396
+    },
+    {
+      "epoch": 0.47779517178874514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.120237802955995e-05,
+      "loss": 0.8902,
+      "step": 17397
+    },
+    {
+      "epoch": 0.4778226360165884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1201494962124662e-05,
+      "loss": 0.9371,
+      "step": 17398
+    },
+    {
+      "epoch": 0.47785010024443164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1200611885182682e-05,
+      "loss": 0.9054,
+      "step": 17399
+    },
+    {
+      "epoch": 0.47787756447227486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1199728798740992e-05,
+      "loss": 0.924,
+      "step": 17400
+    },
+    {
+      "epoch": 0.4779050287001181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1198845702806587e-05,
+      "loss": 0.8168,
+      "step": 17401
+    },
+    {
+      "epoch": 0.4779324929279613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1197962597386449e-05,
+      "loss": 0.8961,
+      "step": 17402
+    },
+    {
+      "epoch": 0.4779599571558046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1197079482487566e-05,
+      "loss": 0.9123,
+      "step": 17403
+    },
+    {
+      "epoch": 0.4779874213836478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1196196358116925e-05,
+      "loss": 0.9084,
+      "step": 17404
+    },
+    {
+      "epoch": 0.478014885611491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1195313224281519e-05,
+      "loss": 0.967,
+      "step": 17405
+    },
+    {
+      "epoch": 0.47804234983933425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1194430080988328e-05,
+      "loss": 0.9196,
+      "step": 17406
+    },
+    {
+      "epoch": 0.4780698140671775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1193546928244343e-05,
+      "loss": 0.8716,
+      "step": 17407
+    },
+    {
+      "epoch": 0.47809727829502074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1192663766056554e-05,
+      "loss": 0.9019,
+      "step": 17408
+    },
+    {
+      "epoch": 0.47812474252286397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1191780594431951e-05,
+      "loss": 0.9495,
+      "step": 17409
+    },
+    {
+      "epoch": 0.4781522067507072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1190897413377513e-05,
+      "loss": 0.973,
+      "step": 17410
+    },
+    {
+      "epoch": 0.47817967097855046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1190014222900237e-05,
+      "loss": 0.8341,
+      "step": 17411
+    },
+    {
+      "epoch": 0.4782071352063937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1189131023007105e-05,
+      "loss": 0.8322,
+      "step": 17412
+    },
+    {
+      "epoch": 0.4782345994342369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1188247813705109e-05,
+      "loss": 0.8954,
+      "step": 17413
+    },
+    {
+      "epoch": 0.47826206366208013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1187364595001235e-05,
+      "loss": 0.8698,
+      "step": 17414
+    },
+    {
+      "epoch": 0.47828952788992335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1186481366902474e-05,
+      "loss": 0.858,
+      "step": 17415
+    },
+    {
+      "epoch": 0.4783169921177666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1185598129415809e-05,
+      "loss": 1.013,
+      "step": 17416
+    },
+    {
+      "epoch": 0.47834445634560985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1184714882548237e-05,
+      "loss": 0.8552,
+      "step": 17417
+    },
+    {
+      "epoch": 0.47837192057345307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.118383162630674e-05,
+      "loss": 0.8721,
+      "step": 17418
+    },
+    {
+      "epoch": 0.4783993848012963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1182948360698309e-05,
+      "loss": 0.8662,
+      "step": 17419
+    },
+    {
+      "epoch": 0.47842684902913957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1182065085729931e-05,
+      "loss": 0.9047,
+      "step": 17420
+    },
+    {
+      "epoch": 0.4784543132569828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1181181801408599e-05,
+      "loss": 0.8925,
+      "step": 17421
+    },
+    {
+      "epoch": 0.478481777484826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1180298507741296e-05,
+      "loss": 0.9169,
+      "step": 17422
+    },
+    {
+      "epoch": 0.47850924171266923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1179415204735015e-05,
+      "loss": 0.994,
+      "step": 17423
+    },
+    {
+      "epoch": 0.4785367059405125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1178531892396745e-05,
+      "loss": 0.8851,
+      "step": 17424
+    },
+    {
+      "epoch": 0.47856417016835573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1177648570733472e-05,
+      "loss": 0.993,
+      "step": 17425
+    },
+    {
+      "epoch": 0.47859163439619895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1176765239752192e-05,
+      "loss": 0.8666,
+      "step": 17426
+    },
+    {
+      "epoch": 0.47861909862404217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1175881899459884e-05,
+      "loss": 0.7865,
+      "step": 17427
+    },
+    {
+      "epoch": 0.4786465628518854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1174998549863545e-05,
+      "loss": 0.809,
+      "step": 17428
+    },
+    {
+      "epoch": 0.47867402707972867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1174115190970166e-05,
+      "loss": 0.804,
+      "step": 17429
+    },
+    {
+      "epoch": 0.4787014913075719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1173231822786727e-05,
+      "loss": 0.9086,
+      "step": 17430
+    },
+    {
+      "epoch": 0.4787289555354151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1172348445320224e-05,
+      "loss": 0.8312,
+      "step": 17431
+    },
+    {
+      "epoch": 0.47875641976325833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1171465058577648e-05,
+      "loss": 0.944,
+      "step": 17432
+    },
+    {
+      "epoch": 0.4787838839911016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1170581662565989e-05,
+      "loss": 0.968,
+      "step": 17433
+    },
+    {
+      "epoch": 0.47881134821894483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.116969825729223e-05,
+      "loss": 0.9857,
+      "step": 17434
+    },
+    {
+      "epoch": 0.47883881244678805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1168814842763366e-05,
+      "loss": 0.802,
+      "step": 17435
+    },
+    {
+      "epoch": 0.4788662766746313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1167931418986389e-05,
+      "loss": 0.9312,
+      "step": 17436
+    },
+    {
+      "epoch": 0.47889374090247455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1167047985968284e-05,
+      "loss": 0.8342,
+      "step": 17437
+    },
+    {
+      "epoch": 0.4789212051303178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1166164543716044e-05,
+      "loss": 0.9297,
+      "step": 17438
+    },
+    {
+      "epoch": 0.478948669358161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1165281092236658e-05,
+      "loss": 0.9399,
+      "step": 17439
+    },
+    {
+      "epoch": 0.4789761335860042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1164397631537116e-05,
+      "loss": 0.9429,
+      "step": 17440
+    },
+    {
+      "epoch": 0.47900359781384744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1163514161624414e-05,
+      "loss": 0.9031,
+      "step": 17441
+    },
+    {
+      "epoch": 0.4790310620416907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1162630682505531e-05,
+      "loss": 0.8284,
+      "step": 17442
+    },
+    {
+      "epoch": 0.47905852626953394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1161747194187464e-05,
+      "loss": 0.9348,
+      "step": 17443
+    },
+    {
+      "epoch": 0.47908599049737716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1160863696677205e-05,
+      "loss": 0.9539,
+      "step": 17444
+    },
+    {
+      "epoch": 0.4791134547252204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1159980189981743e-05,
+      "loss": 0.8632,
+      "step": 17445
+    },
+    {
+      "epoch": 0.47914091895306365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1159096674108068e-05,
+      "loss": 0.9066,
+      "step": 17446
+    },
+    {
+      "epoch": 0.4791683831809069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.115821314906317e-05,
+      "loss": 0.8956,
+      "step": 17447
+    },
+    {
+      "epoch": 0.4791958474087501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1157329614854042e-05,
+      "loss": 0.8912,
+      "step": 17448
+    },
+    {
+      "epoch": 0.4792233116365933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1156446071487675e-05,
+      "loss": 0.8574,
+      "step": 17449
+    },
+    {
+      "epoch": 0.4792507758644366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1155562518971059e-05,
+      "loss": 0.8668,
+      "step": 17450
+    },
+    {
+      "epoch": 0.4792782400922798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.115467895731118e-05,
+      "loss": 0.8912,
+      "step": 17451
+    },
+    {
+      "epoch": 0.47930570432012304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1153795386515037e-05,
+      "loss": 0.9216,
+      "step": 17452
+    },
+    {
+      "epoch": 0.47933316854796626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.115291180658962e-05,
+      "loss": 1.0022,
+      "step": 17453
+    },
+    {
+      "epoch": 0.4793606327758095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1152028217541918e-05,
+      "loss": 1.0054,
+      "step": 17454
+    },
+    {
+      "epoch": 0.47938809700365276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1151144619378917e-05,
+      "loss": 0.8493,
+      "step": 17455
+    },
+    {
+      "epoch": 0.479415561231496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.115026101210762e-05,
+      "loss": 0.8967,
+      "step": 17456
+    },
+    {
+      "epoch": 0.4794430254593392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114937739573501e-05,
+      "loss": 0.903,
+      "step": 17457
+    },
+    {
+      "epoch": 0.4794704896871824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114849377026808e-05,
+      "loss": 0.9619,
+      "step": 17458
+    },
+    {
+      "epoch": 0.4794979539150257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1147610135713822e-05,
+      "loss": 0.7692,
+      "step": 17459
+    },
+    {
+      "epoch": 0.4795254181428689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114672649207923e-05,
+      "loss": 0.937,
+      "step": 17460
+    },
+    {
+      "epoch": 0.47955288237071214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1145842839371293e-05,
+      "loss": 0.8448,
+      "step": 17461
+    },
+    {
+      "epoch": 0.47958034659855536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1144959177597004e-05,
+      "loss": 1.0303,
+      "step": 17462
+    },
+    {
+      "epoch": 0.47960781082639864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1144075506763352e-05,
+      "loss": 0.9937,
+      "step": 17463
+    },
+    {
+      "epoch": 0.47963527505424186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1143191826877335e-05,
+      "loss": 0.8745,
+      "step": 17464
+    },
+    {
+      "epoch": 0.4796627392820851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1142308137945941e-05,
+      "loss": 0.8926,
+      "step": 17465
+    },
+    {
+      "epoch": 0.4796902035099283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114142443997616e-05,
+      "loss": 0.9075,
+      "step": 17466
+    },
+    {
+      "epoch": 0.4797176677377715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114054073297499e-05,
+      "loss": 0.902,
+      "step": 17467
+    },
+    {
+      "epoch": 0.4797451319656148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1139657016949417e-05,
+      "loss": 0.832,
+      "step": 17468
+    },
+    {
+      "epoch": 0.479772596193458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1138773291906438e-05,
+      "loss": 0.7757,
+      "step": 17469
+    },
+    {
+      "epoch": 0.47980006042130124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1137889557853043e-05,
+      "loss": 0.9263,
+      "step": 17470
+    },
+    {
+      "epoch": 0.47982752464914447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1137005814796224e-05,
+      "loss": 0.9273,
+      "step": 17471
+    },
+    {
+      "epoch": 0.47985498887698774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1136122062742973e-05,
+      "loss": 0.8493,
+      "step": 17472
+    },
+    {
+      "epoch": 0.47988245310483096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.113523830170029e-05,
+      "loss": 0.9019,
+      "step": 17473
+    },
+    {
+      "epoch": 0.4799099173326742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1134354531675158e-05,
+      "loss": 0.8544,
+      "step": 17474
+    },
+    {
+      "epoch": 0.4799373815605174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1133470752674572e-05,
+      "loss": 0.9097,
+      "step": 17475
+    },
+    {
+      "epoch": 0.4799648457883607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.113258696470553e-05,
+      "loss": 0.9067,
+      "step": 17476
+    },
+    {
+      "epoch": 0.4799923100162039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1131703167775018e-05,
+      "loss": 0.9322,
+      "step": 17477
+    },
+    {
+      "epoch": 0.4800197742440471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1130819361890033e-05,
+      "loss": 0.9637,
+      "step": 17478
+    },
+    {
+      "epoch": 0.48004723847189035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1129935547057568e-05,
+      "loss": 0.9479,
+      "step": 17479
+    },
+    {
+      "epoch": 0.4800747026997336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1129051723284616e-05,
+      "loss": 0.9449,
+      "step": 17480
+    },
+    {
+      "epoch": 0.48010216692757685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112816789057817e-05,
+      "loss": 0.9212,
+      "step": 17481
+    },
+    {
+      "epoch": 0.48012963115542007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1127284048945222e-05,
+      "loss": 0.8883,
+      "step": 17482
+    },
+    {
+      "epoch": 0.4801570953832633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1126400198392762e-05,
+      "loss": 0.7973,
+      "step": 17483
+    },
+    {
+      "epoch": 0.4801845596111065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1125516338927788e-05,
+      "loss": 0.8112,
+      "step": 17484
+    },
+    {
+      "epoch": 0.4802120238389498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.11246324705573e-05,
+      "loss": 0.8804,
+      "step": 17485
+    },
+    {
+      "epoch": 0.480239488066793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1123748593288278e-05,
+      "loss": 0.8806,
+      "step": 17486
+    },
+    {
+      "epoch": 0.48026695229463623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112286470712772e-05,
+      "loss": 0.9498,
+      "step": 17487
+    },
+    {
+      "epoch": 0.48029441652247945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1121980812082626e-05,
+      "loss": 0.8723,
+      "step": 17488
+    },
+    {
+      "epoch": 0.4803218807503227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1121096908159985e-05,
+      "loss": 0.9761,
+      "step": 17489
+    },
+    {
+      "epoch": 0.48034934497816595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112021299536679e-05,
+      "loss": 0.9043,
+      "step": 17490
+    },
+    {
+      "epoch": 0.48037680920600917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1119329073710037e-05,
+      "loss": 0.9158,
+      "step": 17491
+    },
+    {
+      "epoch": 0.4804042734338524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1118445143196715e-05,
+      "loss": 0.9658,
+      "step": 17492
+    },
+    {
+      "epoch": 0.48043173766169567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1117561203833827e-05,
+      "loss": 0.8901,
+      "step": 17493
+    },
+    {
+      "epoch": 0.4804592018895389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1116677255628359e-05,
+      "loss": 0.9948,
+      "step": 17494
+    },
+    {
+      "epoch": 0.4804866661173821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1115793298587305e-05,
+      "loss": 0.9713,
+      "step": 17495
+    },
+    {
+      "epoch": 0.48051413034522533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1114909332717664e-05,
+      "loss": 0.8976,
+      "step": 17496
+    },
+    {
+      "epoch": 0.48054159457306855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1114025358026435e-05,
+      "loss": 0.9499,
+      "step": 17497
+    },
+    {
+      "epoch": 0.48056905880091183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1113141374520599e-05,
+      "loss": 1.0544,
+      "step": 17498
+    },
+    {
+      "epoch": 0.48059652302875505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1112257382207157e-05,
+      "loss": 0.9149,
+      "step": 17499
+    },
+    {
+      "epoch": 0.4806239872565983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1111373381093107e-05,
+      "loss": 0.8703,
+      "step": 17500
+    },
+    {
+      "epoch": 0.4806514514844415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.111048937118544e-05,
+      "loss": 0.8467,
+      "step": 17501
+    },
+    {
+      "epoch": 0.48067891571228477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1109605352491148e-05,
+      "loss": 0.8764,
+      "step": 17502
+    },
+    {
+      "epoch": 0.480706379940128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1108721325017231e-05,
+      "loss": 1.0196,
+      "step": 17503
+    },
+    {
+      "epoch": 0.4807338441679712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.110783728877068e-05,
+      "loss": 0.9374,
+      "step": 17504
+    },
+    {
+      "epoch": 0.48076130839581444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1106953243758493e-05,
+      "loss": 1.0284,
+      "step": 17505
+    },
+    {
+      "epoch": 0.4807887726236577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1106069189987663e-05,
+      "loss": 0.868,
+      "step": 17506
+    },
+    {
+      "epoch": 0.48081623685150093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1105185127465183e-05,
+      "loss": 0.9426,
+      "step": 17507
+    },
+    {
+      "epoch": 0.48084370107934415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.110430105619805e-05,
+      "loss": 0.9624,
+      "step": 17508
+    },
+    {
+      "epoch": 0.4808711653071874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1103416976193264e-05,
+      "loss": 0.9153,
+      "step": 17509
+    },
+    {
+      "epoch": 0.4808986295350306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1102532887457812e-05,
+      "loss": 0.8475,
+      "step": 17510
+    },
+    {
+      "epoch": 0.4809260937628739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1101648789998692e-05,
+      "loss": 0.9515,
+      "step": 17511
+    },
+    {
+      "epoch": 0.4809535579907171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1100764683822902e-05,
+      "loss": 0.904,
+      "step": 17512
+    },
+    {
+      "epoch": 0.4809810222185603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1099880568937435e-05,
+      "loss": 0.8011,
+      "step": 17513
+    },
+    {
+      "epoch": 0.48100848644640354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1098996445349286e-05,
+      "loss": 0.8697,
+      "step": 17514
+    },
+    {
+      "epoch": 0.4810359506742468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1098112313065452e-05,
+      "loss": 0.8417,
+      "step": 17515
+    },
+    {
+      "epoch": 0.48106341490209004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1097228172092927e-05,
+      "loss": 0.9804,
+      "step": 17516
+    },
+    {
+      "epoch": 0.48109087912993326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1096344022438707e-05,
+      "loss": 1.0244,
+      "step": 17517
+    },
+    {
+      "epoch": 0.4811183433577765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1095459864109792e-05,
+      "loss": 0.8568,
+      "step": 17518
+    },
+    {
+      "epoch": 0.48114580758561976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109457569711317e-05,
+      "loss": 0.8249,
+      "step": 17519
+    },
+    {
+      "epoch": 0.481173271813463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1093691521455845e-05,
+      "loss": 0.9163,
+      "step": 17520
+    },
+    {
+      "epoch": 0.4812007360413062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109280733714481e-05,
+      "loss": 0.8831,
+      "step": 17521
+    },
+    {
+      "epoch": 0.4812282002691494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1091923144187053e-05,
+      "loss": 1.0063,
+      "step": 17522
+    },
+    {
+      "epoch": 0.48125566449699264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1091038942589582e-05,
+      "loss": 0.9235,
+      "step": 17523
+    },
+    {
+      "epoch": 0.4812831287248359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1090154732359389e-05,
+      "loss": 0.9189,
+      "step": 17524
+    },
+    {
+      "epoch": 0.48131059295267914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1089270513503467e-05,
+      "loss": 0.8263,
+      "step": 17525
+    },
+    {
+      "epoch": 0.48133805718052236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1088386286028819e-05,
+      "loss": 0.9019,
+      "step": 17526
+    },
+    {
+      "epoch": 0.4813655214083656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1087502049942432e-05,
+      "loss": 0.942,
+      "step": 17527
+    },
+    {
+      "epoch": 0.48139298563620886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1086617805251307e-05,
+      "loss": 0.8943,
+      "step": 17528
+    },
+    {
+      "epoch": 0.4814204498640521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1085733551962445e-05,
+      "loss": 0.9715,
+      "step": 17529
+    },
+    {
+      "epoch": 0.4814479140918953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1084849290082838e-05,
+      "loss": 0.9045,
+      "step": 17530
+    },
+    {
+      "epoch": 0.4814753783197385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.108396501961948e-05,
+      "loss": 0.9144,
+      "step": 17531
+    },
+    {
+      "epoch": 0.4815028425475818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1083080740579375e-05,
+      "loss": 0.989,
+      "step": 17532
+    },
+    {
+      "epoch": 0.481530306775425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1082196452969513e-05,
+      "loss": 0.9099,
+      "step": 17533
+    },
+    {
+      "epoch": 0.48155777100326824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1081312156796894e-05,
+      "loss": 0.9388,
+      "step": 17534
+    },
+    {
+      "epoch": 0.48158523523111146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1080427852068516e-05,
+      "loss": 0.9451,
+      "step": 17535
+    },
+    {
+      "epoch": 0.4816126994589547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1079543538791372e-05,
+      "loss": 1.0034,
+      "step": 17536
+    },
+    {
+      "epoch": 0.48164016368679796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1078659216972462e-05,
+      "loss": 0.9301,
+      "step": 17537
+    },
+    {
+      "epoch": 0.4816676279146412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1077774886618784e-05,
+      "loss": 1.0634,
+      "step": 17538
+    },
+    {
+      "epoch": 0.4816950921424844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1076890547737331e-05,
+      "loss": 0.9622,
+      "step": 17539
+    },
+    {
+      "epoch": 0.4817225563703276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1076006200335103e-05,
+      "loss": 0.806,
+      "step": 17540
+    },
+    {
+      "epoch": 0.4817500205981709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10751218444191e-05,
+      "loss": 0.8207,
+      "step": 17541
+    },
+    {
+      "epoch": 0.4817774848260141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1074237479996315e-05,
+      "loss": 0.9315,
+      "step": 17542
+    },
+    {
+      "epoch": 0.48180494905385735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1073353107073746e-05,
+      "loss": 1.0109,
+      "step": 17543
+    },
+    {
+      "epoch": 0.48183241328170057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1072468725658391e-05,
+      "loss": 0.8585,
+      "step": 17544
+    },
+    {
+      "epoch": 0.48185987750954384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107158433575725e-05,
+      "loss": 0.8947,
+      "step": 17545
+    },
+    {
+      "epoch": 0.48188734173738706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1070699937377318e-05,
+      "loss": 0.8887,
+      "step": 17546
+    },
+    {
+      "epoch": 0.4819148059652303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1069815530525595e-05,
+      "loss": 0.9787,
+      "step": 17547
+    },
+    {
+      "epoch": 0.4819422701930735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1068931115209071e-05,
+      "loss": 0.9026,
+      "step": 17548
+    },
+    {
+      "epoch": 0.48196973442091673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1068046691434757e-05,
+      "loss": 0.8178,
+      "step": 17549
+    },
+    {
+      "epoch": 0.48199719864876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.106716225920964e-05,
+      "loss": 0.8544,
+      "step": 17550
+    },
+    {
+      "epoch": 0.4820246628766032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1066277818540722e-05,
+      "loss": 0.9606,
+      "step": 17551
+    },
+    {
+      "epoch": 0.48205212710444645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1065393369435002e-05,
+      "loss": 0.9619,
+      "step": 17552
+    },
+    {
+      "epoch": 0.48207959133228967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1064508911899478e-05,
+      "loss": 0.9821,
+      "step": 17553
+    },
+    {
+      "epoch": 0.48210705556013295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1063624445941145e-05,
+      "loss": 0.9411,
+      "step": 17554
+    },
+    {
+      "epoch": 0.48213451978797617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1062739971567004e-05,
+      "loss": 0.812,
+      "step": 17555
+    },
+    {
+      "epoch": 0.4821619840158194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1061855488784054e-05,
+      "loss": 0.8674,
+      "step": 17556
+    },
+    {
+      "epoch": 0.4821894482436626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1060970997599292e-05,
+      "loss": 0.8564,
+      "step": 17557
+    },
+    {
+      "epoch": 0.4822169124715059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1060086498019712e-05,
+      "loss": 0.8763,
+      "step": 17558
+    },
+    {
+      "epoch": 0.4822443766993491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1059201990052321e-05,
+      "loss": 0.8436,
+      "step": 17559
+    },
+    {
+      "epoch": 0.48227184092719233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1058317473704115e-05,
+      "loss": 0.8311,
+      "step": 17560
+    },
+    {
+      "epoch": 0.48229930515503555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.105743294898209e-05,
+      "loss": 0.8609,
+      "step": 17561
+    },
+    {
+      "epoch": 0.4823267693828788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1056548415893244e-05,
+      "loss": 0.8487,
+      "step": 17562
+    },
+    {
+      "epoch": 0.48235423361072205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1055663874444577e-05,
+      "loss": 0.9284,
+      "step": 17563
+    },
+    {
+      "epoch": 0.48238169783856527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.105477932464309e-05,
+      "loss": 0.804,
+      "step": 17564
+    },
+    {
+      "epoch": 0.4824091620664085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1053894766495784e-05,
+      "loss": 0.8889,
+      "step": 17565
+    },
+    {
+      "epoch": 0.4824366262942517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1053010200009649e-05,
+      "loss": 0.9393,
+      "step": 17566
+    },
+    {
+      "epoch": 0.482464090522095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.105212562519169e-05,
+      "loss": 0.9072,
+      "step": 17567
+    },
+    {
+      "epoch": 0.4824915547499382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1051241042048908e-05,
+      "loss": 0.9226,
+      "step": 17568
+    },
+    {
+      "epoch": 0.48251901897778143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10503564505883e-05,
+      "loss": 0.9467,
+      "step": 17569
+    },
+    {
+      "epoch": 0.48254648320562465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1049471850816862e-05,
+      "loss": 0.9179,
+      "step": 17570
+    },
+    {
+      "epoch": 0.48257394743346793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10485872427416e-05,
+      "loss": 0.9609,
+      "step": 17571
+    },
+    {
+      "epoch": 0.48260141166131115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1047702626369506e-05,
+      "loss": 0.9412,
+      "step": 17572
+    },
+    {
+      "epoch": 0.4826288758891544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1046818001707584e-05,
+      "loss": 0.9473,
+      "step": 17573
+    },
+    {
+      "epoch": 0.4826563401169976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1045933368762833e-05,
+      "loss": 0.8515,
+      "step": 17574
+    },
+    {
+      "epoch": 0.48268380434484087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1045048727542251e-05,
+      "loss": 0.7984,
+      "step": 17575
+    },
+    {
+      "epoch": 0.4827112685726841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1044164078052837e-05,
+      "loss": 0.8488,
+      "step": 17576
+    },
+    {
+      "epoch": 0.4827387328005273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1043279420301598e-05,
+      "loss": 0.7602,
+      "step": 17577
+    },
+    {
+      "epoch": 0.48276619702837054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1042394754295524e-05,
+      "loss": 0.8943,
+      "step": 17578
+    },
+    {
+      "epoch": 0.48279366125621376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.104151008004162e-05,
+      "loss": 0.9506,
+      "step": 17579
+    },
+    {
+      "epoch": 0.48282112548405703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1040625397546885e-05,
+      "loss": 0.9413,
+      "step": 17580
+    },
+    {
+      "epoch": 0.48284858971190026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1039740706818318e-05,
+      "loss": 0.892,
+      "step": 17581
+    },
+    {
+      "epoch": 0.4828760539397435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1038856007862922e-05,
+      "loss": 0.8511,
+      "step": 17582
+    },
+    {
+      "epoch": 0.4829035181675867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1037971300687691e-05,
+      "loss": 0.9957,
+      "step": 17583
+    },
+    {
+      "epoch": 0.48293098239543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1037086585299627e-05,
+      "loss": 0.8168,
+      "step": 17584
+    },
+    {
+      "epoch": 0.4829584466232732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.103620186170574e-05,
+      "loss": 0.9257,
+      "step": 17585
+    },
+    {
+      "epoch": 0.4829859108511164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1035317129913018e-05,
+      "loss": 0.8622,
+      "step": 17586
+    },
+    {
+      "epoch": 0.48301337507895964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1034432389928464e-05,
+      "loss": 1.0065,
+      "step": 17587
+    },
+    {
+      "epoch": 0.4830408393068029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1033547641759085e-05,
+      "loss": 0.9091,
+      "step": 17588
+    },
+    {
+      "epoch": 0.48306830353464614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1032662885411876e-05,
+      "loss": 0.8931,
+      "step": 17589
+    },
+    {
+      "epoch": 0.48309576776248936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1031778120893834e-05,
+      "loss": 0.8356,
+      "step": 17590
+    },
+    {
+      "epoch": 0.4831232319903326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1030893348211969e-05,
+      "loss": 0.8678,
+      "step": 17591
+    },
+    {
+      "epoch": 0.4831506962181758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1030008567373271e-05,
+      "loss": 0.8255,
+      "step": 17592
+    },
+    {
+      "epoch": 0.4831781604460191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.102912377838475e-05,
+      "loss": 0.937,
+      "step": 17593
+    },
+    {
+      "epoch": 0.4832056246738623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1028238981253403e-05,
+      "loss": 0.874,
+      "step": 17594
+    },
+    {
+      "epoch": 0.4832330889017055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.102735417598623e-05,
+      "loss": 0.853,
+      "step": 17595
+    },
+    {
+      "epoch": 0.48326055312954874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1026469362590231e-05,
+      "loss": 0.8855,
+      "step": 17596
+    },
+    {
+      "epoch": 0.483288017357392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1025584541072412e-05,
+      "loss": 0.9616,
+      "step": 17597
+    },
+    {
+      "epoch": 0.48331548158523524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1024699711439771e-05,
+      "loss": 0.8761,
+      "step": 17598
+    },
+    {
+      "epoch": 0.48334294581307846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1023814873699306e-05,
+      "loss": 0.8204,
+      "step": 17599
+    },
+    {
+      "epoch": 0.4833704100409217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1022930027858024e-05,
+      "loss": 1.0172,
+      "step": 17600
+    },
+    {
+      "epoch": 0.48339787426876496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1022045173922922e-05,
+      "loss": 0.9451,
+      "step": 17601
+    },
+    {
+      "epoch": 0.4834253384966082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1021160311901e-05,
+      "loss": 0.9732,
+      "step": 17602
+    },
+    {
+      "epoch": 0.4834528027244514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1020275441799267e-05,
+      "loss": 1.0397,
+      "step": 17603
+    },
+    {
+      "epoch": 0.4834802669522946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1019390563624714e-05,
+      "loss": 0.8825,
+      "step": 17604
+    },
+    {
+      "epoch": 0.48350773118013785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1018505677384352e-05,
+      "loss": 0.9197,
+      "step": 17605
+    },
+    {
+      "epoch": 0.4835351954079811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.101762078308518e-05,
+      "loss": 1.0104,
+      "step": 17606
+    },
+    {
+      "epoch": 0.48356265963582434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1016735880734193e-05,
+      "loss": 0.9761,
+      "step": 17607
+    },
+    {
+      "epoch": 0.48359012386366756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10158509703384e-05,
+      "loss": 0.9154,
+      "step": 17608
+    },
+    {
+      "epoch": 0.4836175880915108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1014966051904802e-05,
+      "loss": 0.9061,
+      "step": 17609
+    },
+    {
+      "epoch": 0.48364505231935406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10140811254404e-05,
+      "loss": 0.7865,
+      "step": 17610
+    },
+    {
+      "epoch": 0.4836725165471973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1013196190952193e-05,
+      "loss": 0.9346,
+      "step": 17611
+    },
+    {
+      "epoch": 0.4836999807750405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1012311248447187e-05,
+      "loss": 0.8648,
+      "step": 17612
+    },
+    {
+      "epoch": 0.4837274450028837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1011426297932384e-05,
+      "loss": 0.9344,
+      "step": 17613
+    },
+    {
+      "epoch": 0.483754909230727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.101054133941478e-05,
+      "loss": 0.9227,
+      "step": 17614
+    },
+    {
+      "epoch": 0.4837823734585702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1009656372901383e-05,
+      "loss": 0.8839,
+      "step": 17615
+    },
+    {
+      "epoch": 0.48380983768641345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1008771398399193e-05,
+      "loss": 0.8146,
+      "step": 17616
+    },
+    {
+      "epoch": 0.48383730191425667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1007886415915214e-05,
+      "loss": 0.9142,
+      "step": 17617
+    },
+    {
+      "epoch": 0.4838647661420999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100700142545645e-05,
+      "loss": 1.0508,
+      "step": 17618
+    },
+    {
+      "epoch": 0.48389223036994317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1006116427029895e-05,
+      "loss": 0.9366,
+      "step": 17619
+    },
+    {
+      "epoch": 0.4839196945977864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100523142064256e-05,
+      "loss": 0.8618,
+      "step": 17620
+    },
+    {
+      "epoch": 0.4839471588256296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1004346406301445e-05,
+      "loss": 0.9496,
+      "step": 17621
+    },
+    {
+      "epoch": 0.48397462305347283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1003461384013555e-05,
+      "loss": 0.906,
+      "step": 17622
+    },
+    {
+      "epoch": 0.4840020872813161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1002576353785884e-05,
+      "loss": 0.9006,
+      "step": 17623
+    },
+    {
+      "epoch": 0.48402955150915933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1001691315625444e-05,
+      "loss": 0.8395,
+      "step": 17624
+    },
+    {
+      "epoch": 0.48405701573700255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1000806269539234e-05,
+      "loss": 0.9244,
+      "step": 17625
+    },
+    {
+      "epoch": 0.48408447996484577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0999921215534256e-05,
+      "loss": 0.8803,
+      "step": 17626
+    },
+    {
+      "epoch": 0.48411194419268905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0999036153617515e-05,
+      "loss": 0.8695,
+      "step": 17627
+    },
+    {
+      "epoch": 0.48413940842053227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.099815108379601e-05,
+      "loss": 0.9265,
+      "step": 17628
+    },
+    {
+      "epoch": 0.4841668726483755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0997266006076752e-05,
+      "loss": 0.9028,
+      "step": 17629
+    },
+    {
+      "epoch": 0.4841943368762187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0996380920466739e-05,
+      "loss": 0.9311,
+      "step": 17630
+    },
+    {
+      "epoch": 0.48422180110406193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0995495826972969e-05,
+      "loss": 0.8614,
+      "step": 17631
+    },
+    {
+      "epoch": 0.4842492653319052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0994610725602452e-05,
+      "loss": 0.863,
+      "step": 17632
+    },
+    {
+      "epoch": 0.48427672955974843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0993725616362194e-05,
+      "loss": 0.8992,
+      "step": 17633
+    },
+    {
+      "epoch": 0.48430419378759165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.099284049925919e-05,
+      "loss": 0.8752,
+      "step": 17634
+    },
+    {
+      "epoch": 0.4843316580154349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0991955374300447e-05,
+      "loss": 0.8758,
+      "step": 17635
+    },
+    {
+      "epoch": 0.48435912224327815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.099107024149297e-05,
+      "loss": 0.8453,
+      "step": 17636
+    },
+    {
+      "epoch": 0.48438658647112137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0990185100843763e-05,
+      "loss": 0.8863,
+      "step": 17637
+    },
+    {
+      "epoch": 0.4844140506989646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0989299952359826e-05,
+      "loss": 0.9305,
+      "step": 17638
+    },
+    {
+      "epoch": 0.4844415149268078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0988414796048164e-05,
+      "loss": 0.992,
+      "step": 17639
+    },
+    {
+      "epoch": 0.4844689791546511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0987529631915781e-05,
+      "loss": 0.9615,
+      "step": 17640
+    },
+    {
+      "epoch": 0.4844964433824943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0986644459969685e-05,
+      "loss": 0.9076,
+      "step": 17641
+    },
+    {
+      "epoch": 0.48452390761033753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0985759280216873e-05,
+      "loss": 0.9096,
+      "step": 17642
+    },
+    {
+      "epoch": 0.48455137183818076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.098487409266435e-05,
+      "loss": 0.9161,
+      "step": 17643
+    },
+    {
+      "epoch": 0.484578836066024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0983988897319124e-05,
+      "loss": 0.9194,
+      "step": 17644
+    },
+    {
+      "epoch": 0.48460630029386725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0983103694188198e-05,
+      "loss": 1.0367,
+      "step": 17645
+    },
+    {
+      "epoch": 0.4846337645217105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.098221848327857e-05,
+      "loss": 0.88,
+      "step": 17646
+    },
+    {
+      "epoch": 0.4846612287495537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0981333264597253e-05,
+      "loss": 0.8544,
+      "step": 17647
+    },
+    {
+      "epoch": 0.4846886929773969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0980448038151245e-05,
+      "loss": 0.9612,
+      "step": 17648
+    },
+    {
+      "epoch": 0.4847161572052402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0979562803947554e-05,
+      "loss": 0.9151,
+      "step": 17649
+    },
+    {
+      "epoch": 0.4847436214330834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0978677561993183e-05,
+      "loss": 0.8963,
+      "step": 17650
+    },
+    {
+      "epoch": 0.48477108566092664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0977792312295132e-05,
+      "loss": 0.933,
+      "step": 17651
+    },
+    {
+      "epoch": 0.48479854988876986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0976907054860412e-05,
+      "loss": 0.9662,
+      "step": 17652
+    },
+    {
+      "epoch": 0.48482601411661314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0976021789696025e-05,
+      "loss": 0.934,
+      "step": 17653
+    },
+    {
+      "epoch": 0.48485347834445636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0975136516808977e-05,
+      "loss": 0.9019,
+      "step": 17654
+    },
+    {
+      "epoch": 0.4848809425722996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0974251236206271e-05,
+      "loss": 0.8989,
+      "step": 17655
+    },
+    {
+      "epoch": 0.4849084068001428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.097336594789491e-05,
+      "loss": 0.9608,
+      "step": 17656
+    },
+    {
+      "epoch": 0.484935871027986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0972480651881903e-05,
+      "loss": 0.9455,
+      "step": 17657
+    },
+    {
+      "epoch": 0.4849633352558293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0971595348174248e-05,
+      "loss": 0.8202,
+      "step": 17658
+    },
+    {
+      "epoch": 0.4849907994836725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.097071003677896e-05,
+      "loss": 0.9161,
+      "step": 17659
+    },
+    {
+      "epoch": 0.48501826371151574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0969824717703032e-05,
+      "loss": 0.8806,
+      "step": 17660
+    },
+    {
+      "epoch": 0.48504572793935896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0968939390953479e-05,
+      "loss": 0.886,
+      "step": 17661
+    },
+    {
+      "epoch": 0.48507319216720224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0968054056537303e-05,
+      "loss": 0.8875,
+      "step": 17662
+    },
+    {
+      "epoch": 0.48510065639504546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0967168714461506e-05,
+      "loss": 0.8548,
+      "step": 17663
+    },
+    {
+      "epoch": 0.4851281206228887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0966283364733094e-05,
+      "loss": 0.8706,
+      "step": 17664
+    },
+    {
+      "epoch": 0.4851555848507319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0965398007359078e-05,
+      "loss": 0.9384,
+      "step": 17665
+    },
+    {
+      "epoch": 0.4851830490785752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0964512642346459e-05,
+      "loss": 0.8326,
+      "step": 17666
+    },
+    {
+      "epoch": 0.4852105133064184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0963627269702238e-05,
+      "loss": 0.9406,
+      "step": 17667
+    },
+    {
+      "epoch": 0.4852379775342616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0962741889433427e-05,
+      "loss": 0.9316,
+      "step": 17668
+    },
+    {
+      "epoch": 0.48526544176210484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0961856501547033e-05,
+      "loss": 0.8897,
+      "step": 17669
+    },
+    {
+      "epoch": 0.4852929059899481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0960971106050051e-05,
+      "loss": 0.9081,
+      "step": 17670
+    },
+    {
+      "epoch": 0.48532037021779134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0960085702949496e-05,
+      "loss": 0.9128,
+      "step": 17671
+    },
+    {
+      "epoch": 0.48534783444563456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0959200292252372e-05,
+      "loss": 0.8269,
+      "step": 17672
+    },
+    {
+      "epoch": 0.4853752986734778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0958314873965683e-05,
+      "loss": 0.9093,
+      "step": 17673
+    },
+    {
+      "epoch": 0.485402762901321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0957429448096437e-05,
+      "loss": 0.8974,
+      "step": 17674
+    },
+    {
+      "epoch": 0.4854302271291643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0956544014651635e-05,
+      "loss": 0.8788,
+      "step": 17675
+    },
+    {
+      "epoch": 0.4854576913570075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0955658573638289e-05,
+      "loss": 0.893,
+      "step": 17676
+    },
+    {
+      "epoch": 0.4854851555848507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0954773125063403e-05,
+      "loss": 0.953,
+      "step": 17677
+    },
+    {
+      "epoch": 0.48551261981269395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.095388766893398e-05,
+      "loss": 0.8072,
+      "step": 17678
+    },
+    {
+      "epoch": 0.4855400840405372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0953002205257029e-05,
+      "loss": 0.8866,
+      "step": 17679
+    },
+    {
+      "epoch": 0.48556754826838044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0952116734039555e-05,
+      "loss": 1.0099,
+      "step": 17680
+    },
+    {
+      "epoch": 0.48559501249622367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0951231255288565e-05,
+      "loss": 0.9105,
+      "step": 17681
+    },
+    {
+      "epoch": 0.4856224767240669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0950345769011066e-05,
+      "loss": 0.9027,
+      "step": 17682
+    },
+    {
+      "epoch": 0.48564994095191016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.094946027521406e-05,
+      "loss": 0.7926,
+      "step": 17683
+    },
+    {
+      "epoch": 0.4856774051797534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0948574773904559e-05,
+      "loss": 0.9261,
+      "step": 17684
+    },
+    {
+      "epoch": 0.4857048694075966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0947689265089568e-05,
+      "loss": 1.001,
+      "step": 17685
+    },
+    {
+      "epoch": 0.48573233363543983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0946803748776092e-05,
+      "loss": 0.9546,
+      "step": 17686
+    },
+    {
+      "epoch": 0.48575979786328305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0945918224971135e-05,
+      "loss": 0.8965,
+      "step": 17687
+    },
+    {
+      "epoch": 0.4857872620911263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0945032693681708e-05,
+      "loss": 0.9161,
+      "step": 17688
+    },
+    {
+      "epoch": 0.48581472631896955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0944147154914817e-05,
+      "loss": 0.8133,
+      "step": 17689
+    },
+    {
+      "epoch": 0.48584219054681277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0943261608677467e-05,
+      "loss": 0.789,
+      "step": 17690
+    },
+    {
+      "epoch": 0.485869654774656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0942376054976666e-05,
+      "loss": 0.8345,
+      "step": 17691
+    },
+    {
+      "epoch": 0.48589711900249927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0941490493819422e-05,
+      "loss": 0.8279,
+      "step": 17692
+    },
+    {
+      "epoch": 0.4859245832303425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.094060492521274e-05,
+      "loss": 0.8456,
+      "step": 17693
+    },
+    {
+      "epoch": 0.4859520474581857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0939719349163627e-05,
+      "loss": 0.9213,
+      "step": 17694
+    },
+    {
+      "epoch": 0.48597951168602893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0938833765679091e-05,
+      "loss": 0.7934,
+      "step": 17695
+    },
+    {
+      "epoch": 0.4860069759138722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0937948174766135e-05,
+      "loss": 0.9926,
+      "step": 17696
+    },
+    {
+      "epoch": 0.48603444014171543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0937062576431774e-05,
+      "loss": 0.9791,
+      "step": 17697
+    },
+    {
+      "epoch": 0.48606190436955865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0936176970683011e-05,
+      "loss": 0.8961,
+      "step": 17698
+    },
+    {
+      "epoch": 0.48608936859740187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0935291357526851e-05,
+      "loss": 0.9427,
+      "step": 17699
+    },
+    {
+      "epoch": 0.4861168328252451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0934405736970306e-05,
+      "loss": 0.9392,
+      "step": 17700
+    },
+    {
+      "epoch": 0.48614429705308837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0933520109020378e-05,
+      "loss": 0.8895,
+      "step": 17701
+    },
+    {
+      "epoch": 0.4861717612809316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.093263447368408e-05,
+      "loss": 0.9063,
+      "step": 17702
+    },
+    {
+      "epoch": 0.4861992255087748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0931748830968414e-05,
+      "loss": 0.8914,
+      "step": 17703
+    },
+    {
+      "epoch": 0.48622668973661803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.093086318088039e-05,
+      "loss": 0.8147,
+      "step": 17704
+    },
+    {
+      "epoch": 0.4862541539644613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092997752342702e-05,
+      "loss": 0.953,
+      "step": 17705
+    },
+    {
+      "epoch": 0.48628161819230453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0929091858615307e-05,
+      "loss": 0.8259,
+      "step": 17706
+    },
+    {
+      "epoch": 0.48630908242014775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0928206186452255e-05,
+      "loss": 0.9744,
+      "step": 17707
+    },
+    {
+      "epoch": 0.486336546647991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0927320506944876e-05,
+      "loss": 0.9062,
+      "step": 17708
+    },
+    {
+      "epoch": 0.48636401087583425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0926434820100182e-05,
+      "loss": 0.8898,
+      "step": 17709
+    },
+    {
+      "epoch": 0.4863914751036775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0925549125925176e-05,
+      "loss": 0.9271,
+      "step": 17710
+    },
+    {
+      "epoch": 0.4864189393315207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0924663424426864e-05,
+      "loss": 0.9533,
+      "step": 17711
+    },
+    {
+      "epoch": 0.4864464035593639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0923777715612261e-05,
+      "loss": 0.9156,
+      "step": 17712
+    },
+    {
+      "epoch": 0.48647386778720714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092289199948837e-05,
+      "loss": 0.7749,
+      "step": 17713
+    },
+    {
+      "epoch": 0.4865013320150504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0922006276062198e-05,
+      "loss": 0.9195,
+      "step": 17714
+    },
+    {
+      "epoch": 0.48652879624289364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0921120545340758e-05,
+      "loss": 0.9361,
+      "step": 17715
+    },
+    {
+      "epoch": 0.48655626047073686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0920234807331052e-05,
+      "loss": 0.8745,
+      "step": 17716
+    },
+    {
+      "epoch": 0.4865837246985801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0919349062040093e-05,
+      "loss": 0.9488,
+      "step": 17717
+    },
+    {
+      "epoch": 0.48661118892642335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.091846330947489e-05,
+      "loss": 0.9108,
+      "step": 17718
+    },
+    {
+      "epoch": 0.4866386531542666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0917577549642447e-05,
+      "loss": 0.8786,
+      "step": 17719
+    },
+    {
+      "epoch": 0.4866661173821098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0916691782549773e-05,
+      "loss": 0.8637,
+      "step": 17720
+    },
+    {
+      "epoch": 0.486693581609953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0915806008203881e-05,
+      "loss": 0.8259,
+      "step": 17721
+    },
+    {
+      "epoch": 0.4867210458377963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.091492022661178e-05,
+      "loss": 0.8995,
+      "step": 17722
+    },
+    {
+      "epoch": 0.4867485100656395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0914034437780472e-05,
+      "loss": 0.8772,
+      "step": 17723
+    },
+    {
+      "epoch": 0.48677597429348274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.091314864171697e-05,
+      "loss": 0.934,
+      "step": 17724
+    },
+    {
+      "epoch": 0.48680343852132596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0912262838428285e-05,
+      "loss": 0.9394,
+      "step": 17725
+    },
+    {
+      "epoch": 0.4868309027491692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0911377027921419e-05,
+      "loss": 0.9519,
+      "step": 17726
+    },
+    {
+      "epoch": 0.48685836697701246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.091049121020339e-05,
+      "loss": 0.9737,
+      "step": 17727
+    },
+    {
+      "epoch": 0.4868858312048557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0909605385281195e-05,
+      "loss": 0.8819,
+      "step": 17728
+    },
+    {
+      "epoch": 0.4869132954326989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0908719553161856e-05,
+      "loss": 0.8115,
+      "step": 17729
+    },
+    {
+      "epoch": 0.4869407596605421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0907833713852374e-05,
+      "loss": 0.9692,
+      "step": 17730
+    },
+    {
+      "epoch": 0.4869682238883854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0906947867359758e-05,
+      "loss": 0.9156,
+      "step": 17731
+    },
+    {
+      "epoch": 0.4869956881162286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0906062013691019e-05,
+      "loss": 0.9023,
+      "step": 17732
+    },
+    {
+      "epoch": 0.48702315234407184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0905176152853168e-05,
+      "loss": 0.8646,
+      "step": 17733
+    },
+    {
+      "epoch": 0.48705061657191506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0904290284853214e-05,
+      "loss": 0.9297,
+      "step": 17734
+    },
+    {
+      "epoch": 0.48707808079975834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0903404409698162e-05,
+      "loss": 0.9321,
+      "step": 17735
+    },
+    {
+      "epoch": 0.48710554502760156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0902518527395028e-05,
+      "loss": 0.9334,
+      "step": 17736
+    },
+    {
+      "epoch": 0.4871330092554448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0901632637950816e-05,
+      "loss": 0.888,
+      "step": 17737
+    },
+    {
+      "epoch": 0.487160473483288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0900746741372536e-05,
+      "loss": 0.9123,
+      "step": 17738
+    },
+    {
+      "epoch": 0.4871879377111312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.08998608376672e-05,
+      "loss": 1.0027,
+      "step": 17739
+    },
+    {
+      "epoch": 0.4872154019389745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0898974926841817e-05,
+      "loss": 0.8811,
+      "step": 17740
+    },
+    {
+      "epoch": 0.4872428661668177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0898089008903394e-05,
+      "loss": 0.9117,
+      "step": 17741
+    },
+    {
+      "epoch": 0.48727033039466094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0897203083858946e-05,
+      "loss": 0.8776,
+      "step": 17742
+    },
+    {
+      "epoch": 0.48729779462250417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0896317151715477e-05,
+      "loss": 0.9464,
+      "step": 17743
+    },
+    {
+      "epoch": 0.48732525885034744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.089543121248e-05,
+      "loss": 0.9725,
+      "step": 17744
+    },
+    {
+      "epoch": 0.48735272307819066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0894545266159526e-05,
+      "loss": 0.9124,
+      "step": 17745
+    },
+    {
+      "epoch": 0.4873801873060339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0893659312761063e-05,
+      "loss": 0.9551,
+      "step": 17746
+    },
+    {
+      "epoch": 0.4874076515338771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0892773352291619e-05,
+      "loss": 0.8932,
+      "step": 17747
+    },
+    {
+      "epoch": 0.4874351157617204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0891887384758209e-05,
+      "loss": 0.8291,
+      "step": 17748
+    },
+    {
+      "epoch": 0.4874625799895636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0891001410167842e-05,
+      "loss": 0.9207,
+      "step": 17749
+    },
+    {
+      "epoch": 0.4874900442174068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0890115428527521e-05,
+      "loss": 0.9608,
+      "step": 17750
+    },
+    {
+      "epoch": 0.48751750844525005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0889229439844267e-05,
+      "loss": 0.8813,
+      "step": 17751
+    },
+    {
+      "epoch": 0.48754497267309327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0888343444125083e-05,
+      "loss": 0.8785,
+      "step": 17752
+    },
+    {
+      "epoch": 0.48757243690093655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0887457441376981e-05,
+      "loss": 0.7888,
+      "step": 17753
+    },
+    {
+      "epoch": 0.48759990112877977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0886571431606974e-05,
+      "loss": 0.9771,
+      "step": 17754
+    },
+    {
+      "epoch": 0.487627365356623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0885685414822069e-05,
+      "loss": 0.8899,
+      "step": 17755
+    },
+    {
+      "epoch": 0.4876548295844662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0884799391029277e-05,
+      "loss": 0.9719,
+      "step": 17756
+    },
+    {
+      "epoch": 0.4876822938123095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0883913360235611e-05,
+      "loss": 0.9179,
+      "step": 17757
+    },
+    {
+      "epoch": 0.4877097580401527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0883027322448082e-05,
+      "loss": 0.9221,
+      "step": 17758
+    },
+    {
+      "epoch": 0.48773722226799593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0882141277673696e-05,
+      "loss": 1.0201,
+      "step": 17759
+    },
+    {
+      "epoch": 0.48776468649583915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0881255225919468e-05,
+      "loss": 0.8552,
+      "step": 17760
+    },
+    {
+      "epoch": 0.4877921507236824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0880369167192407e-05,
+      "loss": 0.8908,
+      "step": 17761
+    },
+    {
+      "epoch": 0.48781961495152565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0879483101499524e-05,
+      "loss": 0.8878,
+      "step": 17762
+    },
+    {
+      "epoch": 0.48784707917936887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0878597028847828e-05,
+      "loss": 0.8431,
+      "step": 17763
+    },
+    {
+      "epoch": 0.4878745434072121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0877710949244332e-05,
+      "loss": 0.9235,
+      "step": 17764
+    },
+    {
+      "epoch": 0.48790200763505537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.087682486269605e-05,
+      "loss": 0.9279,
+      "step": 17765
+    },
+    {
+      "epoch": 0.4879294718628986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0875938769209989e-05,
+      "loss": 1.0016,
+      "step": 17766
+    },
+    {
+      "epoch": 0.4879569360907418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.087505266879316e-05,
+      "loss": 0.9633,
+      "step": 17767
+    },
+    {
+      "epoch": 0.48798440031858503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0874166561452577e-05,
+      "loss": 0.9481,
+      "step": 17768
+    },
+    {
+      "epoch": 0.48801186454642825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0873280447195247e-05,
+      "loss": 0.9509,
+      "step": 17769
+    },
+    {
+      "epoch": 0.48803932877427153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0872394326028187e-05,
+      "loss": 0.8774,
+      "step": 17770
+    },
+    {
+      "epoch": 0.48806679300211475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0871508197958403e-05,
+      "loss": 0.9458,
+      "step": 17771
+    },
+    {
+      "epoch": 0.488094257229958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0870622062992908e-05,
+      "loss": 0.9822,
+      "step": 17772
+    },
+    {
+      "epoch": 0.4881217214578012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0869735921138716e-05,
+      "loss": 0.8446,
+      "step": 17773
+    },
+    {
+      "epoch": 0.48814918568564447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0868849772402837e-05,
+      "loss": 0.8943,
+      "step": 17774
+    },
+    {
+      "epoch": 0.4881766499134877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.086796361679228e-05,
+      "loss": 0.8986,
+      "step": 17775
+    },
+    {
+      "epoch": 0.4882041141413309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0867077454314056e-05,
+      "loss": 0.8223,
+      "step": 17776
+    },
+    {
+      "epoch": 0.48823157836917414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0866191284975185e-05,
+      "loss": 0.9176,
+      "step": 17777
+    },
+    {
+      "epoch": 0.4882590425970174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.086530510878267e-05,
+      "loss": 0.9682,
+      "step": 17778
+    },
+    {
+      "epoch": 0.48828650682486063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0864418925743524e-05,
+      "loss": 0.891,
+      "step": 17779
+    },
+    {
+      "epoch": 0.48831397105270385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0863532735864767e-05,
+      "loss": 0.833,
+      "step": 17780
+    },
+    {
+      "epoch": 0.4883414352805471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.08626465391534e-05,
+      "loss": 0.9372,
+      "step": 17781
+    },
+    {
+      "epoch": 0.4883688995083903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0861760335616438e-05,
+      "loss": 0.8808,
+      "step": 17782
+    },
+    {
+      "epoch": 0.4883963637362336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0860874125260898e-05,
+      "loss": 0.905,
+      "step": 17783
+    },
+    {
+      "epoch": 0.4884238279640768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0859987908093786e-05,
+      "loss": 0.91,
+      "step": 17784
+    },
+    {
+      "epoch": 0.48845129219192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0859101684122118e-05,
+      "loss": 0.9326,
+      "step": 17785
+    },
+    {
+      "epoch": 0.48847875641976324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0858215453352904e-05,
+      "loss": 0.9055,
+      "step": 17786
+    },
+    {
+      "epoch": 0.4885062206476065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0857329215793157e-05,
+      "loss": 0.8516,
+      "step": 17787
+    },
+    {
+      "epoch": 0.48853368487544974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0856442971449888e-05,
+      "loss": 0.8796,
+      "step": 17788
+    },
+    {
+      "epoch": 0.48856114910329296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0855556720330111e-05,
+      "loss": 0.8376,
+      "step": 17789
+    },
+    {
+      "epoch": 0.4885886133311362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.085467046244084e-05,
+      "loss": 0.9547,
+      "step": 17790
+    },
+    {
+      "epoch": 0.48861607755897946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0853784197789082e-05,
+      "loss": 0.9258,
+      "step": 17791
+    },
+    {
+      "epoch": 0.4886435417868227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0852897926381856e-05,
+      "loss": 0.8751,
+      "step": 17792
+    },
+    {
+      "epoch": 0.4886710060146659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.085201164822617e-05,
+      "loss": 0.8618,
+      "step": 17793
+    },
+    {
+      "epoch": 0.4886984702425091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0851125363329037e-05,
+      "loss": 0.9892,
+      "step": 17794
+    },
+    {
+      "epoch": 0.48872593447035234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.085023907169747e-05,
+      "loss": 0.9467,
+      "step": 17795
+    },
+    {
+      "epoch": 0.4887533986981956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0849352773338482e-05,
+      "loss": 0.8038,
+      "step": 17796
+    },
+    {
+      "epoch": 0.48878086292603884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0848466468259087e-05,
+      "loss": 0.863,
+      "step": 17797
+    },
+    {
+      "epoch": 0.48880832715388206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0847580156466297e-05,
+      "loss": 0.9081,
+      "step": 17798
+    },
+    {
+      "epoch": 0.4888357913817253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0846693837967123e-05,
+      "loss": 0.8699,
+      "step": 17799
+    },
+    {
+      "epoch": 0.48886325560956856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084580751276858e-05,
+      "loss": 0.9266,
+      "step": 17800
+    },
+    {
+      "epoch": 0.4888907198374118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0844921180877681e-05,
+      "loss": 0.9647,
+      "step": 17801
+    },
+    {
+      "epoch": 0.488918184065255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084403484230144e-05,
+      "loss": 0.9922,
+      "step": 17802
+    },
+    {
+      "epoch": 0.4889456482930982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0843148497046863e-05,
+      "loss": 0.9085,
+      "step": 17803
+    },
+    {
+      "epoch": 0.4889731125209415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0842262145120975e-05,
+      "loss": 0.8896,
+      "step": 17804
+    },
+    {
+      "epoch": 0.4890005767487847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0841375786530779e-05,
+      "loss": 0.966,
+      "step": 17805
+    },
+    {
+      "epoch": 0.48902804097662794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084048942128329e-05,
+      "loss": 0.9016,
+      "step": 17806
+    },
+    {
+      "epoch": 0.48905550520447116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0839603049385528e-05,
+      "loss": 0.8415,
+      "step": 17807
+    },
+    {
+      "epoch": 0.4890829694323144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0838716670844496e-05,
+      "loss": 0.9228,
+      "step": 17808
+    },
+    {
+      "epoch": 0.48911043366015766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0837830285667217e-05,
+      "loss": 0.8806,
+      "step": 17809
+    },
+    {
+      "epoch": 0.4891378978880009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836943893860698e-05,
+      "loss": 0.9383,
+      "step": 17810
+    },
+    {
+      "epoch": 0.4891653621158441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836057495431955e-05,
+      "loss": 0.9117,
+      "step": 17811
+    },
+    {
+      "epoch": 0.4891928263436873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0835171090388002e-05,
+      "loss": 0.9251,
+      "step": 17812
+    },
+    {
+      "epoch": 0.4892202905715306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0834284678735851e-05,
+      "loss": 0.9148,
+      "step": 17813
+    },
+    {
+      "epoch": 0.4892477547993738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083339826048252e-05,
+      "loss": 0.9104,
+      "step": 17814
+    },
+    {
+      "epoch": 0.48927521902721705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0832511835635013e-05,
+      "loss": 0.8861,
+      "step": 17815
+    },
+    {
+      "epoch": 0.48930268325506027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0831625404200355e-05,
+      "loss": 0.9713,
+      "step": 17816
+    },
+    {
+      "epoch": 0.48933014748290354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0830738966185551e-05,
+      "loss": 0.8947,
+      "step": 17817
+    },
+    {
+      "epoch": 0.48935761171074676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0829852521597624e-05,
+      "loss": 0.9605,
+      "step": 17818
+    },
+    {
+      "epoch": 0.48938507593859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0828966070443576e-05,
+      "loss": 0.8264,
+      "step": 17819
+    },
+    {
+      "epoch": 0.4894125401664332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0828079612730429e-05,
+      "loss": 0.8651,
+      "step": 17820
+    },
+    {
+      "epoch": 0.48944000439427643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0827193148465199e-05,
+      "loss": 0.9445,
+      "step": 17821
+    },
+    {
+      "epoch": 0.4894674686221197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0826306677654894e-05,
+      "loss": 0.8578,
+      "step": 17822
+    },
+    {
+      "epoch": 0.4894949328499629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0825420200306529e-05,
+      "loss": 0.8073,
+      "step": 17823
+    },
+    {
+      "epoch": 0.48952239707780615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.082453371642712e-05,
+      "loss": 0.8418,
+      "step": 17824
+    },
+    {
+      "epoch": 0.48954986130564937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.082364722602368e-05,
+      "loss": 0.9986,
+      "step": 17825
+    },
+    {
+      "epoch": 0.48957732553349265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0822760729103228e-05,
+      "loss": 0.8988,
+      "step": 17826
+    },
+    {
+      "epoch": 0.48960478976133587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0821874225672773e-05,
+      "loss": 0.9175,
+      "step": 17827
+    },
+    {
+      "epoch": 0.4896322539891791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0820987715739328e-05,
+      "loss": 1.0377,
+      "step": 17828
+    },
+    {
+      "epoch": 0.4896597182170223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0820101199309913e-05,
+      "loss": 0.9298,
+      "step": 17829
+    },
+    {
+      "epoch": 0.4896871824448656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0819214676391541e-05,
+      "loss": 0.8736,
+      "step": 17830
+    },
+    {
+      "epoch": 0.4897146466727088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0818328146991222e-05,
+      "loss": 1.0497,
+      "step": 17831
+    },
+    {
+      "epoch": 0.48974211090055203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.081744161111597e-05,
+      "loss": 0.8704,
+      "step": 17832
+    },
+    {
+      "epoch": 0.48976957512839525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0816555068772812e-05,
+      "loss": 0.9742,
+      "step": 17833
+    },
+    {
+      "epoch": 0.4897970393562385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.081566851996875e-05,
+      "loss": 0.8019,
+      "step": 17834
+    },
+    {
+      "epoch": 0.48982450358408175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.08147819647108e-05,
+      "loss": 1.0268,
+      "step": 17835
+    },
+    {
+      "epoch": 0.48985196781192497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0813895403005982e-05,
+      "loss": 0.8029,
+      "step": 17836
+    },
+    {
+      "epoch": 0.4898794320397682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0813008834861309e-05,
+      "loss": 0.9222,
+      "step": 17837
+    },
+    {
+      "epoch": 0.4899068962676114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0812122260283795e-05,
+      "loss": 0.7912,
+      "step": 17838
+    },
+    {
+      "epoch": 0.4899343604954547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0811235679280455e-05,
+      "loss": 0.8228,
+      "step": 17839
+    },
+    {
+      "epoch": 0.4899618247232979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0810349091858304e-05,
+      "loss": 0.8791,
+      "step": 17840
+    },
+    {
+      "epoch": 0.48998928895114113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0809462498024356e-05,
+      "loss": 0.8538,
+      "step": 17841
+    },
+    {
+      "epoch": 0.49001675317898435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0808575897785628e-05,
+      "loss": 0.9708,
+      "step": 17842
+    },
+    {
+      "epoch": 0.49004421740682763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0807689291149135e-05,
+      "loss": 0.899,
+      "step": 17843
+    },
+    {
+      "epoch": 0.49007168163467085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080680267812189e-05,
+      "loss": 0.877,
+      "step": 17844
+    },
+    {
+      "epoch": 0.4900991458625141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080591605871091e-05,
+      "loss": 0.8337,
+      "step": 17845
+    },
+    {
+      "epoch": 0.4901266100903573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0805029432923212e-05,
+      "loss": 0.812,
+      "step": 17846
+    },
+    {
+      "epoch": 0.4901540743182005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0804142800765807e-05,
+      "loss": 0.9419,
+      "step": 17847
+    },
+    {
+      "epoch": 0.4901815385460438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0803256162245714e-05,
+      "loss": 0.8446,
+      "step": 17848
+    },
+    {
+      "epoch": 0.490209002773887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0802369517369944e-05,
+      "loss": 0.8454,
+      "step": 17849
+    },
+    {
+      "epoch": 0.49023646700173024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0801482866145518e-05,
+      "loss": 0.9328,
+      "step": 17850
+    },
+    {
+      "epoch": 0.49026393122957346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080059620857945e-05,
+      "loss": 0.9466,
+      "step": 17851
+    },
+    {
+      "epoch": 0.49029139545741673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0799709544678754e-05,
+      "loss": 0.9659,
+      "step": 17852
+    },
+    {
+      "epoch": 0.49031885968525996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0798822874450447e-05,
+      "loss": 0.9437,
+      "step": 17853
+    },
+    {
+      "epoch": 0.4903463239131032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0797936197901544e-05,
+      "loss": 0.9001,
+      "step": 17854
+    },
+    {
+      "epoch": 0.4903737881409464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0797049515039057e-05,
+      "loss": 0.883,
+      "step": 17855
+    },
+    {
+      "epoch": 0.4904012523687897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0796162825870006e-05,
+      "loss": 0.8043,
+      "step": 17856
+    },
+    {
+      "epoch": 0.4904287165966329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079527613040141e-05,
+      "loss": 0.9166,
+      "step": 17857
+    },
+    {
+      "epoch": 0.4904561808244761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0794389428640282e-05,
+      "loss": 0.8614,
+      "step": 17858
+    },
+    {
+      "epoch": 0.49048364505231934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0793502720593634e-05,
+      "loss": 0.8856,
+      "step": 17859
+    },
+    {
+      "epoch": 0.49051110928016256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0792616006268485e-05,
+      "loss": 0.8566,
+      "step": 17860
+    },
+    {
+      "epoch": 0.49053857350800584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0791729285671854e-05,
+      "loss": 0.9206,
+      "step": 17861
+    },
+    {
+      "epoch": 0.49056603773584906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0790842558810753e-05,
+      "loss": 0.8573,
+      "step": 17862
+    },
+    {
+      "epoch": 0.4905935019636923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0789955825692198e-05,
+      "loss": 0.7647,
+      "step": 17863
+    },
+    {
+      "epoch": 0.4906209661915355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0789069086323205e-05,
+      "loss": 0.8112,
+      "step": 17864
+    },
+    {
+      "epoch": 0.4906484304193788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0788182340710796e-05,
+      "loss": 0.9849,
+      "step": 17865
+    },
+    {
+      "epoch": 0.490675894647222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0787295588861981e-05,
+      "loss": 0.8759,
+      "step": 17866
+    },
+    {
+      "epoch": 0.4907033588750652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0786408830783776e-05,
+      "loss": 0.974,
+      "step": 17867
+    },
+    {
+      "epoch": 0.49073082310290844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0785522066483203e-05,
+      "loss": 0.8612,
+      "step": 17868
+    },
+    {
+      "epoch": 0.4907582873307517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0784635295967273e-05,
+      "loss": 0.7768,
+      "step": 17869
+    },
+    {
+      "epoch": 0.49078575155859494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0783748519243007e-05,
+      "loss": 0.9669,
+      "step": 17870
+    },
+    {
+      "epoch": 0.49081321578643816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0782861736317416e-05,
+      "loss": 1.0153,
+      "step": 17871
+    },
+    {
+      "epoch": 0.4908406800142814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0781974947197524e-05,
+      "loss": 0.9821,
+      "step": 17872
+    },
+    {
+      "epoch": 0.49086814424212466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0781088151890339e-05,
+      "loss": 0.8113,
+      "step": 17873
+    },
+    {
+      "epoch": 0.4908956084699679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0780201350402883e-05,
+      "loss": 0.8918,
+      "step": 17874
+    },
+    {
+      "epoch": 0.4909230726978111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0779314542742174e-05,
+      "loss": 0.7904,
+      "step": 17875
+    },
+    {
+      "epoch": 0.4909505369256543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0778427728915223e-05,
+      "loss": 0.8803,
+      "step": 17876
+    },
+    {
+      "epoch": 0.49097800115349755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0777540908929054e-05,
+      "loss": 0.8267,
+      "step": 17877
+    },
+    {
+      "epoch": 0.4910054653813408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0776654082790681e-05,
+      "loss": 0.8638,
+      "step": 17878
+    },
+    {
+      "epoch": 0.49103292960918404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0775767250507115e-05,
+      "loss": 0.9077,
+      "step": 17879
+    },
+    {
+      "epoch": 0.49106039383702726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077488041208538e-05,
+      "loss": 0.964,
+      "step": 17880
+    },
+    {
+      "epoch": 0.4910878580648705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077399356753249e-05,
+      "loss": 0.9445,
+      "step": 17881
+    },
+    {
+      "epoch": 0.49111532229271376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0773106716855466e-05,
+      "loss": 0.9149,
+      "step": 17882
+    },
+    {
+      "epoch": 0.491142786520557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0772219860061322e-05,
+      "loss": 0.8965,
+      "step": 17883
+    },
+    {
+      "epoch": 0.4911702507484002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0771332997157073e-05,
+      "loss": 0.8658,
+      "step": 17884
+    },
+    {
+      "epoch": 0.4911977149762434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0770446128149741e-05,
+      "loss": 0.8541,
+      "step": 17885
+    },
+    {
+      "epoch": 0.4912251792040867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.076955925304634e-05,
+      "loss": 0.9646,
+      "step": 17886
+    },
+    {
+      "epoch": 0.4912526434319299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0768672371853885e-05,
+      "loss": 0.9862,
+      "step": 17887
+    },
+    {
+      "epoch": 0.49128010765977315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0767785484579398e-05,
+      "loss": 0.9546,
+      "step": 17888
+    },
+    {
+      "epoch": 0.49130757188761637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0766898591229898e-05,
+      "loss": 0.9332,
+      "step": 17889
+    },
+    {
+      "epoch": 0.4913350361154596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0766011691812396e-05,
+      "loss": 0.9455,
+      "step": 17890
+    },
+    {
+      "epoch": 0.49136250034330287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0765124786333914e-05,
+      "loss": 0.9734,
+      "step": 17891
+    },
+    {
+      "epoch": 0.4913899645711461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0764237874801468e-05,
+      "loss": 0.8678,
+      "step": 17892
+    },
+    {
+      "epoch": 0.4914174287989893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0763350957222073e-05,
+      "loss": 0.9897,
+      "step": 17893
+    },
+    {
+      "epoch": 0.49144489302683253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0762464033602755e-05,
+      "loss": 0.9026,
+      "step": 17894
+    },
+    {
+      "epoch": 0.4914723572546758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0761577103950525e-05,
+      "loss": 0.9156,
+      "step": 17895
+    },
+    {
+      "epoch": 0.49149982148251903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0760690168272398e-05,
+      "loss": 0.9075,
+      "step": 17896
+    },
+    {
+      "epoch": 0.49152728571036225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0759803226575401e-05,
+      "loss": 1.0016,
+      "step": 17897
+    },
+    {
+      "epoch": 0.49155474993820547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0758916278866545e-05,
+      "loss": 0.8886,
+      "step": 17898
+    },
+    {
+      "epoch": 0.49158221416604875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0758029325152845e-05,
+      "loss": 0.8285,
+      "step": 17899
+    },
+    {
+      "epoch": 0.49160967839389197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0757142365441327e-05,
+      "loss": 0.8215,
+      "step": 17900
+    },
+    {
+      "epoch": 0.4916371426217352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0756255399739006e-05,
+      "loss": 0.9289,
+      "step": 17901
+    },
+    {
+      "epoch": 0.4916646068495784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0755368428052899e-05,
+      "loss": 0.8517,
+      "step": 17902
+    },
+    {
+      "epoch": 0.49169207107742163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0754481450390022e-05,
+      "loss": 0.9,
+      "step": 17903
+    },
+    {
+      "epoch": 0.4917195353052649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0753594466757397e-05,
+      "loss": 0.8694,
+      "step": 17904
+    },
+    {
+      "epoch": 0.49174699953310813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.075270747716204e-05,
+      "loss": 0.9102,
+      "step": 17905
+    },
+    {
+      "epoch": 0.49177446376095135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0751820481610973e-05,
+      "loss": 0.9074,
+      "step": 17906
+    },
+    {
+      "epoch": 0.4918019279887946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.075093348011121e-05,
+      "loss": 0.9196,
+      "step": 17907
+    },
+    {
+      "epoch": 0.49182939221663785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0750046472669766e-05,
+      "loss": 0.8004,
+      "step": 17908
+    },
+    {
+      "epoch": 0.49185685644448107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.074915945929367e-05,
+      "loss": 0.8964,
+      "step": 17909
+    },
+    {
+      "epoch": 0.4918843206723243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0748272439989931e-05,
+      "loss": 0.9412,
+      "step": 17910
+    },
+    {
+      "epoch": 0.4919117849001675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.074738541476557e-05,
+      "loss": 0.8218,
+      "step": 17911
+    },
+    {
+      "epoch": 0.4919392491280108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0746498383627606e-05,
+      "loss": 0.9066,
+      "step": 17912
+    },
+    {
+      "epoch": 0.491966713355854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0745611346583059e-05,
+      "loss": 0.8197,
+      "step": 17913
+    },
+    {
+      "epoch": 0.49199417758369723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0744724303638948e-05,
+      "loss": 0.9538,
+      "step": 17914
+    },
+    {
+      "epoch": 0.49202164181154046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0743837254802284e-05,
+      "loss": 0.8937,
+      "step": 17915
+    },
+    {
+      "epoch": 0.4920491060393837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0742950200080097e-05,
+      "loss": 0.9414,
+      "step": 17916
+    },
+    {
+      "epoch": 0.49207657026722695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0742063139479398e-05,
+      "loss": 0.8532,
+      "step": 17917
+    },
+    {
+      "epoch": 0.4921040344950702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0741176073007209e-05,
+      "loss": 0.8176,
+      "step": 17918
+    },
+    {
+      "epoch": 0.4921314987229134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0740289000670548e-05,
+      "loss": 0.8511,
+      "step": 17919
+    },
+    {
+      "epoch": 0.4921589629507566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.073940192247643e-05,
+      "loss": 0.8994,
+      "step": 17920
+    },
+    {
+      "epoch": 0.4921864271785999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0738514838431882e-05,
+      "loss": 0.9924,
+      "step": 17921
+    },
+    {
+      "epoch": 0.4922138914064431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0737627748543918e-05,
+      "loss": 0.9725,
+      "step": 17922
+    },
+    {
+      "epoch": 0.49224135563428634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0736740652819556e-05,
+      "loss": 0.994,
+      "step": 17923
+    },
+    {
+      "epoch": 0.49226881986212956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0735853551265813e-05,
+      "loss": 0.9999,
+      "step": 17924
+    },
+    {
+      "epoch": 0.49229628408997284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0734966443889718e-05,
+      "loss": 0.9505,
+      "step": 17925
+    },
+    {
+      "epoch": 0.49232374831781606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0734079330698282e-05,
+      "loss": 0.8028,
+      "step": 17926
+    },
+    {
+      "epoch": 0.4923512125456593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0733192211698525e-05,
+      "loss": 0.8962,
+      "step": 17927
+    },
+    {
+      "epoch": 0.4923786767735025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0732305086897468e-05,
+      "loss": 0.8934,
+      "step": 17928
+    },
+    {
+      "epoch": 0.4924061410013457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0731417956302129e-05,
+      "loss": 0.9173,
+      "step": 17929
+    },
+    {
+      "epoch": 0.492433605229189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0730530819919526e-05,
+      "loss": 0.8362,
+      "step": 17930
+    },
+    {
+      "epoch": 0.4924610694570322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0729643677756685e-05,
+      "loss": 0.9411,
+      "step": 17931
+    },
+    {
+      "epoch": 0.49248853368487544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0728756529820617e-05,
+      "loss": 0.7675,
+      "step": 17932
+    },
+    {
+      "epoch": 0.49251599791271866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0727869376118344e-05,
+      "loss": 0.8684,
+      "step": 17933
+    },
+    {
+      "epoch": 0.49254346214056194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072698221665689e-05,
+      "loss": 0.8253,
+      "step": 17934
+    },
+    {
+      "epoch": 0.49257092636840516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0726095051443269e-05,
+      "loss": 0.8211,
+      "step": 17935
+    },
+    {
+      "epoch": 0.4925983905962484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.07252078804845e-05,
+      "loss": 0.8762,
+      "step": 17936
+    },
+    {
+      "epoch": 0.4926258548240916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0724320703787612e-05,
+      "loss": 0.894,
+      "step": 17937
+    },
+    {
+      "epoch": 0.4926533190519349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0723433521359614e-05,
+      "loss": 0.8629,
+      "step": 17938
+    },
+    {
+      "epoch": 0.4926807832797781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072254633320753e-05,
+      "loss": 0.9215,
+      "step": 17939
+    },
+    {
+      "epoch": 0.4927082475076213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072165913933838e-05,
+      "loss": 0.9424,
+      "step": 17940
+    },
+    {
+      "epoch": 0.49273571173546454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0720771939759182e-05,
+      "loss": 0.933,
+      "step": 17941
+    },
+    {
+      "epoch": 0.49276317596330776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071988473447696e-05,
+      "loss": 0.798,
+      "step": 17942
+    },
+    {
+      "epoch": 0.49279064019115104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0718997523498729e-05,
+      "loss": 0.835,
+      "step": 17943
+    },
+    {
+      "epoch": 0.49281810441899426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0718110306831511e-05,
+      "loss": 0.8667,
+      "step": 17944
+    },
+    {
+      "epoch": 0.4928455686468375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0717223084482327e-05,
+      "loss": 1.0068,
+      "step": 17945
+    },
+    {
+      "epoch": 0.4928730328746807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0716335856458195e-05,
+      "loss": 0.8455,
+      "step": 17946
+    },
+    {
+      "epoch": 0.492900497102524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0715448622766137e-05,
+      "loss": 0.9322,
+      "step": 17947
+    },
+    {
+      "epoch": 0.4929279613303672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0714561383413173e-05,
+      "loss": 0.876,
+      "step": 17948
+    },
+    {
+      "epoch": 0.4929554255582104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071367413840632e-05,
+      "loss": 0.8918,
+      "step": 17949
+    },
+    {
+      "epoch": 0.49298288978605365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0712786887752605e-05,
+      "loss": 0.9297,
+      "step": 17950
+    },
+    {
+      "epoch": 0.4930103540138969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0711899631459041e-05,
+      "loss": 0.7971,
+      "step": 17951
+    },
+    {
+      "epoch": 0.49303781824174014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0711012369532649e-05,
+      "loss": 0.8245,
+      "step": 17952
+    },
+    {
+      "epoch": 0.49306528246958337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0710125101980457e-05,
+      "loss": 0.9522,
+      "step": 17953
+    },
+    {
+      "epoch": 0.4930927466974266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0709237828809478e-05,
+      "loss": 0.9204,
+      "step": 17954
+    },
+    {
+      "epoch": 0.4931202109252698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0708350550026733e-05,
+      "loss": 0.918,
+      "step": 17955
+    },
+    {
+      "epoch": 0.4931476751531131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0707463265639244e-05,
+      "loss": 1.032,
+      "step": 17956
+    },
+    {
+      "epoch": 0.4931751393809563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0706575975654035e-05,
+      "loss": 0.9614,
+      "step": 17957
+    },
+    {
+      "epoch": 0.49320260360879953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070568868007812e-05,
+      "loss": 0.9219,
+      "step": 17958
+    },
+    {
+      "epoch": 0.49323006783664275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0704801378918523e-05,
+      "loss": 0.973,
+      "step": 17959
+    },
+    {
+      "epoch": 0.493257532064486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0703914072182267e-05,
+      "loss": 0.9386,
+      "step": 17960
+    },
+    {
+      "epoch": 0.49328499629232925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0703026759876367e-05,
+      "loss": 0.7937,
+      "step": 17961
+    },
+    {
+      "epoch": 0.49331246052017247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070213944200785e-05,
+      "loss": 0.8289,
+      "step": 17962
+    },
+    {
+      "epoch": 0.4933399247480157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0701252118583734e-05,
+      "loss": 0.936,
+      "step": 17963
+    },
+    {
+      "epoch": 0.49336738897585897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0700364789611035e-05,
+      "loss": 0.8896,
+      "step": 17964
+    },
+    {
+      "epoch": 0.4933948532037022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0699477455096783e-05,
+      "loss": 0.9234,
+      "step": 17965
+    },
+    {
+      "epoch": 0.4934223174315454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0698590115047995e-05,
+      "loss": 0.9667,
+      "step": 17966
+    },
+    {
+      "epoch": 0.49344978165938863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0697702769471687e-05,
+      "loss": 0.9617,
+      "step": 17967
+    },
+    {
+      "epoch": 0.4934772458872319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0696815418374887e-05,
+      "loss": 0.9552,
+      "step": 17968
+    },
+    {
+      "epoch": 0.49350471011507513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0695928061764613e-05,
+      "loss": 0.8481,
+      "step": 17969
+    },
+    {
+      "epoch": 0.49353217434291835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0695040699647887e-05,
+      "loss": 0.9736,
+      "step": 17970
+    },
+    {
+      "epoch": 0.49355963857076157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.069415333203173e-05,
+      "loss": 0.9003,
+      "step": 17971
+    },
+    {
+      "epoch": 0.4935871027986048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0693265958923163e-05,
+      "loss": 0.8901,
+      "step": 17972
+    },
+    {
+      "epoch": 0.49361456702644807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0692378580329204e-05,
+      "loss": 0.9088,
+      "step": 17973
+    },
+    {
+      "epoch": 0.4936420312542913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.069149119625688e-05,
+      "loss": 0.9044,
+      "step": 17974
+    },
+    {
+      "epoch": 0.4936694954821345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.069060380671321e-05,
+      "loss": 0.9783,
+      "step": 17975
+    },
+    {
+      "epoch": 0.49369695970997773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0689716411705215e-05,
+      "loss": 0.8977,
+      "step": 17976
+    },
+    {
+      "epoch": 0.493724423937821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0688829011239918e-05,
+      "loss": 0.8836,
+      "step": 17977
+    },
+    {
+      "epoch": 0.49375188816566423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0687941605324339e-05,
+      "loss": 0.8705,
+      "step": 17978
+    },
+    {
+      "epoch": 0.49377935239350745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0687054193965495e-05,
+      "loss": 0.9168,
+      "step": 17979
+    },
+    {
+      "epoch": 0.4938068166213507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0686166777170415e-05,
+      "loss": 0.8141,
+      "step": 17980
+    },
+    {
+      "epoch": 0.49383428084919395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0685279354946119e-05,
+      "loss": 0.9386,
+      "step": 17981
+    },
+    {
+      "epoch": 0.4938617450770372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0684391927299628e-05,
+      "loss": 0.8648,
+      "step": 17982
+    },
+    {
+      "epoch": 0.4938892093048804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.068350449423796e-05,
+      "loss": 0.8939,
+      "step": 17983
+    },
+    {
+      "epoch": 0.4939166735327236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0682617055768141e-05,
+      "loss": 0.8311,
+      "step": 17984
+    },
+    {
+      "epoch": 0.49394413776056684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0681729611897191e-05,
+      "loss": 0.8592,
+      "step": 17985
+    },
+    {
+      "epoch": 0.4939716019884101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0680842162632133e-05,
+      "loss": 0.8194,
+      "step": 17986
+    },
+    {
+      "epoch": 0.49399906621625334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0679954707979989e-05,
+      "loss": 0.8253,
+      "step": 17987
+    },
+    {
+      "epoch": 0.49402653044409656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0679067247947776e-05,
+      "loss": 0.9485,
+      "step": 17988
+    },
+    {
+      "epoch": 0.4940539946719398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0678179782542523e-05,
+      "loss": 0.8594,
+      "step": 17989
+    },
+    {
+      "epoch": 0.49408145889978305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.067729231177125e-05,
+      "loss": 0.9029,
+      "step": 17990
+    },
+    {
+      "epoch": 0.4941089231276263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0676404835640975e-05,
+      "loss": 0.8338,
+      "step": 17991
+    },
+    {
+      "epoch": 0.4941363873554695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0675517354158722e-05,
+      "loss": 0.8211,
+      "step": 17992
+    },
+    {
+      "epoch": 0.4941638515833127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0674629867331518e-05,
+      "loss": 0.9209,
+      "step": 17993
+    },
+    {
+      "epoch": 0.494191315811156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.067374237516638e-05,
+      "loss": 0.8738,
+      "step": 17994
+    },
+    {
+      "epoch": 0.4942187800389992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0672854877670329e-05,
+      "loss": 0.8428,
+      "step": 17995
+    },
+    {
+      "epoch": 0.49424624426684244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0671967374850394e-05,
+      "loss": 0.8578,
+      "step": 17996
+    },
+    {
+      "epoch": 0.49427370849468566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0671079866713588e-05,
+      "loss": 0.8947,
+      "step": 17997
+    },
+    {
+      "epoch": 0.4943011727225289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0670192353266942e-05,
+      "loss": 0.9409,
+      "step": 17998
+    },
+    {
+      "epoch": 0.49432863695037216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0669304834517474e-05,
+      "loss": 0.9203,
+      "step": 17999
+    },
+    {
+      "epoch": 0.4943561011782154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0668417310472203e-05,
+      "loss": 0.9272,
+      "step": 18000
+    },
+    {
+      "epoch": 0.4943835654060586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066752978113816e-05,
+      "loss": 0.9745,
+      "step": 18001
+    },
+    {
+      "epoch": 0.4944110296339018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066664224652236e-05,
+      "loss": 0.9852,
+      "step": 18002
+    },
+    {
+      "epoch": 0.4944384938617451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0665754706631828e-05,
+      "loss": 0.8687,
+      "step": 18003
+    },
+    {
+      "epoch": 0.4944659580895883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0664867161473589e-05,
+      "loss": 0.9081,
+      "step": 18004
+    },
+    {
+      "epoch": 0.49449342231743154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066397961105466e-05,
+      "loss": 0.8962,
+      "step": 18005
+    },
+    {
+      "epoch": 0.49452088654527476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0663092055382071e-05,
+      "loss": 0.8923,
+      "step": 18006
+    },
+    {
+      "epoch": 0.49454835077311804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066220449446284e-05,
+      "loss": 0.8613,
+      "step": 18007
+    },
+    {
+      "epoch": 0.49457581500096126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0661316928303988e-05,
+      "loss": 0.8011,
+      "step": 18008
+    },
+    {
+      "epoch": 0.4946032792288045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0660429356912543e-05,
+      "loss": 0.9474,
+      "step": 18009
+    },
+    {
+      "epoch": 0.4946307434566477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0659541780295525e-05,
+      "loss": 0.9012,
+      "step": 18010
+    },
+    {
+      "epoch": 0.4946582076844909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0658654198459956e-05,
+      "loss": 0.8969,
+      "step": 18011
+    },
+    {
+      "epoch": 0.4946856719123342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0657766611412859e-05,
+      "loss": 0.9422,
+      "step": 18012
+    },
+    {
+      "epoch": 0.4947131361401774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065687901916126e-05,
+      "loss": 0.9725,
+      "step": 18013
+    },
+    {
+      "epoch": 0.49474060036802064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065599142171218e-05,
+      "loss": 0.9878,
+      "step": 18014
+    },
+    {
+      "epoch": 0.49476806459586387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065510381907264e-05,
+      "loss": 0.8898,
+      "step": 18015
+    },
+    {
+      "epoch": 0.49479552882370714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0654216211249664e-05,
+      "loss": 0.8998,
+      "step": 18016
+    },
+    {
+      "epoch": 0.49482299305155036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0653328598250277e-05,
+      "loss": 0.94,
+      "step": 18017
+    },
+    {
+      "epoch": 0.4948504572793936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.06524409800815e-05,
+      "loss": 0.9377,
+      "step": 18018
+    },
+    {
+      "epoch": 0.4948779215072368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065155335675036e-05,
+      "loss": 0.8961,
+      "step": 18019
+    },
+    {
+      "epoch": 0.4949053857350801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0650665728263873e-05,
+      "loss": 0.9959,
+      "step": 18020
+    },
+    {
+      "epoch": 0.4949328499629233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0649778094629071e-05,
+      "loss": 0.9019,
+      "step": 18021
+    },
+    {
+      "epoch": 0.4949603141907665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0648890455852974e-05,
+      "loss": 0.834,
+      "step": 18022
+    },
+    {
+      "epoch": 0.49498777841860975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0648002811942599e-05,
+      "loss": 0.9575,
+      "step": 18023
+    },
+    {
+      "epoch": 0.49501524264645297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0647115162904977e-05,
+      "loss": 0.92,
+      "step": 18024
+    },
+    {
+      "epoch": 0.49504270687429625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0646227508747131e-05,
+      "loss": 0.8377,
+      "step": 18025
+    },
+    {
+      "epoch": 0.49507017110213947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0645339849476081e-05,
+      "loss": 0.8819,
+      "step": 18026
+    },
+    {
+      "epoch": 0.4950976353299827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0644452185098852e-05,
+      "loss": 0.8551,
+      "step": 18027
+    },
+    {
+      "epoch": 0.4951250995578259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064356451562247e-05,
+      "loss": 0.8938,
+      "step": 18028
+    },
+    {
+      "epoch": 0.4951525637856692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0642676841053952e-05,
+      "loss": 0.9102,
+      "step": 18029
+    },
+    {
+      "epoch": 0.4951800280135124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064178916140033e-05,
+      "loss": 0.871,
+      "step": 18030
+    },
+    {
+      "epoch": 0.49520749224135563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064090147666862e-05,
+      "loss": 0.9125,
+      "step": 18031
+    },
+    {
+      "epoch": 0.49523495646919885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0640013786865853e-05,
+      "loss": 0.9605,
+      "step": 18032
+    },
+    {
+      "epoch": 0.4952624206970421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0639126091999048e-05,
+      "loss": 0.9163,
+      "step": 18033
+    },
+    {
+      "epoch": 0.49528988492488535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0638238392075229e-05,
+      "loss": 0.8423,
+      "step": 18034
+    },
+    {
+      "epoch": 0.49531734915272857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.063735068710142e-05,
+      "loss": 0.877,
+      "step": 18035
+    },
+    {
+      "epoch": 0.4953448133805718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0636462977084644e-05,
+      "loss": 0.8595,
+      "step": 18036
+    },
+    {
+      "epoch": 0.495372277608415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.063557526203193e-05,
+      "loss": 0.8928,
+      "step": 18037
+    },
+    {
+      "epoch": 0.4953997418362583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0634687541950297e-05,
+      "loss": 0.8889,
+      "step": 18038
+    },
+    {
+      "epoch": 0.4954272060641015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0633799816846769e-05,
+      "loss": 0.9438,
+      "step": 18039
+    },
+    {
+      "epoch": 0.49545467029194473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0632912086728373e-05,
+      "loss": 0.9525,
+      "step": 18040
+    },
+    {
+      "epoch": 0.49548213451978795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0632024351602131e-05,
+      "loss": 0.9138,
+      "step": 18041
+    },
+    {
+      "epoch": 0.49550959874763123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0631136611475066e-05,
+      "loss": 0.9371,
+      "step": 18042
+    },
+    {
+      "epoch": 0.49553706297547445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0630248866354207e-05,
+      "loss": 0.9217,
+      "step": 18043
+    },
+    {
+      "epoch": 0.4955645272033177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062936111624657e-05,
+      "loss": 0.8849,
+      "step": 18044
+    },
+    {
+      "epoch": 0.4955919914311609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0628473361159188e-05,
+      "loss": 0.8316,
+      "step": 18045
+    },
+    {
+      "epoch": 0.49561945565900417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0627585601099081e-05,
+      "loss": 0.8745,
+      "step": 18046
+    },
+    {
+      "epoch": 0.4956469198868474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0626697836073272e-05,
+      "loss": 0.8236,
+      "step": 18047
+    },
+    {
+      "epoch": 0.4956743841146906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0625810066088787e-05,
+      "loss": 0.9211,
+      "step": 18048
+    },
+    {
+      "epoch": 0.49570184834253384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0624922291152653e-05,
+      "loss": 0.9596,
+      "step": 18049
+    },
+    {
+      "epoch": 0.49572931257037706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062403451127189e-05,
+      "loss": 0.8492,
+      "step": 18050
+    },
+    {
+      "epoch": 0.49575677679822033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0623146726453522e-05,
+      "loss": 0.9097,
+      "step": 18051
+    },
+    {
+      "epoch": 0.49578424102606355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062225893670458e-05,
+      "loss": 0.8441,
+      "step": 18052
+    },
+    {
+      "epoch": 0.4958117052539068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0621371142032079e-05,
+      "loss": 0.7348,
+      "step": 18053
+    },
+    {
+      "epoch": 0.49583916948175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0620483342443053e-05,
+      "loss": 0.839,
+      "step": 18054
+    },
+    {
+      "epoch": 0.4958666337095933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0619595537944523e-05,
+      "loss": 0.9779,
+      "step": 18055
+    },
+    {
+      "epoch": 0.4958940979374365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0618707728543509e-05,
+      "loss": 0.931,
+      "step": 18056
+    },
+    {
+      "epoch": 0.4959215621652797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0617819914247045e-05,
+      "loss": 0.9146,
+      "step": 18057
+    },
+    {
+      "epoch": 0.49594902639312294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0616932095062148e-05,
+      "loss": 0.7918,
+      "step": 18058
+    },
+    {
+      "epoch": 0.4959764906209662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0616044270995845e-05,
+      "loss": 0.8827,
+      "step": 18059
+    },
+    {
+      "epoch": 0.49600395484880944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0615156442055161e-05,
+      "loss": 0.829,
+      "step": 18060
+    },
+    {
+      "epoch": 0.49603141907665266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061426860824712e-05,
+      "loss": 0.8372,
+      "step": 18061
+    },
+    {
+      "epoch": 0.4960588833044959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061338076957875e-05,
+      "loss": 0.8985,
+      "step": 18062
+    },
+    {
+      "epoch": 0.49608634753233916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0612492926057073e-05,
+      "loss": 0.9259,
+      "step": 18063
+    },
+    {
+      "epoch": 0.4961138117601824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0611605077689114e-05,
+      "loss": 0.7769,
+      "step": 18064
+    },
+    {
+      "epoch": 0.4961412759880256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0610717224481896e-05,
+      "loss": 0.829,
+      "step": 18065
+    },
+    {
+      "epoch": 0.4961687402158688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0609829366442452e-05,
+      "loss": 0.8755,
+      "step": 18066
+    },
+    {
+      "epoch": 0.49619620444371204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0608941503577799e-05,
+      "loss": 0.9508,
+      "step": 18067
+    },
+    {
+      "epoch": 0.4962236686715553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0608053635894963e-05,
+      "loss": 0.9055,
+      "step": 18068
+    },
+    {
+      "epoch": 0.49625113289939854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0607165763400978e-05,
+      "loss": 0.9241,
+      "step": 18069
+    },
+    {
+      "epoch": 0.49627859712724176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0606277886102856e-05,
+      "loss": 0.9421,
+      "step": 18070
+    },
+    {
+      "epoch": 0.496306061355085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0605390004007629e-05,
+      "loss": 0.9212,
+      "step": 18071
+    },
+    {
+      "epoch": 0.49633352558292826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0604502117122326e-05,
+      "loss": 0.8819,
+      "step": 18072
+    },
+    {
+      "epoch": 0.4963609898107715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0603614225453965e-05,
+      "loss": 0.8738,
+      "step": 18073
+    },
+    {
+      "epoch": 0.4963884540386147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0602726329009574e-05,
+      "loss": 0.9099,
+      "step": 18074
+    },
+    {
+      "epoch": 0.4964159182664579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.060183842779618e-05,
+      "loss": 0.9635,
+      "step": 18075
+    },
+    {
+      "epoch": 0.4964433824943012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0600950521820806e-05,
+      "loss": 0.8522,
+      "step": 18076
+    },
+    {
+      "epoch": 0.4964708467221444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0600062611090478e-05,
+      "loss": 0.9235,
+      "step": 18077
+    },
+    {
+      "epoch": 0.49649831094998764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0599174695612227e-05,
+      "loss": 0.9164,
+      "step": 18078
+    },
+    {
+      "epoch": 0.49652577517783086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0598286775393069e-05,
+      "loss": 0.8819,
+      "step": 18079
+    },
+    {
+      "epoch": 0.4965532394056741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0597398850440035e-05,
+      "loss": 0.8423,
+      "step": 18080
+    },
+    {
+      "epoch": 0.49658070363351736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059651092076015e-05,
+      "loss": 0.9375,
+      "step": 18081
+    },
+    {
+      "epoch": 0.4966081678613606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059562298636044e-05,
+      "loss": 0.9233,
+      "step": 18082
+    },
+    {
+      "epoch": 0.4966356320892038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0594735047247931e-05,
+      "loss": 0.9223,
+      "step": 18083
+    },
+    {
+      "epoch": 0.496663096317047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0593847103429649e-05,
+      "loss": 0.8085,
+      "step": 18084
+    },
+    {
+      "epoch": 0.4966905605448903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0592959154912614e-05,
+      "loss": 0.8201,
+      "step": 18085
+    },
+    {
+      "epoch": 0.4967180247727335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059207120170386e-05,
+      "loss": 0.9078,
+      "step": 18086
+    },
+    {
+      "epoch": 0.49674548900057675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059118324381041e-05,
+      "loss": 0.9209,
+      "step": 18087
+    },
+    {
+      "epoch": 0.49677295322841997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0590295281239286e-05,
+      "loss": 1.0067,
+      "step": 18088
+    },
+    {
+      "epoch": 0.49680041745626324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0589407313997517e-05,
+      "loss": 0.8265,
+      "step": 18089
+    },
+    {
+      "epoch": 0.49682788168410646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0588519342092136e-05,
+      "loss": 0.9583,
+      "step": 18090
+    },
+    {
+      "epoch": 0.4968553459119497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0587631365530155e-05,
+      "loss": 0.8531,
+      "step": 18091
+    },
+    {
+      "epoch": 0.4968828101397929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0586743384318607e-05,
+      "loss": 0.8849,
+      "step": 18092
+    },
+    {
+      "epoch": 0.49691027436763613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0585855398464522e-05,
+      "loss": 0.8809,
+      "step": 18093
+    },
+    {
+      "epoch": 0.4969377385954794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.058496740797492e-05,
+      "loss": 0.8427,
+      "step": 18094
+    },
+    {
+      "epoch": 0.4969652028233226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0584079412856828e-05,
+      "loss": 0.9961,
+      "step": 18095
+    },
+    {
+      "epoch": 0.49699266705116585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0583191413117275e-05,
+      "loss": 0.8251,
+      "step": 18096
+    },
+    {
+      "epoch": 0.49702013127900907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0582303408763286e-05,
+      "loss": 0.9904,
+      "step": 18097
+    },
+    {
+      "epoch": 0.49704759550685235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0581415399801887e-05,
+      "loss": 0.8847,
+      "step": 18098
+    },
+    {
+      "epoch": 0.49707505973469557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0580527386240104e-05,
+      "loss": 0.973,
+      "step": 18099
+    },
+    {
+      "epoch": 0.4971025239625388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0579639368084962e-05,
+      "loss": 0.8501,
+      "step": 18100
+    },
+    {
+      "epoch": 0.497129988190382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057875134534349e-05,
+      "loss": 0.8693,
+      "step": 18101
+    },
+    {
+      "epoch": 0.4971574524182253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0577863318022714e-05,
+      "loss": 0.981,
+      "step": 18102
+    },
+    {
+      "epoch": 0.4971849166460685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0576975286129658e-05,
+      "loss": 0.8864,
+      "step": 18103
+    },
+    {
+      "epoch": 0.49721238087391173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057608724967135e-05,
+      "loss": 0.8617,
+      "step": 18104
+    },
+    {
+      "epoch": 0.49723984510175495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0575199208654815e-05,
+      "loss": 0.7945,
+      "step": 18105
+    },
+    {
+      "epoch": 0.4972673093295982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0574311163087085e-05,
+      "loss": 0.8491,
+      "step": 18106
+    },
+    {
+      "epoch": 0.49729477355744145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057342311297518e-05,
+      "loss": 0.8957,
+      "step": 18107
+    },
+    {
+      "epoch": 0.49732223778528467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057253505832613e-05,
+      "loss": 0.8359,
+      "step": 18108
+    },
+    {
+      "epoch": 0.4973497020131279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057164699914696e-05,
+      "loss": 0.9421,
+      "step": 18109
+    },
+    {
+      "epoch": 0.4973771662409711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0570758935444699e-05,
+      "loss": 0.768,
+      "step": 18110
+    },
+    {
+      "epoch": 0.4974046304688144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0569870867226373e-05,
+      "loss": 0.9136,
+      "step": 18111
+    },
+    {
+      "epoch": 0.4974320946966576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0568982794499006e-05,
+      "loss": 0.9513,
+      "step": 18112
+    },
+    {
+      "epoch": 0.49745955892450083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0568094717269626e-05,
+      "loss": 0.8698,
+      "step": 18113
+    },
+    {
+      "epoch": 0.49748702315234405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0567206635545264e-05,
+      "loss": 1.0478,
+      "step": 18114
+    },
+    {
+      "epoch": 0.49751448738018733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0566318549332938e-05,
+      "loss": 0.8758,
+      "step": 18115
+    },
+    {
+      "epoch": 0.49754195160803055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0565430458639683e-05,
+      "loss": 0.8738,
+      "step": 18116
+    },
+    {
+      "epoch": 0.4975694158358738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0564542363472526e-05,
+      "loss": 0.9338,
+      "step": 18117
+    },
+    {
+      "epoch": 0.497596880063717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0563654263838487e-05,
+      "loss": 0.8245,
+      "step": 18118
+    },
+    {
+      "epoch": 0.4976243442915602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.05627661597446e-05,
+      "loss": 0.8657,
+      "step": 18119
+    },
+    {
+      "epoch": 0.4976518085194035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0561878051197886e-05,
+      "loss": 0.949,
+      "step": 18120
+    },
+    {
+      "epoch": 0.4976792727472467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0560989938205377e-05,
+      "loss": 0.8793,
+      "step": 18121
+    },
+    {
+      "epoch": 0.49770673697508994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0560101820774103e-05,
+      "loss": 0.9277,
+      "step": 18122
+    },
+    {
+      "epoch": 0.49773420120293316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0559213698911079e-05,
+      "loss": 0.9314,
+      "step": 18123
+    },
+    {
+      "epoch": 0.49776166543077643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0558325572623343e-05,
+      "loss": 0.8353,
+      "step": 18124
+    },
+    {
+      "epoch": 0.49778912965861966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0557437441917917e-05,
+      "loss": 0.8088,
+      "step": 18125
+    },
+    {
+      "epoch": 0.4978165938864629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0556549306801834e-05,
+      "loss": 0.9309,
+      "step": 18126
+    },
+    {
+      "epoch": 0.4978440581143061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0555661167282112e-05,
+      "loss": 0.9039,
+      "step": 18127
+    },
+    {
+      "epoch": 0.4978715223421494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0554773023365787e-05,
+      "loss": 0.9539,
+      "step": 18128
+    },
+    {
+      "epoch": 0.4978989865699926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.055388487505988e-05,
+      "loss": 0.8627,
+      "step": 18129
+    },
+    {
+      "epoch": 0.4979264507978358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0552996722371425e-05,
+      "loss": 0.8965,
+      "step": 18130
+    },
+    {
+      "epoch": 0.49795391502567904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0552108565307446e-05,
+      "loss": 0.9076,
+      "step": 18131
+    },
+    {
+      "epoch": 0.49798137925352226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0551220403874967e-05,
+      "loss": 0.879,
+      "step": 18132
+    },
+    {
+      "epoch": 0.49800884348136554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0550332238081018e-05,
+      "loss": 0.9752,
+      "step": 18133
+    },
+    {
+      "epoch": 0.49803630770920876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0549444067932633e-05,
+      "loss": 1.0081,
+      "step": 18134
+    },
+    {
+      "epoch": 0.498063771937052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0548555893436827e-05,
+      "loss": 0.9705,
+      "step": 18135
+    },
+    {
+      "epoch": 0.4980912361648952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0547667714600637e-05,
+      "loss": 0.8559,
+      "step": 18136
+    },
+    {
+      "epoch": 0.4981187003927385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0546779531431088e-05,
+      "loss": 0.9615,
+      "step": 18137
+    },
+    {
+      "epoch": 0.4981461646205817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0545891343935208e-05,
+      "loss": 0.9912,
+      "step": 18138
+    },
+    {
+      "epoch": 0.4981736288484249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0545003152120024e-05,
+      "loss": 0.9619,
+      "step": 18139
+    },
+    {
+      "epoch": 0.49820109307626814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0544114955992563e-05,
+      "loss": 0.9119,
+      "step": 18140
+    },
+    {
+      "epoch": 0.4982285573041114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0543226755559854e-05,
+      "loss": 0.9215,
+      "step": 18141
+    },
+    {
+      "epoch": 0.49825602153195464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0542338550828925e-05,
+      "loss": 0.9641,
+      "step": 18142
+    },
+    {
+      "epoch": 0.49828348575979786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0541450341806806e-05,
+      "loss": 0.8943,
+      "step": 18143
+    },
+    {
+      "epoch": 0.4983109499876411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0540562128500517e-05,
+      "loss": 0.8873,
+      "step": 18144
+    },
+    {
+      "epoch": 0.4983384142154843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0539673910917092e-05,
+      "loss": 0.8137,
+      "step": 18145
+    },
+    {
+      "epoch": 0.4983658784433276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0538785689063561e-05,
+      "loss": 0.8349,
+      "step": 18146
+    },
+    {
+      "epoch": 0.4983933426711708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.053789746294695e-05,
+      "loss": 0.8773,
+      "step": 18147
+    },
+    {
+      "epoch": 0.498420806899014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.053700923257428e-05,
+      "loss": 0.8323,
+      "step": 18148
+    },
+    {
+      "epoch": 0.49844827112685725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0536120997952587e-05,
+      "loss": 0.9183,
+      "step": 18149
+    },
+    {
+      "epoch": 0.4984757353547005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.05352327590889e-05,
+      "loss": 0.8879,
+      "step": 18150
+    },
+    {
+      "epoch": 0.49850319958254374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0534344515990242e-05,
+      "loss": 0.9366,
+      "step": 18151
+    },
+    {
+      "epoch": 0.49853066381038696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0533456268663645e-05,
+      "loss": 0.8809,
+      "step": 18152
+    },
+    {
+      "epoch": 0.4985581280382302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0532568017116131e-05,
+      "loss": 0.8748,
+      "step": 18153
+    },
+    {
+      "epoch": 0.49858559226607346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0531679761354737e-05,
+      "loss": 0.9656,
+      "step": 18154
+    },
+    {
+      "epoch": 0.4986130564939167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0530791501386485e-05,
+      "loss": 0.925,
+      "step": 18155
+    },
+    {
+      "epoch": 0.4986405207217599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0529903237218404e-05,
+      "loss": 0.815,
+      "step": 18156
+    },
+    {
+      "epoch": 0.4986679849496031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0529014968857526e-05,
+      "loss": 0.9086,
+      "step": 18157
+    },
+    {
+      "epoch": 0.4986954491774464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0528126696310876e-05,
+      "loss": 0.8495,
+      "step": 18158
+    },
+    {
+      "epoch": 0.4987229134052896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0527238419585482e-05,
+      "loss": 0.9011,
+      "step": 18159
+    },
+    {
+      "epoch": 0.49875037763313285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.052635013868837e-05,
+      "loss": 0.9265,
+      "step": 18160
+    },
+    {
+      "epoch": 0.49877784186097607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0525461853626578e-05,
+      "loss": 0.8702,
+      "step": 18161
+    },
+    {
+      "epoch": 0.4988053060888193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0524573564407125e-05,
+      "loss": 1.0104,
+      "step": 18162
+    },
+    {
+      "epoch": 0.49883277031666257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0523685271037043e-05,
+      "loss": 0.822,
+      "step": 18163
+    },
+    {
+      "epoch": 0.4988602345445058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0522796973523362e-05,
+      "loss": 0.8234,
+      "step": 18164
+    },
+    {
+      "epoch": 0.498887698772349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0521908671873107e-05,
+      "loss": 0.9403,
+      "step": 18165
+    },
+    {
+      "epoch": 0.49891516300019223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.052102036609331e-05,
+      "loss": 0.8918,
+      "step": 18166
+    },
+    {
+      "epoch": 0.4989426272280355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0520132056190997e-05,
+      "loss": 0.821,
+      "step": 18167
+    },
+    {
+      "epoch": 0.49897009145587873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0519243742173196e-05,
+      "loss": 0.9143,
+      "step": 18168
+    },
+    {
+      "epoch": 0.49899755568372195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.051835542404694e-05,
+      "loss": 0.879,
+      "step": 18169
+    },
+    {
+      "epoch": 0.49902501991156517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0517467101819254e-05,
+      "loss": 0.8228,
+      "step": 18170
+    },
+    {
+      "epoch": 0.49905248413940845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0516578775497166e-05,
+      "loss": 0.9342,
+      "step": 18171
+    },
+    {
+      "epoch": 0.49907994836725167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0515690445087707e-05,
+      "loss": 0.9299,
+      "step": 18172
+    },
+    {
+      "epoch": 0.4991074125950949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0514802110597909e-05,
+      "loss": 0.8964,
+      "step": 18173
+    },
+    {
+      "epoch": 0.4991348768229381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0513913772034797e-05,
+      "loss": 0.9395,
+      "step": 18174
+    },
+    {
+      "epoch": 0.49916234105078133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0513025429405397e-05,
+      "loss": 0.9102,
+      "step": 18175
+    },
+    {
+      "epoch": 0.4991898052786246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0512137082716742e-05,
+      "loss": 0.9177,
+      "step": 18176
+    },
+    {
+      "epoch": 0.49921726950646783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.051124873197586e-05,
+      "loss": 0.9426,
+      "step": 18177
+    },
+    {
+      "epoch": 0.49924473373431105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0510360377189781e-05,
+      "loss": 0.8371,
+      "step": 18178
+    },
+    {
+      "epoch": 0.4992721979621543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0509472018365534e-05,
+      "loss": 0.8961,
+      "step": 18179
+    },
+    {
+      "epoch": 0.49929966218999755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0508583655510144e-05,
+      "loss": 0.9018,
+      "step": 18180
+    },
+    {
+      "epoch": 0.49932712641784077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0507695288630647e-05,
+      "loss": 0.9594,
+      "step": 18181
+    },
+    {
+      "epoch": 0.499354590645684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0506806917734065e-05,
+      "loss": 0.976,
+      "step": 18182
+    },
+    {
+      "epoch": 0.4993820548735272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0505918542827432e-05,
+      "loss": 0.9166,
+      "step": 18183
+    },
+    {
+      "epoch": 0.4994095191013705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0505030163917773e-05,
+      "loss": 0.8338,
+      "step": 18184
+    },
+    {
+      "epoch": 0.4994369833292137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.050414178101212e-05,
+      "loss": 0.7551,
+      "step": 18185
+    },
+    {
+      "epoch": 0.49946444755705693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0503253394117505e-05,
+      "loss": 0.8721,
+      "step": 18186
+    },
+    {
+      "epoch": 0.49949191178490016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0502365003240955e-05,
+      "loss": 0.9073,
+      "step": 18187
+    },
+    {
+      "epoch": 0.4995193760127434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0501476608389494e-05,
+      "loss": 1.0058,
+      "step": 18188
+    },
+    {
+      "epoch": 0.49954684024058665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0500588209570157e-05,
+      "loss": 1.0013,
+      "step": 18189
+    },
+    {
+      "epoch": 0.4995743044684299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0499699806789976e-05,
+      "loss": 0.9907,
+      "step": 18190
+    },
+    {
+      "epoch": 0.4996017686962731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0498811400055972e-05,
+      "loss": 0.8775,
+      "step": 18191
+    },
+    {
+      "epoch": 0.4996292329241163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049792298937518e-05,
+      "loss": 0.9616,
+      "step": 18192
+    },
+    {
+      "epoch": 0.4996566971519596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049703457475463e-05,
+      "loss": 0.9759,
+      "step": 18193
+    },
+    {
+      "epoch": 0.4996841613798028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049614615620135e-05,
+      "loss": 0.8691,
+      "step": 18194
+    },
+    {
+      "epoch": 0.49971162560764604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0495257733722367e-05,
+      "loss": 0.8334,
+      "step": 18195
+    },
+    {
+      "epoch": 0.49973908983548926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0494369307324715e-05,
+      "loss": 0.9464,
+      "step": 18196
+    },
+    {
+      "epoch": 0.49976655406333254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049348087701542e-05,
+      "loss": 0.8106,
+      "step": 18197
+    },
+    {
+      "epoch": 0.49979401829117576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0492592442801515e-05,
+      "loss": 0.9186,
+      "step": 18198
+    },
+    {
+      "epoch": 0.499821482519019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0491704004690027e-05,
+      "loss": 0.9522,
+      "step": 18199
+    },
+    {
+      "epoch": 0.4998489467468622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0490815562687985e-05,
+      "loss": 0.8457,
+      "step": 18200
+    },
+    {
+      "epoch": 0.4998764109747054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0489927116802421e-05,
+      "loss": 0.8734,
+      "step": 18201
+    },
+    {
+      "epoch": 0.4999038752025487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0489038667040368e-05,
+      "loss": 0.9399,
+      "step": 18202
+    },
+    {
+      "epoch": 0.4999313394303919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0488150213408845e-05,
+      "loss": 0.9372,
+      "step": 18203
+    },
+    {
+      "epoch": 0.49995880365823514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0487261755914891e-05,
+      "loss": 0.8947,
+      "step": 18204
+    },
+    {
+      "epoch": 0.49998626788607836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0486373294565534e-05,
+      "loss": 0.9717,
+      "step": 18205
+    },
+    {
+      "epoch": 0.5000137321139216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0485484829367805e-05,
+      "loss": 0.9983,
+      "step": 18206
+    },
+    {
+      "epoch": 0.5000411963417648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0484596360328727e-05,
+      "loss": 0.8177,
+      "step": 18207
+    },
+    {
+      "epoch": 0.5000686605696081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0483707887455338e-05,
+      "loss": 1.0118,
+      "step": 18208
+    },
+    {
+      "epoch": 0.5000961247974514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0482819410754663e-05,
+      "loss": 0.8365,
+      "step": 18209
+    },
+    {
+      "epoch": 0.5001235890252945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0481930930233737e-05,
+      "loss": 0.9063,
+      "step": 18210
+    },
+    {
+      "epoch": 0.5001510532531378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0481042445899585e-05,
+      "loss": 0.9886,
+      "step": 18211
+    },
+    {
+      "epoch": 0.5001785174809811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0480153957759239e-05,
+      "loss": 0.9111,
+      "step": 18212
+    },
+    {
+      "epoch": 0.5002059817088242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0479265465819726e-05,
+      "loss": 0.9282,
+      "step": 18213
+    },
+    {
+      "epoch": 0.5002334459366675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0478376970088085e-05,
+      "loss": 0.8846,
+      "step": 18214
+    },
+    {
+      "epoch": 0.5002609101645107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0477488470571336e-05,
+      "loss": 0.9147,
+      "step": 18215
+    },
+    {
+      "epoch": 0.500288374392354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0476599967276512e-05,
+      "loss": 0.8714,
+      "step": 18216
+    },
+    {
+      "epoch": 0.5003158386201972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0475711460210646e-05,
+      "loss": 0.8457,
+      "step": 18217
+    },
+    {
+      "epoch": 0.5003433028480404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.047482294938077e-05,
+      "loss": 0.8809,
+      "step": 18218
+    },
+    {
+      "epoch": 0.5003707670758837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0473934434793907e-05,
+      "loss": 0.8191,
+      "step": 18219
+    },
+    {
+      "epoch": 0.5003982313037268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0473045916457092e-05,
+      "loss": 0.8813,
+      "step": 18220
+    },
+    {
+      "epoch": 0.5004256955315701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0472157394377353e-05,
+      "loss": 0.9389,
+      "step": 18221
+    },
+    {
+      "epoch": 0.5004531597594134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0471268868561726e-05,
+      "loss": 0.9283,
+      "step": 18222
+    },
+    {
+      "epoch": 0.5004806239872566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0470380339017235e-05,
+      "loss": 0.8618,
+      "step": 18223
+    },
+    {
+      "epoch": 0.5005080882150998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0469491805750912e-05,
+      "loss": 0.7952,
+      "step": 18224
+    },
+    {
+      "epoch": 0.5005355524429431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0468603268769787e-05,
+      "loss": 0.8822,
+      "step": 18225
+    },
+    {
+      "epoch": 0.5005630166707863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0467714728080895e-05,
+      "loss": 0.9769,
+      "step": 18226
+    },
+    {
+      "epoch": 0.5005904808986296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0466826183691259e-05,
+      "loss": 0.8405,
+      "step": 18227
+    },
+    {
+      "epoch": 0.5006179451264727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0465937635607912e-05,
+      "loss": 0.9046,
+      "step": 18228
+    },
+    {
+      "epoch": 0.500645409354316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0465049083837891e-05,
+      "loss": 0.8509,
+      "step": 18229
+    },
+    {
+      "epoch": 0.5006728735821593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.046416052838822e-05,
+      "loss": 0.8746,
+      "step": 18230
+    },
+    {
+      "epoch": 0.5007003378100024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0463271969265928e-05,
+      "loss": 0.842,
+      "step": 18231
+    },
+    {
+      "epoch": 0.5007278020378457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0462383406478054e-05,
+      "loss": 0.7703,
+      "step": 18232
+    },
+    {
+      "epoch": 0.5007552662656889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0461494840031619e-05,
+      "loss": 0.8939,
+      "step": 18233
+    },
+    {
+      "epoch": 0.5007827304935322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.046060626993366e-05,
+      "loss": 0.9,
+      "step": 18234
+    },
+    {
+      "epoch": 0.5008101947213754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0459717696191206e-05,
+      "loss": 0.8631,
+      "step": 18235
+    },
+    {
+      "epoch": 0.5008376589492186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0458829118811287e-05,
+      "loss": 0.8873,
+      "step": 18236
+    },
+    {
+      "epoch": 0.5008651231770619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0457940537800932e-05,
+      "loss": 0.7814,
+      "step": 18237
+    },
+    {
+      "epoch": 0.5008925874049052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.045705195316718e-05,
+      "loss": 0.8897,
+      "step": 18238
+    },
+    {
+      "epoch": 0.5009200516327483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0456163364917049e-05,
+      "loss": 0.9673,
+      "step": 18239
+    },
+    {
+      "epoch": 0.5009475158605916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0455274773057583e-05,
+      "loss": 0.8654,
+      "step": 18240
+    },
+    {
+      "epoch": 0.5009749800884348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0454386177595802e-05,
+      "loss": 1.0077,
+      "step": 18241
+    },
+    {
+      "epoch": 0.501002444316278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0453497578538742e-05,
+      "loss": 0.9336,
+      "step": 18242
+    },
+    {
+      "epoch": 0.5010299085441213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0452608975893437e-05,
+      "loss": 0.9997,
+      "step": 18243
+    },
+    {
+      "epoch": 0.5010573727719645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.045172036966691e-05,
+      "loss": 0.9182,
+      "step": 18244
+    },
+    {
+      "epoch": 0.5010848369998078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0450831759866198e-05,
+      "loss": 0.9383,
+      "step": 18245
+    },
+    {
+      "epoch": 0.5011123012276509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0449943146498333e-05,
+      "loss": 0.8153,
+      "step": 18246
+    },
+    {
+      "epoch": 0.5011397654554942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0449054529570339e-05,
+      "loss": 0.7769,
+      "step": 18247
+    },
+    {
+      "epoch": 0.5011672296833375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0448165909089254e-05,
+      "loss": 0.813,
+      "step": 18248
+    },
+    {
+      "epoch": 0.5011946939111807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0447277285062107e-05,
+      "loss": 0.9088,
+      "step": 18249
+    },
+    {
+      "epoch": 0.5012221581390239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.044638865749593e-05,
+      "loss": 0.9613,
+      "step": 18250
+    },
+    {
+      "epoch": 0.5012496223668672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.044550002639775e-05,
+      "loss": 0.8377,
+      "step": 18251
+    },
+    {
+      "epoch": 0.5012770865947104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0444611391774603e-05,
+      "loss": 0.9286,
+      "step": 18252
+    },
+    {
+      "epoch": 0.5013045508225537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0443722753633519e-05,
+      "loss": 0.8673,
+      "step": 18253
+    },
+    {
+      "epoch": 0.5013320150503968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0442834111981527e-05,
+      "loss": 0.9506,
+      "step": 18254
+    },
+    {
+      "epoch": 0.5013594792782401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0441945466825662e-05,
+      "loss": 0.8581,
+      "step": 18255
+    },
+    {
+      "epoch": 0.5013869435060834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0441056818172953e-05,
+      "loss": 0.818,
+      "step": 18256
+    },
+    {
+      "epoch": 0.5014144077339265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.044016816603043e-05,
+      "loss": 0.8419,
+      "step": 18257
+    },
+    {
+      "epoch": 0.5014418719617698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.043927951040513e-05,
+      "loss": 0.9472,
+      "step": 18258
+    },
+    {
+      "epoch": 0.501469336189613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0438390851304074e-05,
+      "loss": 0.9812,
+      "step": 18259
+    },
+    {
+      "epoch": 0.5014968004174563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0437502188734303e-05,
+      "loss": 0.9494,
+      "step": 18260
+    },
+    {
+      "epoch": 0.5015242646452995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0436613522702846e-05,
+      "loss": 0.8728,
+      "step": 18261
+    },
+    {
+      "epoch": 0.5015517288731427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0435724853216733e-05,
+      "loss": 0.8169,
+      "step": 18262
+    },
+    {
+      "epoch": 0.501579193100986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0434836180282995e-05,
+      "loss": 0.9544,
+      "step": 18263
+    },
+    {
+      "epoch": 0.5016066573288293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0433947503908668e-05,
+      "loss": 0.8705,
+      "step": 18264
+    },
+    {
+      "epoch": 0.5016341215566724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0433058824100774e-05,
+      "loss": 0.9135,
+      "step": 18265
+    },
+    {
+      "epoch": 0.5016615857845157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0432170140866358e-05,
+      "loss": 0.809,
+      "step": 18266
+    },
+    {
+      "epoch": 0.5016890500123589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.043128145421244e-05,
+      "loss": 1.0042,
+      "step": 18267
+    },
+    {
+      "epoch": 0.5017165142402021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0430392764146058e-05,
+      "loss": 0.9204,
+      "step": 18268
+    },
+    {
+      "epoch": 0.5017439784680454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0429504070674239e-05,
+      "loss": 0.9647,
+      "step": 18269
+    },
+    {
+      "epoch": 0.5017714426958886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0428615373804023e-05,
+      "loss": 0.9301,
+      "step": 18270
+    },
+    {
+      "epoch": 0.5017989069237319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.042772667354243e-05,
+      "loss": 0.8582,
+      "step": 18271
+    },
+    {
+      "epoch": 0.501826371151575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0426837969896502e-05,
+      "loss": 0.828,
+      "step": 18272
+    },
+    {
+      "epoch": 0.5018538353794183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0425949262873265e-05,
+      "loss": 0.8171,
+      "step": 18273
+    },
+    {
+      "epoch": 0.5018812996072616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0425060552479754e-05,
+      "loss": 0.7902,
+      "step": 18274
+    },
+    {
+      "epoch": 0.5019087638351047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0424171838722997e-05,
+      "loss": 0.8666,
+      "step": 18275
+    },
+    {
+      "epoch": 0.501936228062948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0423283121610029e-05,
+      "loss": 0.9041,
+      "step": 18276
+    },
+    {
+      "epoch": 0.5019636922907913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0422394401147879e-05,
+      "loss": 0.9135,
+      "step": 18277
+    },
+    {
+      "epoch": 0.5019911565186345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0421505677343585e-05,
+      "loss": 0.8314,
+      "step": 18278
+    },
+    {
+      "epoch": 0.5020186207464777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0420616950204173e-05,
+      "loss": 0.8556,
+      "step": 18279
+    },
+    {
+      "epoch": 0.5020460849743209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0419728219736675e-05,
+      "loss": 0.8873,
+      "step": 18280
+    },
+    {
+      "epoch": 0.5020735492021642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0418839485948124e-05,
+      "loss": 0.9268,
+      "step": 18281
+    },
+    {
+      "epoch": 0.5021010134300075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0417950748845558e-05,
+      "loss": 0.9575,
+      "step": 18282
+    },
+    {
+      "epoch": 0.5021284776578506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0417062008436001e-05,
+      "loss": 0.8245,
+      "step": 18283
+    },
+    {
+      "epoch": 0.5021559418856939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0416173264726487e-05,
+      "loss": 0.8304,
+      "step": 18284
+    },
+    {
+      "epoch": 0.5021834061135371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0415284517724049e-05,
+      "loss": 0.8383,
+      "step": 18285
+    },
+    {
+      "epoch": 0.5022108703413803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0414395767435719e-05,
+      "loss": 0.9514,
+      "step": 18286
+    },
+    {
+      "epoch": 0.5022383345692236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0413507013868528e-05,
+      "loss": 1.0002,
+      "step": 18287
+    },
+    {
+      "epoch": 0.5022657987970668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0412618257029512e-05,
+      "loss": 0.9741,
+      "step": 18288
+    },
+    {
+      "epoch": 0.5022932630249101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0411729496925697e-05,
+      "loss": 0.9666,
+      "step": 18289
+    },
+    {
+      "epoch": 0.5023207272527533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0410840733564122e-05,
+      "loss": 0.9032,
+      "step": 18290
+    },
+    {
+      "epoch": 0.5023481914805965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0409951966951818e-05,
+      "loss": 0.9776,
+      "step": 18291
+    },
+    {
+      "epoch": 0.5023756557084398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.040906319709581e-05,
+      "loss": 0.9465,
+      "step": 18292
+    },
+    {
+      "epoch": 0.502403119936283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0408174424003137e-05,
+      "loss": 0.9678,
+      "step": 18293
+    },
+    {
+      "epoch": 0.5024305841641262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0407285647680834e-05,
+      "loss": 0.9337,
+      "step": 18294
+    },
+    {
+      "epoch": 0.5024580483919695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0406396868135924e-05,
+      "loss": 0.8741,
+      "step": 18295
+    },
+    {
+      "epoch": 0.5024855126198127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0405508085375447e-05,
+      "loss": 0.8643,
+      "step": 18296
+    },
+    {
+      "epoch": 0.502512976847656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0404619299406434e-05,
+      "loss": 0.8446,
+      "step": 18297
+    },
+    {
+      "epoch": 0.5025404410754991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0403730510235914e-05,
+      "loss": 0.8152,
+      "step": 18298
+    },
+    {
+      "epoch": 0.5025679053033424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0402841717870921e-05,
+      "loss": 0.9695,
+      "step": 18299
+    },
+    {
+      "epoch": 0.5025953695311857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0401952922318493e-05,
+      "loss": 0.8806,
+      "step": 18300
+    },
+    {
+      "epoch": 0.5026228337590288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0401064123585652e-05,
+      "loss": 0.9759,
+      "step": 18301
+    },
+    {
+      "epoch": 0.5026502979868721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0400175321679443e-05,
+      "loss": 0.9353,
+      "step": 18302
+    },
+    {
+      "epoch": 0.5026777622147154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0399286516606886e-05,
+      "loss": 0.8901,
+      "step": 18303
+    },
+    {
+      "epoch": 0.5027052264425586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0398397708375018e-05,
+      "loss": 0.8765,
+      "step": 18304
+    },
+    {
+      "epoch": 0.5027326906704018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0397508896990878e-05,
+      "loss": 0.9611,
+      "step": 18305
+    },
+    {
+      "epoch": 0.502760154898245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0396620082461493e-05,
+      "loss": 0.8995,
+      "step": 18306
+    },
+    {
+      "epoch": 0.5027876191260883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0395731264793893e-05,
+      "loss": 0.9382,
+      "step": 18307
+    },
+    {
+      "epoch": 0.5028150833539315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0394842443995119e-05,
+      "loss": 0.9215,
+      "step": 18308
+    },
+    {
+      "epoch": 0.5028425475817747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0393953620072192e-05,
+      "loss": 0.8578,
+      "step": 18309
+    },
+    {
+      "epoch": 0.502870011809618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0393064793032158e-05,
+      "loss": 0.8496,
+      "step": 18310
+    },
+    {
+      "epoch": 0.5028974760374612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.039217596288204e-05,
+      "loss": 0.8408,
+      "step": 18311
+    },
+    {
+      "epoch": 0.5029249402653044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0391287129628873e-05,
+      "loss": 0.8929,
+      "step": 18312
+    },
+    {
+      "epoch": 0.5029524044931477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.039039829327969e-05,
+      "loss": 0.8061,
+      "step": 18313
+    },
+    {
+      "epoch": 0.5029798687209909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0389509453841531e-05,
+      "loss": 0.9152,
+      "step": 18314
+    },
+    {
+      "epoch": 0.5030073329488342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0388620611321415e-05,
+      "loss": 0.9175,
+      "step": 18315
+    },
+    {
+      "epoch": 0.5030347971766774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0387731765726384e-05,
+      "loss": 0.8582,
+      "step": 18316
+    },
+    {
+      "epoch": 0.5030622614045206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0386842917063472e-05,
+      "loss": 0.8543,
+      "step": 18317
+    },
+    {
+      "epoch": 0.5030897256323639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0385954065339706e-05,
+      "loss": 0.8413,
+      "step": 18318
+    },
+    {
+      "epoch": 0.503117189860207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0385065210562122e-05,
+      "loss": 0.829,
+      "step": 18319
+    },
+    {
+      "epoch": 0.5031446540880503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0384176352737754e-05,
+      "loss": 0.8718,
+      "step": 18320
+    },
+    {
+      "epoch": 0.5031721183158936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0383287491873632e-05,
+      "loss": 0.8413,
+      "step": 18321
+    },
+    {
+      "epoch": 0.5031995825437368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0382398627976793e-05,
+      "loss": 0.9848,
+      "step": 18322
+    },
+    {
+      "epoch": 0.50322704677158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0381509761054267e-05,
+      "loss": 0.915,
+      "step": 18323
+    },
+    {
+      "epoch": 0.5032545109994232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0380620891113089e-05,
+      "loss": 0.7998,
+      "step": 18324
+    },
+    {
+      "epoch": 0.5032819752272665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0379732018160289e-05,
+      "loss": 0.813,
+      "step": 18325
+    },
+    {
+      "epoch": 0.5033094394551098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0378843142202906e-05,
+      "loss": 0.9818,
+      "step": 18326
+    },
+    {
+      "epoch": 0.5033369036829529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0377954263247964e-05,
+      "loss": 0.877,
+      "step": 18327
+    },
+    {
+      "epoch": 0.5033643679107962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0377065381302505e-05,
+      "loss": 0.9624,
+      "step": 18328
+    },
+    {
+      "epoch": 0.5033918321386395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0376176496373559e-05,
+      "loss": 0.849,
+      "step": 18329
+    },
+    {
+      "epoch": 0.5034192963664826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0375287608468159e-05,
+      "loss": 0.9268,
+      "step": 18330
+    },
+    {
+      "epoch": 0.5034467605943259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0374398717593334e-05,
+      "loss": 0.8868,
+      "step": 18331
+    },
+    {
+      "epoch": 0.5034742248221691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0373509823756126e-05,
+      "loss": 0.9167,
+      "step": 18332
+    },
+    {
+      "epoch": 0.5035016890500124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.037262092696356e-05,
+      "loss": 0.8611,
+      "step": 18333
+    },
+    {
+      "epoch": 0.5035291532778556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0371732027222677e-05,
+      "loss": 0.7924,
+      "step": 18334
+    },
+    {
+      "epoch": 0.5035566175056988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0370843124540504e-05,
+      "loss": 0.9902,
+      "step": 18335
+    },
+    {
+      "epoch": 0.5035840817335421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0369954218924074e-05,
+      "loss": 0.9236,
+      "step": 18336
+    },
+    {
+      "epoch": 0.5036115459613852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0369065310380424e-05,
+      "loss": 0.8197,
+      "step": 18337
+    },
+    {
+      "epoch": 0.5036390101892285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0368176398916592e-05,
+      "loss": 0.7811,
+      "step": 18338
+    },
+    {
+      "epoch": 0.5036664744170718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.03672874845396e-05,
+      "loss": 0.9783,
+      "step": 18339
+    },
+    {
+      "epoch": 0.503693938644915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0366398567256487e-05,
+      "loss": 0.8863,
+      "step": 18340
+    },
+    {
+      "epoch": 0.5037214028727582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0365509647074288e-05,
+      "loss": 0.932,
+      "step": 18341
+    },
+    {
+      "epoch": 0.5037488671006015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0364620724000036e-05,
+      "loss": 0.8757,
+      "step": 18342
+    },
+    {
+      "epoch": 0.5037763313284447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0363731798040762e-05,
+      "loss": 0.8804,
+      "step": 18343
+    },
+    {
+      "epoch": 0.503803795556288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0362842869203503e-05,
+      "loss": 0.891,
+      "step": 18344
+    },
+    {
+      "epoch": 0.5038312597841311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0361953937495288e-05,
+      "loss": 0.9176,
+      "step": 18345
+    },
+    {
+      "epoch": 0.5038587240119744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0361065002923154e-05,
+      "loss": 0.9645,
+      "step": 18346
+    },
+    {
+      "epoch": 0.5038861882398177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0360176065494136e-05,
+      "loss": 0.8855,
+      "step": 18347
+    },
+    {
+      "epoch": 0.5039136524676608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0359287125215263e-05,
+      "loss": 0.8201,
+      "step": 18348
+    },
+    {
+      "epoch": 0.5039411166955041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0358398182093573e-05,
+      "loss": 0.9447,
+      "step": 18349
+    },
+    {
+      "epoch": 0.5039685809233473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0357509236136099e-05,
+      "loss": 0.8987,
+      "step": 18350
+    },
+    {
+      "epoch": 0.5039960451511906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0356620287349871e-05,
+      "loss": 0.9276,
+      "step": 18351
+    },
+    {
+      "epoch": 0.5040235093790338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0355731335741923e-05,
+      "loss": 0.7244,
+      "step": 18352
+    },
+    {
+      "epoch": 0.504050973606877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0354842381319294e-05,
+      "loss": 0.9814,
+      "step": 18353
+    },
+    {
+      "epoch": 0.5040784378347203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0353953424089015e-05,
+      "loss": 0.9696,
+      "step": 18354
+    },
+    {
+      "epoch": 0.5041059020625636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0353064464058116e-05,
+      "loss": 0.9573,
+      "step": 18355
+    },
+    {
+      "epoch": 0.5041333662904067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035217550123364e-05,
+      "loss": 0.9318,
+      "step": 18356
+    },
+    {
+      "epoch": 0.50416083051825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035128653562261e-05,
+      "loss": 0.8216,
+      "step": 18357
+    },
+    {
+      "epoch": 0.5041882947460932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0350397567232066e-05,
+      "loss": 0.8875,
+      "step": 18358
+    },
+    {
+      "epoch": 0.5042157589739364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0349508596069042e-05,
+      "loss": 0.8563,
+      "step": 18359
+    },
+    {
+      "epoch": 0.5042432232017797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0348619622140568e-05,
+      "loss": 0.9961,
+      "step": 18360
+    },
+    {
+      "epoch": 0.5042706874296229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0347730645453685e-05,
+      "loss": 0.8442,
+      "step": 18361
+    },
+    {
+      "epoch": 0.5042981516574662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.034684166601542e-05,
+      "loss": 0.8911,
+      "step": 18362
+    },
+    {
+      "epoch": 0.5043256158853093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0345952683832808e-05,
+      "loss": 0.933,
+      "step": 18363
+    },
+    {
+      "epoch": 0.5043530801131526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0345063698912884e-05,
+      "loss": 0.9372,
+      "step": 18364
+    },
+    {
+      "epoch": 0.5043805443409959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0344174711262684e-05,
+      "loss": 0.883,
+      "step": 18365
+    },
+    {
+      "epoch": 0.504408008568839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.034328572088924e-05,
+      "loss": 0.929,
+      "step": 18366
+    },
+    {
+      "epoch": 0.5044354727966823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0342396727799589e-05,
+      "loss": 0.8393,
+      "step": 18367
+    },
+    {
+      "epoch": 0.5044629370245256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0341507732000757e-05,
+      "loss": 0.9308,
+      "step": 18368
+    },
+    {
+      "epoch": 0.5044904012523688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0340618733499785e-05,
+      "loss": 0.8707,
+      "step": 18369
+    },
+    {
+      "epoch": 0.504517865480212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0339729732303711e-05,
+      "loss": 0.9005,
+      "step": 18370
+    },
+    {
+      "epoch": 0.5045453297080552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0338840728419559e-05,
+      "loss": 0.9818,
+      "step": 18371
+    },
+    {
+      "epoch": 0.5045727939358985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0337951721854367e-05,
+      "loss": 0.8718,
+      "step": 18372
+    },
+    {
+      "epoch": 0.5046002581637418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0337062712615174e-05,
+      "loss": 0.7833,
+      "step": 18373
+    },
+    {
+      "epoch": 0.5046277223915849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0336173700709009e-05,
+      "loss": 0.9507,
+      "step": 18374
+    },
+    {
+      "epoch": 0.5046551866194282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0335284686142905e-05,
+      "loss": 0.8338,
+      "step": 18375
+    },
+    {
+      "epoch": 0.5046826508472714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0334395668923903e-05,
+      "loss": 0.9075,
+      "step": 18376
+    },
+    {
+      "epoch": 0.5047101150751147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0333506649059026e-05,
+      "loss": 0.8291,
+      "step": 18377
+    },
+    {
+      "epoch": 0.5047375793029579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0332617626555321e-05,
+      "loss": 0.9282,
+      "step": 18378
+    },
+    {
+      "epoch": 0.5047650435308011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0331728601419816e-05,
+      "loss": 0.8025,
+      "step": 18379
+    },
+    {
+      "epoch": 0.5047925077586444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0330839573659543e-05,
+      "loss": 0.9338,
+      "step": 18380
+    },
+    {
+      "epoch": 0.5048199719864876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.032995054328154e-05,
+      "loss": 0.9485,
+      "step": 18381
+    },
+    {
+      "epoch": 0.5048474362143308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0329061510292847e-05,
+      "loss": 0.8412,
+      "step": 18382
+    },
+    {
+      "epoch": 0.5048749004421741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0328172474700481e-05,
+      "loss": 1.0362,
+      "step": 18383
+    },
+    {
+      "epoch": 0.5049023646700173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0327283436511493e-05,
+      "loss": 0.8935,
+      "step": 18384
+    },
+    {
+      "epoch": 0.5049298288978605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0326394395732912e-05,
+      "loss": 0.9985,
+      "step": 18385
+    },
+    {
+      "epoch": 0.5049572931257038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0325505352371773e-05,
+      "loss": 0.9221,
+      "step": 18386
+    },
+    {
+      "epoch": 0.504984757353547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0324616306435107e-05,
+      "loss": 0.9587,
+      "step": 18387
+    },
+    {
+      "epoch": 0.5050122215813903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0323727257929952e-05,
+      "loss": 0.8403,
+      "step": 18388
+    },
+    {
+      "epoch": 0.5050396858092334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.032283820686334e-05,
+      "loss": 0.9505,
+      "step": 18389
+    },
+    {
+      "epoch": 0.5050671500370767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.032194915324231e-05,
+      "loss": 0.8399,
+      "step": 18390
+    },
+    {
+      "epoch": 0.50509461426492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0321060097073891e-05,
+      "loss": 0.9753,
+      "step": 18391
+    },
+    {
+      "epoch": 0.5051220784927631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0320171038365121e-05,
+      "loss": 0.8909,
+      "step": 18392
+    },
+    {
+      "epoch": 0.5051495427206064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0319281977123033e-05,
+      "loss": 0.8628,
+      "step": 18393
+    },
+    {
+      "epoch": 0.5051770069484497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0318392913354667e-05,
+      "loss": 0.9608,
+      "step": 18394
+    },
+    {
+      "epoch": 0.5052044711762929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0317503847067048e-05,
+      "loss": 0.7971,
+      "step": 18395
+    },
+    {
+      "epoch": 0.5052319354041361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0316614778267215e-05,
+      "loss": 0.8632,
+      "step": 18396
+    },
+    {
+      "epoch": 0.5052593996319793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0315725706962206e-05,
+      "loss": 0.7924,
+      "step": 18397
+    },
+    {
+      "epoch": 0.5052868638598226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0314836633159053e-05,
+      "loss": 0.9428,
+      "step": 18398
+    },
+    {
+      "epoch": 0.5053143280876659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0313947556864787e-05,
+      "loss": 0.8668,
+      "step": 18399
+    },
+    {
+      "epoch": 0.505341792315509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.031305847808645e-05,
+      "loss": 0.8385,
+      "step": 18400
+    },
+    {
+      "epoch": 0.5053692565433523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0312169396831071e-05,
+      "loss": 0.8835,
+      "step": 18401
+    },
+    {
+      "epoch": 0.5053967207711955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.031128031310569e-05,
+      "loss": 0.8403,
+      "step": 18402
+    },
+    {
+      "epoch": 0.5054241849990387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0310391226917336e-05,
+      "loss": 0.9586,
+      "step": 18403
+    },
+    {
+      "epoch": 0.505451649226882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0309502138273044e-05,
+      "loss": 0.9344,
+      "step": 18404
+    },
+    {
+      "epoch": 0.5054791134547252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0308613047179854e-05,
+      "loss": 0.8615,
+      "step": 18405
+    },
+    {
+      "epoch": 0.5055065776825685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.03077239536448e-05,
+      "loss": 0.8801,
+      "step": 18406
+    },
+    {
+      "epoch": 0.5055340419104117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0306834857674912e-05,
+      "loss": 0.8698,
+      "step": 18407
+    },
+    {
+      "epoch": 0.5055615061382549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0305945759277227e-05,
+      "loss": 0.8423,
+      "step": 18408
+    },
+    {
+      "epoch": 0.5055889703660982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0305056658458783e-05,
+      "loss": 0.9875,
+      "step": 18409
+    },
+    {
+      "epoch": 0.5056164345939413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0304167555226612e-05,
+      "loss": 0.9342,
+      "step": 18410
+    },
+    {
+      "epoch": 0.5056438988217846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0303278449587748e-05,
+      "loss": 0.9485,
+      "step": 18411
+    },
+    {
+      "epoch": 0.5056713630496279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.030238934154923e-05,
+      "loss": 0.8375,
+      "step": 18412
+    },
+    {
+      "epoch": 0.5056988272774711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0301500231118087e-05,
+      "loss": 1.0045,
+      "step": 18413
+    },
+    {
+      "epoch": 0.5057262915053143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.030061111830136e-05,
+      "loss": 0.9181,
+      "step": 18414
+    },
+    {
+      "epoch": 0.5057537557331576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0299722003106083e-05,
+      "loss": 0.7709,
+      "step": 18415
+    },
+    {
+      "epoch": 0.5057812199610008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0298832885539285e-05,
+      "loss": 0.818,
+      "step": 18416
+    },
+    {
+      "epoch": 0.5058086841888441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0297943765608007e-05,
+      "loss": 0.923,
+      "step": 18417
+    },
+    {
+      "epoch": 0.5058361484166872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0297054643319286e-05,
+      "loss": 0.9637,
+      "step": 18418
+    },
+    {
+      "epoch": 0.5058636126445305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029616551868015e-05,
+      "loss": 0.8729,
+      "step": 18419
+    },
+    {
+      "epoch": 0.5058910768723738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0295276391697637e-05,
+      "loss": 0.937,
+      "step": 18420
+    },
+    {
+      "epoch": 0.505918541100217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0294387262378784e-05,
+      "loss": 0.8343,
+      "step": 18421
+    },
+    {
+      "epoch": 0.5059460053280602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0293498130730626e-05,
+      "loss": 0.8785,
+      "step": 18422
+    },
+    {
+      "epoch": 0.5059734695559034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0292608996760199e-05,
+      "loss": 0.7916,
+      "step": 18423
+    },
+    {
+      "epoch": 0.5060009337837467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0291719860474532e-05,
+      "loss": 0.9033,
+      "step": 18424
+    },
+    {
+      "epoch": 0.5060283980115899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0290830721880665e-05,
+      "loss": 0.925,
+      "step": 18425
+    },
+    {
+      "epoch": 0.5060558622394331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0289941580985635e-05,
+      "loss": 0.7696,
+      "step": 18426
+    },
+    {
+      "epoch": 0.5060833264672764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0289052437796473e-05,
+      "loss": 0.9275,
+      "step": 18427
+    },
+    {
+      "epoch": 0.5061107906951197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0288163292320214e-05,
+      "loss": 0.8693,
+      "step": 18428
+    },
+    {
+      "epoch": 0.5061382549229628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0287274144563902e-05,
+      "loss": 0.8473,
+      "step": 18429
+    },
+    {
+      "epoch": 0.5061657191508061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0286384994534562e-05,
+      "loss": 0.8375,
+      "step": 18430
+    },
+    {
+      "epoch": 0.5061931833786493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.028549584223923e-05,
+      "loss": 0.9481,
+      "step": 18431
+    },
+    {
+      "epoch": 0.5062206476064925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0284606687684948e-05,
+      "loss": 0.9037,
+      "step": 18432
+    },
+    {
+      "epoch": 0.5062481118343358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0283717530878748e-05,
+      "loss": 0.9227,
+      "step": 18433
+    },
+    {
+      "epoch": 0.506275576062179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0282828371827663e-05,
+      "loss": 0.9172,
+      "step": 18434
+    },
+    {
+      "epoch": 0.5063030402900223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0281939210538732e-05,
+      "loss": 0.8489,
+      "step": 18435
+    },
+    {
+      "epoch": 0.5063305045178654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0281050047018986e-05,
+      "loss": 0.878,
+      "step": 18436
+    },
+    {
+      "epoch": 0.5063579687457087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0280160881275465e-05,
+      "loss": 0.988,
+      "step": 18437
+    },
+    {
+      "epoch": 0.506385432973552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0279271713315202e-05,
+      "loss": 0.7909,
+      "step": 18438
+    },
+    {
+      "epoch": 0.5064128972013952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0278382543145236e-05,
+      "loss": 0.8325,
+      "step": 18439
+    },
+    {
+      "epoch": 0.5064403614292384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0277493370772596e-05,
+      "loss": 0.9071,
+      "step": 18440
+    },
+    {
+      "epoch": 0.5064678256570817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0276604196204323e-05,
+      "loss": 0.969,
+      "step": 18441
+    },
+    {
+      "epoch": 0.5064952898849249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.027571501944745e-05,
+      "loss": 0.8728,
+      "step": 18442
+    },
+    {
+      "epoch": 0.5065227541127681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.027482584050901e-05,
+      "loss": 0.9339,
+      "step": 18443
+    },
+    {
+      "epoch": 0.5065502183406113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0273936659396047e-05,
+      "loss": 0.957,
+      "step": 18444
+    },
+    {
+      "epoch": 0.5065776825684546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0273047476115585e-05,
+      "loss": 0.8324,
+      "step": 18445
+    },
+    {
+      "epoch": 0.5066051467962979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.027215829067467e-05,
+      "loss": 0.8961,
+      "step": 18446
+    },
+    {
+      "epoch": 0.506632611024141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0271269103080333e-05,
+      "loss": 0.9528,
+      "step": 18447
+    },
+    {
+      "epoch": 0.5066600752519843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0270379913339608e-05,
+      "loss": 0.9766,
+      "step": 18448
+    },
+    {
+      "epoch": 0.5066875394798275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.026949072145953e-05,
+      "loss": 0.8912,
+      "step": 18449
+    },
+    {
+      "epoch": 0.5067150037076708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0268601527447145e-05,
+      "loss": 0.8757,
+      "step": 18450
+    },
+    {
+      "epoch": 0.506742467935514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0267712331309474e-05,
+      "loss": 0.8891,
+      "step": 18451
+    },
+    {
+      "epoch": 0.5067699321633572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0266823133053558e-05,
+      "loss": 0.8832,
+      "step": 18452
+    },
+    {
+      "epoch": 0.5067973963912005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0265933932686438e-05,
+      "loss": 0.8485,
+      "step": 18453
+    },
+    {
+      "epoch": 0.5068248606190437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0265044730215145e-05,
+      "loss": 0.7814,
+      "step": 18454
+    },
+    {
+      "epoch": 0.5068523248468869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0264155525646715e-05,
+      "loss": 0.7757,
+      "step": 18455
+    },
+    {
+      "epoch": 0.5068797890747302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0263266318988186e-05,
+      "loss": 0.9971,
+      "step": 18456
+    },
+    {
+      "epoch": 0.5069072533025734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0262377110246589e-05,
+      "loss": 0.8748,
+      "step": 18457
+    },
+    {
+      "epoch": 0.5069347175304166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0261487899428963e-05,
+      "loss": 0.8734,
+      "step": 18458
+    },
+    {
+      "epoch": 0.5069621817582599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0260598686542346e-05,
+      "loss": 1.0431,
+      "step": 18459
+    },
+    {
+      "epoch": 0.5069896459861031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0259709471593768e-05,
+      "loss": 0.836,
+      "step": 18460
+    },
+    {
+      "epoch": 0.5070171102139464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025882025459027e-05,
+      "loss": 0.887,
+      "step": 18461
+    },
+    {
+      "epoch": 0.5070445744417895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0257931035538889e-05,
+      "loss": 0.9018,
+      "step": 18462
+    },
+    {
+      "epoch": 0.5070720386696328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025704181444665e-05,
+      "loss": 0.9557,
+      "step": 18463
+    },
+    {
+      "epoch": 0.5070995028974761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0256152591320601e-05,
+      "loss": 0.8963,
+      "step": 18464
+    },
+    {
+      "epoch": 0.5071269671253192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0255263366167773e-05,
+      "loss": 0.9383,
+      "step": 18465
+    },
+    {
+      "epoch": 0.5071544313531625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0254374138995205e-05,
+      "loss": 0.9404,
+      "step": 18466
+    },
+    {
+      "epoch": 0.5071818955810058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0253484909809927e-05,
+      "loss": 0.8323,
+      "step": 18467
+    },
+    {
+      "epoch": 0.507209359808849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025259567861898e-05,
+      "loss": 0.8445,
+      "step": 18468
+    },
+    {
+      "epoch": 0.5072368240366922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0251706445429394e-05,
+      "loss": 0.8423,
+      "step": 18469
+    },
+    {
+      "epoch": 0.5072642882645354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0250817210248214e-05,
+      "loss": 0.806,
+      "step": 18470
+    },
+    {
+      "epoch": 0.5072917524923787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0249927973082472e-05,
+      "loss": 0.8916,
+      "step": 18471
+    },
+    {
+      "epoch": 0.507319216720222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0249038733939199e-05,
+      "loss": 0.9263,
+      "step": 18472
+    },
+    {
+      "epoch": 0.5073466809480651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0248149492825434e-05,
+      "loss": 0.8271,
+      "step": 18473
+    },
+    {
+      "epoch": 0.5073741451759084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247260249748222e-05,
+      "loss": 0.9054,
+      "step": 18474
+    },
+    {
+      "epoch": 0.5074016094037516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0246371004714583e-05,
+      "loss": 0.7372,
+      "step": 18475
+    },
+    {
+      "epoch": 0.5074290736315948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0245481757731563e-05,
+      "loss": 0.984,
+      "step": 18476
+    },
+    {
+      "epoch": 0.5074565378594381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.02445925088062e-05,
+      "loss": 0.9496,
+      "step": 18477
+    },
+    {
+      "epoch": 0.5074840020872813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0243703257945524e-05,
+      "loss": 0.9316,
+      "step": 18478
+    },
+    {
+      "epoch": 0.5075114663151246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.024281400515657e-05,
+      "loss": 0.9128,
+      "step": 18479
+    },
+    {
+      "epoch": 0.5075389305429678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0241924750446382e-05,
+      "loss": 0.9239,
+      "step": 18480
+    },
+    {
+      "epoch": 0.507566394770811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0241035493821988e-05,
+      "loss": 0.8327,
+      "step": 18481
+    },
+    {
+      "epoch": 0.5075938589986543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0240146235290431e-05,
+      "loss": 0.8921,
+      "step": 18482
+    },
+    {
+      "epoch": 0.5076213232264974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0239256974858745e-05,
+      "loss": 0.9391,
+      "step": 18483
+    },
+    {
+      "epoch": 0.5076487874543407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0238367712533961e-05,
+      "loss": 0.889,
+      "step": 18484
+    },
+    {
+      "epoch": 0.507676251682184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0237478448323122e-05,
+      "loss": 0.8959,
+      "step": 18485
+    },
+    {
+      "epoch": 0.5077037159100272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0236589182233263e-05,
+      "loss": 0.975,
+      "step": 18486
+    },
+    {
+      "epoch": 0.5077311801378704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0235699914271415e-05,
+      "loss": 0.8231,
+      "step": 18487
+    },
+    {
+      "epoch": 0.5077586443657136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0234810644444621e-05,
+      "loss": 0.9115,
+      "step": 18488
+    },
+    {
+      "epoch": 0.5077861085935569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0233921372759913e-05,
+      "loss": 0.8819,
+      "step": 18489
+    },
+    {
+      "epoch": 0.5078135728214002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0233032099224327e-05,
+      "loss": 0.8237,
+      "step": 18490
+    },
+    {
+      "epoch": 0.5078410370492433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0232142823844903e-05,
+      "loss": 0.8406,
+      "step": 18491
+    },
+    {
+      "epoch": 0.5078685012770866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0231253546628672e-05,
+      "loss": 0.9632,
+      "step": 18492
+    },
+    {
+      "epoch": 0.5078959655049299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0230364267582676e-05,
+      "loss": 0.8452,
+      "step": 18493
+    },
+    {
+      "epoch": 0.507923429732773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.022947498671395e-05,
+      "loss": 0.8124,
+      "step": 18494
+    },
+    {
+      "epoch": 0.5079508939606163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0228585704029528e-05,
+      "loss": 0.9649,
+      "step": 18495
+    },
+    {
+      "epoch": 0.5079783581884595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0227696419536443e-05,
+      "loss": 0.8851,
+      "step": 18496
+    },
+    {
+      "epoch": 0.5080058224163028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0226807133241741e-05,
+      "loss": 0.903,
+      "step": 18497
+    },
+    {
+      "epoch": 0.508033286644146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0225917845152452e-05,
+      "loss": 0.8977,
+      "step": 18498
+    },
+    {
+      "epoch": 0.5080607508719892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0225028555275614e-05,
+      "loss": 0.8862,
+      "step": 18499
+    },
+    {
+      "epoch": 0.5080882150998325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.022413926361826e-05,
+      "loss": 0.9138,
+      "step": 18500
+    },
+    {
+      "epoch": 0.5081156793276757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0223249970187433e-05,
+      "loss": 0.8712,
+      "step": 18501
+    },
+    {
+      "epoch": 0.5081431435555189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0222360674990164e-05,
+      "loss": 0.834,
+      "step": 18502
+    },
+    {
+      "epoch": 0.5081706077833622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0221471378033492e-05,
+      "loss": 0.9368,
+      "step": 18503
+    },
+    {
+      "epoch": 0.5081980720112054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.022058207932445e-05,
+      "loss": 0.9491,
+      "step": 18504
+    },
+    {
+      "epoch": 0.5082255362390486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0219692778870076e-05,
+      "loss": 0.8789,
+      "step": 18505
+    },
+    {
+      "epoch": 0.5082530004668919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0218803476677413e-05,
+      "loss": 0.7535,
+      "step": 18506
+    },
+    {
+      "epoch": 0.5082804646947351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0217914172753489e-05,
+      "loss": 0.9014,
+      "step": 18507
+    },
+    {
+      "epoch": 0.5083079289225784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0217024867105346e-05,
+      "loss": 0.8605,
+      "step": 18508
+    },
+    {
+      "epoch": 0.5083353931504215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0216135559740015e-05,
+      "loss": 0.8666,
+      "step": 18509
+    },
+    {
+      "epoch": 0.5083628573782648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.021524625066454e-05,
+      "loss": 0.9039,
+      "step": 18510
+    },
+    {
+      "epoch": 0.5083903216061081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0214356939885947e-05,
+      "loss": 0.9418,
+      "step": 18511
+    },
+    {
+      "epoch": 0.5084177858339513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0213467627411285e-05,
+      "loss": 0.8638,
+      "step": 18512
+    },
+    {
+      "epoch": 0.5084452500617945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.021257831324758e-05,
+      "loss": 0.9291,
+      "step": 18513
+    },
+    {
+      "epoch": 0.5084727142896377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0211688997401876e-05,
+      "loss": 0.9314,
+      "step": 18514
+    },
+    {
+      "epoch": 0.508500178517481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0210799679881206e-05,
+      "loss": 0.8416,
+      "step": 18515
+    },
+    {
+      "epoch": 0.5085276427453242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0209910360692605e-05,
+      "loss": 0.9186,
+      "step": 18516
+    },
+    {
+      "epoch": 0.5085551069731674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0209021039843114e-05,
+      "loss": 0.9963,
+      "step": 18517
+    },
+    {
+      "epoch": 0.5085825712010107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0208131717339768e-05,
+      "loss": 0.8679,
+      "step": 18518
+    },
+    {
+      "epoch": 0.508610035428854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0207242393189602e-05,
+      "loss": 0.9861,
+      "step": 18519
+    },
+    {
+      "epoch": 0.5086374996566971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0206353067399654e-05,
+      "loss": 0.9596,
+      "step": 18520
+    },
+    {
+      "epoch": 0.5086649638845404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0205463739976963e-05,
+      "loss": 0.8579,
+      "step": 18521
+    },
+    {
+      "epoch": 0.5086924281123836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.020457441092856e-05,
+      "loss": 0.8059,
+      "step": 18522
+    },
+    {
+      "epoch": 0.5087198923402269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0203685080261487e-05,
+      "loss": 0.9241,
+      "step": 18523
+    },
+    {
+      "epoch": 0.5087473565680701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.020279574798278e-05,
+      "loss": 0.9506,
+      "step": 18524
+    },
+    {
+      "epoch": 0.5087748207959133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0201906414099469e-05,
+      "loss": 0.8273,
+      "step": 18525
+    },
+    {
+      "epoch": 0.5088022850237566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0201017078618602e-05,
+      "loss": 0.9327,
+      "step": 18526
+    },
+    {
+      "epoch": 0.5088297492515997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0200127741547208e-05,
+      "loss": 0.8636,
+      "step": 18527
+    },
+    {
+      "epoch": 0.508857213479443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0199238402892324e-05,
+      "loss": 0.859,
+      "step": 18528
+    },
+    {
+      "epoch": 0.5088846777072863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0198349062660991e-05,
+      "loss": 0.8975,
+      "step": 18529
+    },
+    {
+      "epoch": 0.5089121419351295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0197459720860244e-05,
+      "loss": 0.8864,
+      "step": 18530
+    },
+    {
+      "epoch": 0.5089396061629727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0196570377497117e-05,
+      "loss": 0.9381,
+      "step": 18531
+    },
+    {
+      "epoch": 0.508967070390816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0195681032578648e-05,
+      "loss": 0.8689,
+      "step": 18532
+    },
+    {
+      "epoch": 0.5089945346186592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0194791686111878e-05,
+      "loss": 0.9525,
+      "step": 18533
+    },
+    {
+      "epoch": 0.5090219988465025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0193902338103841e-05,
+      "loss": 0.9374,
+      "step": 18534
+    },
+    {
+      "epoch": 0.5090494630743456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019301298856157e-05,
+      "loss": 0.9005,
+      "step": 18535
+    },
+    {
+      "epoch": 0.5090769273021889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0192123637492109e-05,
+      "loss": 0.98,
+      "step": 18536
+    },
+    {
+      "epoch": 0.5091043915300322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0191234284902488e-05,
+      "loss": 0.9575,
+      "step": 18537
+    },
+    {
+      "epoch": 0.5091318557578753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0190344930799749e-05,
+      "loss": 0.9752,
+      "step": 18538
+    },
+    {
+      "epoch": 0.5091593199857186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.018945557519093e-05,
+      "loss": 0.8322,
+      "step": 18539
+    },
+    {
+      "epoch": 0.5091867842135618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.018856621808306e-05,
+      "loss": 0.9355,
+      "step": 18540
+    },
+    {
+      "epoch": 0.5092142484414051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0187676859483181e-05,
+      "loss": 0.9098,
+      "step": 18541
+    },
+    {
+      "epoch": 0.5092417126692483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0186787499398337e-05,
+      "loss": 0.791,
+      "step": 18542
+    },
+    {
+      "epoch": 0.5092691768970915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0185898137835552e-05,
+      "loss": 0.7939,
+      "step": 18543
+    },
+    {
+      "epoch": 0.5092966411249348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.018500877480187e-05,
+      "loss": 0.9301,
+      "step": 18544
+    },
+    {
+      "epoch": 0.5093241053527781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0184119410304327e-05,
+      "loss": 0.825,
+      "step": 18545
+    },
+    {
+      "epoch": 0.5093515695806212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.018323004434996e-05,
+      "loss": 0.8352,
+      "step": 18546
+    },
+    {
+      "epoch": 0.5093790338084645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0182340676945808e-05,
+      "loss": 0.9903,
+      "step": 18547
+    },
+    {
+      "epoch": 0.5094064980363077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0181451308098902e-05,
+      "loss": 0.9601,
+      "step": 18548
+    },
+    {
+      "epoch": 0.5094339622641509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0180561937816284e-05,
+      "loss": 0.9666,
+      "step": 18549
+    },
+    {
+      "epoch": 0.5094614264919942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0179672566104992e-05,
+      "loss": 0.8462,
+      "step": 18550
+    },
+    {
+      "epoch": 0.5094888907198374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.017878319297206e-05,
+      "loss": 0.8009,
+      "step": 18551
+    },
+    {
+      "epoch": 0.5095163549476807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0177893818424525e-05,
+      "loss": 0.9229,
+      "step": 18552
+    },
+    {
+      "epoch": 0.5095438191755238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0177004442469427e-05,
+      "loss": 0.8844,
+      "step": 18553
+    },
+    {
+      "epoch": 0.5095712834033671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0176115065113803e-05,
+      "loss": 0.9554,
+      "step": 18554
+    },
+    {
+      "epoch": 0.5095987476312104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0175225686364686e-05,
+      "loss": 0.8046,
+      "step": 18555
+    },
+    {
+      "epoch": 0.5096262118590535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0174336306229113e-05,
+      "loss": 0.945,
+      "step": 18556
+    },
+    {
+      "epoch": 0.5096536760868968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0173446924714126e-05,
+      "loss": 0.8148,
+      "step": 18557
+    },
+    {
+      "epoch": 0.5096811403147401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.017255754182676e-05,
+      "loss": 0.8837,
+      "step": 18558
+    },
+    {
+      "epoch": 0.5097086045425833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0171668157574053e-05,
+      "loss": 0.8897,
+      "step": 18559
+    },
+    {
+      "epoch": 0.5097360687704265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0170778771963039e-05,
+      "loss": 0.929,
+      "step": 18560
+    },
+    {
+      "epoch": 0.5097635329982697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0169889385000754e-05,
+      "loss": 0.8841,
+      "step": 18561
+    },
+    {
+      "epoch": 0.509790997226113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0168999996694246e-05,
+      "loss": 0.9388,
+      "step": 18562
+    },
+    {
+      "epoch": 0.5098184614539563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.016811060705054e-05,
+      "loss": 0.9573,
+      "step": 18563
+    },
+    {
+      "epoch": 0.5098459256817994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0167221216076679e-05,
+      "loss": 0.9018,
+      "step": 18564
+    },
+    {
+      "epoch": 0.5098733899096427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0166331823779698e-05,
+      "loss": 0.9395,
+      "step": 18565
+    },
+    {
+      "epoch": 0.5099008541374859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0165442430166639e-05,
+      "loss": 0.945,
+      "step": 18566
+    },
+    {
+      "epoch": 0.5099283183653291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0164553035244528e-05,
+      "loss": 0.8759,
+      "step": 18567
+    },
+    {
+      "epoch": 0.5099557825931724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0163663639020417e-05,
+      "loss": 0.9113,
+      "step": 18568
+    },
+    {
+      "epoch": 0.5099832468210156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.016277424150133e-05,
+      "loss": 0.8194,
+      "step": 18569
+    },
+    {
+      "epoch": 0.5100107110488589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0161884842694314e-05,
+      "loss": 1.0092,
+      "step": 18570
+    },
+    {
+      "epoch": 0.5100381752767021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0160995442606404e-05,
+      "loss": 0.9324,
+      "step": 18571
+    },
+    {
+      "epoch": 0.5100656395045453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.016010604124463e-05,
+      "loss": 0.8946,
+      "step": 18572
+    },
+    {
+      "epoch": 0.5100931037323886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0159216638616037e-05,
+      "loss": 0.9472,
+      "step": 18573
+    },
+    {
+      "epoch": 0.5101205679602318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0158327234727662e-05,
+      "loss": 0.934,
+      "step": 18574
+    },
+    {
+      "epoch": 0.510148032188075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.015743782958654e-05,
+      "loss": 1.0441,
+      "step": 18575
+    },
+    {
+      "epoch": 0.5101754964159183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0156548423199708e-05,
+      "loss": 0.8252,
+      "step": 18576
+    },
+    {
+      "epoch": 0.5102029606437615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0155659015574206e-05,
+      "loss": 0.8435,
+      "step": 18577
+    },
+    {
+      "epoch": 0.5102304248716047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0154769606717069e-05,
+      "loss": 0.7979,
+      "step": 18578
+    },
+    {
+      "epoch": 0.5102578890994479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0153880196635332e-05,
+      "loss": 0.9652,
+      "step": 18579
+    },
+    {
+      "epoch": 0.5102853533272912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0152990785336042e-05,
+      "loss": 0.8949,
+      "step": 18580
+    },
+    {
+      "epoch": 0.5103128175551345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0152101372826222e-05,
+      "loss": 0.8971,
+      "step": 18581
+    },
+    {
+      "epoch": 0.5103402817829776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0151211959112922e-05,
+      "loss": 0.7914,
+      "step": 18582
+    },
+    {
+      "epoch": 0.5103677460108209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0150322544203173e-05,
+      "loss": 0.8765,
+      "step": 18583
+    },
+    {
+      "epoch": 0.5103952102386642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0149433128104014e-05,
+      "loss": 0.8973,
+      "step": 18584
+    },
+    {
+      "epoch": 0.5104226744665074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014854371082248e-05,
+      "loss": 0.8696,
+      "step": 18585
+    },
+    {
+      "epoch": 0.5104501386943506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0147654292365613e-05,
+      "loss": 0.8475,
+      "step": 18586
+    },
+    {
+      "epoch": 0.5104776029221938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0146764872740449e-05,
+      "loss": 0.9632,
+      "step": 18587
+    },
+    {
+      "epoch": 0.5105050671500371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0145875451954021e-05,
+      "loss": 0.9026,
+      "step": 18588
+    },
+    {
+      "epoch": 0.5105325313778803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0144986030013373e-05,
+      "loss": 0.8401,
+      "step": 18589
+    },
+    {
+      "epoch": 0.5105599956057235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014409660692554e-05,
+      "loss": 0.8756,
+      "step": 18590
+    },
+    {
+      "epoch": 0.5105874598335668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0143207182697553e-05,
+      "loss": 0.8908,
+      "step": 18591
+    },
+    {
+      "epoch": 0.51061492406141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0142317757336462e-05,
+      "loss": 0.9087,
+      "step": 18592
+    },
+    {
+      "epoch": 0.5106423882892532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0141428330849295e-05,
+      "loss": 0.9226,
+      "step": 18593
+    },
+    {
+      "epoch": 0.5106698525170965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0140538903243092e-05,
+      "loss": 0.8403,
+      "step": 18594
+    },
+    {
+      "epoch": 0.5106973167449397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.013964947452489e-05,
+      "loss": 0.8653,
+      "step": 18595
+    },
+    {
+      "epoch": 0.510724780972783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.013876004470173e-05,
+      "loss": 0.9012,
+      "step": 18596
+    },
+    {
+      "epoch": 0.5107522452006262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0137870613780642e-05,
+      "loss": 0.9908,
+      "step": 18597
+    },
+    {
+      "epoch": 0.5107797094284694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0136981181768676e-05,
+      "loss": 0.9264,
+      "step": 18598
+    },
+    {
+      "epoch": 0.5108071736563127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0136091748672855e-05,
+      "loss": 0.9279,
+      "step": 18599
+    },
+    {
+      "epoch": 0.5108346378841558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0135202314500226e-05,
+      "loss": 0.9484,
+      "step": 18600
+    },
+    {
+      "epoch": 0.5108621021119991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0134312879257824e-05,
+      "loss": 0.7694,
+      "step": 18601
+    },
+    {
+      "epoch": 0.5108895663398424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0133423442952686e-05,
+      "loss": 0.8244,
+      "step": 18602
+    },
+    {
+      "epoch": 0.5109170305676856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0132534005591852e-05,
+      "loss": 0.8348,
+      "step": 18603
+    },
+    {
+      "epoch": 0.5109444947955288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0131644567182355e-05,
+      "loss": 0.87,
+      "step": 18604
+    },
+    {
+      "epoch": 0.5109719590233721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0130755127731233e-05,
+      "loss": 1.0229,
+      "step": 18605
+    },
+    {
+      "epoch": 0.5109994232512153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.012986568724553e-05,
+      "loss": 0.8956,
+      "step": 18606
+    },
+    {
+      "epoch": 0.5110268874790586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0128976245732278e-05,
+      "loss": 0.9034,
+      "step": 18607
+    },
+    {
+      "epoch": 0.5110543517069017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0128086803198515e-05,
+      "loss": 0.8134,
+      "step": 18608
+    },
+    {
+      "epoch": 0.511081815934745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127197359651282e-05,
+      "loss": 0.8107,
+      "step": 18609
+    },
+    {
+      "epoch": 0.5111092801625883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0126307915097615e-05,
+      "loss": 1.0059,
+      "step": 18610
+    },
+    {
+      "epoch": 0.5111367443904314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0125418469544546e-05,
+      "loss": 0.9353,
+      "step": 18611
+    },
+    {
+      "epoch": 0.5111642086182747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0124529022999122e-05,
+      "loss": 1.0255,
+      "step": 18612
+    },
+    {
+      "epoch": 0.5111916728461179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0123639575468373e-05,
+      "loss": 0.9288,
+      "step": 18613
+    },
+    {
+      "epoch": 0.5112191370739612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0122750126959341e-05,
+      "loss": 0.9683,
+      "step": 18614
+    },
+    {
+      "epoch": 0.5112466013018044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0121860677479065e-05,
+      "loss": 0.8546,
+      "step": 18615
+    },
+    {
+      "epoch": 0.5112740655296476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0120971227034575e-05,
+      "loss": 0.8434,
+      "step": 18616
+    },
+    {
+      "epoch": 0.5113015297574909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0120081775632914e-05,
+      "loss": 0.8738,
+      "step": 18617
+    },
+    {
+      "epoch": 0.5113289939853342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0119192323281123e-05,
+      "loss": 0.9229,
+      "step": 18618
+    },
+    {
+      "epoch": 0.5113564582131773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0118302869986239e-05,
+      "loss": 0.8545,
+      "step": 18619
+    },
+    {
+      "epoch": 0.5113839224410206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.011741341575529e-05,
+      "loss": 0.9475,
+      "step": 18620
+    },
+    {
+      "epoch": 0.5114113866688638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0116523960595325e-05,
+      "loss": 0.8032,
+      "step": 18621
+    },
+    {
+      "epoch": 0.511438850896707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0115634504513376e-05,
+      "loss": 0.8431,
+      "step": 18622
+    },
+    {
+      "epoch": 0.5114663151245503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0114745047516481e-05,
+      "loss": 0.8933,
+      "step": 18623
+    },
+    {
+      "epoch": 0.5114937793523935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0113855589611681e-05,
+      "loss": 0.9185,
+      "step": 18624
+    },
+    {
+      "epoch": 0.5115212435802368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0112966130806007e-05,
+      "loss": 0.9451,
+      "step": 18625
+    },
+    {
+      "epoch": 0.5115487078080799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0112076671106505e-05,
+      "loss": 0.8404,
+      "step": 18626
+    },
+    {
+      "epoch": 0.5115761720359232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.011118721052021e-05,
+      "loss": 0.8459,
+      "step": 18627
+    },
+    {
+      "epoch": 0.5116036362637665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0110297749054157e-05,
+      "loss": 0.8132,
+      "step": 18628
+    },
+    {
+      "epoch": 0.5116311004916096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0109408286715383e-05,
+      "loss": 0.8859,
+      "step": 18629
+    },
+    {
+      "epoch": 0.5116585647194529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0108518823510931e-05,
+      "loss": 0.9293,
+      "step": 18630
+    },
+    {
+      "epoch": 0.5116860289472962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0107629359447837e-05,
+      "loss": 0.8926,
+      "step": 18631
+    },
+    {
+      "epoch": 0.5117134931751394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0106739894533135e-05,
+      "loss": 0.7854,
+      "step": 18632
+    },
+    {
+      "epoch": 0.5117409574029826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0105850428773869e-05,
+      "loss": 0.9493,
+      "step": 18633
+    },
+    {
+      "epoch": 0.5117684216308258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0104960962177073e-05,
+      "loss": 0.8647,
+      "step": 18634
+    },
+    {
+      "epoch": 0.5117958858586691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.010407149474978e-05,
+      "loss": 0.9754,
+      "step": 18635
+    },
+    {
+      "epoch": 0.5118233500865124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0103182026499039e-05,
+      "loss": 0.8893,
+      "step": 18636
+    },
+    {
+      "epoch": 0.5118508143143555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0102292557431879e-05,
+      "loss": 0.8381,
+      "step": 18637
+    },
+    {
+      "epoch": 0.5118782785421988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0101403087555342e-05,
+      "loss": 0.9435,
+      "step": 18638
+    },
+    {
+      "epoch": 0.511905742770042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0100513616876463e-05,
+      "loss": 0.9172,
+      "step": 18639
+    },
+    {
+      "epoch": 0.5119332069978852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0099624145402282e-05,
+      "loss": 0.9612,
+      "step": 18640
+    },
+    {
+      "epoch": 0.5119606712257285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0098734673139836e-05,
+      "loss": 0.8951,
+      "step": 18641
+    },
+    {
+      "epoch": 0.5119881354535717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0097845200096163e-05,
+      "loss": 0.9706,
+      "step": 18642
+    },
+    {
+      "epoch": 0.512015599681415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0096955726278304e-05,
+      "loss": 0.8948,
+      "step": 18643
+    },
+    {
+      "epoch": 0.5120430639092582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0096066251693289e-05,
+      "loss": 0.8728,
+      "step": 18644
+    },
+    {
+      "epoch": 0.5120705281371014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0095176776348164e-05,
+      "loss": 0.916,
+      "step": 18645
+    },
+    {
+      "epoch": 0.5120979923649447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0094287300249961e-05,
+      "loss": 0.8013,
+      "step": 18646
+    },
+    {
+      "epoch": 0.5121254565927879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0093397823405721e-05,
+      "loss": 1.0045,
+      "step": 18647
+    },
+    {
+      "epoch": 0.5121529208206311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0092508345822482e-05,
+      "loss": 0.9513,
+      "step": 18648
+    },
+    {
+      "epoch": 0.5121803850484744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.009161886750728e-05,
+      "loss": 0.8548,
+      "step": 18649
+    },
+    {
+      "epoch": 0.5122078492763176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0090729388467156e-05,
+      "loss": 0.8943,
+      "step": 18650
+    },
+    {
+      "epoch": 0.5122353135041608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0089839908709144e-05,
+      "loss": 0.8769,
+      "step": 18651
+    },
+    {
+      "epoch": 0.512262777732004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0088950428240284e-05,
+      "loss": 0.9264,
+      "step": 18652
+    },
+    {
+      "epoch": 0.5122902419598473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0088060947067613e-05,
+      "loss": 0.8627,
+      "step": 18653
+    },
+    {
+      "epoch": 0.5123177061876906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0087171465198172e-05,
+      "loss": 0.9734,
+      "step": 18654
+    },
+    {
+      "epoch": 0.5123451704155337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0086281982638997e-05,
+      "loss": 0.8346,
+      "step": 18655
+    },
+    {
+      "epoch": 0.512372634643377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0085392499397122e-05,
+      "loss": 0.8161,
+      "step": 18656
+    },
+    {
+      "epoch": 0.5124000988712203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0084503015479593e-05,
+      "loss": 0.942,
+      "step": 18657
+    },
+    {
+      "epoch": 0.5124275630990635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0083613530893441e-05,
+      "loss": 0.8873,
+      "step": 18658
+    },
+    {
+      "epoch": 0.5124550273269067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0082724045645704e-05,
+      "loss": 0.7651,
+      "step": 18659
+    },
+    {
+      "epoch": 0.5124824915547499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0081834559743426e-05,
+      "loss": 0.771,
+      "step": 18660
+    },
+    {
+      "epoch": 0.5125099557825932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0080945073193638e-05,
+      "loss": 0.8961,
+      "step": 18661
+    },
+    {
+      "epoch": 0.5125374200104365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0080055586003385e-05,
+      "loss": 0.9374,
+      "step": 18662
+    },
+    {
+      "epoch": 0.5125648842382796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.00791660981797e-05,
+      "loss": 0.8537,
+      "step": 18663
+    },
+    {
+      "epoch": 0.5125923484661229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.007827660972962e-05,
+      "loss": 0.8789,
+      "step": 18664
+    },
+    {
+      "epoch": 0.5126198126939661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0077387120660188e-05,
+      "loss": 0.864,
+      "step": 18665
+    },
+    {
+      "epoch": 0.5126472769218093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0076497630978437e-05,
+      "loss": 0.8659,
+      "step": 18666
+    },
+    {
+      "epoch": 0.5126747411496526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.007560814069141e-05,
+      "loss": 0.814,
+      "step": 18667
+    },
+    {
+      "epoch": 0.5127022053774958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.007471864980614e-05,
+      "loss": 0.8702,
+      "step": 18668
+    },
+    {
+      "epoch": 0.5127296696053391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0073829158329664e-05,
+      "loss": 0.9069,
+      "step": 18669
+    },
+    {
+      "epoch": 0.5127571338331823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0072939666269028e-05,
+      "loss": 0.9163,
+      "step": 18670
+    },
+    {
+      "epoch": 0.5127845980610255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0072050173631262e-05,
+      "loss": 0.7829,
+      "step": 18671
+    },
+    {
+      "epoch": 0.5128120622888688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0071160680423408e-05,
+      "loss": 0.9672,
+      "step": 18672
+    },
+    {
+      "epoch": 0.5128395265167119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0070271186652502e-05,
+      "loss": 0.8877,
+      "step": 18673
+    },
+    {
+      "epoch": 0.5128669907445552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0069381692325587e-05,
+      "loss": 0.9119,
+      "step": 18674
+    },
+    {
+      "epoch": 0.5128944549723985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0068492197449696e-05,
+      "loss": 0.9022,
+      "step": 18675
+    },
+    {
+      "epoch": 0.5129219192002417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0067602702031864e-05,
+      "loss": 0.9506,
+      "step": 18676
+    },
+    {
+      "epoch": 0.5129493834280849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0066713206079137e-05,
+      "loss": 1.0026,
+      "step": 18677
+    },
+    {
+      "epoch": 0.5129768476559281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.006582370959855e-05,
+      "loss": 0.9495,
+      "step": 18678
+    },
+    {
+      "epoch": 0.5130043118837714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064934212597137e-05,
+      "loss": 0.8498,
+      "step": 18679
+    },
+    {
+      "epoch": 0.5130317761116147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064044715081944e-05,
+      "loss": 0.8882,
+      "step": 18680
+    },
+    {
+      "epoch": 0.5130592403394578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.006315521706e-05,
+      "loss": 0.9066,
+      "step": 18681
+    },
+    {
+      "epoch": 0.5130867045673011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0062265718538348e-05,
+      "loss": 0.8368,
+      "step": 18682
+    },
+    {
+      "epoch": 0.5131141687951444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0061376219524029e-05,
+      "loss": 0.8649,
+      "step": 18683
+    },
+    {
+      "epoch": 0.5131416330229875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0060486720024073e-05,
+      "loss": 0.8656,
+      "step": 18684
+    },
+    {
+      "epoch": 0.5131690972508308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0059597220045525e-05,
+      "loss": 0.8129,
+      "step": 18685
+    },
+    {
+      "epoch": 0.513196561478674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0058707719595422e-05,
+      "loss": 0.8253,
+      "step": 18686
+    },
+    {
+      "epoch": 0.5132240257065173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.00578182186808e-05,
+      "loss": 0.8202,
+      "step": 18687
+    },
+    {
+      "epoch": 0.5132514899343605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0056928717308694e-05,
+      "loss": 0.8864,
+      "step": 18688
+    },
+    {
+      "epoch": 0.5132789541622037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.005603921548615e-05,
+      "loss": 0.9608,
+      "step": 18689
+    },
+    {
+      "epoch": 0.513306418390047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0055149713220202e-05,
+      "loss": 0.8512,
+      "step": 18690
+    },
+    {
+      "epoch": 0.5133338826178901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0054260210517888e-05,
+      "loss": 1.0894,
+      "step": 18691
+    },
+    {
+      "epoch": 0.5133613468457334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0053370707386244e-05,
+      "loss": 0.9415,
+      "step": 18692
+    },
+    {
+      "epoch": 0.5133888110735767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0052481203832311e-05,
+      "loss": 0.83,
+      "step": 18693
+    },
+    {
+      "epoch": 0.5134162753014199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0051591699863128e-05,
+      "loss": 0.8839,
+      "step": 18694
+    },
+    {
+      "epoch": 0.5134437395292631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0050702195485733e-05,
+      "loss": 0.9839,
+      "step": 18695
+    },
+    {
+      "epoch": 0.5134712037571064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0049812690707156e-05,
+      "loss": 0.8762,
+      "step": 18696
+    },
+    {
+      "epoch": 0.5134986679849496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0048923185534447e-05,
+      "loss": 0.8953,
+      "step": 18697
+    },
+    {
+      "epoch": 0.5135261322127929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0048033679974638e-05,
+      "loss": 0.881,
+      "step": 18698
+    },
+    {
+      "epoch": 0.513553596440636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0047144174034768e-05,
+      "loss": 0.87,
+      "step": 18699
+    },
+    {
+      "epoch": 0.5135810606684793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0046254667721873e-05,
+      "loss": 0.8602,
+      "step": 18700
+    },
+    {
+      "epoch": 0.5136085248963226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0045365161042998e-05,
+      "loss": 0.8805,
+      "step": 18701
+    },
+    {
+      "epoch": 0.5136359891241657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0044475654005172e-05,
+      "loss": 0.91,
+      "step": 18702
+    },
+    {
+      "epoch": 0.513663453352009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0043586146615438e-05,
+      "loss": 0.8379,
+      "step": 18703
+    },
+    {
+      "epoch": 0.5136909175798522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0042696638880833e-05,
+      "loss": 0.7903,
+      "step": 18704
+    },
+    {
+      "epoch": 0.5137183818076955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0041807130808396e-05,
+      "loss": 0.9584,
+      "step": 18705
+    },
+    {
+      "epoch": 0.5137458460355387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0040917622405167e-05,
+      "loss": 0.9327,
+      "step": 18706
+    },
+    {
+      "epoch": 0.5137733102633819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0040028113678181e-05,
+      "loss": 0.8549,
+      "step": 18707
+    },
+    {
+      "epoch": 0.5138007744912252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0039138604634475e-05,
+      "loss": 0.9008,
+      "step": 18708
+    },
+    {
+      "epoch": 0.5138282387190685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.003824909528109e-05,
+      "loss": 0.9262,
+      "step": 18709
+    },
+    {
+      "epoch": 0.5138557029469116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0037359585625064e-05,
+      "loss": 0.979,
+      "step": 18710
+    },
+    {
+      "epoch": 0.5138831671747549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0036470075673436e-05,
+      "loss": 0.9063,
+      "step": 18711
+    },
+    {
+      "epoch": 0.5139106314025981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.003558056543324e-05,
+      "loss": 0.8697,
+      "step": 18712
+    },
+    {
+      "epoch": 0.5139380956304413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0034691054911519e-05,
+      "loss": 0.9457,
+      "step": 18713
+    },
+    {
+      "epoch": 0.5139655598582846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0033801544115307e-05,
+      "loss": 0.8983,
+      "step": 18714
+    },
+    {
+      "epoch": 0.5139930240861278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0032912033051645e-05,
+      "loss": 0.888,
+      "step": 18715
+    },
+    {
+      "epoch": 0.5140204883139711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.003202252172757e-05,
+      "loss": 0.8257,
+      "step": 18716
+    },
+    {
+      "epoch": 0.5140479525418142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0031133010150121e-05,
+      "loss": 0.8864,
+      "step": 18717
+    },
+    {
+      "epoch": 0.5140754167696575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0030243498326336e-05,
+      "loss": 0.9062,
+      "step": 18718
+    },
+    {
+      "epoch": 0.5141028809975008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.002935398626325e-05,
+      "loss": 0.8602,
+      "step": 18719
+    },
+    {
+      "epoch": 0.514130345225344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0028464473967905e-05,
+      "loss": 0.843,
+      "step": 18720
+    },
+    {
+      "epoch": 0.5141578094531872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0027574961447338e-05,
+      "loss": 0.8654,
+      "step": 18721
+    },
+    {
+      "epoch": 0.5141852736810305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0026685448708588e-05,
+      "loss": 0.8596,
+      "step": 18722
+    },
+    {
+      "epoch": 0.5142127379088737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0025795935758696e-05,
+      "loss": 0.7864,
+      "step": 18723
+    },
+    {
+      "epoch": 0.514240202136717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0024906422604693e-05,
+      "loss": 0.826,
+      "step": 18724
+    },
+    {
+      "epoch": 0.5142676663645601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0024016909253617e-05,
+      "loss": 0.8927,
+      "step": 18725
+    },
+    {
+      "epoch": 0.5142951305924034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0023127395712515e-05,
+      "loss": 0.9923,
+      "step": 18726
+    },
+    {
+      "epoch": 0.5143225948202467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0022237881988421e-05,
+      "loss": 0.9352,
+      "step": 18727
+    },
+    {
+      "epoch": 0.5143500590480898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0021348368088369e-05,
+      "loss": 0.9183,
+      "step": 18728
+    },
+    {
+      "epoch": 0.5143775232759331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0020458854019399e-05,
+      "loss": 0.8613,
+      "step": 18729
+    },
+    {
+      "epoch": 0.5144049875037763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0019569339788556e-05,
+      "loss": 0.7481,
+      "step": 18730
+    },
+    {
+      "epoch": 0.5144324517316196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0018679825402869e-05,
+      "loss": 0.7361,
+      "step": 18731
+    },
+    {
+      "epoch": 0.5144599159594628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.001779031086938e-05,
+      "loss": 0.9116,
+      "step": 18732
+    },
+    {
+      "epoch": 0.514487380187306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0016900796195128e-05,
+      "loss": 0.877,
+      "step": 18733
+    },
+    {
+      "epoch": 0.5145148444151493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0016011281387151e-05,
+      "loss": 0.9146,
+      "step": 18734
+    },
+    {
+      "epoch": 0.5145423086429926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0015121766452485e-05,
+      "loss": 0.9791,
+      "step": 18735
+    },
+    {
+      "epoch": 0.5145697728708357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.001423225139817e-05,
+      "loss": 0.7772,
+      "step": 18736
+    },
+    {
+      "epoch": 0.514597237098679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0013342736231244e-05,
+      "loss": 0.9175,
+      "step": 18737
+    },
+    {
+      "epoch": 0.5146247013265222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0012453220958747e-05,
+      "loss": 0.8776,
+      "step": 18738
+    },
+    {
+      "epoch": 0.5146521655543654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0011563705587715e-05,
+      "loss": 0.8253,
+      "step": 18739
+    },
+    {
+      "epoch": 0.5146796297822087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0010674190125184e-05,
+      "loss": 1.0053,
+      "step": 18740
+    },
+    {
+      "epoch": 0.5147070940100519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0009784674578193e-05,
+      "loss": 0.9166,
+      "step": 18741
+    },
+    {
+      "epoch": 0.5147345582378952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0008895158953786e-05,
+      "loss": 0.979,
+      "step": 18742
+    },
+    {
+      "epoch": 0.5147620224657383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0008005643258997e-05,
+      "loss": 1.0109,
+      "step": 18743
+    },
+    {
+      "epoch": 0.5147894866935816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0007116127500862e-05,
+      "loss": 1.0033,
+      "step": 18744
+    },
+    {
+      "epoch": 0.5148169509214249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0006226611686425e-05,
+      "loss": 0.907,
+      "step": 18745
+    },
+    {
+      "epoch": 0.514844415149268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0005337095822716e-05,
+      "loss": 0.9256,
+      "step": 18746
+    },
+    {
+      "epoch": 0.5148718793771113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0004447579916781e-05,
+      "loss": 0.8923,
+      "step": 18747
+    },
+    {
+      "epoch": 0.5148993436049546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0003558063975655e-05,
+      "loss": 0.8809,
+      "step": 18748
+    },
+    {
+      "epoch": 0.5149268078327978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0002668548006375e-05,
+      "loss": 1.0017,
+      "step": 18749
+    },
+    {
+      "epoch": 0.514954272060641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.000177903201598e-05,
+      "loss": 0.8491,
+      "step": 18750
+    },
+    {
+      "epoch": 0.5149817362884842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0000889516011512e-05,
+      "loss": 0.8701,
+      "step": 18751
+    },
+    {
+      "epoch": 0.5150092005163275,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.9933,
+      "step": 18752
+    },
+    {
+      "epoch": 0.5150366647441708,
+      "grad_norm": 0.0,
+      "learning_rate": 9.999110483988491e-06,
+      "loss": 0.8959,
+      "step": 18753
+    },
+    {
+      "epoch": 0.5150641289720139,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99822096798402e-06,
+      "loss": 0.9578,
+      "step": 18754
+    },
+    {
+      "epoch": 0.5150915931998572,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99733145199363e-06,
+      "loss": 0.8162,
+      "step": 18755
+    },
+    {
+      "epoch": 0.5151190574277004,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99644193602435e-06,
+      "loss": 0.8285,
+      "step": 18756
+    },
+    {
+      "epoch": 0.5151465216555436,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99555242008322e-06,
+      "loss": 0.9097,
+      "step": 18757
+    },
+    {
+      "epoch": 0.5151739858833869,
+      "grad_norm": 0.0,
+      "learning_rate": 9.994662904177285e-06,
+      "loss": 1.0172,
+      "step": 18758
+    },
+    {
+      "epoch": 0.5152014501112301,
+      "grad_norm": 0.0,
+      "learning_rate": 9.993773388313578e-06,
+      "loss": 0.9265,
+      "step": 18759
+    },
+    {
+      "epoch": 0.5152289143390734,
+      "grad_norm": 0.0,
+      "learning_rate": 9.992883872499137e-06,
+      "loss": 0.8806,
+      "step": 18760
+    },
+    {
+      "epoch": 0.5152563785669166,
+      "grad_norm": 0.0,
+      "learning_rate": 9.991994356741008e-06,
+      "loss": 0.9003,
+      "step": 18761
+    },
+    {
+      "epoch": 0.5152838427947598,
+      "grad_norm": 0.0,
+      "learning_rate": 9.991104841046216e-06,
+      "loss": 0.8775,
+      "step": 18762
+    },
+    {
+      "epoch": 0.5153113070226031,
+      "grad_norm": 0.0,
+      "learning_rate": 9.990215325421809e-06,
+      "loss": 0.9421,
+      "step": 18763
+    },
+    {
+      "epoch": 0.5153387712504462,
+      "grad_norm": 0.0,
+      "learning_rate": 9.98932580987482e-06,
+      "loss": 0.8745,
+      "step": 18764
+    },
+    {
+      "epoch": 0.5153662354782895,
+      "grad_norm": 0.0,
+      "learning_rate": 9.988436294412289e-06,
+      "loss": 0.8753,
+      "step": 18765
+    },
+    {
+      "epoch": 0.5153936997061328,
+      "grad_norm": 0.0,
+      "learning_rate": 9.987546779041256e-06,
+      "loss": 0.8956,
+      "step": 18766
+    },
+    {
+      "epoch": 0.515421163933976,
+      "grad_norm": 0.0,
+      "learning_rate": 9.986657263768761e-06,
+      "loss": 0.8467,
+      "step": 18767
+    },
+    {
+      "epoch": 0.5154486281618192,
+      "grad_norm": 0.0,
+      "learning_rate": 9.985767748601833e-06,
+      "loss": 0.8729,
+      "step": 18768
+    },
+    {
+      "epoch": 0.5154760923896624,
+      "grad_norm": 0.0,
+      "learning_rate": 9.984878233547519e-06,
+      "loss": 0.9449,
+      "step": 18769
+    },
+    {
+      "epoch": 0.5155035566175057,
+      "grad_norm": 0.0,
+      "learning_rate": 9.983988718612854e-06,
+      "loss": 0.9011,
+      "step": 18770
+    },
+    {
+      "epoch": 0.515531020845349,
+      "grad_norm": 0.0,
+      "learning_rate": 9.983099203804875e-06,
+      "loss": 0.919,
+      "step": 18771
+    },
+    {
+      "epoch": 0.5155584850731921,
+      "grad_norm": 0.0,
+      "learning_rate": 9.98220968913062e-06,
+      "loss": 0.909,
+      "step": 18772
+    },
+    {
+      "epoch": 0.5155859493010354,
+      "grad_norm": 0.0,
+      "learning_rate": 9.981320174597136e-06,
+      "loss": 0.8473,
+      "step": 18773
+    },
+    {
+      "epoch": 0.5156134135288787,
+      "grad_norm": 0.0,
+      "learning_rate": 9.980430660211449e-06,
+      "loss": 0.8111,
+      "step": 18774
+    },
+    {
+      "epoch": 0.5156408777567218,
+      "grad_norm": 0.0,
+      "learning_rate": 9.979541145980603e-06,
+      "loss": 0.9858,
+      "step": 18775
+    },
+    {
+      "epoch": 0.5156683419845651,
+      "grad_norm": 0.0,
+      "learning_rate": 9.978651631911635e-06,
+      "loss": 0.9963,
+      "step": 18776
+    },
+    {
+      "epoch": 0.5156958062124083,
+      "grad_norm": 0.0,
+      "learning_rate": 9.977762118011582e-06,
+      "loss": 0.8831,
+      "step": 18777
+    },
+    {
+      "epoch": 0.5157232704402516,
+      "grad_norm": 0.0,
+      "learning_rate": 9.976872604287487e-06,
+      "loss": 0.9156,
+      "step": 18778
+    },
+    {
+      "epoch": 0.5157507346680948,
+      "grad_norm": 0.0,
+      "learning_rate": 9.975983090746386e-06,
+      "loss": 0.9006,
+      "step": 18779
+    },
+    {
+      "epoch": 0.515778198895938,
+      "grad_norm": 0.0,
+      "learning_rate": 9.975093577395312e-06,
+      "loss": 0.9764,
+      "step": 18780
+    },
+    {
+      "epoch": 0.5158056631237813,
+      "grad_norm": 0.0,
+      "learning_rate": 9.97420406424131e-06,
+      "loss": 0.9674,
+      "step": 18781
+    },
+    {
+      "epoch": 0.5158331273516245,
+      "grad_norm": 0.0,
+      "learning_rate": 9.973314551291414e-06,
+      "loss": 0.8582,
+      "step": 18782
+    },
+    {
+      "epoch": 0.5158605915794677,
+      "grad_norm": 0.0,
+      "learning_rate": 9.972425038552662e-06,
+      "loss": 0.8712,
+      "step": 18783
+    },
+    {
+      "epoch": 0.515888055807311,
+      "grad_norm": 0.0,
+      "learning_rate": 9.971535526032094e-06,
+      "loss": 0.9774,
+      "step": 18784
+    },
+    {
+      "epoch": 0.5159155200351542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.970646013736753e-06,
+      "loss": 0.8764,
+      "step": 18785
+    },
+    {
+      "epoch": 0.5159429842629975,
+      "grad_norm": 0.0,
+      "learning_rate": 9.96975650167367e-06,
+      "loss": 0.8478,
+      "step": 18786
+    },
+    {
+      "epoch": 0.5159704484908407,
+      "grad_norm": 0.0,
+      "learning_rate": 9.968866989849884e-06,
+      "loss": 0.9064,
+      "step": 18787
+    },
+    {
+      "epoch": 0.5159979127186839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.967977478272433e-06,
+      "loss": 0.9544,
+      "step": 18788
+    },
+    {
+      "epoch": 0.5160253769465272,
+      "grad_norm": 0.0,
+      "learning_rate": 9.967087966948356e-06,
+      "loss": 0.8871,
+      "step": 18789
+    },
+    {
+      "epoch": 0.5160528411743703,
+      "grad_norm": 0.0,
+      "learning_rate": 9.966198455884695e-06,
+      "loss": 0.9762,
+      "step": 18790
+    },
+    {
+      "epoch": 0.5160803054022136,
+      "grad_norm": 0.0,
+      "learning_rate": 9.965308945088486e-06,
+      "loss": 0.8023,
+      "step": 18791
+    },
+    {
+      "epoch": 0.5161077696300569,
+      "grad_norm": 0.0,
+      "learning_rate": 9.964419434566763e-06,
+      "loss": 0.903,
+      "step": 18792
+    },
+    {
+      "epoch": 0.5161352338579,
+      "grad_norm": 0.0,
+      "learning_rate": 9.963529924326567e-06,
+      "loss": 0.9314,
+      "step": 18793
+    },
+    {
+      "epoch": 0.5161626980857433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.96264041437494e-06,
+      "loss": 0.9098,
+      "step": 18794
+    },
+    {
+      "epoch": 0.5161901623135865,
+      "grad_norm": 0.0,
+      "learning_rate": 9.961750904718911e-06,
+      "loss": 0.8339,
+      "step": 18795
+    },
+    {
+      "epoch": 0.5162176265414298,
+      "grad_norm": 0.0,
+      "learning_rate": 9.960861395365525e-06,
+      "loss": 0.9079,
+      "step": 18796
+    },
+    {
+      "epoch": 0.516245090769273,
+      "grad_norm": 0.0,
+      "learning_rate": 9.959971886321824e-06,
+      "loss": 0.8575,
+      "step": 18797
+    },
+    {
+      "epoch": 0.5162725549971162,
+      "grad_norm": 0.0,
+      "learning_rate": 9.959082377594835e-06,
+      "loss": 0.8838,
+      "step": 18798
+    },
+    {
+      "epoch": 0.5163000192249595,
+      "grad_norm": 0.0,
+      "learning_rate": 9.958192869191606e-06,
+      "loss": 0.9509,
+      "step": 18799
+    },
+    {
+      "epoch": 0.5163274834528028,
+      "grad_norm": 0.0,
+      "learning_rate": 9.957303361119168e-06,
+      "loss": 0.8399,
+      "step": 18800
+    },
+    {
+      "epoch": 0.5163549476806459,
+      "grad_norm": 0.0,
+      "learning_rate": 9.956413853384563e-06,
+      "loss": 0.8731,
+      "step": 18801
+    },
+    {
+      "epoch": 0.5163824119084892,
+      "grad_norm": 0.0,
+      "learning_rate": 9.95552434599483e-06,
+      "loss": 0.9075,
+      "step": 18802
+    },
+    {
+      "epoch": 0.5164098761363324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.954634838957009e-06,
+      "loss": 0.8576,
+      "step": 18803
+    },
+    {
+      "epoch": 0.5164373403641757,
+      "grad_norm": 0.0,
+      "learning_rate": 9.953745332278129e-06,
+      "loss": 0.9247,
+      "step": 18804
+    },
+    {
+      "epoch": 0.5164648045920189,
+      "grad_norm": 0.0,
+      "learning_rate": 9.952855825965235e-06,
+      "loss": 0.9032,
+      "step": 18805
+    },
+    {
+      "epoch": 0.5164922688198621,
+      "grad_norm": 0.0,
+      "learning_rate": 9.951966320025364e-06,
+      "loss": 1.0318,
+      "step": 18806
+    },
+    {
+      "epoch": 0.5165197330477054,
+      "grad_norm": 0.0,
+      "learning_rate": 9.951076814465555e-06,
+      "loss": 0.8457,
+      "step": 18807
+    },
+    {
+      "epoch": 0.5165471972755487,
+      "grad_norm": 0.0,
+      "learning_rate": 9.950187309292842e-06,
+      "loss": 0.8688,
+      "step": 18808
+    },
+    {
+      "epoch": 0.5165746615033918,
+      "grad_norm": 0.0,
+      "learning_rate": 9.949297804514273e-06,
+      "loss": 0.9716,
+      "step": 18809
+    },
+    {
+      "epoch": 0.5166021257312351,
+      "grad_norm": 0.0,
+      "learning_rate": 9.948408300136875e-06,
+      "loss": 0.8612,
+      "step": 18810
+    },
+    {
+      "epoch": 0.5166295899590783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.94751879616769e-06,
+      "loss": 0.8519,
+      "step": 18811
+    },
+    {
+      "epoch": 0.5166570541869215,
+      "grad_norm": 0.0,
+      "learning_rate": 9.946629292613757e-06,
+      "loss": 0.9067,
+      "step": 18812
+    },
+    {
+      "epoch": 0.5166845184147648,
+      "grad_norm": 0.0,
+      "learning_rate": 9.945739789482114e-06,
+      "loss": 0.9349,
+      "step": 18813
+    },
+    {
+      "epoch": 0.516711982642608,
+      "grad_norm": 0.0,
+      "learning_rate": 9.944850286779803e-06,
+      "loss": 0.9917,
+      "step": 18814
+    },
+    {
+      "epoch": 0.5167394468704513,
+      "grad_norm": 0.0,
+      "learning_rate": 9.943960784513854e-06,
+      "loss": 0.8934,
+      "step": 18815
+    },
+    {
+      "epoch": 0.5167669110982944,
+      "grad_norm": 0.0,
+      "learning_rate": 9.943071282691307e-06,
+      "loss": 0.8655,
+      "step": 18816
+    },
+    {
+      "epoch": 0.5167943753261377,
+      "grad_norm": 0.0,
+      "learning_rate": 9.942181781319204e-06,
+      "loss": 0.9054,
+      "step": 18817
+    },
+    {
+      "epoch": 0.516821839553981,
+      "grad_norm": 0.0,
+      "learning_rate": 9.941292280404583e-06,
+      "loss": 0.9437,
+      "step": 18818
+    },
+    {
+      "epoch": 0.5168493037818241,
+      "grad_norm": 0.0,
+      "learning_rate": 9.940402779954476e-06,
+      "loss": 0.8682,
+      "step": 18819
+    },
+    {
+      "epoch": 0.5168767680096674,
+      "grad_norm": 0.0,
+      "learning_rate": 9.93951327997593e-06,
+      "loss": 0.8665,
+      "step": 18820
+    },
+    {
+      "epoch": 0.5169042322375107,
+      "grad_norm": 0.0,
+      "learning_rate": 9.938623780475976e-06,
+      "loss": 0.9828,
+      "step": 18821
+    },
+    {
+      "epoch": 0.5169316964653539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.937734281461653e-06,
+      "loss": 0.8617,
+      "step": 18822
+    },
+    {
+      "epoch": 0.5169591606931971,
+      "grad_norm": 0.0,
+      "learning_rate": 9.936844782940002e-06,
+      "loss": 0.9727,
+      "step": 18823
+    },
+    {
+      "epoch": 0.5169866249210403,
+      "grad_norm": 0.0,
+      "learning_rate": 9.93595528491806e-06,
+      "loss": 0.8775,
+      "step": 18824
+    },
+    {
+      "epoch": 0.5170140891488836,
+      "grad_norm": 0.0,
+      "learning_rate": 9.935065787402863e-06,
+      "loss": 0.9369,
+      "step": 18825
+    },
+    {
+      "epoch": 0.5170415533767269,
+      "grad_norm": 0.0,
+      "learning_rate": 9.934176290401455e-06,
+      "loss": 0.9736,
+      "step": 18826
+    },
+    {
+      "epoch": 0.51706901760457,
+      "grad_norm": 0.0,
+      "learning_rate": 9.933286793920866e-06,
+      "loss": 0.876,
+      "step": 18827
+    },
+    {
+      "epoch": 0.5170964818324133,
+      "grad_norm": 0.0,
+      "learning_rate": 9.932397297968139e-06,
+      "loss": 0.7874,
+      "step": 18828
+    },
+    {
+      "epoch": 0.5171239460602565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.931507802550308e-06,
+      "loss": 0.8778,
+      "step": 18829
+    },
+    {
+      "epoch": 0.5171514102880997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.930618307674417e-06,
+      "loss": 0.8175,
+      "step": 18830
+    },
+    {
+      "epoch": 0.517178874515943,
+      "grad_norm": 0.0,
+      "learning_rate": 9.929728813347498e-06,
+      "loss": 0.9572,
+      "step": 18831
+    },
+    {
+      "epoch": 0.5172063387437862,
+      "grad_norm": 0.0,
+      "learning_rate": 9.928839319576597e-06,
+      "loss": 0.9752,
+      "step": 18832
+    },
+    {
+      "epoch": 0.5172338029716295,
+      "grad_norm": 0.0,
+      "learning_rate": 9.927949826368741e-06,
+      "loss": 0.9057,
+      "step": 18833
+    },
+    {
+      "epoch": 0.5172612671994727,
+      "grad_norm": 0.0,
+      "learning_rate": 9.927060333730975e-06,
+      "loss": 0.8449,
+      "step": 18834
+    },
+    {
+      "epoch": 0.5172887314273159,
+      "grad_norm": 0.0,
+      "learning_rate": 9.926170841670339e-06,
+      "loss": 0.8785,
+      "step": 18835
+    },
+    {
+      "epoch": 0.5173161956551592,
+      "grad_norm": 0.0,
+      "learning_rate": 9.925281350193864e-06,
+      "loss": 0.8224,
+      "step": 18836
+    },
+    {
+      "epoch": 0.5173436598830023,
+      "grad_norm": 0.0,
+      "learning_rate": 9.924391859308592e-06,
+      "loss": 0.8833,
+      "step": 18837
+    },
+    {
+      "epoch": 0.5173711241108456,
+      "grad_norm": 0.0,
+      "learning_rate": 9.923502369021567e-06,
+      "loss": 0.9197,
+      "step": 18838
+    },
+    {
+      "epoch": 0.5173985883386889,
+      "grad_norm": 0.0,
+      "learning_rate": 9.922612879339817e-06,
+      "loss": 0.8964,
+      "step": 18839
+    },
+    {
+      "epoch": 0.5174260525665321,
+      "grad_norm": 0.0,
+      "learning_rate": 9.921723390270383e-06,
+      "loss": 0.7841,
+      "step": 18840
+    },
+    {
+      "epoch": 0.5174535167943753,
+      "grad_norm": 0.0,
+      "learning_rate": 9.920833901820302e-06,
+      "loss": 0.9715,
+      "step": 18841
+    },
+    {
+      "epoch": 0.5174809810222185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.919944413996618e-06,
+      "loss": 0.7802,
+      "step": 18842
+    },
+    {
+      "epoch": 0.5175084452500618,
+      "grad_norm": 0.0,
+      "learning_rate": 9.919054926806362e-06,
+      "loss": 0.805,
+      "step": 18843
+    },
+    {
+      "epoch": 0.5175359094779051,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91816544025658e-06,
+      "loss": 0.928,
+      "step": 18844
+    },
+    {
+      "epoch": 0.5175633737057482,
+      "grad_norm": 0.0,
+      "learning_rate": 9.9172759543543e-06,
+      "loss": 0.8823,
+      "step": 18845
+    },
+    {
+      "epoch": 0.5175908379335915,
+      "grad_norm": 0.0,
+      "learning_rate": 9.916386469106562e-06,
+      "loss": 0.8382,
+      "step": 18846
+    },
+    {
+      "epoch": 0.5176183021614348,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91549698452041e-06,
+      "loss": 0.8601,
+      "step": 18847
+    },
+    {
+      "epoch": 0.517645766389278,
+      "grad_norm": 0.0,
+      "learning_rate": 9.914607500602878e-06,
+      "loss": 0.8856,
+      "step": 18848
+    },
+    {
+      "epoch": 0.5176732306171212,
+      "grad_norm": 0.0,
+      "learning_rate": 9.913718017361003e-06,
+      "loss": 0.949,
+      "step": 18849
+    },
+    {
+      "epoch": 0.5177006948449644,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91282853480183e-06,
+      "loss": 0.8341,
+      "step": 18850
+    },
+    {
+      "epoch": 0.5177281590728077,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91193905293239e-06,
+      "loss": 0.8644,
+      "step": 18851
+    },
+    {
+      "epoch": 0.517755623300651,
+      "grad_norm": 0.0,
+      "learning_rate": 9.911049571759719e-06,
+      "loss": 0.8079,
+      "step": 18852
+    },
+    {
+      "epoch": 0.5177830875284941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.910160091290857e-06,
+      "loss": 0.9489,
+      "step": 18853
+    },
+    {
+      "epoch": 0.5178105517563374,
+      "grad_norm": 0.0,
+      "learning_rate": 9.909270611532848e-06,
+      "loss": 0.8722,
+      "step": 18854
+    },
+    {
+      "epoch": 0.5178380159841806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.908381132492722e-06,
+      "loss": 0.742,
+      "step": 18855
+    },
+    {
+      "epoch": 0.5178654802120238,
+      "grad_norm": 0.0,
+      "learning_rate": 9.907491654177523e-06,
+      "loss": 0.8982,
+      "step": 18856
+    },
+    {
+      "epoch": 0.5178929444398671,
+      "grad_norm": 0.0,
+      "learning_rate": 9.906602176594282e-06,
+      "loss": 0.8187,
+      "step": 18857
+    },
+    {
+      "epoch": 0.5179204086677103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90571269975004e-06,
+      "loss": 0.8877,
+      "step": 18858
+    },
+    {
+      "epoch": 0.5179478728955536,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90482322365184e-06,
+      "loss": 0.9897,
+      "step": 18859
+    },
+    {
+      "epoch": 0.5179753371233968,
+      "grad_norm": 0.0,
+      "learning_rate": 9.903933748306713e-06,
+      "loss": 0.9069,
+      "step": 18860
+    },
+    {
+      "epoch": 0.51800280135124,
+      "grad_norm": 0.0,
+      "learning_rate": 9.903044273721697e-06,
+      "loss": 0.9642,
+      "step": 18861
+    },
+    {
+      "epoch": 0.5180302655790833,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90215479990384e-06,
+      "loss": 0.9369,
+      "step": 18862
+    },
+    {
+      "epoch": 0.5180577298069264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.901265326860167e-06,
+      "loss": 0.9145,
+      "step": 18863
+    },
+    {
+      "epoch": 0.5180851940347697,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90037585459772e-06,
+      "loss": 0.9383,
+      "step": 18864
+    },
+    {
+      "epoch": 0.518112658262613,
+      "grad_norm": 0.0,
+      "learning_rate": 9.899486383123538e-06,
+      "loss": 0.8205,
+      "step": 18865
+    },
+    {
+      "epoch": 0.5181401224904562,
+      "grad_norm": 0.0,
+      "learning_rate": 9.898596912444661e-06,
+      "loss": 0.8448,
+      "step": 18866
+    },
+    {
+      "epoch": 0.5181675867182994,
+      "grad_norm": 0.0,
+      "learning_rate": 9.897707442568123e-06,
+      "loss": 0.8508,
+      "step": 18867
+    },
+    {
+      "epoch": 0.5181950509461426,
+      "grad_norm": 0.0,
+      "learning_rate": 9.896817973500965e-06,
+      "loss": 0.9194,
+      "step": 18868
+    },
+    {
+      "epoch": 0.5182225151739859,
+      "grad_norm": 0.0,
+      "learning_rate": 9.895928505250221e-06,
+      "loss": 0.8702,
+      "step": 18869
+    },
+    {
+      "epoch": 0.5182499794018292,
+      "grad_norm": 0.0,
+      "learning_rate": 9.89503903782293e-06,
+      "loss": 0.9315,
+      "step": 18870
+    },
+    {
+      "epoch": 0.5182774436296723,
+      "grad_norm": 0.0,
+      "learning_rate": 9.894149571226134e-06,
+      "loss": 0.8565,
+      "step": 18871
+    },
+    {
+      "epoch": 0.5183049078575156,
+      "grad_norm": 0.0,
+      "learning_rate": 9.893260105466865e-06,
+      "loss": 0.9292,
+      "step": 18872
+    },
+    {
+      "epoch": 0.5183323720853589,
+      "grad_norm": 0.0,
+      "learning_rate": 9.892370640552168e-06,
+      "loss": 0.8552,
+      "step": 18873
+    },
+    {
+      "epoch": 0.518359836313202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.89148117648907e-06,
+      "loss": 0.8764,
+      "step": 18874
+    },
+    {
+      "epoch": 0.5183873005410453,
+      "grad_norm": 0.0,
+      "learning_rate": 9.890591713284619e-06,
+      "loss": 0.9395,
+      "step": 18875
+    },
+    {
+      "epoch": 0.5184147647688885,
+      "grad_norm": 0.0,
+      "learning_rate": 9.889702250945847e-06,
+      "loss": 0.7763,
+      "step": 18876
+    },
+    {
+      "epoch": 0.5184422289967318,
+      "grad_norm": 0.0,
+      "learning_rate": 9.888812789479793e-06,
+      "loss": 0.9199,
+      "step": 18877
+    },
+    {
+      "epoch": 0.518469693224575,
+      "grad_norm": 0.0,
+      "learning_rate": 9.887923328893497e-06,
+      "loss": 0.8417,
+      "step": 18878
+    },
+    {
+      "epoch": 0.5184971574524182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.887033869193996e-06,
+      "loss": 0.841,
+      "step": 18879
+    },
+    {
+      "epoch": 0.5185246216802615,
+      "grad_norm": 0.0,
+      "learning_rate": 9.886144410388324e-06,
+      "loss": 0.9136,
+      "step": 18880
+    },
+    {
+      "epoch": 0.5185520859081046,
+      "grad_norm": 0.0,
+      "learning_rate": 9.885254952483522e-06,
+      "loss": 0.8643,
+      "step": 18881
+    },
+    {
+      "epoch": 0.5185795501359479,
+      "grad_norm": 0.0,
+      "learning_rate": 9.884365495486626e-06,
+      "loss": 0.8013,
+      "step": 18882
+    },
+    {
+      "epoch": 0.5186070143637912,
+      "grad_norm": 0.0,
+      "learning_rate": 9.883476039404678e-06,
+      "loss": 0.8921,
+      "step": 18883
+    },
+    {
+      "epoch": 0.5186344785916344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.88258658424471e-06,
+      "loss": 0.9512,
+      "step": 18884
+    },
+    {
+      "epoch": 0.5186619428194776,
+      "grad_norm": 0.0,
+      "learning_rate": 9.881697130013768e-06,
+      "loss": 0.9603,
+      "step": 18885
+    },
+    {
+      "epoch": 0.5186894070473209,
+      "grad_norm": 0.0,
+      "learning_rate": 9.880807676718878e-06,
+      "loss": 0.9293,
+      "step": 18886
+    },
+    {
+      "epoch": 0.5187168712751641,
+      "grad_norm": 0.0,
+      "learning_rate": 9.879918224367087e-06,
+      "loss": 0.9074,
+      "step": 18887
+    },
+    {
+      "epoch": 0.5187443355030074,
+      "grad_norm": 0.0,
+      "learning_rate": 9.879028772965428e-06,
+      "loss": 0.8638,
+      "step": 18888
+    },
+    {
+      "epoch": 0.5187717997308505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.878139322520938e-06,
+      "loss": 0.8788,
+      "step": 18889
+    },
+    {
+      "epoch": 0.5187992639586938,
+      "grad_norm": 0.0,
+      "learning_rate": 9.87724987304066e-06,
+      "loss": 0.8205,
+      "step": 18890
+    },
+    {
+      "epoch": 0.5188267281865371,
+      "grad_norm": 0.0,
+      "learning_rate": 9.876360424531632e-06,
+      "loss": 0.7977,
+      "step": 18891
+    },
+    {
+      "epoch": 0.5188541924143802,
+      "grad_norm": 0.0,
+      "learning_rate": 9.875470977000883e-06,
+      "loss": 0.9638,
+      "step": 18892
+    },
+    {
+      "epoch": 0.5188816566422235,
+      "grad_norm": 0.0,
+      "learning_rate": 9.874581530455456e-06,
+      "loss": 0.8257,
+      "step": 18893
+    },
+    {
+      "epoch": 0.5189091208700667,
+      "grad_norm": 0.0,
+      "learning_rate": 9.873692084902388e-06,
+      "loss": 0.8822,
+      "step": 18894
+    },
+    {
+      "epoch": 0.51893658509791,
+      "grad_norm": 0.0,
+      "learning_rate": 9.872802640348721e-06,
+      "loss": 0.8512,
+      "step": 18895
+    },
+    {
+      "epoch": 0.5189640493257532,
+      "grad_norm": 0.0,
+      "learning_rate": 9.871913196801485e-06,
+      "loss": 0.8619,
+      "step": 18896
+    },
+    {
+      "epoch": 0.5189915135535964,
+      "grad_norm": 0.0,
+      "learning_rate": 9.871023754267726e-06,
+      "loss": 0.8201,
+      "step": 18897
+    },
+    {
+      "epoch": 0.5190189777814397,
+      "grad_norm": 0.0,
+      "learning_rate": 9.870134312754473e-06,
+      "loss": 0.9349,
+      "step": 18898
+    },
+    {
+      "epoch": 0.519046442009283,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86924487226877e-06,
+      "loss": 0.8879,
+      "step": 18899
+    },
+    {
+      "epoch": 0.5190739062371261,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86835543281765e-06,
+      "loss": 0.7848,
+      "step": 18900
+    },
+    {
+      "epoch": 0.5191013704649694,
+      "grad_norm": 0.0,
+      "learning_rate": 9.867465994408151e-06,
+      "loss": 0.9821,
+      "step": 18901
+    },
+    {
+      "epoch": 0.5191288346928126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.866576557047315e-06,
+      "loss": 0.8941,
+      "step": 18902
+    },
+    {
+      "epoch": 0.5191562989206558,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86568712074218e-06,
+      "loss": 0.9807,
+      "step": 18903
+    },
+    {
+      "epoch": 0.5191837631484991,
+      "grad_norm": 0.0,
+      "learning_rate": 9.864797685499779e-06,
+      "loss": 0.9055,
+      "step": 18904
+    },
+    {
+      "epoch": 0.5192112273763423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.863908251327148e-06,
+      "loss": 0.8445,
+      "step": 18905
+    },
+    {
+      "epoch": 0.5192386916041856,
+      "grad_norm": 0.0,
+      "learning_rate": 9.863018818231326e-06,
+      "loss": 0.8871,
+      "step": 18906
+    },
+    {
+      "epoch": 0.5192661558320287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.862129386219357e-06,
+      "loss": 0.8581,
+      "step": 18907
+    },
+    {
+      "epoch": 0.519293620059872,
+      "grad_norm": 0.0,
+      "learning_rate": 9.861239955298272e-06,
+      "loss": 0.9286,
+      "step": 18908
+    },
+    {
+      "epoch": 0.5193210842877153,
+      "grad_norm": 0.0,
+      "learning_rate": 9.860350525475113e-06,
+      "loss": 0.9619,
+      "step": 18909
+    },
+    {
+      "epoch": 0.5193485485155585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.859461096756912e-06,
+      "loss": 1.0211,
+      "step": 18910
+    },
+    {
+      "epoch": 0.5193760127434017,
+      "grad_norm": 0.0,
+      "learning_rate": 9.858571669150709e-06,
+      "loss": 0.8645,
+      "step": 18911
+    },
+    {
+      "epoch": 0.519403476971245,
+      "grad_norm": 0.0,
+      "learning_rate": 9.857682242663541e-06,
+      "loss": 0.9044,
+      "step": 18912
+    },
+    {
+      "epoch": 0.5194309411990882,
+      "grad_norm": 0.0,
+      "learning_rate": 9.856792817302447e-06,
+      "loss": 0.9274,
+      "step": 18913
+    },
+    {
+      "epoch": 0.5194584054269314,
+      "grad_norm": 0.0,
+      "learning_rate": 9.855903393074463e-06,
+      "loss": 0.8949,
+      "step": 18914
+    },
+    {
+      "epoch": 0.5194858696547746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.855013969986632e-06,
+      "loss": 0.9283,
+      "step": 18915
+    },
+    {
+      "epoch": 0.5195133338826179,
+      "grad_norm": 0.0,
+      "learning_rate": 9.854124548045982e-06,
+      "loss": 0.9096,
+      "step": 18916
+    },
+    {
+      "epoch": 0.5195407981104612,
+      "grad_norm": 0.0,
+      "learning_rate": 9.853235127259555e-06,
+      "loss": 0.7938,
+      "step": 18917
+    },
+    {
+      "epoch": 0.5195682623383043,
+      "grad_norm": 0.0,
+      "learning_rate": 9.852345707634389e-06,
+      "loss": 0.8702,
+      "step": 18918
+    },
+    {
+      "epoch": 0.5195957265661476,
+      "grad_norm": 0.0,
+      "learning_rate": 9.851456289177521e-06,
+      "loss": 0.9939,
+      "step": 18919
+    },
+    {
+      "epoch": 0.5196231907939908,
+      "grad_norm": 0.0,
+      "learning_rate": 9.850566871895987e-06,
+      "loss": 1.0241,
+      "step": 18920
+    },
+    {
+      "epoch": 0.519650655021834,
+      "grad_norm": 0.0,
+      "learning_rate": 9.849677455796832e-06,
+      "loss": 0.9013,
+      "step": 18921
+    },
+    {
+      "epoch": 0.5196781192496773,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84878804088708e-06,
+      "loss": 0.9303,
+      "step": 18922
+    },
+    {
+      "epoch": 0.5197055834775205,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84789862717378e-06,
+      "loss": 0.8697,
+      "step": 18923
+    },
+    {
+      "epoch": 0.5197330477053638,
+      "grad_norm": 0.0,
+      "learning_rate": 9.847009214663963e-06,
+      "loss": 0.8934,
+      "step": 18924
+    },
+    {
+      "epoch": 0.519760511933207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.846119803364666e-06,
+      "loss": 0.9404,
+      "step": 18925
+    },
+    {
+      "epoch": 0.5197879761610502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.845230393282933e-06,
+      "loss": 0.9159,
+      "step": 18926
+    },
+    {
+      "epoch": 0.5198154403888935,
+      "grad_norm": 0.0,
+      "learning_rate": 9.844340984425799e-06,
+      "loss": 0.9543,
+      "step": 18927
+    },
+    {
+      "epoch": 0.5198429046167367,
+      "grad_norm": 0.0,
+      "learning_rate": 9.843451576800295e-06,
+      "loss": 0.8762,
+      "step": 18928
+    },
+    {
+      "epoch": 0.5198703688445799,
+      "grad_norm": 0.0,
+      "learning_rate": 9.842562170413462e-06,
+      "loss": 0.786,
+      "step": 18929
+    },
+    {
+      "epoch": 0.5198978330724232,
+      "grad_norm": 0.0,
+      "learning_rate": 9.841672765272341e-06,
+      "loss": 0.9064,
+      "step": 18930
+    },
+    {
+      "epoch": 0.5199252973002664,
+      "grad_norm": 0.0,
+      "learning_rate": 9.840783361383963e-06,
+      "loss": 0.9849,
+      "step": 18931
+    },
+    {
+      "epoch": 0.5199527615281097,
+      "grad_norm": 0.0,
+      "learning_rate": 9.83989395875537e-06,
+      "loss": 0.843,
+      "step": 18932
+    },
+    {
+      "epoch": 0.5199802257559528,
+      "grad_norm": 0.0,
+      "learning_rate": 9.839004557393601e-06,
+      "loss": 0.8988,
+      "step": 18933
+    },
+    {
+      "epoch": 0.5200076899837961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.838115157305688e-06,
+      "loss": 0.8583,
+      "step": 18934
+    },
+    {
+      "epoch": 0.5200351542116394,
+      "grad_norm": 0.0,
+      "learning_rate": 9.837225758498671e-06,
+      "loss": 0.9787,
+      "step": 18935
+    },
+    {
+      "epoch": 0.5200626184394825,
+      "grad_norm": 0.0,
+      "learning_rate": 9.836336360979587e-06,
+      "loss": 0.9572,
+      "step": 18936
+    },
+    {
+      "epoch": 0.5200900826673258,
+      "grad_norm": 0.0,
+      "learning_rate": 9.83544696475547e-06,
+      "loss": 0.8734,
+      "step": 18937
+    },
+    {
+      "epoch": 0.5201175468951691,
+      "grad_norm": 0.0,
+      "learning_rate": 9.834557569833366e-06,
+      "loss": 0.9121,
+      "step": 18938
+    },
+    {
+      "epoch": 0.5201450111230123,
+      "grad_norm": 0.0,
+      "learning_rate": 9.833668176220305e-06,
+      "loss": 0.9015,
+      "step": 18939
+    },
+    {
+      "epoch": 0.5201724753508555,
+      "grad_norm": 0.0,
+      "learning_rate": 9.832778783923324e-06,
+      "loss": 0.9201,
+      "step": 18940
+    },
+    {
+      "epoch": 0.5201999395786987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.831889392949462e-06,
+      "loss": 0.9391,
+      "step": 18941
+    },
+    {
+      "epoch": 0.520227403806542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.831000003305757e-06,
+      "loss": 0.9341,
+      "step": 18942
+    },
+    {
+      "epoch": 0.5202548680343853,
+      "grad_norm": 0.0,
+      "learning_rate": 9.830110614999244e-06,
+      "loss": 0.8536,
+      "step": 18943
+    },
+    {
+      "epoch": 0.5202823322622284,
+      "grad_norm": 0.0,
+      "learning_rate": 9.829221228036966e-06,
+      "loss": 0.9727,
+      "step": 18944
+    },
+    {
+      "epoch": 0.5203097964900717,
+      "grad_norm": 0.0,
+      "learning_rate": 9.828331842425952e-06,
+      "loss": 0.8103,
+      "step": 18945
+    },
+    {
+      "epoch": 0.5203372607179149,
+      "grad_norm": 0.0,
+      "learning_rate": 9.827442458173242e-06,
+      "loss": 0.9531,
+      "step": 18946
+    },
+    {
+      "epoch": 0.5203647249457581,
+      "grad_norm": 0.0,
+      "learning_rate": 9.826553075285877e-06,
+      "loss": 0.9641,
+      "step": 18947
+    },
+    {
+      "epoch": 0.5203921891736014,
+      "grad_norm": 0.0,
+      "learning_rate": 9.825663693770889e-06,
+      "loss": 0.977,
+      "step": 18948
+    },
+    {
+      "epoch": 0.5204196534014446,
+      "grad_norm": 0.0,
+      "learning_rate": 9.824774313635316e-06,
+      "loss": 0.9265,
+      "step": 18949
+    },
+    {
+      "epoch": 0.5204471176292879,
+      "grad_norm": 0.0,
+      "learning_rate": 9.823884934886204e-06,
+      "loss": 0.97,
+      "step": 18950
+    },
+    {
+      "epoch": 0.5204745818571311,
+      "grad_norm": 0.0,
+      "learning_rate": 9.822995557530576e-06,
+      "loss": 0.9191,
+      "step": 18951
+    },
+    {
+      "epoch": 0.5205020460849743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.822106181575477e-06,
+      "loss": 0.8104,
+      "step": 18952
+    },
+    {
+      "epoch": 0.5205295103128176,
+      "grad_norm": 0.0,
+      "learning_rate": 9.821216807027943e-06,
+      "loss": 0.882,
+      "step": 18953
+    },
+    {
+      "epoch": 0.5205569745406607,
+      "grad_norm": 0.0,
+      "learning_rate": 9.820327433895011e-06,
+      "loss": 0.8555,
+      "step": 18954
+    },
+    {
+      "epoch": 0.520584438768504,
+      "grad_norm": 0.0,
+      "learning_rate": 9.819438062183718e-06,
+      "loss": 0.8222,
+      "step": 18955
+    },
+    {
+      "epoch": 0.5206119029963473,
+      "grad_norm": 0.0,
+      "learning_rate": 9.818548691901103e-06,
+      "loss": 0.8871,
+      "step": 18956
+    },
+    {
+      "epoch": 0.5206393672241905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.817659323054197e-06,
+      "loss": 0.8747,
+      "step": 18957
+    },
+    {
+      "epoch": 0.5206668314520337,
+      "grad_norm": 0.0,
+      "learning_rate": 9.816769955650041e-06,
+      "loss": 0.8776,
+      "step": 18958
+    },
+    {
+      "epoch": 0.5206942956798769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.815880589695675e-06,
+      "loss": 0.946,
+      "step": 18959
+    },
+    {
+      "epoch": 0.5207217599077202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.814991225198131e-06,
+      "loss": 0.899,
+      "step": 18960
+    },
+    {
+      "epoch": 0.5207492241355635,
+      "grad_norm": 0.0,
+      "learning_rate": 9.81410186216445e-06,
+      "loss": 0.8877,
+      "step": 18961
+    },
+    {
+      "epoch": 0.5207766883634066,
+      "grad_norm": 0.0,
+      "learning_rate": 9.813212500601668e-06,
+      "loss": 0.8605,
+      "step": 18962
+    },
+    {
+      "epoch": 0.5208041525912499,
+      "grad_norm": 0.0,
+      "learning_rate": 9.81232314051682e-06,
+      "loss": 0.835,
+      "step": 18963
+    },
+    {
+      "epoch": 0.5208316168190932,
+      "grad_norm": 0.0,
+      "learning_rate": 9.811433781916943e-06,
+      "loss": 1.0491,
+      "step": 18964
+    },
+    {
+      "epoch": 0.5208590810469363,
+      "grad_norm": 0.0,
+      "learning_rate": 9.810544424809074e-06,
+      "loss": 0.8808,
+      "step": 18965
+    },
+    {
+      "epoch": 0.5208865452747796,
+      "grad_norm": 0.0,
+      "learning_rate": 9.809655069200253e-06,
+      "loss": 0.9129,
+      "step": 18966
+    },
+    {
+      "epoch": 0.5209140095026228,
+      "grad_norm": 0.0,
+      "learning_rate": 9.808765715097512e-06,
+      "loss": 0.9284,
+      "step": 18967
+    },
+    {
+      "epoch": 0.5209414737304661,
+      "grad_norm": 0.0,
+      "learning_rate": 9.807876362507896e-06,
+      "loss": 0.8731,
+      "step": 18968
+    },
+    {
+      "epoch": 0.5209689379583093,
+      "grad_norm": 0.0,
+      "learning_rate": 9.806987011438432e-06,
+      "loss": 0.845,
+      "step": 18969
+    },
+    {
+      "epoch": 0.5209964021861525,
+      "grad_norm": 0.0,
+      "learning_rate": 9.806097661896162e-06,
+      "loss": 0.901,
+      "step": 18970
+    },
+    {
+      "epoch": 0.5210238664139958,
+      "grad_norm": 0.0,
+      "learning_rate": 9.805208313888126e-06,
+      "loss": 0.8753,
+      "step": 18971
+    },
+    {
+      "epoch": 0.521051330641839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.804318967421352e-06,
+      "loss": 0.9692,
+      "step": 18972
+    },
+    {
+      "epoch": 0.5210787948696822,
+      "grad_norm": 0.0,
+      "learning_rate": 9.803429622502883e-06,
+      "loss": 0.8627,
+      "step": 18973
+    },
+    {
+      "epoch": 0.5211062590975255,
+      "grad_norm": 0.0,
+      "learning_rate": 9.802540279139761e-06,
+      "loss": 0.8242,
+      "step": 18974
+    },
+    {
+      "epoch": 0.5211337233253687,
+      "grad_norm": 0.0,
+      "learning_rate": 9.801650937339014e-06,
+      "loss": 0.9238,
+      "step": 18975
+    },
+    {
+      "epoch": 0.521161187553212,
+      "grad_norm": 0.0,
+      "learning_rate": 9.800761597107678e-06,
+      "loss": 0.8207,
+      "step": 18976
+    },
+    {
+      "epoch": 0.5211886517810552,
+      "grad_norm": 0.0,
+      "learning_rate": 9.799872258452796e-06,
+      "loss": 0.8346,
+      "step": 18977
+    },
+    {
+      "epoch": 0.5212161160088984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.798982921381401e-06,
+      "loss": 0.9307,
+      "step": 18978
+    },
+    {
+      "epoch": 0.5212435802367417,
+      "grad_norm": 0.0,
+      "learning_rate": 9.798093585900531e-06,
+      "loss": 0.9832,
+      "step": 18979
+    },
+    {
+      "epoch": 0.5212710444645848,
+      "grad_norm": 0.0,
+      "learning_rate": 9.797204252017226e-06,
+      "loss": 0.8261,
+      "step": 18980
+    },
+    {
+      "epoch": 0.5212985086924281,
+      "grad_norm": 0.0,
+      "learning_rate": 9.796314919738516e-06,
+      "loss": 0.8811,
+      "step": 18981
+    },
+    {
+      "epoch": 0.5213259729202714,
+      "grad_norm": 0.0,
+      "learning_rate": 9.795425589071441e-06,
+      "loss": 0.7733,
+      "step": 18982
+    },
+    {
+      "epoch": 0.5213534371481146,
+      "grad_norm": 0.0,
+      "learning_rate": 9.79453626002304e-06,
+      "loss": 0.9245,
+      "step": 18983
+    },
+    {
+      "epoch": 0.5213809013759578,
+      "grad_norm": 0.0,
+      "learning_rate": 9.793646932600347e-06,
+      "loss": 0.8376,
+      "step": 18984
+    },
+    {
+      "epoch": 0.521408365603801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.792757606810398e-06,
+      "loss": 0.9541,
+      "step": 18985
+    },
+    {
+      "epoch": 0.5214358298316443,
+      "grad_norm": 0.0,
+      "learning_rate": 9.791868282660235e-06,
+      "loss": 0.8378,
+      "step": 18986
+    },
+    {
+      "epoch": 0.5214632940594875,
+      "grad_norm": 0.0,
+      "learning_rate": 9.790978960156889e-06,
+      "loss": 0.9257,
+      "step": 18987
+    },
+    {
+      "epoch": 0.5214907582873307,
+      "grad_norm": 0.0,
+      "learning_rate": 9.790089639307397e-06,
+      "loss": 0.7925,
+      "step": 18988
+    },
+    {
+      "epoch": 0.521518222515174,
+      "grad_norm": 0.0,
+      "learning_rate": 9.789200320118797e-06,
+      "loss": 0.95,
+      "step": 18989
+    },
+    {
+      "epoch": 0.5215456867430173,
+      "grad_norm": 0.0,
+      "learning_rate": 9.788311002598127e-06,
+      "loss": 0.9131,
+      "step": 18990
+    },
+    {
+      "epoch": 0.5215731509708604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78742168675242e-06,
+      "loss": 0.9405,
+      "step": 18991
+    },
+    {
+      "epoch": 0.5216006151987037,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78653237258872e-06,
+      "loss": 0.7855,
+      "step": 18992
+    },
+    {
+      "epoch": 0.5216280794265469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.785643060114055e-06,
+      "loss": 0.9568,
+      "step": 18993
+    },
+    {
+      "epoch": 0.5216555436543902,
+      "grad_norm": 0.0,
+      "learning_rate": 9.784753749335464e-06,
+      "loss": 0.8583,
+      "step": 18994
+    },
+    {
+      "epoch": 0.5216830078822334,
+      "grad_norm": 0.0,
+      "learning_rate": 9.783864440259987e-06,
+      "loss": 0.9779,
+      "step": 18995
+    },
+    {
+      "epoch": 0.5217104721100766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.782975132894656e-06,
+      "loss": 0.9158,
+      "step": 18996
+    },
+    {
+      "epoch": 0.5217379363379199,
+      "grad_norm": 0.0,
+      "learning_rate": 9.782085827246515e-06,
+      "loss": 0.9224,
+      "step": 18997
+    },
+    {
+      "epoch": 0.5217654005657631,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78119652332259e-06,
+      "loss": 0.922,
+      "step": 18998
+    },
+    {
+      "epoch": 0.5217928647936063,
+      "grad_norm": 0.0,
+      "learning_rate": 9.780307221129926e-06,
+      "loss": 0.8525,
+      "step": 18999
+    },
+    {
+      "epoch": 0.5218203290214496,
+      "grad_norm": 0.0,
+      "learning_rate": 9.779417920675553e-06,
+      "loss": 0.9451,
+      "step": 19000
+    },
+    {
+      "epoch": 0.5218477932492928,
+      "grad_norm": 0.0,
+      "learning_rate": 9.778528621966513e-06,
+      "loss": 0.9012,
+      "step": 19001
+    },
+    {
+      "epoch": 0.521875257477136,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77763932500984e-06,
+      "loss": 0.836,
+      "step": 19002
+    },
+    {
+      "epoch": 0.5219027217049793,
+      "grad_norm": 0.0,
+      "learning_rate": 9.776750029812572e-06,
+      "loss": 0.8936,
+      "step": 19003
+    },
+    {
+      "epoch": 0.5219301859328225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.775860736381743e-06,
+      "loss": 0.8875,
+      "step": 19004
+    },
+    {
+      "epoch": 0.5219576501606658,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77497144472439e-06,
+      "loss": 0.9103,
+      "step": 19005
+    },
+    {
+      "epoch": 0.5219851143885089,
+      "grad_norm": 0.0,
+      "learning_rate": 9.774082154847551e-06,
+      "loss": 0.8792,
+      "step": 19006
+    },
+    {
+      "epoch": 0.5220125786163522,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77319286675826e-06,
+      "loss": 0.9989,
+      "step": 19007
+    },
+    {
+      "epoch": 0.5220400428441955,
+      "grad_norm": 0.0,
+      "learning_rate": 9.772303580463557e-06,
+      "loss": 0.9356,
+      "step": 19008
+    },
+    {
+      "epoch": 0.5220675070720386,
+      "grad_norm": 0.0,
+      "learning_rate": 9.771414295970479e-06,
+      "loss": 0.7873,
+      "step": 19009
+    },
+    {
+      "epoch": 0.5220949712998819,
+      "grad_norm": 0.0,
+      "learning_rate": 9.770525013286055e-06,
+      "loss": 1.0036,
+      "step": 19010
+    },
+    {
+      "epoch": 0.5221224355277252,
+      "grad_norm": 0.0,
+      "learning_rate": 9.769635732417327e-06,
+      "loss": 0.8333,
+      "step": 19011
+    },
+    {
+      "epoch": 0.5221498997555684,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76874645337133e-06,
+      "loss": 0.9591,
+      "step": 19012
+    },
+    {
+      "epoch": 0.5221773639834116,
+      "grad_norm": 0.0,
+      "learning_rate": 9.7678571761551e-06,
+      "loss": 0.9163,
+      "step": 19013
+    },
+    {
+      "epoch": 0.5222048282112548,
+      "grad_norm": 0.0,
+      "learning_rate": 9.766967900775676e-06,
+      "loss": 0.9087,
+      "step": 19014
+    },
+    {
+      "epoch": 0.5222322924390981,
+      "grad_norm": 0.0,
+      "learning_rate": 9.766078627240092e-06,
+      "loss": 0.8284,
+      "step": 19015
+    },
+    {
+      "epoch": 0.5222597566669414,
+      "grad_norm": 0.0,
+      "learning_rate": 9.765189355555384e-06,
+      "loss": 0.9265,
+      "step": 19016
+    },
+    {
+      "epoch": 0.5222872208947845,
+      "grad_norm": 0.0,
+      "learning_rate": 9.764300085728588e-06,
+      "loss": 0.8787,
+      "step": 19017
+    },
+    {
+      "epoch": 0.5223146851226278,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76341081776674e-06,
+      "loss": 0.9304,
+      "step": 19018
+    },
+    {
+      "epoch": 0.522342149350471,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76252155167688e-06,
+      "loss": 0.8336,
+      "step": 19019
+    },
+    {
+      "epoch": 0.5223696135783142,
+      "grad_norm": 0.0,
+      "learning_rate": 9.761632287466039e-06,
+      "loss": 0.8718,
+      "step": 19020
+    },
+    {
+      "epoch": 0.5223970778061575,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76074302514126e-06,
+      "loss": 0.8849,
+      "step": 19021
+    },
+    {
+      "epoch": 0.5224245420340007,
+      "grad_norm": 0.0,
+      "learning_rate": 9.759853764709572e-06,
+      "loss": 0.8593,
+      "step": 19022
+    },
+    {
+      "epoch": 0.522452006261844,
+      "grad_norm": 0.0,
+      "learning_rate": 9.758964506178015e-06,
+      "loss": 1.0421,
+      "step": 19023
+    },
+    {
+      "epoch": 0.5224794704896872,
+      "grad_norm": 0.0,
+      "learning_rate": 9.758075249553622e-06,
+      "loss": 0.9608,
+      "step": 19024
+    },
+    {
+      "epoch": 0.5225069347175304,
+      "grad_norm": 0.0,
+      "learning_rate": 9.75718599484343e-06,
+      "loss": 0.8141,
+      "step": 19025
+    },
+    {
+      "epoch": 0.5225343989453737,
+      "grad_norm": 0.0,
+      "learning_rate": 9.756296742054481e-06,
+      "loss": 0.8124,
+      "step": 19026
+    },
+    {
+      "epoch": 0.5225618631732168,
+      "grad_norm": 0.0,
+      "learning_rate": 9.755407491193807e-06,
+      "loss": 0.9278,
+      "step": 19027
+    },
+    {
+      "epoch": 0.5225893274010601,
+      "grad_norm": 0.0,
+      "learning_rate": 9.75451824226844e-06,
+      "loss": 0.8615,
+      "step": 19028
+    },
+    {
+      "epoch": 0.5226167916289034,
+      "grad_norm": 0.0,
+      "learning_rate": 9.753628995285418e-06,
+      "loss": 0.8656,
+      "step": 19029
+    },
+    {
+      "epoch": 0.5226442558567466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.752739750251782e-06,
+      "loss": 0.782,
+      "step": 19030
+    },
+    {
+      "epoch": 0.5226717200845898,
+      "grad_norm": 0.0,
+      "learning_rate": 9.751850507174566e-06,
+      "loss": 0.9966,
+      "step": 19031
+    },
+    {
+      "epoch": 0.522699184312433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.750961266060803e-06,
+      "loss": 0.8445,
+      "step": 19032
+    },
+    {
+      "epoch": 0.5227266485402763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.750072026917535e-06,
+      "loss": 0.7804,
+      "step": 19033
+    },
+    {
+      "epoch": 0.5227541127681196,
+      "grad_norm": 0.0,
+      "learning_rate": 9.749182789751787e-06,
+      "loss": 0.8705,
+      "step": 19034
+    },
+    {
+      "epoch": 0.5227815769959627,
+      "grad_norm": 0.0,
+      "learning_rate": 9.748293554570608e-06,
+      "loss": 0.818,
+      "step": 19035
+    },
+    {
+      "epoch": 0.522809041223806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.747404321381022e-06,
+      "loss": 0.8165,
+      "step": 19036
+    },
+    {
+      "epoch": 0.5228365054516493,
+      "grad_norm": 0.0,
+      "learning_rate": 9.746515090190074e-06,
+      "loss": 0.7627,
+      "step": 19037
+    },
+    {
+      "epoch": 0.5228639696794924,
+      "grad_norm": 0.0,
+      "learning_rate": 9.745625861004798e-06,
+      "loss": 0.9305,
+      "step": 19038
+    },
+    {
+      "epoch": 0.5228914339073357,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74473663383223e-06,
+      "loss": 0.845,
+      "step": 19039
+    },
+    {
+      "epoch": 0.5229188981351789,
+      "grad_norm": 0.0,
+      "learning_rate": 9.743847408679402e-06,
+      "loss": 0.8393,
+      "step": 19040
+    },
+    {
+      "epoch": 0.5229463623630222,
+      "grad_norm": 0.0,
+      "learning_rate": 9.742958185553351e-06,
+      "loss": 0.8783,
+      "step": 19041
+    },
+    {
+      "epoch": 0.5229738265908654,
+      "grad_norm": 0.0,
+      "learning_rate": 9.742068964461116e-06,
+      "loss": 0.9113,
+      "step": 19042
+    },
+    {
+      "epoch": 0.5230012908187086,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74117974540973e-06,
+      "loss": 1.0361,
+      "step": 19043
+    },
+    {
+      "epoch": 0.5230287550465519,
+      "grad_norm": 0.0,
+      "learning_rate": 9.740290528406232e-06,
+      "loss": 0.7756,
+      "step": 19044
+    },
+    {
+      "epoch": 0.523056219274395,
+      "grad_norm": 0.0,
+      "learning_rate": 9.739401313457659e-06,
+      "loss": 0.812,
+      "step": 19045
+    },
+    {
+      "epoch": 0.5230836835022383,
+      "grad_norm": 0.0,
+      "learning_rate": 9.738512100571039e-06,
+      "loss": 0.8628,
+      "step": 19046
+    },
+    {
+      "epoch": 0.5231111477300816,
+      "grad_norm": 0.0,
+      "learning_rate": 9.737622889753415e-06,
+      "loss": 0.8819,
+      "step": 19047
+    },
+    {
+      "epoch": 0.5231386119579248,
+      "grad_norm": 0.0,
+      "learning_rate": 9.736733681011818e-06,
+      "loss": 0.8544,
+      "step": 19048
+    },
+    {
+      "epoch": 0.523166076185768,
+      "grad_norm": 0.0,
+      "learning_rate": 9.735844474353285e-06,
+      "loss": 0.9679,
+      "step": 19049
+    },
+    {
+      "epoch": 0.5231935404136113,
+      "grad_norm": 0.0,
+      "learning_rate": 9.734955269784857e-06,
+      "loss": 0.886,
+      "step": 19050
+    },
+    {
+      "epoch": 0.5232210046414545,
+      "grad_norm": 0.0,
+      "learning_rate": 9.734066067313567e-06,
+      "loss": 0.946,
+      "step": 19051
+    },
+    {
+      "epoch": 0.5232484688692978,
+      "grad_norm": 0.0,
+      "learning_rate": 9.733176866946444e-06,
+      "loss": 0.9685,
+      "step": 19052
+    },
+    {
+      "epoch": 0.5232759330971409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73228766869053e-06,
+      "loss": 0.8865,
+      "step": 19053
+    },
+    {
+      "epoch": 0.5233033973249842,
+      "grad_norm": 0.0,
+      "learning_rate": 9.731398472552858e-06,
+      "loss": 0.8651,
+      "step": 19054
+    },
+    {
+      "epoch": 0.5233308615528275,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73050927854047e-06,
+      "loss": 0.875,
+      "step": 19055
+    },
+    {
+      "epoch": 0.5233583257806707,
+      "grad_norm": 0.0,
+      "learning_rate": 9.729620086660399e-06,
+      "loss": 0.8729,
+      "step": 19056
+    },
+    {
+      "epoch": 0.5233857900085139,
+      "grad_norm": 0.0,
+      "learning_rate": 9.728730896919672e-06,
+      "loss": 0.9277,
+      "step": 19057
+    },
+    {
+      "epoch": 0.5234132542363571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.727841709325333e-06,
+      "loss": 0.9489,
+      "step": 19058
+    },
+    {
+      "epoch": 0.5234407184642004,
+      "grad_norm": 0.0,
+      "learning_rate": 9.726952523884417e-06,
+      "loss": 0.9112,
+      "step": 19059
+    },
+    {
+      "epoch": 0.5234681826920436,
+      "grad_norm": 0.0,
+      "learning_rate": 9.726063340603958e-06,
+      "loss": 0.9182,
+      "step": 19060
+    },
+    {
+      "epoch": 0.5234956469198868,
+      "grad_norm": 0.0,
+      "learning_rate": 9.72517415949099e-06,
+      "loss": 0.8429,
+      "step": 19061
+    },
+    {
+      "epoch": 0.5235231111477301,
+      "grad_norm": 0.0,
+      "learning_rate": 9.724284980552556e-06,
+      "loss": 0.9128,
+      "step": 19062
+    },
+    {
+      "epoch": 0.5235505753755734,
+      "grad_norm": 0.0,
+      "learning_rate": 9.723395803795683e-06,
+      "loss": 0.8753,
+      "step": 19063
+    },
+    {
+      "epoch": 0.5235780396034165,
+      "grad_norm": 0.0,
+      "learning_rate": 9.722506629227407e-06,
+      "loss": 0.8891,
+      "step": 19064
+    },
+    {
+      "epoch": 0.5236055038312598,
+      "grad_norm": 0.0,
+      "learning_rate": 9.721617456854767e-06,
+      "loss": 0.8242,
+      "step": 19065
+    },
+    {
+      "epoch": 0.523632968059103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.7207282866848e-06,
+      "loss": 0.9455,
+      "step": 19066
+    },
+    {
+      "epoch": 0.5236604322869463,
+      "grad_norm": 0.0,
+      "learning_rate": 9.719839118724537e-06,
+      "loss": 0.9146,
+      "step": 19067
+    },
+    {
+      "epoch": 0.5236878965147895,
+      "grad_norm": 0.0,
+      "learning_rate": 9.718949952981018e-06,
+      "loss": 0.8326,
+      "step": 19068
+    },
+    {
+      "epoch": 0.5237153607426327,
+      "grad_norm": 0.0,
+      "learning_rate": 9.718060789461273e-06,
+      "loss": 0.8533,
+      "step": 19069
+    },
+    {
+      "epoch": 0.523742824970476,
+      "grad_norm": 0.0,
+      "learning_rate": 9.717171628172339e-06,
+      "loss": 0.9294,
+      "step": 19070
+    },
+    {
+      "epoch": 0.5237702891983191,
+      "grad_norm": 0.0,
+      "learning_rate": 9.716282469121257e-06,
+      "loss": 0.9576,
+      "step": 19071
+    },
+    {
+      "epoch": 0.5237977534261624,
+      "grad_norm": 0.0,
+      "learning_rate": 9.715393312315054e-06,
+      "loss": 0.822,
+      "step": 19072
+    },
+    {
+      "epoch": 0.5238252176540057,
+      "grad_norm": 0.0,
+      "learning_rate": 9.71450415776077e-06,
+      "loss": 0.8796,
+      "step": 19073
+    },
+    {
+      "epoch": 0.5238526818818489,
+      "grad_norm": 0.0,
+      "learning_rate": 9.713615005465444e-06,
+      "loss": 0.8972,
+      "step": 19074
+    },
+    {
+      "epoch": 0.5238801461096921,
+      "grad_norm": 0.0,
+      "learning_rate": 9.712725855436105e-06,
+      "loss": 0.9398,
+      "step": 19075
+    },
+    {
+      "epoch": 0.5239076103375354,
+      "grad_norm": 0.0,
+      "learning_rate": 9.711836707679787e-06,
+      "loss": 0.9592,
+      "step": 19076
+    },
+    {
+      "epoch": 0.5239350745653786,
+      "grad_norm": 0.0,
+      "learning_rate": 9.71094756220353e-06,
+      "loss": 0.8441,
+      "step": 19077
+    },
+    {
+      "epoch": 0.5239625387932219,
+      "grad_norm": 0.0,
+      "learning_rate": 9.710058419014369e-06,
+      "loss": 0.811,
+      "step": 19078
+    },
+    {
+      "epoch": 0.523990003021065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.709169278119337e-06,
+      "loss": 0.9362,
+      "step": 19079
+    },
+    {
+      "epoch": 0.5240174672489083,
+      "grad_norm": 0.0,
+      "learning_rate": 9.708280139525473e-06,
+      "loss": 0.9634,
+      "step": 19080
+    },
+    {
+      "epoch": 0.5240449314767516,
+      "grad_norm": 0.0,
+      "learning_rate": 9.707391003239806e-06,
+      "loss": 0.9969,
+      "step": 19081
+    },
+    {
+      "epoch": 0.5240723957045947,
+      "grad_norm": 0.0,
+      "learning_rate": 9.706501869269375e-06,
+      "loss": 0.9746,
+      "step": 19082
+    },
+    {
+      "epoch": 0.524099859932438,
+      "grad_norm": 0.0,
+      "learning_rate": 9.705612737621219e-06,
+      "loss": 1.0021,
+      "step": 19083
+    },
+    {
+      "epoch": 0.5241273241602812,
+      "grad_norm": 0.0,
+      "learning_rate": 9.704723608302364e-06,
+      "loss": 0.7801,
+      "step": 19084
+    },
+    {
+      "epoch": 0.5241547883881245,
+      "grad_norm": 0.0,
+      "learning_rate": 9.703834481319852e-06,
+      "loss": 0.9125,
+      "step": 19085
+    },
+    {
+      "epoch": 0.5241822526159677,
+      "grad_norm": 0.0,
+      "learning_rate": 9.702945356680719e-06,
+      "loss": 0.8644,
+      "step": 19086
+    },
+    {
+      "epoch": 0.5242097168438109,
+      "grad_norm": 0.0,
+      "learning_rate": 9.702056234391995e-06,
+      "loss": 0.8897,
+      "step": 19087
+    },
+    {
+      "epoch": 0.5242371810716542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.701167114460717e-06,
+      "loss": 0.8958,
+      "step": 19088
+    },
+    {
+      "epoch": 0.5242646452994975,
+      "grad_norm": 0.0,
+      "learning_rate": 9.700277996893919e-06,
+      "loss": 0.8751,
+      "step": 19089
+    },
+    {
+      "epoch": 0.5242921095273406,
+      "grad_norm": 0.0,
+      "learning_rate": 9.699388881698641e-06,
+      "loss": 0.9492,
+      "step": 19090
+    },
+    {
+      "epoch": 0.5243195737551839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.698499768881912e-06,
+      "loss": 0.9492,
+      "step": 19091
+    },
+    {
+      "epoch": 0.5243470379830271,
+      "grad_norm": 0.0,
+      "learning_rate": 9.697610658450774e-06,
+      "loss": 0.8448,
+      "step": 19092
+    },
+    {
+      "epoch": 0.5243745022108703,
+      "grad_norm": 0.0,
+      "learning_rate": 9.696721550412255e-06,
+      "loss": 0.9479,
+      "step": 19093
+    },
+    {
+      "epoch": 0.5244019664387136,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69583244477339e-06,
+      "loss": 0.9459,
+      "step": 19094
+    },
+    {
+      "epoch": 0.5244294306665568,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69494334154122e-06,
+      "loss": 0.8556,
+      "step": 19095
+    },
+    {
+      "epoch": 0.5244568948944001,
+      "grad_norm": 0.0,
+      "learning_rate": 9.694054240722774e-06,
+      "loss": 0.8291,
+      "step": 19096
+    },
+    {
+      "epoch": 0.5244843591222432,
+      "grad_norm": 0.0,
+      "learning_rate": 9.693165142325088e-06,
+      "loss": 0.821,
+      "step": 19097
+    },
+    {
+      "epoch": 0.5245118233500865,
+      "grad_norm": 0.0,
+      "learning_rate": 9.692276046355204e-06,
+      "loss": 0.913,
+      "step": 19098
+    },
+    {
+      "epoch": 0.5245392875779298,
+      "grad_norm": 0.0,
+      "learning_rate": 9.691386952820149e-06,
+      "loss": 0.9162,
+      "step": 19099
+    },
+    {
+      "epoch": 0.524566751805773,
+      "grad_norm": 0.0,
+      "learning_rate": 9.690497861726958e-06,
+      "loss": 0.9646,
+      "step": 19100
+    },
+    {
+      "epoch": 0.5245942160336162,
+      "grad_norm": 0.0,
+      "learning_rate": 9.689608773082668e-06,
+      "loss": 0.8632,
+      "step": 19101
+    },
+    {
+      "epoch": 0.5246216802614595,
+      "grad_norm": 0.0,
+      "learning_rate": 9.688719686894314e-06,
+      "loss": 1.0255,
+      "step": 19102
+    },
+    {
+      "epoch": 0.5246491444893027,
+      "grad_norm": 0.0,
+      "learning_rate": 9.687830603168929e-06,
+      "loss": 0.8259,
+      "step": 19103
+    },
+    {
+      "epoch": 0.5246766087171459,
+      "grad_norm": 0.0,
+      "learning_rate": 9.686941521913554e-06,
+      "loss": 0.8338,
+      "step": 19104
+    },
+    {
+      "epoch": 0.5247040729449891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.686052443135216e-06,
+      "loss": 0.8972,
+      "step": 19105
+    },
+    {
+      "epoch": 0.5247315371728324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.68516336684095e-06,
+      "loss": 1.0116,
+      "step": 19106
+    },
+    {
+      "epoch": 0.5247590014006757,
+      "grad_norm": 0.0,
+      "learning_rate": 9.684274293037798e-06,
+      "loss": 0.8282,
+      "step": 19107
+    },
+    {
+      "epoch": 0.5247864656285188,
+      "grad_norm": 0.0,
+      "learning_rate": 9.683385221732787e-06,
+      "loss": 0.9265,
+      "step": 19108
+    },
+    {
+      "epoch": 0.5248139298563621,
+      "grad_norm": 0.0,
+      "learning_rate": 9.682496152932954e-06,
+      "loss": 0.8902,
+      "step": 19109
+    },
+    {
+      "epoch": 0.5248413940842053,
+      "grad_norm": 0.0,
+      "learning_rate": 9.68160708664534e-06,
+      "loss": 0.932,
+      "step": 19110
+    },
+    {
+      "epoch": 0.5248688583120485,
+      "grad_norm": 0.0,
+      "learning_rate": 9.68071802287697e-06,
+      "loss": 0.9122,
+      "step": 19111
+    },
+    {
+      "epoch": 0.5248963225398918,
+      "grad_norm": 0.0,
+      "learning_rate": 9.679828961634882e-06,
+      "loss": 0.8024,
+      "step": 19112
+    },
+    {
+      "epoch": 0.524923786767735,
+      "grad_norm": 0.0,
+      "learning_rate": 9.67893990292611e-06,
+      "loss": 0.8496,
+      "step": 19113
+    },
+    {
+      "epoch": 0.5249512509955783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.678050846757693e-06,
+      "loss": 0.7768,
+      "step": 19114
+    },
+    {
+      "epoch": 0.5249787152234215,
+      "grad_norm": 0.0,
+      "learning_rate": 9.677161793136662e-06,
+      "loss": 0.9001,
+      "step": 19115
+    },
+    {
+      "epoch": 0.5250061794512647,
+      "grad_norm": 0.0,
+      "learning_rate": 9.676272742070053e-06,
+      "loss": 0.9037,
+      "step": 19116
+    },
+    {
+      "epoch": 0.525033643679108,
+      "grad_norm": 0.0,
+      "learning_rate": 9.675383693564896e-06,
+      "loss": 0.9822,
+      "step": 19117
+    },
+    {
+      "epoch": 0.5250611079069512,
+      "grad_norm": 0.0,
+      "learning_rate": 9.67449464762823e-06,
+      "loss": 0.7936,
+      "step": 19118
+    },
+    {
+      "epoch": 0.5250885721347944,
+      "grad_norm": 0.0,
+      "learning_rate": 9.673605604267091e-06,
+      "loss": 0.9575,
+      "step": 19119
+    },
+    {
+      "epoch": 0.5251160363626377,
+      "grad_norm": 0.0,
+      "learning_rate": 9.672716563488507e-06,
+      "loss": 0.8526,
+      "step": 19120
+    },
+    {
+      "epoch": 0.5251435005904809,
+      "grad_norm": 0.0,
+      "learning_rate": 9.671827525299522e-06,
+      "loss": 0.8525,
+      "step": 19121
+    },
+    {
+      "epoch": 0.5251709648183241,
+      "grad_norm": 0.0,
+      "learning_rate": 9.67093848970716e-06,
+      "loss": 0.887,
+      "step": 19122
+    },
+    {
+      "epoch": 0.5251984290461673,
+      "grad_norm": 0.0,
+      "learning_rate": 9.670049456718462e-06,
+      "loss": 0.8962,
+      "step": 19123
+    },
+    {
+      "epoch": 0.5252258932740106,
+      "grad_norm": 0.0,
+      "learning_rate": 9.669160426340459e-06,
+      "loss": 0.8362,
+      "step": 19124
+    },
+    {
+      "epoch": 0.5252533575018539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.668271398580187e-06,
+      "loss": 0.8303,
+      "step": 19125
+    },
+    {
+      "epoch": 0.525280821729697,
+      "grad_norm": 0.0,
+      "learning_rate": 9.66738237344468e-06,
+      "loss": 0.8411,
+      "step": 19126
+    },
+    {
+      "epoch": 0.5253082859575403,
+      "grad_norm": 0.0,
+      "learning_rate": 9.666493350940975e-06,
+      "loss": 0.8698,
+      "step": 19127
+    },
+    {
+      "epoch": 0.5253357501853836,
+      "grad_norm": 0.0,
+      "learning_rate": 9.665604331076104e-06,
+      "loss": 0.9539,
+      "step": 19128
+    },
+    {
+      "epoch": 0.5253632144132268,
+      "grad_norm": 0.0,
+      "learning_rate": 9.664715313857098e-06,
+      "loss": 0.9093,
+      "step": 19129
+    },
+    {
+      "epoch": 0.52539067864107,
+      "grad_norm": 0.0,
+      "learning_rate": 9.663826299290994e-06,
+      "loss": 0.8774,
+      "step": 19130
+    },
+    {
+      "epoch": 0.5254181428689132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.66293728738483e-06,
+      "loss": 0.9172,
+      "step": 19131
+    },
+    {
+      "epoch": 0.5254456070967565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.662048278145633e-06,
+      "loss": 0.9343,
+      "step": 19132
+    },
+    {
+      "epoch": 0.5254730713245997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.661159271580446e-06,
+      "loss": 0.8819,
+      "step": 19133
+    },
+    {
+      "epoch": 0.5255005355524429,
+      "grad_norm": 0.0,
+      "learning_rate": 9.660270267696294e-06,
+      "loss": 0.9434,
+      "step": 19134
+    },
+    {
+      "epoch": 0.5255279997802862,
+      "grad_norm": 0.0,
+      "learning_rate": 9.659381266500217e-06,
+      "loss": 0.8252,
+      "step": 19135
+    },
+    {
+      "epoch": 0.5255554640081294,
+      "grad_norm": 0.0,
+      "learning_rate": 9.658492267999245e-06,
+      "loss": 0.9466,
+      "step": 19136
+    },
+    {
+      "epoch": 0.5255829282359726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.657603272200414e-06,
+      "loss": 0.871,
+      "step": 19137
+    },
+    {
+      "epoch": 0.5256103924638159,
+      "grad_norm": 0.0,
+      "learning_rate": 9.656714279110761e-06,
+      "loss": 0.8708,
+      "step": 19138
+    },
+    {
+      "epoch": 0.5256378566916591,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65582528873732e-06,
+      "loss": 0.8401,
+      "step": 19139
+    },
+    {
+      "epoch": 0.5256653209195024,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65493630108712e-06,
+      "loss": 0.9014,
+      "step": 19140
+    },
+    {
+      "epoch": 0.5256927851473456,
+      "grad_norm": 0.0,
+      "learning_rate": 9.654047316167195e-06,
+      "loss": 0.9814,
+      "step": 19141
+    },
+    {
+      "epoch": 0.5257202493751888,
+      "grad_norm": 0.0,
+      "learning_rate": 9.653158333984584e-06,
+      "loss": 0.8499,
+      "step": 19142
+    },
+    {
+      "epoch": 0.5257477136030321,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65226935454632e-06,
+      "loss": 0.9164,
+      "step": 19143
+    },
+    {
+      "epoch": 0.5257751778308752,
+      "grad_norm": 0.0,
+      "learning_rate": 9.651380377859433e-06,
+      "loss": 0.8715,
+      "step": 19144
+    },
+    {
+      "epoch": 0.5258026420587185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.650491403930963e-06,
+      "loss": 0.8185,
+      "step": 19145
+    },
+    {
+      "epoch": 0.5258301062865618,
+      "grad_norm": 0.0,
+      "learning_rate": 9.649602432767937e-06,
+      "loss": 0.9156,
+      "step": 19146
+    },
+    {
+      "epoch": 0.525857570514405,
+      "grad_norm": 0.0,
+      "learning_rate": 9.648713464377396e-06,
+      "loss": 0.9895,
+      "step": 19147
+    },
+    {
+      "epoch": 0.5258850347422482,
+      "grad_norm": 0.0,
+      "learning_rate": 9.647824498766366e-06,
+      "loss": 0.9109,
+      "step": 19148
+    },
+    {
+      "epoch": 0.5259124989700914,
+      "grad_norm": 0.0,
+      "learning_rate": 9.646935535941884e-06,
+      "loss": 0.8643,
+      "step": 19149
+    },
+    {
+      "epoch": 0.5259399631979347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.646046575910987e-06,
+      "loss": 0.8344,
+      "step": 19150
+    },
+    {
+      "epoch": 0.525967427425778,
+      "grad_norm": 0.0,
+      "learning_rate": 9.64515761868071e-06,
+      "loss": 0.826,
+      "step": 19151
+    },
+    {
+      "epoch": 0.5259948916536211,
+      "grad_norm": 0.0,
+      "learning_rate": 9.64426866425808e-06,
+      "loss": 0.905,
+      "step": 19152
+    },
+    {
+      "epoch": 0.5260223558814644,
+      "grad_norm": 0.0,
+      "learning_rate": 9.643379712650134e-06,
+      "loss": 0.8883,
+      "step": 19153
+    },
+    {
+      "epoch": 0.5260498201093077,
+      "grad_norm": 0.0,
+      "learning_rate": 9.642490763863904e-06,
+      "loss": 0.882,
+      "step": 19154
+    },
+    {
+      "epoch": 0.5260772843371508,
+      "grad_norm": 0.0,
+      "learning_rate": 9.641601817906429e-06,
+      "loss": 0.8386,
+      "step": 19155
+    },
+    {
+      "epoch": 0.5261047485649941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.640712874784737e-06,
+      "loss": 0.9842,
+      "step": 19156
+    },
+    {
+      "epoch": 0.5261322127928373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.639823934505869e-06,
+      "loss": 0.883,
+      "step": 19157
+    },
+    {
+      "epoch": 0.5261596770206806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.638934997076848e-06,
+      "loss": 0.8902,
+      "step": 19158
+    },
+    {
+      "epoch": 0.5261871412485238,
+      "grad_norm": 0.0,
+      "learning_rate": 9.638046062504715e-06,
+      "loss": 0.9915,
+      "step": 19159
+    },
+    {
+      "epoch": 0.526214605476367,
+      "grad_norm": 0.0,
+      "learning_rate": 9.6371571307965e-06,
+      "loss": 0.9757,
+      "step": 19160
+    },
+    {
+      "epoch": 0.5262420697042103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.636268201959238e-06,
+      "loss": 0.8867,
+      "step": 19161
+    },
+    {
+      "epoch": 0.5262695339320534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.635379275999967e-06,
+      "loss": 0.8864,
+      "step": 19162
+    },
+    {
+      "epoch": 0.5262969981598967,
+      "grad_norm": 0.0,
+      "learning_rate": 9.634490352925715e-06,
+      "loss": 0.8816,
+      "step": 19163
+    },
+    {
+      "epoch": 0.52632446238774,
+      "grad_norm": 0.0,
+      "learning_rate": 9.633601432743517e-06,
+      "loss": 1.058,
+      "step": 19164
+    },
+    {
+      "epoch": 0.5263519266155832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.632712515460404e-06,
+      "loss": 0.8325,
+      "step": 19165
+    },
+    {
+      "epoch": 0.5263793908434264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.631823601083412e-06,
+      "loss": 0.8499,
+      "step": 19166
+    },
+    {
+      "epoch": 0.5264068550712697,
+      "grad_norm": 0.0,
+      "learning_rate": 9.630934689619576e-06,
+      "loss": 0.8382,
+      "step": 19167
+    },
+    {
+      "epoch": 0.5264343192991129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.630045781075926e-06,
+      "loss": 0.8286,
+      "step": 19168
+    },
+    {
+      "epoch": 0.5264617835269562,
+      "grad_norm": 0.0,
+      "learning_rate": 9.629156875459502e-06,
+      "loss": 0.913,
+      "step": 19169
+    },
+    {
+      "epoch": 0.5264892477547993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.628267972777328e-06,
+      "loss": 0.9146,
+      "step": 19170
+    },
+    {
+      "epoch": 0.5265167119826426,
+      "grad_norm": 0.0,
+      "learning_rate": 9.627379073036444e-06,
+      "loss": 0.9167,
+      "step": 19171
+    },
+    {
+      "epoch": 0.5265441762104859,
+      "grad_norm": 0.0,
+      "learning_rate": 9.626490176243878e-06,
+      "loss": 0.9036,
+      "step": 19172
+    },
+    {
+      "epoch": 0.526571640438329,
+      "grad_norm": 0.0,
+      "learning_rate": 9.625601282406667e-06,
+      "loss": 0.8518,
+      "step": 19173
+    },
+    {
+      "epoch": 0.5265991046661723,
+      "grad_norm": 0.0,
+      "learning_rate": 9.624712391531848e-06,
+      "loss": 0.8752,
+      "step": 19174
+    },
+    {
+      "epoch": 0.5266265688940155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.623823503626445e-06,
+      "loss": 0.8782,
+      "step": 19175
+    },
+    {
+      "epoch": 0.5266540331218588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.622934618697498e-06,
+      "loss": 0.8443,
+      "step": 19176
+    },
+    {
+      "epoch": 0.526681497349702,
+      "grad_norm": 0.0,
+      "learning_rate": 9.622045736752038e-06,
+      "loss": 0.9548,
+      "step": 19177
+    },
+    {
+      "epoch": 0.5267089615775452,
+      "grad_norm": 0.0,
+      "learning_rate": 9.621156857797098e-06,
+      "loss": 0.8634,
+      "step": 19178
+    },
+    {
+      "epoch": 0.5267364258053885,
+      "grad_norm": 0.0,
+      "learning_rate": 9.620267981839713e-06,
+      "loss": 0.8737,
+      "step": 19179
+    },
+    {
+      "epoch": 0.5267638900332318,
+      "grad_norm": 0.0,
+      "learning_rate": 9.619379108886918e-06,
+      "loss": 0.8691,
+      "step": 19180
+    },
+    {
+      "epoch": 0.5267913542610749,
+      "grad_norm": 0.0,
+      "learning_rate": 9.618490238945737e-06,
+      "loss": 0.8987,
+      "step": 19181
+    },
+    {
+      "epoch": 0.5268188184889182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61760137202321e-06,
+      "loss": 0.8295,
+      "step": 19182
+    },
+    {
+      "epoch": 0.5268462827167614,
+      "grad_norm": 0.0,
+      "learning_rate": 9.616712508126371e-06,
+      "loss": 0.8558,
+      "step": 19183
+    },
+    {
+      "epoch": 0.5268737469446046,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61582364726225e-06,
+      "loss": 0.917,
+      "step": 19184
+    },
+    {
+      "epoch": 0.5269012111724479,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61493478943788e-06,
+      "loss": 0.9577,
+      "step": 19185
+    },
+    {
+      "epoch": 0.5269286754002911,
+      "grad_norm": 0.0,
+      "learning_rate": 9.6140459346603e-06,
+      "loss": 0.9419,
+      "step": 19186
+    },
+    {
+      "epoch": 0.5269561396281344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.613157082936535e-06,
+      "loss": 0.857,
+      "step": 19187
+    },
+    {
+      "epoch": 0.5269836038559775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61226823427362e-06,
+      "loss": 0.9441,
+      "step": 19188
+    },
+    {
+      "epoch": 0.5270110680838208,
+      "grad_norm": 0.0,
+      "learning_rate": 9.611379388678588e-06,
+      "loss": 0.8949,
+      "step": 19189
+    },
+    {
+      "epoch": 0.5270385323116641,
+      "grad_norm": 0.0,
+      "learning_rate": 9.610490546158472e-06,
+      "loss": 0.9806,
+      "step": 19190
+    },
+    {
+      "epoch": 0.5270659965395073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.609601706720311e-06,
+      "loss": 0.8986,
+      "step": 19191
+    },
+    {
+      "epoch": 0.5270934607673505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.608712870371132e-06,
+      "loss": 0.9483,
+      "step": 19192
+    },
+    {
+      "epoch": 0.5271209249951938,
+      "grad_norm": 0.0,
+      "learning_rate": 9.607824037117965e-06,
+      "loss": 0.8973,
+      "step": 19193
+    },
+    {
+      "epoch": 0.527148389223037,
+      "grad_norm": 0.0,
+      "learning_rate": 9.606935206967846e-06,
+      "loss": 0.7576,
+      "step": 19194
+    },
+    {
+      "epoch": 0.5271758534508802,
+      "grad_norm": 0.0,
+      "learning_rate": 9.60604637992781e-06,
+      "loss": 0.8245,
+      "step": 19195
+    },
+    {
+      "epoch": 0.5272033176787234,
+      "grad_norm": 0.0,
+      "learning_rate": 9.605157556004885e-06,
+      "loss": 0.9099,
+      "step": 19196
+    },
+    {
+      "epoch": 0.5272307819065667,
+      "grad_norm": 0.0,
+      "learning_rate": 9.604268735206107e-06,
+      "loss": 0.8928,
+      "step": 19197
+    },
+    {
+      "epoch": 0.52725824613441,
+      "grad_norm": 0.0,
+      "learning_rate": 9.603379917538512e-06,
+      "loss": 0.9315,
+      "step": 19198
+    },
+    {
+      "epoch": 0.5272857103622531,
+      "grad_norm": 0.0,
+      "learning_rate": 9.602491103009127e-06,
+      "loss": 0.88,
+      "step": 19199
+    },
+    {
+      "epoch": 0.5273131745900964,
+      "grad_norm": 0.0,
+      "learning_rate": 9.601602291624984e-06,
+      "loss": 0.8568,
+      "step": 19200
+    },
+    {
+      "epoch": 0.5273406388179397,
+      "grad_norm": 0.0,
+      "learning_rate": 9.600713483393117e-06,
+      "loss": 0.9718,
+      "step": 19201
+    },
+    {
+      "epoch": 0.5273681030457829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59982467832056e-06,
+      "loss": 0.8994,
+      "step": 19202
+    },
+    {
+      "epoch": 0.5273955672736261,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59893587641435e-06,
+      "loss": 0.9335,
+      "step": 19203
+    },
+    {
+      "epoch": 0.5274230315014693,
+      "grad_norm": 0.0,
+      "learning_rate": 9.598047077681514e-06,
+      "loss": 0.8681,
+      "step": 19204
+    },
+    {
+      "epoch": 0.5274504957293126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59715828212908e-06,
+      "loss": 0.9707,
+      "step": 19205
+    },
+    {
+      "epoch": 0.5274779599571559,
+      "grad_norm": 0.0,
+      "learning_rate": 9.596269489764088e-06,
+      "loss": 0.9097,
+      "step": 19206
+    },
+    {
+      "epoch": 0.527505424184999,
+      "grad_norm": 0.0,
+      "learning_rate": 9.595380700593571e-06,
+      "loss": 0.9329,
+      "step": 19207
+    },
+    {
+      "epoch": 0.5275328884128423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.594491914624553e-06,
+      "loss": 0.8333,
+      "step": 19208
+    },
+    {
+      "epoch": 0.5275603526406855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.593603131864076e-06,
+      "loss": 1.0142,
+      "step": 19209
+    },
+    {
+      "epoch": 0.5275878168685287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59271435231917e-06,
+      "loss": 0.8395,
+      "step": 19210
+    },
+    {
+      "epoch": 0.527615281096372,
+      "grad_norm": 0.0,
+      "learning_rate": 9.591825575996867e-06,
+      "loss": 0.86,
+      "step": 19211
+    },
+    {
+      "epoch": 0.5276427453242152,
+      "grad_norm": 0.0,
+      "learning_rate": 9.590936802904191e-06,
+      "loss": 0.9939,
+      "step": 19212
+    },
+    {
+      "epoch": 0.5276702095520585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.590048033048184e-06,
+      "loss": 0.9057,
+      "step": 19213
+    },
+    {
+      "epoch": 0.5276976737799017,
+      "grad_norm": 0.0,
+      "learning_rate": 9.589159266435876e-06,
+      "loss": 0.8608,
+      "step": 19214
+    },
+    {
+      "epoch": 0.5277251380077449,
+      "grad_norm": 0.0,
+      "learning_rate": 9.588270503074301e-06,
+      "loss": 0.8985,
+      "step": 19215
+    },
+    {
+      "epoch": 0.5277526022355882,
+      "grad_norm": 0.0,
+      "learning_rate": 9.587381742970493e-06,
+      "loss": 0.9534,
+      "step": 19216
+    },
+    {
+      "epoch": 0.5277800664634313,
+      "grad_norm": 0.0,
+      "learning_rate": 9.586492986131473e-06,
+      "loss": 0.8483,
+      "step": 19217
+    },
+    {
+      "epoch": 0.5278075306912746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.585604232564283e-06,
+      "loss": 0.9067,
+      "step": 19218
+    },
+    {
+      "epoch": 0.5278349949191179,
+      "grad_norm": 0.0,
+      "learning_rate": 9.584715482275955e-06,
+      "loss": 0.9068,
+      "step": 19219
+    },
+    {
+      "epoch": 0.5278624591469611,
+      "grad_norm": 0.0,
+      "learning_rate": 9.583826735273515e-06,
+      "loss": 0.9477,
+      "step": 19220
+    },
+    {
+      "epoch": 0.5278899233748043,
+      "grad_norm": 0.0,
+      "learning_rate": 9.582937991564e-06,
+      "loss": 0.7888,
+      "step": 19221
+    },
+    {
+      "epoch": 0.5279173876026475,
+      "grad_norm": 0.0,
+      "learning_rate": 9.582049251154447e-06,
+      "loss": 0.8339,
+      "step": 19222
+    },
+    {
+      "epoch": 0.5279448518304908,
+      "grad_norm": 0.0,
+      "learning_rate": 9.581160514051878e-06,
+      "loss": 1.0325,
+      "step": 19223
+    },
+    {
+      "epoch": 0.5279723160583341,
+      "grad_norm": 0.0,
+      "learning_rate": 9.580271780263328e-06,
+      "loss": 0.8405,
+      "step": 19224
+    },
+    {
+      "epoch": 0.5279997802861772,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57938304979583e-06,
+      "loss": 0.9213,
+      "step": 19225
+    },
+    {
+      "epoch": 0.5280272445140205,
+      "grad_norm": 0.0,
+      "learning_rate": 9.578494322656419e-06,
+      "loss": 0.89,
+      "step": 19226
+    },
+    {
+      "epoch": 0.5280547087418638,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57760559885212e-06,
+      "loss": 0.8069,
+      "step": 19227
+    },
+    {
+      "epoch": 0.5280821729697069,
+      "grad_norm": 0.0,
+      "learning_rate": 9.576716878389976e-06,
+      "loss": 0.9329,
+      "step": 19228
+    },
+    {
+      "epoch": 0.5281096371975502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.575828161277008e-06,
+      "loss": 0.8183,
+      "step": 19229
+    },
+    {
+      "epoch": 0.5281371014253934,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57493944752025e-06,
+      "loss": 0.9577,
+      "step": 19230
+    },
+    {
+      "epoch": 0.5281645656532367,
+      "grad_norm": 0.0,
+      "learning_rate": 9.574050737126738e-06,
+      "loss": 0.9159,
+      "step": 19231
+    },
+    {
+      "epoch": 0.5281920298810799,
+      "grad_norm": 0.0,
+      "learning_rate": 9.5731620301035e-06,
+      "loss": 0.8433,
+      "step": 19232
+    },
+    {
+      "epoch": 0.5282194941089231,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57227332645757e-06,
+      "loss": 0.9382,
+      "step": 19233
+    },
+    {
+      "epoch": 0.5282469583367664,
+      "grad_norm": 0.0,
+      "learning_rate": 9.571384626195982e-06,
+      "loss": 0.8821,
+      "step": 19234
+    },
+    {
+      "epoch": 0.5282744225646095,
+      "grad_norm": 0.0,
+      "learning_rate": 9.570495929325763e-06,
+      "loss": 0.9055,
+      "step": 19235
+    },
+    {
+      "epoch": 0.5283018867924528,
+      "grad_norm": 0.0,
+      "learning_rate": 9.569607235853945e-06,
+      "loss": 0.8485,
+      "step": 19236
+    },
+    {
+      "epoch": 0.5283293510202961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.568718545787561e-06,
+      "loss": 0.8165,
+      "step": 19237
+    },
+    {
+      "epoch": 0.5283568152481393,
+      "grad_norm": 0.0,
+      "learning_rate": 9.567829859133644e-06,
+      "loss": 1.0383,
+      "step": 19238
+    },
+    {
+      "epoch": 0.5283842794759825,
+      "grad_norm": 0.0,
+      "learning_rate": 9.566941175899228e-06,
+      "loss": 0.854,
+      "step": 19239
+    },
+    {
+      "epoch": 0.5284117437038258,
+      "grad_norm": 0.0,
+      "learning_rate": 9.566052496091338e-06,
+      "loss": 0.8658,
+      "step": 19240
+    },
+    {
+      "epoch": 0.528439207931669,
+      "grad_norm": 0.0,
+      "learning_rate": 9.565163819717007e-06,
+      "loss": 0.9057,
+      "step": 19241
+    },
+    {
+      "epoch": 0.5284666721595123,
+      "grad_norm": 0.0,
+      "learning_rate": 9.56427514678327e-06,
+      "loss": 0.8139,
+      "step": 19242
+    },
+    {
+      "epoch": 0.5284941363873554,
+      "grad_norm": 0.0,
+      "learning_rate": 9.563386477297158e-06,
+      "loss": 0.9797,
+      "step": 19243
+    },
+    {
+      "epoch": 0.5285216006151987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.562497811265699e-06,
+      "loss": 0.9209,
+      "step": 19244
+    },
+    {
+      "epoch": 0.528549064843042,
+      "grad_norm": 0.0,
+      "learning_rate": 9.56160914869593e-06,
+      "loss": 0.9921,
+      "step": 19245
+    },
+    {
+      "epoch": 0.5285765290708851,
+      "grad_norm": 0.0,
+      "learning_rate": 9.560720489594875e-06,
+      "loss": 0.8138,
+      "step": 19246
+    },
+    {
+      "epoch": 0.5286039932987284,
+      "grad_norm": 0.0,
+      "learning_rate": 9.559831833969574e-06,
+      "loss": 0.8744,
+      "step": 19247
+    },
+    {
+      "epoch": 0.5286314575265716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.55894318182705e-06,
+      "loss": 0.8725,
+      "step": 19248
+    },
+    {
+      "epoch": 0.5286589217544149,
+      "grad_norm": 0.0,
+      "learning_rate": 9.55805453317434e-06,
+      "loss": 0.8401,
+      "step": 19249
+    },
+    {
+      "epoch": 0.5286863859822581,
+      "grad_norm": 0.0,
+      "learning_rate": 9.557165888018474e-06,
+      "loss": 0.8506,
+      "step": 19250
+    },
+    {
+      "epoch": 0.5287138502101013,
+      "grad_norm": 0.0,
+      "learning_rate": 9.556277246366484e-06,
+      "loss": 0.8727,
+      "step": 19251
+    },
+    {
+      "epoch": 0.5287413144379446,
+      "grad_norm": 0.0,
+      "learning_rate": 9.555388608225402e-06,
+      "loss": 0.8498,
+      "step": 19252
+    },
+    {
+      "epoch": 0.5287687786657879,
+      "grad_norm": 0.0,
+      "learning_rate": 9.554499973602252e-06,
+      "loss": 0.8182,
+      "step": 19253
+    },
+    {
+      "epoch": 0.528796242893631,
+      "grad_norm": 0.0,
+      "learning_rate": 9.553611342504072e-06,
+      "loss": 0.7795,
+      "step": 19254
+    },
+    {
+      "epoch": 0.5288237071214743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.552722714937896e-06,
+      "loss": 0.8155,
+      "step": 19255
+    },
+    {
+      "epoch": 0.5288511713493175,
+      "grad_norm": 0.0,
+      "learning_rate": 9.551834090910747e-06,
+      "loss": 1.051,
+      "step": 19256
+    },
+    {
+      "epoch": 0.5288786355771607,
+      "grad_norm": 0.0,
+      "learning_rate": 9.550945470429665e-06,
+      "loss": 0.8065,
+      "step": 19257
+    },
+    {
+      "epoch": 0.528906099805004,
+      "grad_norm": 0.0,
+      "learning_rate": 9.550056853501672e-06,
+      "loss": 0.8662,
+      "step": 19258
+    },
+    {
+      "epoch": 0.5289335640328472,
+      "grad_norm": 0.0,
+      "learning_rate": 9.549168240133806e-06,
+      "loss": 0.8522,
+      "step": 19259
+    },
+    {
+      "epoch": 0.5289610282606905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.548279630333092e-06,
+      "loss": 0.8825,
+      "step": 19260
+    },
+    {
+      "epoch": 0.5289884924885336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.547391024106568e-06,
+      "loss": 0.9702,
+      "step": 19261
+    },
+    {
+      "epoch": 0.5290159567163769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54650242146126e-06,
+      "loss": 0.9093,
+      "step": 19262
+    },
+    {
+      "epoch": 0.5290434209442202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.545613822404203e-06,
+      "loss": 0.9089,
+      "step": 19263
+    },
+    {
+      "epoch": 0.5290708851720634,
+      "grad_norm": 0.0,
+      "learning_rate": 9.544725226942424e-06,
+      "loss": 0.8166,
+      "step": 19264
+    },
+    {
+      "epoch": 0.5290983493999066,
+      "grad_norm": 0.0,
+      "learning_rate": 9.543836635082954e-06,
+      "loss": 0.8434,
+      "step": 19265
+    },
+    {
+      "epoch": 0.5291258136277499,
+      "grad_norm": 0.0,
+      "learning_rate": 9.542948046832824e-06,
+      "loss": 1.0457,
+      "step": 19266
+    },
+    {
+      "epoch": 0.5291532778555931,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54205946219907e-06,
+      "loss": 0.8433,
+      "step": 19267
+    },
+    {
+      "epoch": 0.5291807420834364,
+      "grad_norm": 0.0,
+      "learning_rate": 9.541170881188715e-06,
+      "loss": 0.6924,
+      "step": 19268
+    },
+    {
+      "epoch": 0.5292082063112795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.540282303808799e-06,
+      "loss": 0.8718,
+      "step": 19269
+    },
+    {
+      "epoch": 0.5292356705391228,
+      "grad_norm": 0.0,
+      "learning_rate": 9.539393730066343e-06,
+      "loss": 0.9049,
+      "step": 19270
+    },
+    {
+      "epoch": 0.5292631347669661,
+      "grad_norm": 0.0,
+      "learning_rate": 9.538505159968385e-06,
+      "loss": 0.8693,
+      "step": 19271
+    },
+    {
+      "epoch": 0.5292905989948092,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53761659352195e-06,
+      "loss": 0.86,
+      "step": 19272
+    },
+    {
+      "epoch": 0.5293180632226525,
+      "grad_norm": 0.0,
+      "learning_rate": 9.536728030734072e-06,
+      "loss": 0.8252,
+      "step": 19273
+    },
+    {
+      "epoch": 0.5293455274504957,
+      "grad_norm": 0.0,
+      "learning_rate": 9.535839471611783e-06,
+      "loss": 0.9527,
+      "step": 19274
+    },
+    {
+      "epoch": 0.529372991678339,
+      "grad_norm": 0.0,
+      "learning_rate": 9.534950916162112e-06,
+      "loss": 0.9373,
+      "step": 19275
+    },
+    {
+      "epoch": 0.5294004559061822,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53406236439209e-06,
+      "loss": 0.9076,
+      "step": 19276
+    },
+    {
+      "epoch": 0.5294279201340254,
+      "grad_norm": 0.0,
+      "learning_rate": 9.533173816308745e-06,
+      "loss": 0.8996,
+      "step": 19277
+    },
+    {
+      "epoch": 0.5294553843618687,
+      "grad_norm": 0.0,
+      "learning_rate": 9.532285271919108e-06,
+      "loss": 0.904,
+      "step": 19278
+    },
+    {
+      "epoch": 0.529482848589712,
+      "grad_norm": 0.0,
+      "learning_rate": 9.531396731230215e-06,
+      "loss": 0.8082,
+      "step": 19279
+    },
+    {
+      "epoch": 0.5295103128175551,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53050819424909e-06,
+      "loss": 0.8916,
+      "step": 19280
+    },
+    {
+      "epoch": 0.5295377770453984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.52961966098277e-06,
+      "loss": 0.9088,
+      "step": 19281
+    },
+    {
+      "epoch": 0.5295652412732416,
+      "grad_norm": 0.0,
+      "learning_rate": 9.528731131438277e-06,
+      "loss": 0.8746,
+      "step": 19282
+    },
+    {
+      "epoch": 0.5295927055010848,
+      "grad_norm": 0.0,
+      "learning_rate": 9.527842605622649e-06,
+      "loss": 0.9886,
+      "step": 19283
+    },
+    {
+      "epoch": 0.5296201697289281,
+      "grad_norm": 0.0,
+      "learning_rate": 9.52695408354291e-06,
+      "loss": 0.9442,
+      "step": 19284
+    },
+    {
+      "epoch": 0.5296476339567713,
+      "grad_norm": 0.0,
+      "learning_rate": 9.526065565206095e-06,
+      "loss": 0.8319,
+      "step": 19285
+    },
+    {
+      "epoch": 0.5296750981846146,
+      "grad_norm": 0.0,
+      "learning_rate": 9.525177050619233e-06,
+      "loss": 0.9479,
+      "step": 19286
+    },
+    {
+      "epoch": 0.5297025624124577,
+      "grad_norm": 0.0,
+      "learning_rate": 9.524288539789356e-06,
+      "loss": 0.8789,
+      "step": 19287
+    },
+    {
+      "epoch": 0.529730026640301,
+      "grad_norm": 0.0,
+      "learning_rate": 9.52340003272349e-06,
+      "loss": 0.8146,
+      "step": 19288
+    },
+    {
+      "epoch": 0.5297574908681443,
+      "grad_norm": 0.0,
+      "learning_rate": 9.522511529428668e-06,
+      "loss": 0.8187,
+      "step": 19289
+    },
+    {
+      "epoch": 0.5297849550959874,
+      "grad_norm": 0.0,
+      "learning_rate": 9.521623029911917e-06,
+      "loss": 0.947,
+      "step": 19290
+    },
+    {
+      "epoch": 0.5298124193238307,
+      "grad_norm": 0.0,
+      "learning_rate": 9.520734534180274e-06,
+      "loss": 0.9342,
+      "step": 19291
+    },
+    {
+      "epoch": 0.529839883551674,
+      "grad_norm": 0.0,
+      "learning_rate": 9.519846042240763e-06,
+      "loss": 0.7854,
+      "step": 19292
+    },
+    {
+      "epoch": 0.5298673477795172,
+      "grad_norm": 0.0,
+      "learning_rate": 9.51895755410042e-06,
+      "loss": 0.8931,
+      "step": 19293
+    },
+    {
+      "epoch": 0.5298948120073604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.518069069766267e-06,
+      "loss": 0.8426,
+      "step": 19294
+    },
+    {
+      "epoch": 0.5299222762352036,
+      "grad_norm": 0.0,
+      "learning_rate": 9.517180589245339e-06,
+      "loss": 0.8578,
+      "step": 19295
+    },
+    {
+      "epoch": 0.5299497404630469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.516292112544663e-06,
+      "loss": 0.8843,
+      "step": 19296
+    },
+    {
+      "epoch": 0.5299772046908902,
+      "grad_norm": 0.0,
+      "learning_rate": 9.515403639671274e-06,
+      "loss": 0.9247,
+      "step": 19297
+    },
+    {
+      "epoch": 0.5300046689187333,
+      "grad_norm": 0.0,
+      "learning_rate": 9.514515170632202e-06,
+      "loss": 0.8842,
+      "step": 19298
+    },
+    {
+      "epoch": 0.5300321331465766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.513626705434469e-06,
+      "loss": 0.8745,
+      "step": 19299
+    },
+    {
+      "epoch": 0.5300595973744198,
+      "grad_norm": 0.0,
+      "learning_rate": 9.512738244085112e-06,
+      "loss": 0.822,
+      "step": 19300
+    },
+    {
+      "epoch": 0.530087061602263,
+      "grad_norm": 0.0,
+      "learning_rate": 9.511849786591158e-06,
+      "loss": 0.9581,
+      "step": 19301
+    },
+    {
+      "epoch": 0.5301145258301063,
+      "grad_norm": 0.0,
+      "learning_rate": 9.510961332959637e-06,
+      "loss": 0.8567,
+      "step": 19302
+    },
+    {
+      "epoch": 0.5301419900579495,
+      "grad_norm": 0.0,
+      "learning_rate": 9.51007288319758e-06,
+      "loss": 0.9628,
+      "step": 19303
+    },
+    {
+      "epoch": 0.5301694542857928,
+      "grad_norm": 0.0,
+      "learning_rate": 9.50918443731202e-06,
+      "loss": 0.9073,
+      "step": 19304
+    },
+    {
+      "epoch": 0.530196918513636,
+      "grad_norm": 0.0,
+      "learning_rate": 9.508295995309978e-06,
+      "loss": 0.9869,
+      "step": 19305
+    },
+    {
+      "epoch": 0.5302243827414792,
+      "grad_norm": 0.0,
+      "learning_rate": 9.507407557198489e-06,
+      "loss": 0.9417,
+      "step": 19306
+    },
+    {
+      "epoch": 0.5302518469693225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.506519122984583e-06,
+      "loss": 0.8508,
+      "step": 19307
+    },
+    {
+      "epoch": 0.5302793111971656,
+      "grad_norm": 0.0,
+      "learning_rate": 9.505630692675289e-06,
+      "loss": 0.9156,
+      "step": 19308
+    },
+    {
+      "epoch": 0.5303067754250089,
+      "grad_norm": 0.0,
+      "learning_rate": 9.504742266277633e-06,
+      "loss": 1.006,
+      "step": 19309
+    },
+    {
+      "epoch": 0.5303342396528522,
+      "grad_norm": 0.0,
+      "learning_rate": 9.503853843798656e-06,
+      "loss": 0.9243,
+      "step": 19310
+    },
+    {
+      "epoch": 0.5303617038806954,
+      "grad_norm": 0.0,
+      "learning_rate": 9.502965425245372e-06,
+      "loss": 0.895,
+      "step": 19311
+    },
+    {
+      "epoch": 0.5303891681085386,
+      "grad_norm": 0.0,
+      "learning_rate": 9.502077010624823e-06,
+      "loss": 0.8737,
+      "step": 19312
+    },
+    {
+      "epoch": 0.5304166323363818,
+      "grad_norm": 0.0,
+      "learning_rate": 9.50118859994403e-06,
+      "loss": 0.8969,
+      "step": 19313
+    },
+    {
+      "epoch": 0.5304440965642251,
+      "grad_norm": 0.0,
+      "learning_rate": 9.500300193210027e-06,
+      "loss": 0.9359,
+      "step": 19314
+    },
+    {
+      "epoch": 0.5304715607920684,
+      "grad_norm": 0.0,
+      "learning_rate": 9.499411790429844e-06,
+      "loss": 0.9147,
+      "step": 19315
+    },
+    {
+      "epoch": 0.5304990250199115,
+      "grad_norm": 0.0,
+      "learning_rate": 9.498523391610511e-06,
+      "loss": 0.8651,
+      "step": 19316
+    },
+    {
+      "epoch": 0.5305264892477548,
+      "grad_norm": 0.0,
+      "learning_rate": 9.497634996759051e-06,
+      "loss": 0.8985,
+      "step": 19317
+    },
+    {
+      "epoch": 0.5305539534755981,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496746605882497e-06,
+      "loss": 0.8977,
+      "step": 19318
+    },
+    {
+      "epoch": 0.5305814177034412,
+      "grad_norm": 0.0,
+      "learning_rate": 9.495858218987882e-06,
+      "loss": 0.8698,
+      "step": 19319
+    },
+    {
+      "epoch": 0.5306088819312845,
+      "grad_norm": 0.0,
+      "learning_rate": 9.494969836082229e-06,
+      "loss": 0.9131,
+      "step": 19320
+    },
+    {
+      "epoch": 0.5306363461591277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.49408145717257e-06,
+      "loss": 0.7542,
+      "step": 19321
+    },
+    {
+      "epoch": 0.530663810386971,
+      "grad_norm": 0.0,
+      "learning_rate": 9.493193082265939e-06,
+      "loss": 0.8212,
+      "step": 19322
+    },
+    {
+      "epoch": 0.5306912746148142,
+      "grad_norm": 0.0,
+      "learning_rate": 9.492304711369357e-06,
+      "loss": 0.8628,
+      "step": 19323
+    },
+    {
+      "epoch": 0.5307187388426574,
+      "grad_norm": 0.0,
+      "learning_rate": 9.491416344489859e-06,
+      "loss": 0.7747,
+      "step": 19324
+    },
+    {
+      "epoch": 0.5307462030705007,
+      "grad_norm": 0.0,
+      "learning_rate": 9.490527981634469e-06,
+      "loss": 0.8909,
+      "step": 19325
+    },
+    {
+      "epoch": 0.5307736672983439,
+      "grad_norm": 0.0,
+      "learning_rate": 9.489639622810219e-06,
+      "loss": 0.8797,
+      "step": 19326
+    },
+    {
+      "epoch": 0.5308011315261871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.48875126802414e-06,
+      "loss": 0.9689,
+      "step": 19327
+    },
+    {
+      "epoch": 0.5308285957540304,
+      "grad_norm": 0.0,
+      "learning_rate": 9.487862917283263e-06,
+      "loss": 0.9132,
+      "step": 19328
+    },
+    {
+      "epoch": 0.5308560599818736,
+      "grad_norm": 0.0,
+      "learning_rate": 9.486974570594605e-06,
+      "loss": 0.7828,
+      "step": 19329
+    },
+    {
+      "epoch": 0.5308835242097169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.486086227965207e-06,
+      "loss": 0.9895,
+      "step": 19330
+    },
+    {
+      "epoch": 0.5309109884375601,
+      "grad_norm": 0.0,
+      "learning_rate": 9.485197889402093e-06,
+      "loss": 0.9223,
+      "step": 19331
+    },
+    {
+      "epoch": 0.5309384526654033,
+      "grad_norm": 0.0,
+      "learning_rate": 9.484309554912291e-06,
+      "loss": 0.9191,
+      "step": 19332
+    },
+    {
+      "epoch": 0.5309659168932466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.483421224502834e-06,
+      "loss": 0.8876,
+      "step": 19333
+    },
+    {
+      "epoch": 0.5309933811210897,
+      "grad_norm": 0.0,
+      "learning_rate": 9.482532898180751e-06,
+      "loss": 0.846,
+      "step": 19334
+    },
+    {
+      "epoch": 0.531020845348933,
+      "grad_norm": 0.0,
+      "learning_rate": 9.481644575953062e-06,
+      "loss": 0.902,
+      "step": 19335
+    },
+    {
+      "epoch": 0.5310483095767763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.480756257826807e-06,
+      "loss": 0.8416,
+      "step": 19336
+    },
+    {
+      "epoch": 0.5310757738046195,
+      "grad_norm": 0.0,
+      "learning_rate": 9.479867943809006e-06,
+      "loss": 0.919,
+      "step": 19337
+    },
+    {
+      "epoch": 0.5311032380324627,
+      "grad_norm": 0.0,
+      "learning_rate": 9.478979633906692e-06,
+      "loss": 0.8255,
+      "step": 19338
+    },
+    {
+      "epoch": 0.5311307022603059,
+      "grad_norm": 0.0,
+      "learning_rate": 9.478091328126895e-06,
+      "loss": 0.9097,
+      "step": 19339
+    },
+    {
+      "epoch": 0.5311581664881492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.477203026476643e-06,
+      "loss": 0.8895,
+      "step": 19340
+    },
+    {
+      "epoch": 0.5311856307159925,
+      "grad_norm": 0.0,
+      "learning_rate": 9.476314728962959e-06,
+      "loss": 0.9563,
+      "step": 19341
+    },
+    {
+      "epoch": 0.5312130949438356,
+      "grad_norm": 0.0,
+      "learning_rate": 9.475426435592877e-06,
+      "loss": 0.9353,
+      "step": 19342
+    },
+    {
+      "epoch": 0.5312405591716789,
+      "grad_norm": 0.0,
+      "learning_rate": 9.474538146373425e-06,
+      "loss": 0.9336,
+      "step": 19343
+    },
+    {
+      "epoch": 0.5312680233995222,
+      "grad_norm": 0.0,
+      "learning_rate": 9.473649861311628e-06,
+      "loss": 0.8875,
+      "step": 19344
+    },
+    {
+      "epoch": 0.5312954876273653,
+      "grad_norm": 0.0,
+      "learning_rate": 9.47276158041452e-06,
+      "loss": 0.8692,
+      "step": 19345
+    },
+    {
+      "epoch": 0.5313229518552086,
+      "grad_norm": 0.0,
+      "learning_rate": 9.47187330368913e-06,
+      "loss": 0.9072,
+      "step": 19346
+    },
+    {
+      "epoch": 0.5313504160830518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.47098503114248e-06,
+      "loss": 0.8107,
+      "step": 19347
+    },
+    {
+      "epoch": 0.5313778803108951,
+      "grad_norm": 0.0,
+      "learning_rate": 9.470096762781597e-06,
+      "loss": 0.9667,
+      "step": 19348
+    },
+    {
+      "epoch": 0.5314053445387383,
+      "grad_norm": 0.0,
+      "learning_rate": 9.469208498613516e-06,
+      "loss": 0.8987,
+      "step": 19349
+    },
+    {
+      "epoch": 0.5314328087665815,
+      "grad_norm": 0.0,
+      "learning_rate": 9.468320238645263e-06,
+      "loss": 0.921,
+      "step": 19350
+    },
+    {
+      "epoch": 0.5314602729944248,
+      "grad_norm": 0.0,
+      "learning_rate": 9.467431982883869e-06,
+      "loss": 0.9042,
+      "step": 19351
+    },
+    {
+      "epoch": 0.5314877372222679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.46654373133636e-06,
+      "loss": 1.0161,
+      "step": 19352
+    },
+    {
+      "epoch": 0.5315152014501112,
+      "grad_norm": 0.0,
+      "learning_rate": 9.465655484009761e-06,
+      "loss": 0.8776,
+      "step": 19353
+    },
+    {
+      "epoch": 0.5315426656779545,
+      "grad_norm": 0.0,
+      "learning_rate": 9.464767240911101e-06,
+      "loss": 0.9407,
+      "step": 19354
+    },
+    {
+      "epoch": 0.5315701299057977,
+      "grad_norm": 0.0,
+      "learning_rate": 9.463879002047414e-06,
+      "loss": 0.897,
+      "step": 19355
+    },
+    {
+      "epoch": 0.5315975941336409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.462990767425721e-06,
+      "loss": 0.7011,
+      "step": 19356
+    },
+    {
+      "epoch": 0.5316250583614842,
+      "grad_norm": 0.0,
+      "learning_rate": 9.462102537053057e-06,
+      "loss": 0.9754,
+      "step": 19357
+    },
+    {
+      "epoch": 0.5316525225893274,
+      "grad_norm": 0.0,
+      "learning_rate": 9.461214310936442e-06,
+      "loss": 0.8319,
+      "step": 19358
+    },
+    {
+      "epoch": 0.5316799868171707,
+      "grad_norm": 0.0,
+      "learning_rate": 9.46032608908291e-06,
+      "loss": 0.8852,
+      "step": 19359
+    },
+    {
+      "epoch": 0.5317074510450138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.459437871499485e-06,
+      "loss": 0.8775,
+      "step": 19360
+    },
+    {
+      "epoch": 0.5317349152728571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.458549658193198e-06,
+      "loss": 0.907,
+      "step": 19361
+    },
+    {
+      "epoch": 0.5317623795007004,
+      "grad_norm": 0.0,
+      "learning_rate": 9.457661449171075e-06,
+      "loss": 0.8096,
+      "step": 19362
+    },
+    {
+      "epoch": 0.5317898437285435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.456773244440148e-06,
+      "loss": 0.8187,
+      "step": 19363
+    },
+    {
+      "epoch": 0.5318173079563868,
+      "grad_norm": 0.0,
+      "learning_rate": 9.455885044007442e-06,
+      "loss": 0.815,
+      "step": 19364
+    },
+    {
+      "epoch": 0.53184477218423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.45499684787998e-06,
+      "loss": 0.8526,
+      "step": 19365
+    },
+    {
+      "epoch": 0.5318722364120733,
+      "grad_norm": 0.0,
+      "learning_rate": 9.454108656064794e-06,
+      "loss": 0.8681,
+      "step": 19366
+    },
+    {
+      "epoch": 0.5318997006399165,
+      "grad_norm": 0.0,
+      "learning_rate": 9.453220468568914e-06,
+      "loss": 0.8929,
+      "step": 19367
+    },
+    {
+      "epoch": 0.5319271648677597,
+      "grad_norm": 0.0,
+      "learning_rate": 9.452332285399364e-06,
+      "loss": 0.9008,
+      "step": 19368
+    },
+    {
+      "epoch": 0.531954629095603,
+      "grad_norm": 0.0,
+      "learning_rate": 9.451444106563176e-06,
+      "loss": 0.8972,
+      "step": 19369
+    },
+    {
+      "epoch": 0.5319820933234463,
+      "grad_norm": 0.0,
+      "learning_rate": 9.450555932067372e-06,
+      "loss": 0.8946,
+      "step": 19370
+    },
+    {
+      "epoch": 0.5320095575512894,
+      "grad_norm": 0.0,
+      "learning_rate": 9.449667761918984e-06,
+      "loss": 0.7762,
+      "step": 19371
+    },
+    {
+      "epoch": 0.5320370217791327,
+      "grad_norm": 0.0,
+      "learning_rate": 9.448779596125037e-06,
+      "loss": 0.8253,
+      "step": 19372
+    },
+    {
+      "epoch": 0.5320644860069759,
+      "grad_norm": 0.0,
+      "learning_rate": 9.447891434692557e-06,
+      "loss": 0.9696,
+      "step": 19373
+    },
+    {
+      "epoch": 0.5320919502348191,
+      "grad_norm": 0.0,
+      "learning_rate": 9.447003277628575e-06,
+      "loss": 0.9717,
+      "step": 19374
+    },
+    {
+      "epoch": 0.5321194144626624,
+      "grad_norm": 0.0,
+      "learning_rate": 9.446115124940121e-06,
+      "loss": 0.8808,
+      "step": 19375
+    },
+    {
+      "epoch": 0.5321468786905056,
+      "grad_norm": 0.0,
+      "learning_rate": 9.445226976634218e-06,
+      "loss": 1.008,
+      "step": 19376
+    },
+    {
+      "epoch": 0.5321743429183489,
+      "grad_norm": 0.0,
+      "learning_rate": 9.444338832717892e-06,
+      "loss": 0.8609,
+      "step": 19377
+    },
+    {
+      "epoch": 0.532201807146192,
+      "grad_norm": 0.0,
+      "learning_rate": 9.44345069319817e-06,
+      "loss": 0.8834,
+      "step": 19378
+    },
+    {
+      "epoch": 0.5322292713740353,
+      "grad_norm": 0.0,
+      "learning_rate": 9.442562558082085e-06,
+      "loss": 0.8788,
+      "step": 19379
+    },
+    {
+      "epoch": 0.5322567356018786,
+      "grad_norm": 0.0,
+      "learning_rate": 9.441674427376659e-06,
+      "loss": 0.791,
+      "step": 19380
+    },
+    {
+      "epoch": 0.5322841998297217,
+      "grad_norm": 0.0,
+      "learning_rate": 9.440786301088926e-06,
+      "loss": 0.7552,
+      "step": 19381
+    },
+    {
+      "epoch": 0.532311664057565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.439898179225904e-06,
+      "loss": 0.8804,
+      "step": 19382
+    },
+    {
+      "epoch": 0.5323391282854083,
+      "grad_norm": 0.0,
+      "learning_rate": 9.439010061794626e-06,
+      "loss": 0.8936,
+      "step": 19383
+    },
+    {
+      "epoch": 0.5323665925132515,
+      "grad_norm": 0.0,
+      "learning_rate": 9.438121948802116e-06,
+      "loss": 0.955,
+      "step": 19384
+    },
+    {
+      "epoch": 0.5323940567410947,
+      "grad_norm": 0.0,
+      "learning_rate": 9.437233840255402e-06,
+      "loss": 0.8662,
+      "step": 19385
+    },
+    {
+      "epoch": 0.5324215209689379,
+      "grad_norm": 0.0,
+      "learning_rate": 9.436345736161514e-06,
+      "loss": 0.8625,
+      "step": 19386
+    },
+    {
+      "epoch": 0.5324489851967812,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43545763652748e-06,
+      "loss": 0.8319,
+      "step": 19387
+    },
+    {
+      "epoch": 0.5324764494246245,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43456954136032e-06,
+      "loss": 0.9055,
+      "step": 19388
+    },
+    {
+      "epoch": 0.5325039136524676,
+      "grad_norm": 0.0,
+      "learning_rate": 9.433681450667063e-06,
+      "loss": 0.8048,
+      "step": 19389
+    },
+    {
+      "epoch": 0.5325313778803109,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43279336445474e-06,
+      "loss": 0.866,
+      "step": 19390
+    },
+    {
+      "epoch": 0.5325588421081542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.431905282730376e-06,
+      "loss": 0.9927,
+      "step": 19391
+    },
+    {
+      "epoch": 0.5325863063359974,
+      "grad_norm": 0.0,
+      "learning_rate": 9.431017205500996e-06,
+      "loss": 0.9571,
+      "step": 19392
+    },
+    {
+      "epoch": 0.5326137705638406,
+      "grad_norm": 0.0,
+      "learning_rate": 9.430129132773632e-06,
+      "loss": 0.8367,
+      "step": 19393
+    },
+    {
+      "epoch": 0.5326412347916838,
+      "grad_norm": 0.0,
+      "learning_rate": 9.429241064555303e-06,
+      "loss": 0.8838,
+      "step": 19394
+    },
+    {
+      "epoch": 0.5326686990195271,
+      "grad_norm": 0.0,
+      "learning_rate": 9.428353000853043e-06,
+      "loss": 0.9012,
+      "step": 19395
+    },
+    {
+      "epoch": 0.5326961632473703,
+      "grad_norm": 0.0,
+      "learning_rate": 9.427464941673872e-06,
+      "loss": 0.8982,
+      "step": 19396
+    },
+    {
+      "epoch": 0.5327236274752135,
+      "grad_norm": 0.0,
+      "learning_rate": 9.426576887024822e-06,
+      "loss": 0.8558,
+      "step": 19397
+    },
+    {
+      "epoch": 0.5327510917030568,
+      "grad_norm": 0.0,
+      "learning_rate": 9.425688836912917e-06,
+      "loss": 0.9055,
+      "step": 19398
+    },
+    {
+      "epoch": 0.5327785559309,
+      "grad_norm": 0.0,
+      "learning_rate": 9.424800791345186e-06,
+      "loss": 0.8938,
+      "step": 19399
+    },
+    {
+      "epoch": 0.5328060201587432,
+      "grad_norm": 0.0,
+      "learning_rate": 9.423912750328655e-06,
+      "loss": 0.9318,
+      "step": 19400
+    },
+    {
+      "epoch": 0.5328334843865865,
+      "grad_norm": 0.0,
+      "learning_rate": 9.423024713870345e-06,
+      "loss": 0.9481,
+      "step": 19401
+    },
+    {
+      "epoch": 0.5328609486144297,
+      "grad_norm": 0.0,
+      "learning_rate": 9.42213668197729e-06,
+      "loss": 0.8934,
+      "step": 19402
+    },
+    {
+      "epoch": 0.532888412842273,
+      "grad_norm": 0.0,
+      "learning_rate": 9.421248654656513e-06,
+      "loss": 0.8637,
+      "step": 19403
+    },
+    {
+      "epoch": 0.5329158770701162,
+      "grad_norm": 0.0,
+      "learning_rate": 9.420360631915038e-06,
+      "loss": 0.8431,
+      "step": 19404
+    },
+    {
+      "epoch": 0.5329433412979594,
+      "grad_norm": 0.0,
+      "learning_rate": 9.419472613759901e-06,
+      "loss": 0.7591,
+      "step": 19405
+    },
+    {
+      "epoch": 0.5329708055258027,
+      "grad_norm": 0.0,
+      "learning_rate": 9.418584600198117e-06,
+      "loss": 0.9005,
+      "step": 19406
+    },
+    {
+      "epoch": 0.5329982697536458,
+      "grad_norm": 0.0,
+      "learning_rate": 9.417696591236717e-06,
+      "loss": 0.8336,
+      "step": 19407
+    },
+    {
+      "epoch": 0.5330257339814891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.416808586882727e-06,
+      "loss": 0.8675,
+      "step": 19408
+    },
+    {
+      "epoch": 0.5330531982093324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.415920587143173e-06,
+      "loss": 0.9172,
+      "step": 19409
+    },
+    {
+      "epoch": 0.5330806624371756,
+      "grad_norm": 0.0,
+      "learning_rate": 9.415032592025082e-06,
+      "loss": 0.8949,
+      "step": 19410
+    },
+    {
+      "epoch": 0.5331081266650188,
+      "grad_norm": 0.0,
+      "learning_rate": 9.414144601535481e-06,
+      "loss": 0.9019,
+      "step": 19411
+    },
+    {
+      "epoch": 0.533135590892862,
+      "grad_norm": 0.0,
+      "learning_rate": 9.413256615681395e-06,
+      "loss": 0.7769,
+      "step": 19412
+    },
+    {
+      "epoch": 0.5331630551207053,
+      "grad_norm": 0.0,
+      "learning_rate": 9.412368634469849e-06,
+      "loss": 0.8683,
+      "step": 19413
+    },
+    {
+      "epoch": 0.5331905193485486,
+      "grad_norm": 0.0,
+      "learning_rate": 9.411480657907868e-06,
+      "loss": 0.9203,
+      "step": 19414
+    },
+    {
+      "epoch": 0.5332179835763917,
+      "grad_norm": 0.0,
+      "learning_rate": 9.410592686002483e-06,
+      "loss": 0.8697,
+      "step": 19415
+    },
+    {
+      "epoch": 0.533245447804235,
+      "grad_norm": 0.0,
+      "learning_rate": 9.409704718760719e-06,
+      "loss": 0.9259,
+      "step": 19416
+    },
+    {
+      "epoch": 0.5332729120320783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.408816756189595e-06,
+      "loss": 0.8945,
+      "step": 19417
+    },
+    {
+      "epoch": 0.5333003762599214,
+      "grad_norm": 0.0,
+      "learning_rate": 9.407928798296142e-06,
+      "loss": 0.9284,
+      "step": 19418
+    },
+    {
+      "epoch": 0.5333278404877647,
+      "grad_norm": 0.0,
+      "learning_rate": 9.407040845087388e-06,
+      "loss": 0.9227,
+      "step": 19419
+    },
+    {
+      "epoch": 0.5333553047156079,
+      "grad_norm": 0.0,
+      "learning_rate": 9.406152896570355e-06,
+      "loss": 0.9769,
+      "step": 19420
+    },
+    {
+      "epoch": 0.5333827689434512,
+      "grad_norm": 0.0,
+      "learning_rate": 9.40526495275207e-06,
+      "loss": 0.7832,
+      "step": 19421
+    },
+    {
+      "epoch": 0.5334102331712944,
+      "grad_norm": 0.0,
+      "learning_rate": 9.404377013639563e-06,
+      "loss": 0.9162,
+      "step": 19422
+    },
+    {
+      "epoch": 0.5334376973991376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.403489079239853e-06,
+      "loss": 0.827,
+      "step": 19423
+    },
+    {
+      "epoch": 0.5334651616269809,
+      "grad_norm": 0.0,
+      "learning_rate": 9.402601149559969e-06,
+      "loss": 0.8818,
+      "step": 19424
+    },
+    {
+      "epoch": 0.533492625854824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.401713224606934e-06,
+      "loss": 0.8813,
+      "step": 19425
+    },
+    {
+      "epoch": 0.5335200900826673,
+      "grad_norm": 0.0,
+      "learning_rate": 9.400825304387776e-06,
+      "loss": 0.919,
+      "step": 19426
+    },
+    {
+      "epoch": 0.5335475543105106,
+      "grad_norm": 0.0,
+      "learning_rate": 9.399937388909522e-06,
+      "loss": 0.8311,
+      "step": 19427
+    },
+    {
+      "epoch": 0.5335750185383538,
+      "grad_norm": 0.0,
+      "learning_rate": 9.3990494781792e-06,
+      "loss": 1.029,
+      "step": 19428
+    },
+    {
+      "epoch": 0.533602482766197,
+      "grad_norm": 0.0,
+      "learning_rate": 9.398161572203825e-06,
+      "loss": 0.8387,
+      "step": 19429
+    },
+    {
+      "epoch": 0.5336299469940403,
+      "grad_norm": 0.0,
+      "learning_rate": 9.397273670990429e-06,
+      "loss": 0.9609,
+      "step": 19430
+    },
+    {
+      "epoch": 0.5336574112218835,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39638577454604e-06,
+      "loss": 0.8998,
+      "step": 19431
+    },
+    {
+      "epoch": 0.5336848754497268,
+      "grad_norm": 0.0,
+      "learning_rate": 9.395497882877678e-06,
+      "loss": 0.9356,
+      "step": 19432
+    },
+    {
+      "epoch": 0.5337123396775699,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39460999599237e-06,
+      "loss": 0.8011,
+      "step": 19433
+    },
+    {
+      "epoch": 0.5337398039054132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.393722113897147e-06,
+      "loss": 0.8072,
+      "step": 19434
+    },
+    {
+      "epoch": 0.5337672681332565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.392834236599027e-06,
+      "loss": 0.8488,
+      "step": 19435
+    },
+    {
+      "epoch": 0.5337947323610996,
+      "grad_norm": 0.0,
+      "learning_rate": 9.391946364105039e-06,
+      "loss": 0.8043,
+      "step": 19436
+    },
+    {
+      "epoch": 0.5338221965889429,
+      "grad_norm": 0.0,
+      "learning_rate": 9.391058496422204e-06,
+      "loss": 0.826,
+      "step": 19437
+    },
+    {
+      "epoch": 0.5338496608167861,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39017063355755e-06,
+      "loss": 0.8874,
+      "step": 19438
+    },
+    {
+      "epoch": 0.5338771250446294,
+      "grad_norm": 0.0,
+      "learning_rate": 9.389282775518105e-06,
+      "loss": 0.8833,
+      "step": 19439
+    },
+    {
+      "epoch": 0.5339045892724726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.388394922310893e-06,
+      "loss": 0.8952,
+      "step": 19440
+    },
+    {
+      "epoch": 0.5339320535003158,
+      "grad_norm": 0.0,
+      "learning_rate": 9.387507073942932e-06,
+      "loss": 0.8748,
+      "step": 19441
+    },
+    {
+      "epoch": 0.5339595177281591,
+      "grad_norm": 0.0,
+      "learning_rate": 9.386619230421254e-06,
+      "loss": 0.8806,
+      "step": 19442
+    },
+    {
+      "epoch": 0.5339869819560024,
+      "grad_norm": 0.0,
+      "learning_rate": 9.385731391752884e-06,
+      "loss": 0.9253,
+      "step": 19443
+    },
+    {
+      "epoch": 0.5340144461838455,
+      "grad_norm": 0.0,
+      "learning_rate": 9.384843557944842e-06,
+      "loss": 0.9306,
+      "step": 19444
+    },
+    {
+      "epoch": 0.5340419104116888,
+      "grad_norm": 0.0,
+      "learning_rate": 9.383955729004156e-06,
+      "loss": 0.8644,
+      "step": 19445
+    },
+    {
+      "epoch": 0.534069374639532,
+      "grad_norm": 0.0,
+      "learning_rate": 9.383067904937857e-06,
+      "loss": 0.9358,
+      "step": 19446
+    },
+    {
+      "epoch": 0.5340968388673752,
+      "grad_norm": 0.0,
+      "learning_rate": 9.382180085752958e-06,
+      "loss": 0.7136,
+      "step": 19447
+    },
+    {
+      "epoch": 0.5341243030952185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.381292271456493e-06,
+      "loss": 0.8838,
+      "step": 19448
+    },
+    {
+      "epoch": 0.5341517673230617,
+      "grad_norm": 0.0,
+      "learning_rate": 9.38040446205548e-06,
+      "loss": 0.8839,
+      "step": 19449
+    },
+    {
+      "epoch": 0.534179231550905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.379516657556948e-06,
+      "loss": 1.0085,
+      "step": 19450
+    },
+    {
+      "epoch": 0.5342066957787481,
+      "grad_norm": 0.0,
+      "learning_rate": 9.378628857967921e-06,
+      "loss": 0.8684,
+      "step": 19451
+    },
+    {
+      "epoch": 0.5342341600065914,
+      "grad_norm": 0.0,
+      "learning_rate": 9.377741063295425e-06,
+      "loss": 0.9128,
+      "step": 19452
+    },
+    {
+      "epoch": 0.5342616242344347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.37685327354648e-06,
+      "loss": 0.9695,
+      "step": 19453
+    },
+    {
+      "epoch": 0.5342890884622779,
+      "grad_norm": 0.0,
+      "learning_rate": 9.375965488728113e-06,
+      "loss": 0.8808,
+      "step": 19454
+    },
+    {
+      "epoch": 0.5343165526901211,
+      "grad_norm": 0.0,
+      "learning_rate": 9.375077708847352e-06,
+      "loss": 0.8927,
+      "step": 19455
+    },
+    {
+      "epoch": 0.5343440169179644,
+      "grad_norm": 0.0,
+      "learning_rate": 9.374189933911215e-06,
+      "loss": 0.8962,
+      "step": 19456
+    },
+    {
+      "epoch": 0.5343714811458076,
+      "grad_norm": 0.0,
+      "learning_rate": 9.373302163926728e-06,
+      "loss": 0.8386,
+      "step": 19457
+    },
+    {
+      "epoch": 0.5343989453736508,
+      "grad_norm": 0.0,
+      "learning_rate": 9.372414398900924e-06,
+      "loss": 0.8524,
+      "step": 19458
+    },
+    {
+      "epoch": 0.534426409601494,
+      "grad_norm": 0.0,
+      "learning_rate": 9.371526638840813e-06,
+      "loss": 0.8928,
+      "step": 19459
+    },
+    {
+      "epoch": 0.5344538738293373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.370638883753432e-06,
+      "loss": 0.9291,
+      "step": 19460
+    },
+    {
+      "epoch": 0.5344813380571806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.369751133645796e-06,
+      "loss": 0.8197,
+      "step": 19461
+    },
+    {
+      "epoch": 0.5345088022850237,
+      "grad_norm": 0.0,
+      "learning_rate": 9.368863388524934e-06,
+      "loss": 0.8624,
+      "step": 19462
+    },
+    {
+      "epoch": 0.534536266512867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.36797564839787e-06,
+      "loss": 0.863,
+      "step": 19463
+    },
+    {
+      "epoch": 0.5345637307407102,
+      "grad_norm": 0.0,
+      "learning_rate": 9.367087913271632e-06,
+      "loss": 0.9185,
+      "step": 19464
+    },
+    {
+      "epoch": 0.5345911949685535,
+      "grad_norm": 0.0,
+      "learning_rate": 9.366200183153233e-06,
+      "loss": 0.9512,
+      "step": 19465
+    },
+    {
+      "epoch": 0.5346186591963967,
+      "grad_norm": 0.0,
+      "learning_rate": 9.365312458049706e-06,
+      "loss": 0.8134,
+      "step": 19466
+    },
+    {
+      "epoch": 0.5346461234242399,
+      "grad_norm": 0.0,
+      "learning_rate": 9.364424737968074e-06,
+      "loss": 0.9353,
+      "step": 19467
+    },
+    {
+      "epoch": 0.5346735876520832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.363537022915357e-06,
+      "loss": 0.8492,
+      "step": 19468
+    },
+    {
+      "epoch": 0.5347010518799264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.36264931289858e-06,
+      "loss": 0.8644,
+      "step": 19469
+    },
+    {
+      "epoch": 0.5347285161077696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.361761607924775e-06,
+      "loss": 0.7389,
+      "step": 19470
+    },
+    {
+      "epoch": 0.5347559803356129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.360873908000956e-06,
+      "loss": 0.7678,
+      "step": 19471
+    },
+    {
+      "epoch": 0.5347834445634561,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35998621313415e-06,
+      "loss": 0.9391,
+      "step": 19472
+    },
+    {
+      "epoch": 0.5348109087912993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.359098523331381e-06,
+      "loss": 0.9026,
+      "step": 19473
+    },
+    {
+      "epoch": 0.5348383730191426,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35821083859967e-06,
+      "loss": 0.9102,
+      "step": 19474
+    },
+    {
+      "epoch": 0.5348658372469858,
+      "grad_norm": 0.0,
+      "learning_rate": 9.357323158946048e-06,
+      "loss": 0.901,
+      "step": 19475
+    },
+    {
+      "epoch": 0.534893301474829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.356435484377536e-06,
+      "loss": 0.7902,
+      "step": 19476
+    },
+    {
+      "epoch": 0.5349207657026722,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35554781490115e-06,
+      "loss": 0.86,
+      "step": 19477
+    },
+    {
+      "epoch": 0.5349482299305155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35466015052392e-06,
+      "loss": 0.963,
+      "step": 19478
+    },
+    {
+      "epoch": 0.5349756941583588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.353772491252872e-06,
+      "loss": 0.8033,
+      "step": 19479
+    },
+    {
+      "epoch": 0.5350031583862019,
+      "grad_norm": 0.0,
+      "learning_rate": 9.352884837095024e-06,
+      "loss": 0.9025,
+      "step": 19480
+    },
+    {
+      "epoch": 0.5350306226140452,
+      "grad_norm": 0.0,
+      "learning_rate": 9.351997188057406e-06,
+      "loss": 0.9452,
+      "step": 19481
+    },
+    {
+      "epoch": 0.5350580868418885,
+      "grad_norm": 0.0,
+      "learning_rate": 9.351109544147032e-06,
+      "loss": 0.9168,
+      "step": 19482
+    },
+    {
+      "epoch": 0.5350855510697317,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35022190537093e-06,
+      "loss": 0.9276,
+      "step": 19483
+    },
+    {
+      "epoch": 0.5351130152975749,
+      "grad_norm": 0.0,
+      "learning_rate": 9.349334271736128e-06,
+      "loss": 0.8817,
+      "step": 19484
+    },
+    {
+      "epoch": 0.5351404795254181,
+      "grad_norm": 0.0,
+      "learning_rate": 9.348446643249644e-06,
+      "loss": 0.8575,
+      "step": 19485
+    },
+    {
+      "epoch": 0.5351679437532614,
+      "grad_norm": 0.0,
+      "learning_rate": 9.3475590199185e-06,
+      "loss": 0.9469,
+      "step": 19486
+    },
+    {
+      "epoch": 0.5351954079811047,
+      "grad_norm": 0.0,
+      "learning_rate": 9.346671401749728e-06,
+      "loss": 0.8409,
+      "step": 19487
+    },
+    {
+      "epoch": 0.5352228722089478,
+      "grad_norm": 0.0,
+      "learning_rate": 9.345783788750341e-06,
+      "loss": 0.7759,
+      "step": 19488
+    },
+    {
+      "epoch": 0.5352503364367911,
+      "grad_norm": 0.0,
+      "learning_rate": 9.344896180927366e-06,
+      "loss": 0.9347,
+      "step": 19489
+    },
+    {
+      "epoch": 0.5352778006646343,
+      "grad_norm": 0.0,
+      "learning_rate": 9.344008578287824e-06,
+      "loss": 0.8405,
+      "step": 19490
+    },
+    {
+      "epoch": 0.5353052648924775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.343120980838744e-06,
+      "loss": 0.8846,
+      "step": 19491
+    },
+    {
+      "epoch": 0.5353327291203208,
+      "grad_norm": 0.0,
+      "learning_rate": 9.342233388587143e-06,
+      "loss": 0.8646,
+      "step": 19492
+    },
+    {
+      "epoch": 0.535360193348164,
+      "grad_norm": 0.0,
+      "learning_rate": 9.341345801540049e-06,
+      "loss": 0.8963,
+      "step": 19493
+    },
+    {
+      "epoch": 0.5353876575760073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.34045821970448e-06,
+      "loss": 0.862,
+      "step": 19494
+    },
+    {
+      "epoch": 0.5354151218038505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.339570643087462e-06,
+      "loss": 0.9107,
+      "step": 19495
+    },
+    {
+      "epoch": 0.5354425860316937,
+      "grad_norm": 0.0,
+      "learning_rate": 9.338683071696014e-06,
+      "loss": 0.8519,
+      "step": 19496
+    },
+    {
+      "epoch": 0.535470050259537,
+      "grad_norm": 0.0,
+      "learning_rate": 9.337795505537163e-06,
+      "loss": 0.8787,
+      "step": 19497
+    },
+    {
+      "epoch": 0.5354975144873801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.336907944617929e-06,
+      "loss": 0.8802,
+      "step": 19498
+    },
+    {
+      "epoch": 0.5355249787152234,
+      "grad_norm": 0.0,
+      "learning_rate": 9.336020388945342e-06,
+      "loss": 0.8755,
+      "step": 19499
+    },
+    {
+      "epoch": 0.5355524429430667,
+      "grad_norm": 0.0,
+      "learning_rate": 9.335132838526416e-06,
+      "loss": 0.9139,
+      "step": 19500
+    },
+    {
+      "epoch": 0.5355799071709099,
+      "grad_norm": 0.0,
+      "learning_rate": 9.334245293368175e-06,
+      "loss": 0.8835,
+      "step": 19501
+    },
+    {
+      "epoch": 0.5356073713987531,
+      "grad_norm": 0.0,
+      "learning_rate": 9.333357753477643e-06,
+      "loss": 0.7914,
+      "step": 19502
+    },
+    {
+      "epoch": 0.5356348356265963,
+      "grad_norm": 0.0,
+      "learning_rate": 9.332470218861844e-06,
+      "loss": 0.9409,
+      "step": 19503
+    },
+    {
+      "epoch": 0.5356622998544396,
+      "grad_norm": 0.0,
+      "learning_rate": 9.331582689527799e-06,
+      "loss": 0.8705,
+      "step": 19504
+    },
+    {
+      "epoch": 0.5356897640822829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.330695165482533e-06,
+      "loss": 0.83,
+      "step": 19505
+    },
+    {
+      "epoch": 0.535717228310126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.329807646733063e-06,
+      "loss": 0.9104,
+      "step": 19506
+    },
+    {
+      "epoch": 0.5357446925379693,
+      "grad_norm": 0.0,
+      "learning_rate": 9.328920133286416e-06,
+      "loss": 0.8698,
+      "step": 19507
+    },
+    {
+      "epoch": 0.5357721567658126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.32803262514961e-06,
+      "loss": 0.8787,
+      "step": 19508
+    },
+    {
+      "epoch": 0.5357996209936557,
+      "grad_norm": 0.0,
+      "learning_rate": 9.327145122329671e-06,
+      "loss": 0.838,
+      "step": 19509
+    },
+    {
+      "epoch": 0.535827085221499,
+      "grad_norm": 0.0,
+      "learning_rate": 9.32625762483362e-06,
+      "loss": 0.8481,
+      "step": 19510
+    },
+    {
+      "epoch": 0.5358545494493422,
+      "grad_norm": 0.0,
+      "learning_rate": 9.325370132668484e-06,
+      "loss": 0.9264,
+      "step": 19511
+    },
+    {
+      "epoch": 0.5358820136771855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.32448264584128e-06,
+      "loss": 0.9272,
+      "step": 19512
+    },
+    {
+      "epoch": 0.5359094779050287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.323595164359027e-06,
+      "loss": 0.9413,
+      "step": 19513
+    },
+    {
+      "epoch": 0.5359369421328719,
+      "grad_norm": 0.0,
+      "learning_rate": 9.322707688228753e-06,
+      "loss": 0.9341,
+      "step": 19514
+    },
+    {
+      "epoch": 0.5359644063607152,
+      "grad_norm": 0.0,
+      "learning_rate": 9.321820217457478e-06,
+      "loss": 0.8547,
+      "step": 19515
+    },
+    {
+      "epoch": 0.5359918705885584,
+      "grad_norm": 0.0,
+      "learning_rate": 9.320932752052224e-06,
+      "loss": 0.8141,
+      "step": 19516
+    },
+    {
+      "epoch": 0.5360193348164016,
+      "grad_norm": 0.0,
+      "learning_rate": 9.320045292020016e-06,
+      "loss": 0.9097,
+      "step": 19517
+    },
+    {
+      "epoch": 0.5360467990442449,
+      "grad_norm": 0.0,
+      "learning_rate": 9.319157837367869e-06,
+      "loss": 0.9645,
+      "step": 19518
+    },
+    {
+      "epoch": 0.5360742632720881,
+      "grad_norm": 0.0,
+      "learning_rate": 9.318270388102812e-06,
+      "loss": 0.8539,
+      "step": 19519
+    },
+    {
+      "epoch": 0.5361017274999313,
+      "grad_norm": 0.0,
+      "learning_rate": 9.31738294423186e-06,
+      "loss": 0.855,
+      "step": 19520
+    },
+    {
+      "epoch": 0.5361291917277746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.31649550576204e-06,
+      "loss": 0.8507,
+      "step": 19521
+    },
+    {
+      "epoch": 0.5361566559556178,
+      "grad_norm": 0.0,
+      "learning_rate": 9.315608072700373e-06,
+      "loss": 0.9443,
+      "step": 19522
+    },
+    {
+      "epoch": 0.5361841201834611,
+      "grad_norm": 0.0,
+      "learning_rate": 9.314720645053882e-06,
+      "loss": 0.9478,
+      "step": 19523
+    },
+    {
+      "epoch": 0.5362115844113042,
+      "grad_norm": 0.0,
+      "learning_rate": 9.313833222829587e-06,
+      "loss": 0.8595,
+      "step": 19524
+    },
+    {
+      "epoch": 0.5362390486391475,
+      "grad_norm": 0.0,
+      "learning_rate": 9.312945806034506e-06,
+      "loss": 0.9605,
+      "step": 19525
+    },
+    {
+      "epoch": 0.5362665128669908,
+      "grad_norm": 0.0,
+      "learning_rate": 9.312058394675664e-06,
+      "loss": 0.892,
+      "step": 19526
+    },
+    {
+      "epoch": 0.536293977094834,
+      "grad_norm": 0.0,
+      "learning_rate": 9.311170988760085e-06,
+      "loss": 0.9255,
+      "step": 19527
+    },
+    {
+      "epoch": 0.5363214413226772,
+      "grad_norm": 0.0,
+      "learning_rate": 9.310283588294785e-06,
+      "loss": 0.8957,
+      "step": 19528
+    },
+    {
+      "epoch": 0.5363489055505204,
+      "grad_norm": 0.0,
+      "learning_rate": 9.309396193286793e-06,
+      "loss": 0.7945,
+      "step": 19529
+    },
+    {
+      "epoch": 0.5363763697783637,
+      "grad_norm": 0.0,
+      "learning_rate": 9.308508803743122e-06,
+      "loss": 0.7222,
+      "step": 19530
+    },
+    {
+      "epoch": 0.536403834006207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.307621419670798e-06,
+      "loss": 0.8423,
+      "step": 19531
+    },
+    {
+      "epoch": 0.5364312982340501,
+      "grad_norm": 0.0,
+      "learning_rate": 9.30673404107684e-06,
+      "loss": 0.8936,
+      "step": 19532
+    },
+    {
+      "epoch": 0.5364587624618934,
+      "grad_norm": 0.0,
+      "learning_rate": 9.305846667968272e-06,
+      "loss": 0.8581,
+      "step": 19533
+    },
+    {
+      "epoch": 0.5364862266897367,
+      "grad_norm": 0.0,
+      "learning_rate": 9.304959300352116e-06,
+      "loss": 0.8878,
+      "step": 19534
+    },
+    {
+      "epoch": 0.5365136909175798,
+      "grad_norm": 0.0,
+      "learning_rate": 9.30407193823539e-06,
+      "loss": 0.9095,
+      "step": 19535
+    },
+    {
+      "epoch": 0.5365411551454231,
+      "grad_norm": 0.0,
+      "learning_rate": 9.303184581625118e-06,
+      "loss": 0.8764,
+      "step": 19536
+    },
+    {
+      "epoch": 0.5365686193732663,
+      "grad_norm": 0.0,
+      "learning_rate": 9.302297230528315e-06,
+      "loss": 0.936,
+      "step": 19537
+    },
+    {
+      "epoch": 0.5365960836011096,
+      "grad_norm": 0.0,
+      "learning_rate": 9.301409884952009e-06,
+      "loss": 0.8393,
+      "step": 19538
+    },
+    {
+      "epoch": 0.5366235478289528,
+      "grad_norm": 0.0,
+      "learning_rate": 9.300522544903219e-06,
+      "loss": 1.0137,
+      "step": 19539
+    },
+    {
+      "epoch": 0.536651012056796,
+      "grad_norm": 0.0,
+      "learning_rate": 9.299635210388968e-06,
+      "loss": 0.8714,
+      "step": 19540
+    },
+    {
+      "epoch": 0.5366784762846393,
+      "grad_norm": 0.0,
+      "learning_rate": 9.298747881416271e-06,
+      "loss": 0.9505,
+      "step": 19541
+    },
+    {
+      "epoch": 0.5367059405124824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.297860557992153e-06,
+      "loss": 0.8823,
+      "step": 19542
+    },
+    {
+      "epoch": 0.5367334047403257,
+      "grad_norm": 0.0,
+      "learning_rate": 9.296973240123634e-06,
+      "loss": 0.917,
+      "step": 19543
+    },
+    {
+      "epoch": 0.536760868968169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.296085927817735e-06,
+      "loss": 0.7527,
+      "step": 19544
+    },
+    {
+      "epoch": 0.5367883331960122,
+      "grad_norm": 0.0,
+      "learning_rate": 9.295198621081476e-06,
+      "loss": 0.9525,
+      "step": 19545
+    },
+    {
+      "epoch": 0.5368157974238554,
+      "grad_norm": 0.0,
+      "learning_rate": 9.294311319921883e-06,
+      "loss": 0.8016,
+      "step": 19546
+    },
+    {
+      "epoch": 0.5368432616516987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.293424024345969e-06,
+      "loss": 0.7845,
+      "step": 19547
+    },
+    {
+      "epoch": 0.5368707258795419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.292536734360758e-06,
+      "loss": 0.8617,
+      "step": 19548
+    },
+    {
+      "epoch": 0.5368981901073852,
+      "grad_norm": 0.0,
+      "learning_rate": 9.291649449973268e-06,
+      "loss": 0.8739,
+      "step": 19549
+    },
+    {
+      "epoch": 0.5369256543352283,
+      "grad_norm": 0.0,
+      "learning_rate": 9.290762171190524e-06,
+      "loss": 0.9657,
+      "step": 19550
+    },
+    {
+      "epoch": 0.5369531185630716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.289874898019546e-06,
+      "loss": 0.8427,
+      "step": 19551
+    },
+    {
+      "epoch": 0.5369805827909149,
+      "grad_norm": 0.0,
+      "learning_rate": 9.288987630467354e-06,
+      "loss": 0.8965,
+      "step": 19552
+    },
+    {
+      "epoch": 0.537008047018758,
+      "grad_norm": 0.0,
+      "learning_rate": 9.288100368540964e-06,
+      "loss": 0.8898,
+      "step": 19553
+    },
+    {
+      "epoch": 0.5370355112466013,
+      "grad_norm": 0.0,
+      "learning_rate": 9.2872131122474e-06,
+      "loss": 1.0143,
+      "step": 19554
+    },
+    {
+      "epoch": 0.5370629754744445,
+      "grad_norm": 0.0,
+      "learning_rate": 9.286325861593684e-06,
+      "loss": 0.8689,
+      "step": 19555
+    },
+    {
+      "epoch": 0.5370904397022878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.28543861658683e-06,
+      "loss": 0.8408,
+      "step": 19556
+    },
+    {
+      "epoch": 0.537117903930131,
+      "grad_norm": 0.0,
+      "learning_rate": 9.284551377233865e-06,
+      "loss": 0.9527,
+      "step": 19557
+    },
+    {
+      "epoch": 0.5371453681579742,
+      "grad_norm": 0.0,
+      "learning_rate": 9.28366414354181e-06,
+      "loss": 0.9532,
+      "step": 19558
+    },
+    {
+      "epoch": 0.5371728323858175,
+      "grad_norm": 0.0,
+      "learning_rate": 9.282776915517677e-06,
+      "loss": 0.9713,
+      "step": 19559
+    },
+    {
+      "epoch": 0.5372002966136608,
+      "grad_norm": 0.0,
+      "learning_rate": 9.281889693168492e-06,
+      "loss": 0.8797,
+      "step": 19560
+    },
+    {
+      "epoch": 0.5372277608415039,
+      "grad_norm": 0.0,
+      "learning_rate": 9.281002476501275e-06,
+      "loss": 0.9281,
+      "step": 19561
+    },
+    {
+      "epoch": 0.5372552250693472,
+      "grad_norm": 0.0,
+      "learning_rate": 9.280115265523042e-06,
+      "loss": 0.9028,
+      "step": 19562
+    },
+    {
+      "epoch": 0.5372826892971904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.27922806024082e-06,
+      "loss": 0.8357,
+      "step": 19563
+    },
+    {
+      "epoch": 0.5373101535250336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.278340860661626e-06,
+      "loss": 0.8247,
+      "step": 19564
+    },
+    {
+      "epoch": 0.5373376177528769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.277453666792475e-06,
+      "loss": 0.9158,
+      "step": 19565
+    },
+    {
+      "epoch": 0.5373650819807201,
+      "grad_norm": 0.0,
+      "learning_rate": 9.276566478640389e-06,
+      "loss": 0.7932,
+      "step": 19566
+    },
+    {
+      "epoch": 0.5373925462085634,
+      "grad_norm": 0.0,
+      "learning_rate": 9.275679296212391e-06,
+      "loss": 0.7856,
+      "step": 19567
+    },
+    {
+      "epoch": 0.5374200104364065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.2747921195155e-06,
+      "loss": 0.8758,
+      "step": 19568
+    },
+    {
+      "epoch": 0.5374474746642498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.273904948556733e-06,
+      "loss": 0.811,
+      "step": 19569
+    },
+    {
+      "epoch": 0.5374749388920931,
+      "grad_norm": 0.0,
+      "learning_rate": 9.273017783343114e-06,
+      "loss": 0.8552,
+      "step": 19570
+    },
+    {
+      "epoch": 0.5375024031199362,
+      "grad_norm": 0.0,
+      "learning_rate": 9.272130623881657e-06,
+      "loss": 0.7654,
+      "step": 19571
+    },
+    {
+      "epoch": 0.5375298673477795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.271243470179388e-06,
+      "loss": 0.8407,
+      "step": 19572
+    },
+    {
+      "epoch": 0.5375573315756228,
+      "grad_norm": 0.0,
+      "learning_rate": 9.270356322243319e-06,
+      "loss": 0.7962,
+      "step": 19573
+    },
+    {
+      "epoch": 0.537584795803466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.269469180080472e-06,
+      "loss": 0.8596,
+      "step": 19574
+    },
+    {
+      "epoch": 0.5376122600313092,
+      "grad_norm": 0.0,
+      "learning_rate": 9.268582043697873e-06,
+      "loss": 0.999,
+      "step": 19575
+    },
+    {
+      "epoch": 0.5376397242591524,
+      "grad_norm": 0.0,
+      "learning_rate": 9.267694913102537e-06,
+      "loss": 0.8779,
+      "step": 19576
+    },
+    {
+      "epoch": 0.5376671884869957,
+      "grad_norm": 0.0,
+      "learning_rate": 9.266807788301478e-06,
+      "loss": 0.9579,
+      "step": 19577
+    },
+    {
+      "epoch": 0.537694652714839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.26592066930172e-06,
+      "loss": 0.998,
+      "step": 19578
+    },
+    {
+      "epoch": 0.5377221169426821,
+      "grad_norm": 0.0,
+      "learning_rate": 9.265033556110285e-06,
+      "loss": 0.9092,
+      "step": 19579
+    },
+    {
+      "epoch": 0.5377495811705254,
+      "grad_norm": 0.0,
+      "learning_rate": 9.264146448734185e-06,
+      "loss": 0.7771,
+      "step": 19580
+    },
+    {
+      "epoch": 0.5377770453983686,
+      "grad_norm": 0.0,
+      "learning_rate": 9.263259347180446e-06,
+      "loss": 0.968,
+      "step": 19581
+    },
+    {
+      "epoch": 0.5378045096262118,
+      "grad_norm": 0.0,
+      "learning_rate": 9.262372251456087e-06,
+      "loss": 0.9546,
+      "step": 19582
+    },
+    {
+      "epoch": 0.5378319738540551,
+      "grad_norm": 0.0,
+      "learning_rate": 9.261485161568121e-06,
+      "loss": 0.9447,
+      "step": 19583
+    },
+    {
+      "epoch": 0.5378594380818983,
+      "grad_norm": 0.0,
+      "learning_rate": 9.260598077523573e-06,
+      "loss": 0.8918,
+      "step": 19584
+    },
+    {
+      "epoch": 0.5378869023097416,
+      "grad_norm": 0.0,
+      "learning_rate": 9.259710999329455e-06,
+      "loss": 0.8779,
+      "step": 19585
+    },
+    {
+      "epoch": 0.5379143665375848,
+      "grad_norm": 0.0,
+      "learning_rate": 9.258823926992793e-06,
+      "loss": 0.8561,
+      "step": 19586
+    },
+    {
+      "epoch": 0.537941830765428,
+      "grad_norm": 0.0,
+      "learning_rate": 9.257936860520604e-06,
+      "loss": 0.8292,
+      "step": 19587
+    },
+    {
+      "epoch": 0.5379692949932713,
+      "grad_norm": 0.0,
+      "learning_rate": 9.257049799919908e-06,
+      "loss": 0.9589,
+      "step": 19588
+    },
+    {
+      "epoch": 0.5379967592211145,
+      "grad_norm": 0.0,
+      "learning_rate": 9.256162745197717e-06,
+      "loss": 0.8587,
+      "step": 19589
+    },
+    {
+      "epoch": 0.5380242234489577,
+      "grad_norm": 0.0,
+      "learning_rate": 9.255275696361057e-06,
+      "loss": 0.8725,
+      "step": 19590
+    },
+    {
+      "epoch": 0.538051687676801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.254388653416943e-06,
+      "loss": 0.7927,
+      "step": 19591
+    },
+    {
+      "epoch": 0.5380791519046442,
+      "grad_norm": 0.0,
+      "learning_rate": 9.253501616372396e-06,
+      "loss": 0.8349,
+      "step": 19592
+    },
+    {
+      "epoch": 0.5381066161324874,
+      "grad_norm": 0.0,
+      "learning_rate": 9.25261458523443e-06,
+      "loss": 0.8446,
+      "step": 19593
+    },
+    {
+      "epoch": 0.5381340803603307,
+      "grad_norm": 0.0,
+      "learning_rate": 9.251727560010074e-06,
+      "loss": 0.8582,
+      "step": 19594
+    },
+    {
+      "epoch": 0.5381615445881739,
+      "grad_norm": 0.0,
+      "learning_rate": 9.250840540706334e-06,
+      "loss": 0.9193,
+      "step": 19595
+    },
+    {
+      "epoch": 0.5381890088160172,
+      "grad_norm": 0.0,
+      "learning_rate": 9.249953527330235e-06,
+      "loss": 0.9125,
+      "step": 19596
+    },
+    {
+      "epoch": 0.5382164730438603,
+      "grad_norm": 0.0,
+      "learning_rate": 9.249066519888794e-06,
+      "loss": 0.9213,
+      "step": 19597
+    },
+    {
+      "epoch": 0.5382439372717036,
+      "grad_norm": 0.0,
+      "learning_rate": 9.248179518389029e-06,
+      "loss": 0.9897,
+      "step": 19598
+    },
+    {
+      "epoch": 0.5382714014995469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.247292522837961e-06,
+      "loss": 0.8489,
+      "step": 19599
+    },
+    {
+      "epoch": 0.53829886572739,
+      "grad_norm": 0.0,
+      "learning_rate": 9.246405533242608e-06,
+      "loss": 0.8683,
+      "step": 19600
+    },
+    {
+      "epoch": 0.5383263299552333,
+      "grad_norm": 0.0,
+      "learning_rate": 9.245518549609982e-06,
+      "loss": 0.8526,
+      "step": 19601
+    },
+    {
+      "epoch": 0.5383537941830765,
+      "grad_norm": 0.0,
+      "learning_rate": 9.244631571947105e-06,
+      "loss": 0.9451,
+      "step": 19602
+    },
+    {
+      "epoch": 0.5383812584109198,
+      "grad_norm": 0.0,
+      "learning_rate": 9.243744600260997e-06,
+      "loss": 1.0036,
+      "step": 19603
+    },
+    {
+      "epoch": 0.538408722638763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.242857634558674e-06,
+      "loss": 0.8327,
+      "step": 19604
+    },
+    {
+      "epoch": 0.5384361868666062,
+      "grad_norm": 0.0,
+      "learning_rate": 9.24197067484716e-06,
+      "loss": 0.9425,
+      "step": 19605
+    },
+    {
+      "epoch": 0.5384636510944495,
+      "grad_norm": 0.0,
+      "learning_rate": 9.241083721133462e-06,
+      "loss": 0.8153,
+      "step": 19606
+    },
+    {
+      "epoch": 0.5384911153222928,
+      "grad_norm": 0.0,
+      "learning_rate": 9.240196773424602e-06,
+      "loss": 0.9357,
+      "step": 19607
+    },
+    {
+      "epoch": 0.5385185795501359,
+      "grad_norm": 0.0,
+      "learning_rate": 9.239309831727603e-06,
+      "loss": 0.8827,
+      "step": 19608
+    },
+    {
+      "epoch": 0.5385460437779792,
+      "grad_norm": 0.0,
+      "learning_rate": 9.238422896049478e-06,
+      "loss": 0.8679,
+      "step": 19609
+    },
+    {
+      "epoch": 0.5385735080058224,
+      "grad_norm": 0.0,
+      "learning_rate": 9.237535966397245e-06,
+      "loss": 0.8326,
+      "step": 19610
+    },
+    {
+      "epoch": 0.5386009722336657,
+      "grad_norm": 0.0,
+      "learning_rate": 9.236649042777928e-06,
+      "loss": 0.8804,
+      "step": 19611
+    },
+    {
+      "epoch": 0.5386284364615089,
+      "grad_norm": 0.0,
+      "learning_rate": 9.235762125198537e-06,
+      "loss": 0.8743,
+      "step": 19612
+    },
+    {
+      "epoch": 0.5386559006893521,
+      "grad_norm": 0.0,
+      "learning_rate": 9.23487521366609e-06,
+      "loss": 0.8494,
+      "step": 19613
+    },
+    {
+      "epoch": 0.5386833649171954,
+      "grad_norm": 0.0,
+      "learning_rate": 9.233988308187607e-06,
+      "loss": 0.8749,
+      "step": 19614
+    },
+    {
+      "epoch": 0.5387108291450385,
+      "grad_norm": 0.0,
+      "learning_rate": 9.233101408770107e-06,
+      "loss": 0.9503,
+      "step": 19615
+    },
+    {
+      "epoch": 0.5387382933728818,
+      "grad_norm": 0.0,
+      "learning_rate": 9.232214515420602e-06,
+      "loss": 0.9461,
+      "step": 19616
+    },
+    {
+      "epoch": 0.5387657576007251,
+      "grad_norm": 0.0,
+      "learning_rate": 9.23132762814612e-06,
+      "loss": 1.0282,
+      "step": 19617
+    },
+    {
+      "epoch": 0.5387932218285683,
+      "grad_norm": 0.0,
+      "learning_rate": 9.230440746953667e-06,
+      "loss": 0.8027,
+      "step": 19618
+    },
+    {
+      "epoch": 0.5388206860564115,
+      "grad_norm": 0.0,
+      "learning_rate": 9.229553871850262e-06,
+      "loss": 0.9807,
+      "step": 19619
+    },
+    {
+      "epoch": 0.5388481502842548,
+      "grad_norm": 0.0,
+      "learning_rate": 9.22866700284293e-06,
+      "loss": 0.9383,
+      "step": 19620
+    },
+    {
+      "epoch": 0.538875614512098,
+      "grad_norm": 0.0,
+      "learning_rate": 9.227780139938681e-06,
+      "loss": 0.9144,
+      "step": 19621
+    },
+    {
+      "epoch": 0.5389030787399413,
+      "grad_norm": 0.0,
+      "learning_rate": 9.226893283144534e-06,
+      "loss": 0.9451,
+      "step": 19622
+    },
+    {
+      "epoch": 0.5389305429677844,
+      "grad_norm": 0.0,
+      "learning_rate": 9.226006432467512e-06,
+      "loss": 0.8862,
+      "step": 19623
+    },
+    {
+      "epoch": 0.5389580071956277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.225119587914624e-06,
+      "loss": 0.8887,
+      "step": 19624
+    },
+    {
+      "epoch": 0.538985471423471,
+      "grad_norm": 0.0,
+      "learning_rate": 9.224232749492888e-06,
+      "loss": 0.847,
+      "step": 19625
+    },
+    {
+      "epoch": 0.5390129356513141,
+      "grad_norm": 0.0,
+      "learning_rate": 9.223345917209324e-06,
+      "loss": 0.8196,
+      "step": 19626
+    },
+    {
+      "epoch": 0.5390403998791574,
+      "grad_norm": 0.0,
+      "learning_rate": 9.22245909107095e-06,
+      "loss": 0.8631,
+      "step": 19627
+    },
+    {
+      "epoch": 0.5390678641070006,
+      "grad_norm": 0.0,
+      "learning_rate": 9.221572271084777e-06,
+      "loss": 0.9622,
+      "step": 19628
+    },
+    {
+      "epoch": 0.5390953283348439,
+      "grad_norm": 0.0,
+      "learning_rate": 9.22068545725783e-06,
+      "loss": 0.8947,
+      "step": 19629
+    },
+    {
+      "epoch": 0.5391227925626871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.219798649597118e-06,
+      "loss": 0.976,
+      "step": 19630
+    },
+    {
+      "epoch": 0.5391502567905303,
+      "grad_norm": 0.0,
+      "learning_rate": 9.218911848109663e-06,
+      "loss": 0.9928,
+      "step": 19631
+    },
+    {
+      "epoch": 0.5391777210183736,
+      "grad_norm": 0.0,
+      "learning_rate": 9.21802505280248e-06,
+      "loss": 0.868,
+      "step": 19632
+    },
+    {
+      "epoch": 0.5392051852462169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.217138263682584e-06,
+      "loss": 1.0662,
+      "step": 19633
+    },
+    {
+      "epoch": 0.53923264947406,
+      "grad_norm": 0.0,
+      "learning_rate": 9.216251480756993e-06,
+      "loss": 0.9359,
+      "step": 19634
+    },
+    {
+      "epoch": 0.5392601137019033,
+      "grad_norm": 0.0,
+      "learning_rate": 9.215364704032728e-06,
+      "loss": 0.7998,
+      "step": 19635
+    },
+    {
+      "epoch": 0.5392875779297465,
+      "grad_norm": 0.0,
+      "learning_rate": 9.214477933516802e-06,
+      "loss": 0.8569,
+      "step": 19636
+    },
+    {
+      "epoch": 0.5393150421575897,
+      "grad_norm": 0.0,
+      "learning_rate": 9.213591169216225e-06,
+      "loss": 0.8969,
+      "step": 19637
+    },
+    {
+      "epoch": 0.539342506385433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.212704411138022e-06,
+      "loss": 0.8836,
+      "step": 19638
+    },
+    {
+      "epoch": 0.5393699706132762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.211817659289208e-06,
+      "loss": 0.7607,
+      "step": 19639
+    },
+    {
+      "epoch": 0.5393974348411195,
+      "grad_norm": 0.0,
+      "learning_rate": 9.210930913676794e-06,
+      "loss": 0.8903,
+      "step": 19640
+    },
+    {
+      "epoch": 0.5394248990689626,
+      "grad_norm": 0.0,
+      "learning_rate": 9.210044174307807e-06,
+      "loss": 0.8084,
+      "step": 19641
+    },
+    {
+      "epoch": 0.5394523632968059,
+      "grad_norm": 0.0,
+      "learning_rate": 9.20915744118925e-06,
+      "loss": 0.9034,
+      "step": 19642
+    },
+    {
+      "epoch": 0.5394798275246492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.208270714328147e-06,
+      "loss": 0.8887,
+      "step": 19643
+    },
+    {
+      "epoch": 0.5395072917524923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.207383993731516e-06,
+      "loss": 0.8492,
+      "step": 19644
+    },
+    {
+      "epoch": 0.5395347559803356,
+      "grad_norm": 0.0,
+      "learning_rate": 9.206497279406366e-06,
+      "loss": 0.8337,
+      "step": 19645
+    },
+    {
+      "epoch": 0.5395622202081789,
+      "grad_norm": 0.0,
+      "learning_rate": 9.20561057135972e-06,
+      "loss": 0.904,
+      "step": 19646
+    },
+    {
+      "epoch": 0.5395896844360221,
+      "grad_norm": 0.0,
+      "learning_rate": 9.20472386959859e-06,
+      "loss": 0.8978,
+      "step": 19647
+    },
+    {
+      "epoch": 0.5396171486638653,
+      "grad_norm": 0.0,
+      "learning_rate": 9.203837174129995e-06,
+      "loss": 0.8354,
+      "step": 19648
+    },
+    {
+      "epoch": 0.5396446128917085,
+      "grad_norm": 0.0,
+      "learning_rate": 9.202950484960945e-06,
+      "loss": 0.8011,
+      "step": 19649
+    },
+    {
+      "epoch": 0.5396720771195518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.20206380209846e-06,
+      "loss": 0.9149,
+      "step": 19650
+    },
+    {
+      "epoch": 0.5396995413473951,
+      "grad_norm": 0.0,
+      "learning_rate": 9.201177125549556e-06,
+      "loss": 0.8467,
+      "step": 19651
+    },
+    {
+      "epoch": 0.5397270055752382,
+      "grad_norm": 0.0,
+      "learning_rate": 9.200290455321246e-06,
+      "loss": 0.868,
+      "step": 19652
+    },
+    {
+      "epoch": 0.5397544698030815,
+      "grad_norm": 0.0,
+      "learning_rate": 9.199403791420554e-06,
+      "loss": 0.8807,
+      "step": 19653
+    },
+    {
+      "epoch": 0.5397819340309247,
+      "grad_norm": 0.0,
+      "learning_rate": 9.198517133854484e-06,
+      "loss": 0.921,
+      "step": 19654
+    },
+    {
+      "epoch": 0.539809398258768,
+      "grad_norm": 0.0,
+      "learning_rate": 9.197630482630058e-06,
+      "loss": 0.8685,
+      "step": 19655
+    },
+    {
+      "epoch": 0.5398368624866112,
+      "grad_norm": 0.0,
+      "learning_rate": 9.19674383775429e-06,
+      "loss": 0.8406,
+      "step": 19656
+    },
+    {
+      "epoch": 0.5398643267144544,
+      "grad_norm": 0.0,
+      "learning_rate": 9.195857199234194e-06,
+      "loss": 0.9005,
+      "step": 19657
+    },
+    {
+      "epoch": 0.5398917909422977,
+      "grad_norm": 0.0,
+      "learning_rate": 9.194970567076793e-06,
+      "loss": 0.9074,
+      "step": 19658
+    },
+    {
+      "epoch": 0.5399192551701409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.194083941289092e-06,
+      "loss": 0.8233,
+      "step": 19659
+    },
+    {
+      "epoch": 0.5399467193979841,
+      "grad_norm": 0.0,
+      "learning_rate": 9.193197321878114e-06,
+      "loss": 0.8462,
+      "step": 19660
+    },
+    {
+      "epoch": 0.5399741836258274,
+      "grad_norm": 0.0,
+      "learning_rate": 9.192310708850868e-06,
+      "loss": 0.9326,
+      "step": 19661
+    },
+    {
+      "epoch": 0.5400016478536706,
+      "grad_norm": 0.0,
+      "learning_rate": 9.191424102214373e-06,
+      "loss": 0.7485,
+      "step": 19662
+    },
+    {
+      "epoch": 0.5400291120815138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.190537501975645e-06,
+      "loss": 0.8149,
+      "step": 19663
+    },
+    {
+      "epoch": 0.5400565763093571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.189650908141702e-06,
+      "loss": 0.906,
+      "step": 19664
+    },
+    {
+      "epoch": 0.5400840405372003,
+      "grad_norm": 0.0,
+      "learning_rate": 9.188764320719548e-06,
+      "loss": 0.879,
+      "step": 19665
+    },
+    {
+      "epoch": 0.5401115047650435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.187877739716208e-06,
+      "loss": 0.8812,
+      "step": 19666
+    },
+    {
+      "epoch": 0.5401389689928867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.186991165138694e-06,
+      "loss": 1.0289,
+      "step": 19667
+    },
+    {
+      "epoch": 0.54016643322073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.18610459699402e-06,
+      "loss": 0.8568,
+      "step": 19668
+    },
+    {
+      "epoch": 0.5401938974485733,
+      "grad_norm": 0.0,
+      "learning_rate": 9.185218035289199e-06,
+      "loss": 0.8833,
+      "step": 19669
+    },
+    {
+      "epoch": 0.5402213616764164,
+      "grad_norm": 0.0,
+      "learning_rate": 9.184331480031255e-06,
+      "loss": 0.8507,
+      "step": 19670
+    },
+    {
+      "epoch": 0.5402488259042597,
+      "grad_norm": 0.0,
+      "learning_rate": 9.183444931227192e-06,
+      "loss": 0.9383,
+      "step": 19671
+    },
+    {
+      "epoch": 0.540276290132103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.182558388884032e-06,
+      "loss": 0.8513,
+      "step": 19672
+    },
+    {
+      "epoch": 0.5403037543599462,
+      "grad_norm": 0.0,
+      "learning_rate": 9.181671853008783e-06,
+      "loss": 0.7676,
+      "step": 19673
+    },
+    {
+      "epoch": 0.5403312185877894,
+      "grad_norm": 0.0,
+      "learning_rate": 9.180785323608462e-06,
+      "loss": 0.9757,
+      "step": 19674
+    },
+    {
+      "epoch": 0.5403586828156326,
+      "grad_norm": 0.0,
+      "learning_rate": 9.179898800690089e-06,
+      "loss": 0.8229,
+      "step": 19675
+    },
+    {
+      "epoch": 0.5403861470434759,
+      "grad_norm": 0.0,
+      "learning_rate": 9.179012284260677e-06,
+      "loss": 0.8595,
+      "step": 19676
+    },
+    {
+      "epoch": 0.5404136112713191,
+      "grad_norm": 0.0,
+      "learning_rate": 9.178125774327232e-06,
+      "loss": 0.9306,
+      "step": 19677
+    },
+    {
+      "epoch": 0.5404410754991623,
+      "grad_norm": 0.0,
+      "learning_rate": 9.177239270896775e-06,
+      "loss": 0.9339,
+      "step": 19678
+    },
+    {
+      "epoch": 0.5404685397270056,
+      "grad_norm": 0.0,
+      "learning_rate": 9.176352773976321e-06,
+      "loss": 0.8683,
+      "step": 19679
+    },
+    {
+      "epoch": 0.5404960039548488,
+      "grad_norm": 0.0,
+      "learning_rate": 9.175466283572883e-06,
+      "loss": 0.9561,
+      "step": 19680
+    },
+    {
+      "epoch": 0.540523468182692,
+      "grad_norm": 0.0,
+      "learning_rate": 9.174579799693473e-06,
+      "loss": 0.8279,
+      "step": 19681
+    },
+    {
+      "epoch": 0.5405509324105353,
+      "grad_norm": 0.0,
+      "learning_rate": 9.173693322345113e-06,
+      "loss": 1.0242,
+      "step": 19682
+    },
+    {
+      "epoch": 0.5405783966383785,
+      "grad_norm": 0.0,
+      "learning_rate": 9.172806851534806e-06,
+      "loss": 0.8895,
+      "step": 19683
+    },
+    {
+      "epoch": 0.5406058608662218,
+      "grad_norm": 0.0,
+      "learning_rate": 9.171920387269573e-06,
+      "loss": 0.9035,
+      "step": 19684
+    },
+    {
+      "epoch": 0.540633325094065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.171033929556426e-06,
+      "loss": 0.8749,
+      "step": 19685
+    },
+    {
+      "epoch": 0.5406607893219082,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17014747840238e-06,
+      "loss": 1.0063,
+      "step": 19686
+    },
+    {
+      "epoch": 0.5406882535497515,
+      "grad_norm": 0.0,
+      "learning_rate": 9.16926103381445e-06,
+      "loss": 0.9633,
+      "step": 19687
+    },
+    {
+      "epoch": 0.5407157177775946,
+      "grad_norm": 0.0,
+      "learning_rate": 9.16837459579965e-06,
+      "loss": 1.0453,
+      "step": 19688
+    },
+    {
+      "epoch": 0.5407431820054379,
+      "grad_norm": 0.0,
+      "learning_rate": 9.167488164364989e-06,
+      "loss": 0.9063,
+      "step": 19689
+    },
+    {
+      "epoch": 0.5407706462332812,
+      "grad_norm": 0.0,
+      "learning_rate": 9.166601739517484e-06,
+      "loss": 0.915,
+      "step": 19690
+    },
+    {
+      "epoch": 0.5407981104611244,
+      "grad_norm": 0.0,
+      "learning_rate": 9.16571532126415e-06,
+      "loss": 0.8948,
+      "step": 19691
+    },
+    {
+      "epoch": 0.5408255746889676,
+      "grad_norm": 0.0,
+      "learning_rate": 9.164828909612e-06,
+      "loss": 0.8421,
+      "step": 19692
+    },
+    {
+      "epoch": 0.5408530389168108,
+      "grad_norm": 0.0,
+      "learning_rate": 9.163942504568045e-06,
+      "loss": 0.8658,
+      "step": 19693
+    },
+    {
+      "epoch": 0.5408805031446541,
+      "grad_norm": 0.0,
+      "learning_rate": 9.163056106139305e-06,
+      "loss": 0.9528,
+      "step": 19694
+    },
+    {
+      "epoch": 0.5409079673724974,
+      "grad_norm": 0.0,
+      "learning_rate": 9.162169714332787e-06,
+      "loss": 0.9271,
+      "step": 19695
+    },
+    {
+      "epoch": 0.5409354316003405,
+      "grad_norm": 0.0,
+      "learning_rate": 9.161283329155507e-06,
+      "loss": 0.92,
+      "step": 19696
+    },
+    {
+      "epoch": 0.5409628958281838,
+      "grad_norm": 0.0,
+      "learning_rate": 9.160396950614476e-06,
+      "loss": 0.8602,
+      "step": 19697
+    },
+    {
+      "epoch": 0.5409903600560271,
+      "grad_norm": 0.0,
+      "learning_rate": 9.159510578716711e-06,
+      "loss": 0.9463,
+      "step": 19698
+    },
+    {
+      "epoch": 0.5410178242838702,
+      "grad_norm": 0.0,
+      "learning_rate": 9.158624213469224e-06,
+      "loss": 0.8379,
+      "step": 19699
+    },
+    {
+      "epoch": 0.5410452885117135,
+      "grad_norm": 0.0,
+      "learning_rate": 9.157737854879032e-06,
+      "loss": 0.7983,
+      "step": 19700
+    },
+    {
+      "epoch": 0.5410727527395567,
+      "grad_norm": 0.0,
+      "learning_rate": 9.156851502953139e-06,
+      "loss": 0.7879,
+      "step": 19701
+    },
+    {
+      "epoch": 0.5411002169674,
+      "grad_norm": 0.0,
+      "learning_rate": 9.155965157698564e-06,
+      "loss": 0.9285,
+      "step": 19702
+    },
+    {
+      "epoch": 0.5411276811952432,
+      "grad_norm": 0.0,
+      "learning_rate": 9.155078819122322e-06,
+      "loss": 0.9336,
+      "step": 19703
+    },
+    {
+      "epoch": 0.5411551454230864,
+      "grad_norm": 0.0,
+      "learning_rate": 9.154192487231421e-06,
+      "loss": 0.8546,
+      "step": 19704
+    },
+    {
+      "epoch": 0.5411826096509297,
+      "grad_norm": 0.0,
+      "learning_rate": 9.153306162032877e-06,
+      "loss": 1.0142,
+      "step": 19705
+    },
+    {
+      "epoch": 0.5412100738787728,
+      "grad_norm": 0.0,
+      "learning_rate": 9.152419843533706e-06,
+      "loss": 0.9056,
+      "step": 19706
+    },
+    {
+      "epoch": 0.5412375381066161,
+      "grad_norm": 0.0,
+      "learning_rate": 9.151533531740915e-06,
+      "loss": 0.9506,
+      "step": 19707
+    },
+    {
+      "epoch": 0.5412650023344594,
+      "grad_norm": 0.0,
+      "learning_rate": 9.150647226661521e-06,
+      "loss": 0.7968,
+      "step": 19708
+    },
+    {
+      "epoch": 0.5412924665623026,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149760928302532e-06,
+      "loss": 0.8925,
+      "step": 19709
+    },
+    {
+      "epoch": 0.5413199307901458,
+      "grad_norm": 0.0,
+      "learning_rate": 9.148874636670965e-06,
+      "loss": 0.9288,
+      "step": 19710
+    },
+    {
+      "epoch": 0.5413473950179891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.147988351773833e-06,
+      "loss": 1.0159,
+      "step": 19711
+    },
+    {
+      "epoch": 0.5413748592458323,
+      "grad_norm": 0.0,
+      "learning_rate": 9.147102073618149e-06,
+      "loss": 0.8767,
+      "step": 19712
+    },
+    {
+      "epoch": 0.5414023234736756,
+      "grad_norm": 0.0,
+      "learning_rate": 9.14621580221092e-06,
+      "loss": 0.8354,
+      "step": 19713
+    },
+    {
+      "epoch": 0.5414297877015187,
+      "grad_norm": 0.0,
+      "learning_rate": 9.145329537559161e-06,
+      "loss": 0.8699,
+      "step": 19714
+    },
+    {
+      "epoch": 0.541457251929362,
+      "grad_norm": 0.0,
+      "learning_rate": 9.14444327966989e-06,
+      "loss": 0.8865,
+      "step": 19715
+    },
+    {
+      "epoch": 0.5414847161572053,
+      "grad_norm": 0.0,
+      "learning_rate": 9.143557028550114e-06,
+      "loss": 0.8555,
+      "step": 19716
+    },
+    {
+      "epoch": 0.5415121803850484,
+      "grad_norm": 0.0,
+      "learning_rate": 9.14267078420685e-06,
+      "loss": 0.8673,
+      "step": 19717
+    },
+    {
+      "epoch": 0.5415396446128917,
+      "grad_norm": 0.0,
+      "learning_rate": 9.1417845466471e-06,
+      "loss": 0.8069,
+      "step": 19718
+    },
+    {
+      "epoch": 0.5415671088407349,
+      "grad_norm": 0.0,
+      "learning_rate": 9.140898315877885e-06,
+      "loss": 0.8342,
+      "step": 19719
+    },
+    {
+      "epoch": 0.5415945730685782,
+      "grad_norm": 0.0,
+      "learning_rate": 9.140012091906217e-06,
+      "loss": 0.8184,
+      "step": 19720
+    },
+    {
+      "epoch": 0.5416220372964214,
+      "grad_norm": 0.0,
+      "learning_rate": 9.139125874739104e-06,
+      "loss": 0.9134,
+      "step": 19721
+    },
+    {
+      "epoch": 0.5416495015242646,
+      "grad_norm": 0.0,
+      "learning_rate": 9.138239664383562e-06,
+      "loss": 0.8252,
+      "step": 19722
+    },
+    {
+      "epoch": 0.5416769657521079,
+      "grad_norm": 0.0,
+      "learning_rate": 9.137353460846604e-06,
+      "loss": 0.7562,
+      "step": 19723
+    },
+    {
+      "epoch": 0.5417044299799512,
+      "grad_norm": 0.0,
+      "learning_rate": 9.136467264135238e-06,
+      "loss": 0.9486,
+      "step": 19724
+    },
+    {
+      "epoch": 0.5417318942077943,
+      "grad_norm": 0.0,
+      "learning_rate": 9.135581074256477e-06,
+      "loss": 0.9223,
+      "step": 19725
+    },
+    {
+      "epoch": 0.5417593584356376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.134694891217331e-06,
+      "loss": 0.7778,
+      "step": 19726
+    },
+    {
+      "epoch": 0.5417868226634808,
+      "grad_norm": 0.0,
+      "learning_rate": 9.133808715024818e-06,
+      "loss": 0.9157,
+      "step": 19727
+    },
+    {
+      "epoch": 0.541814286891324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.132922545685944e-06,
+      "loss": 1.0031,
+      "step": 19728
+    },
+    {
+      "epoch": 0.5418417511191673,
+      "grad_norm": 0.0,
+      "learning_rate": 9.132036383207726e-06,
+      "loss": 0.8921,
+      "step": 19729
+    },
+    {
+      "epoch": 0.5418692153470105,
+      "grad_norm": 0.0,
+      "learning_rate": 9.131150227597168e-06,
+      "loss": 0.8994,
+      "step": 19730
+    },
+    {
+      "epoch": 0.5418966795748538,
+      "grad_norm": 0.0,
+      "learning_rate": 9.130264078861286e-06,
+      "loss": 0.8727,
+      "step": 19731
+    },
+    {
+      "epoch": 0.5419241438026969,
+      "grad_norm": 0.0,
+      "learning_rate": 9.129377937007094e-06,
+      "loss": 0.9404,
+      "step": 19732
+    },
+    {
+      "epoch": 0.5419516080305402,
+      "grad_norm": 0.0,
+      "learning_rate": 9.128491802041599e-06,
+      "loss": 0.9254,
+      "step": 19733
+    },
+    {
+      "epoch": 0.5419790722583835,
+      "grad_norm": 0.0,
+      "learning_rate": 9.127605673971815e-06,
+      "loss": 0.8435,
+      "step": 19734
+    },
+    {
+      "epoch": 0.5420065364862267,
+      "grad_norm": 0.0,
+      "learning_rate": 9.126719552804758e-06,
+      "loss": 0.8534,
+      "step": 19735
+    },
+    {
+      "epoch": 0.5420340007140699,
+      "grad_norm": 0.0,
+      "learning_rate": 9.12583343854743e-06,
+      "loss": 0.8131,
+      "step": 19736
+    },
+    {
+      "epoch": 0.5420614649419132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.124947331206843e-06,
+      "loss": 0.985,
+      "step": 19737
+    },
+    {
+      "epoch": 0.5420889291697564,
+      "grad_norm": 0.0,
+      "learning_rate": 9.124061230790015e-06,
+      "loss": 0.8954,
+      "step": 19738
+    },
+    {
+      "epoch": 0.5421163933975996,
+      "grad_norm": 0.0,
+      "learning_rate": 9.123175137303954e-06,
+      "loss": 0.8872,
+      "step": 19739
+    },
+    {
+      "epoch": 0.5421438576254428,
+      "grad_norm": 0.0,
+      "learning_rate": 9.122289050755668e-06,
+      "loss": 0.8964,
+      "step": 19740
+    },
+    {
+      "epoch": 0.5421713218532861,
+      "grad_norm": 0.0,
+      "learning_rate": 9.121402971152177e-06,
+      "loss": 0.8527,
+      "step": 19741
+    },
+    {
+      "epoch": 0.5421987860811294,
+      "grad_norm": 0.0,
+      "learning_rate": 9.120516898500482e-06,
+      "loss": 0.925,
+      "step": 19742
+    },
+    {
+      "epoch": 0.5422262503089725,
+      "grad_norm": 0.0,
+      "learning_rate": 9.119630832807597e-06,
+      "loss": 0.9908,
+      "step": 19743
+    },
+    {
+      "epoch": 0.5422537145368158,
+      "grad_norm": 0.0,
+      "learning_rate": 9.118744774080537e-06,
+      "loss": 0.9398,
+      "step": 19744
+    },
+    {
+      "epoch": 0.542281178764659,
+      "grad_norm": 0.0,
+      "learning_rate": 9.117858722326305e-06,
+      "loss": 0.8624,
+      "step": 19745
+    },
+    {
+      "epoch": 0.5423086429925023,
+      "grad_norm": 0.0,
+      "learning_rate": 9.11697267755192e-06,
+      "loss": 0.8744,
+      "step": 19746
+    },
+    {
+      "epoch": 0.5423361072203455,
+      "grad_norm": 0.0,
+      "learning_rate": 9.116086639764392e-06,
+      "loss": 0.927,
+      "step": 19747
+    },
+    {
+      "epoch": 0.5423635714481887,
+      "grad_norm": 0.0,
+      "learning_rate": 9.115200608970726e-06,
+      "loss": 0.9576,
+      "step": 19748
+    },
+    {
+      "epoch": 0.542391035676032,
+      "grad_norm": 0.0,
+      "learning_rate": 9.114314585177935e-06,
+      "loss": 0.9223,
+      "step": 19749
+    },
+    {
+      "epoch": 0.5424184999038753,
+      "grad_norm": 0.0,
+      "learning_rate": 9.113428568393028e-06,
+      "loss": 0.925,
+      "step": 19750
+    },
+    {
+      "epoch": 0.5424459641317184,
+      "grad_norm": 0.0,
+      "learning_rate": 9.11254255862302e-06,
+      "loss": 0.895,
+      "step": 19751
+    },
+    {
+      "epoch": 0.5424734283595617,
+      "grad_norm": 0.0,
+      "learning_rate": 9.111656555874919e-06,
+      "loss": 0.8443,
+      "step": 19752
+    },
+    {
+      "epoch": 0.5425008925874049,
+      "grad_norm": 0.0,
+      "learning_rate": 9.110770560155738e-06,
+      "loss": 0.8364,
+      "step": 19753
+    },
+    {
+      "epoch": 0.5425283568152481,
+      "grad_norm": 0.0,
+      "learning_rate": 9.10988457147248e-06,
+      "loss": 0.8452,
+      "step": 19754
+    },
+    {
+      "epoch": 0.5425558210430914,
+      "grad_norm": 0.0,
+      "learning_rate": 9.108998589832162e-06,
+      "loss": 0.7245,
+      "step": 19755
+    },
+    {
+      "epoch": 0.5425832852709346,
+      "grad_norm": 0.0,
+      "learning_rate": 9.108112615241795e-06,
+      "loss": 0.9455,
+      "step": 19756
+    },
+    {
+      "epoch": 0.5426107494987779,
+      "grad_norm": 0.0,
+      "learning_rate": 9.107226647708381e-06,
+      "loss": 0.8375,
+      "step": 19757
+    },
+    {
+      "epoch": 0.542638213726621,
+      "grad_norm": 0.0,
+      "learning_rate": 9.106340687238939e-06,
+      "loss": 0.9059,
+      "step": 19758
+    },
+    {
+      "epoch": 0.5426656779544643,
+      "grad_norm": 0.0,
+      "learning_rate": 9.105454733840477e-06,
+      "loss": 0.7859,
+      "step": 19759
+    },
+    {
+      "epoch": 0.5426931421823076,
+      "grad_norm": 0.0,
+      "learning_rate": 9.104568787520002e-06,
+      "loss": 0.8489,
+      "step": 19760
+    },
+    {
+      "epoch": 0.5427206064101507,
+      "grad_norm": 0.0,
+      "learning_rate": 9.103682848284525e-06,
+      "loss": 0.8306,
+      "step": 19761
+    },
+    {
+      "epoch": 0.542748070637994,
+      "grad_norm": 0.0,
+      "learning_rate": 9.102796916141057e-06,
+      "loss": 0.9128,
+      "step": 19762
+    },
+    {
+      "epoch": 0.5427755348658373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.101910991096608e-06,
+      "loss": 0.8679,
+      "step": 19763
+    },
+    {
+      "epoch": 0.5428029990936805,
+      "grad_norm": 0.0,
+      "learning_rate": 9.101025073158185e-06,
+      "loss": 0.8622,
+      "step": 19764
+    },
+    {
+      "epoch": 0.5428304633215237,
+      "grad_norm": 0.0,
+      "learning_rate": 9.100139162332805e-06,
+      "loss": 1.0379,
+      "step": 19765
+    },
+    {
+      "epoch": 0.5428579275493669,
+      "grad_norm": 0.0,
+      "learning_rate": 9.099253258627468e-06,
+      "loss": 0.8692,
+      "step": 19766
+    },
+    {
+      "epoch": 0.5428853917772102,
+      "grad_norm": 0.0,
+      "learning_rate": 9.098367362049187e-06,
+      "loss": 0.9765,
+      "step": 19767
+    },
+    {
+      "epoch": 0.5429128560050535,
+      "grad_norm": 0.0,
+      "learning_rate": 9.097481472604976e-06,
+      "loss": 0.8626,
+      "step": 19768
+    },
+    {
+      "epoch": 0.5429403202328966,
+      "grad_norm": 0.0,
+      "learning_rate": 9.096595590301838e-06,
+      "loss": 0.9728,
+      "step": 19769
+    },
+    {
+      "epoch": 0.5429677844607399,
+      "grad_norm": 0.0,
+      "learning_rate": 9.095709715146786e-06,
+      "loss": 1.0402,
+      "step": 19770
+    },
+    {
+      "epoch": 0.5429952486885831,
+      "grad_norm": 0.0,
+      "learning_rate": 9.094823847146834e-06,
+      "loss": 0.8403,
+      "step": 19771
+    },
+    {
+      "epoch": 0.5430227129164263,
+      "grad_norm": 0.0,
+      "learning_rate": 9.093937986308983e-06,
+      "loss": 0.9387,
+      "step": 19772
+    },
+    {
+      "epoch": 0.5430501771442696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.093052132640245e-06,
+      "loss": 0.8651,
+      "step": 19773
+    },
+    {
+      "epoch": 0.5430776413721128,
+      "grad_norm": 0.0,
+      "learning_rate": 9.092166286147631e-06,
+      "loss": 0.8391,
+      "step": 19774
+    },
+    {
+      "epoch": 0.5431051055999561,
+      "grad_norm": 0.0,
+      "learning_rate": 9.091280446838148e-06,
+      "loss": 0.8047,
+      "step": 19775
+    },
+    {
+      "epoch": 0.5431325698277993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.090394614718805e-06,
+      "loss": 0.8953,
+      "step": 19776
+    },
+    {
+      "epoch": 0.5431600340556425,
+      "grad_norm": 0.0,
+      "learning_rate": 9.089508789796617e-06,
+      "loss": 0.865,
+      "step": 19777
+    },
+    {
+      "epoch": 0.5431874982834858,
+      "grad_norm": 0.0,
+      "learning_rate": 9.088622972078583e-06,
+      "loss": 0.9011,
+      "step": 19778
+    },
+    {
+      "epoch": 0.543214962511329,
+      "grad_norm": 0.0,
+      "learning_rate": 9.087737161571718e-06,
+      "loss": 0.833,
+      "step": 19779
+    },
+    {
+      "epoch": 0.5432424267391722,
+      "grad_norm": 0.0,
+      "learning_rate": 9.086851358283033e-06,
+      "loss": 0.9347,
+      "step": 19780
+    },
+    {
+      "epoch": 0.5432698909670155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.08596556221953e-06,
+      "loss": 0.7866,
+      "step": 19781
+    },
+    {
+      "epoch": 0.5432973551948587,
+      "grad_norm": 0.0,
+      "learning_rate": 9.085079773388226e-06,
+      "loss": 0.8491,
+      "step": 19782
+    },
+    {
+      "epoch": 0.5433248194227019,
+      "grad_norm": 0.0,
+      "learning_rate": 9.08419399179612e-06,
+      "loss": 0.8561,
+      "step": 19783
+    },
+    {
+      "epoch": 0.5433522836505452,
+      "grad_norm": 0.0,
+      "learning_rate": 9.083308217450229e-06,
+      "loss": 0.878,
+      "step": 19784
+    },
+    {
+      "epoch": 0.5433797478783884,
+      "grad_norm": 0.0,
+      "learning_rate": 9.082422450357557e-06,
+      "loss": 0.857,
+      "step": 19785
+    },
+    {
+      "epoch": 0.5434072121062317,
+      "grad_norm": 0.0,
+      "learning_rate": 9.081536690525114e-06,
+      "loss": 0.8433,
+      "step": 19786
+    },
+    {
+      "epoch": 0.5434346763340748,
+      "grad_norm": 0.0,
+      "learning_rate": 9.08065093795991e-06,
+      "loss": 0.8118,
+      "step": 19787
+    },
+    {
+      "epoch": 0.5434621405619181,
+      "grad_norm": 0.0,
+      "learning_rate": 9.079765192668955e-06,
+      "loss": 0.8743,
+      "step": 19788
+    },
+    {
+      "epoch": 0.5434896047897614,
+      "grad_norm": 0.0,
+      "learning_rate": 9.078879454659249e-06,
+      "loss": 0.9264,
+      "step": 19789
+    },
+    {
+      "epoch": 0.5435170690176045,
+      "grad_norm": 0.0,
+      "learning_rate": 9.077993723937806e-06,
+      "loss": 0.8654,
+      "step": 19790
+    },
+    {
+      "epoch": 0.5435445332454478,
+      "grad_norm": 0.0,
+      "learning_rate": 9.077108000511633e-06,
+      "loss": 0.874,
+      "step": 19791
+    },
+    {
+      "epoch": 0.543571997473291,
+      "grad_norm": 0.0,
+      "learning_rate": 9.07622228438774e-06,
+      "loss": 0.8136,
+      "step": 19792
+    },
+    {
+      "epoch": 0.5435994617011343,
+      "grad_norm": 0.0,
+      "learning_rate": 9.075336575573134e-06,
+      "loss": 0.9339,
+      "step": 19793
+    },
+    {
+      "epoch": 0.5436269259289775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.074450874074829e-06,
+      "loss": 0.8372,
+      "step": 19794
+    },
+    {
+      "epoch": 0.5436543901568207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.07356517989982e-06,
+      "loss": 0.894,
+      "step": 19795
+    },
+    {
+      "epoch": 0.543681854384664,
+      "grad_norm": 0.0,
+      "learning_rate": 9.072679493055125e-06,
+      "loss": 0.8696,
+      "step": 19796
+    },
+    {
+      "epoch": 0.5437093186125073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.071793813547749e-06,
+      "loss": 0.9082,
+      "step": 19797
+    },
+    {
+      "epoch": 0.5437367828403504,
+      "grad_norm": 0.0,
+      "learning_rate": 9.070908141384698e-06,
+      "loss": 0.9755,
+      "step": 19798
+    },
+    {
+      "epoch": 0.5437642470681937,
+      "grad_norm": 0.0,
+      "learning_rate": 9.070022476572982e-06,
+      "loss": 0.889,
+      "step": 19799
+    },
+    {
+      "epoch": 0.5437917112960369,
+      "grad_norm": 0.0,
+      "learning_rate": 9.069136819119615e-06,
+      "loss": 0.9205,
+      "step": 19800
+    },
+    {
+      "epoch": 0.5438191755238801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.06825116903159e-06,
+      "loss": 0.8308,
+      "step": 19801
+    },
+    {
+      "epoch": 0.5438466397517234,
+      "grad_norm": 0.0,
+      "learning_rate": 9.067365526315924e-06,
+      "loss": 0.8731,
+      "step": 19802
+    },
+    {
+      "epoch": 0.5438741039795666,
+      "grad_norm": 0.0,
+      "learning_rate": 9.066479890979623e-06,
+      "loss": 0.8627,
+      "step": 19803
+    },
+    {
+      "epoch": 0.5439015682074099,
+      "grad_norm": 0.0,
+      "learning_rate": 9.065594263029697e-06,
+      "loss": 0.9201,
+      "step": 19804
+    },
+    {
+      "epoch": 0.543929032435253,
+      "grad_norm": 0.0,
+      "learning_rate": 9.06470864247315e-06,
+      "loss": 0.8976,
+      "step": 19805
+    },
+    {
+      "epoch": 0.5439564966630963,
+      "grad_norm": 0.0,
+      "learning_rate": 9.063823029316994e-06,
+      "loss": 0.9884,
+      "step": 19806
+    },
+    {
+      "epoch": 0.5439839608909396,
+      "grad_norm": 0.0,
+      "learning_rate": 9.062937423568228e-06,
+      "loss": 0.8298,
+      "step": 19807
+    },
+    {
+      "epoch": 0.5440114251187828,
+      "grad_norm": 0.0,
+      "learning_rate": 9.062051825233867e-06,
+      "loss": 0.8946,
+      "step": 19808
+    },
+    {
+      "epoch": 0.544038889346626,
+      "grad_norm": 0.0,
+      "learning_rate": 9.061166234320914e-06,
+      "loss": 0.8873,
+      "step": 19809
+    },
+    {
+      "epoch": 0.5440663535744693,
+      "grad_norm": 0.0,
+      "learning_rate": 9.060280650836375e-06,
+      "loss": 0.9672,
+      "step": 19810
+    },
+    {
+      "epoch": 0.5440938178023125,
+      "grad_norm": 0.0,
+      "learning_rate": 9.059395074787263e-06,
+      "loss": 0.8996,
+      "step": 19811
+    },
+    {
+      "epoch": 0.5441212820301558,
+      "grad_norm": 0.0,
+      "learning_rate": 9.058509506180583e-06,
+      "loss": 0.8505,
+      "step": 19812
+    },
+    {
+      "epoch": 0.5441487462579989,
+      "grad_norm": 0.0,
+      "learning_rate": 9.057623945023337e-06,
+      "loss": 0.8251,
+      "step": 19813
+    },
+    {
+      "epoch": 0.5441762104858422,
+      "grad_norm": 0.0,
+      "learning_rate": 9.056738391322535e-06,
+      "loss": 0.8588,
+      "step": 19814
+    },
+    {
+      "epoch": 0.5442036747136855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.055852845085186e-06,
+      "loss": 0.9191,
+      "step": 19815
+    },
+    {
+      "epoch": 0.5442311389415286,
+      "grad_norm": 0.0,
+      "learning_rate": 9.054967306318293e-06,
+      "loss": 0.9405,
+      "step": 19816
+    },
+    {
+      "epoch": 0.5442586031693719,
+      "grad_norm": 0.0,
+      "learning_rate": 9.054081775028867e-06,
+      "loss": 0.7486,
+      "step": 19817
+    },
+    {
+      "epoch": 0.5442860673972151,
+      "grad_norm": 0.0,
+      "learning_rate": 9.053196251223915e-06,
+      "loss": 0.9524,
+      "step": 19818
+    },
+    {
+      "epoch": 0.5443135316250584,
+      "grad_norm": 0.0,
+      "learning_rate": 9.052310734910436e-06,
+      "loss": 0.8896,
+      "step": 19819
+    },
+    {
+      "epoch": 0.5443409958529016,
+      "grad_norm": 0.0,
+      "learning_rate": 9.051425226095445e-06,
+      "loss": 0.9504,
+      "step": 19820
+    },
+    {
+      "epoch": 0.5443684600807448,
+      "grad_norm": 0.0,
+      "learning_rate": 9.050539724785941e-06,
+      "loss": 0.8595,
+      "step": 19821
+    },
+    {
+      "epoch": 0.5443959243085881,
+      "grad_norm": 0.0,
+      "learning_rate": 9.049654230988936e-06,
+      "loss": 0.8749,
+      "step": 19822
+    },
+    {
+      "epoch": 0.5444233885364314,
+      "grad_norm": 0.0,
+      "learning_rate": 9.048768744711436e-06,
+      "loss": 0.8973,
+      "step": 19823
+    },
+    {
+      "epoch": 0.5444508527642745,
+      "grad_norm": 0.0,
+      "learning_rate": 9.04788326596045e-06,
+      "loss": 0.9453,
+      "step": 19824
+    },
+    {
+      "epoch": 0.5444783169921178,
+      "grad_norm": 0.0,
+      "learning_rate": 9.046997794742974e-06,
+      "loss": 0.884,
+      "step": 19825
+    },
+    {
+      "epoch": 0.544505781219961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.046112331066023e-06,
+      "loss": 0.8636,
+      "step": 19826
+    },
+    {
+      "epoch": 0.5445332454478042,
+      "grad_norm": 0.0,
+      "learning_rate": 9.045226874936602e-06,
+      "loss": 0.9265,
+      "step": 19827
+    },
+    {
+      "epoch": 0.5445607096756475,
+      "grad_norm": 0.0,
+      "learning_rate": 9.044341426361713e-06,
+      "loss": 0.8225,
+      "step": 19828
+    },
+    {
+      "epoch": 0.5445881739034907,
+      "grad_norm": 0.0,
+      "learning_rate": 9.043455985348363e-06,
+      "loss": 0.8416,
+      "step": 19829
+    },
+    {
+      "epoch": 0.544615638131334,
+      "grad_norm": 0.0,
+      "learning_rate": 9.042570551903568e-06,
+      "loss": 0.9722,
+      "step": 19830
+    },
+    {
+      "epoch": 0.5446431023591771,
+      "grad_norm": 0.0,
+      "learning_rate": 9.041685126034319e-06,
+      "loss": 0.9325,
+      "step": 19831
+    },
+    {
+      "epoch": 0.5446705665870204,
+      "grad_norm": 0.0,
+      "learning_rate": 9.040799707747631e-06,
+      "loss": 0.814,
+      "step": 19832
+    },
+    {
+      "epoch": 0.5446980308148637,
+      "grad_norm": 0.0,
+      "learning_rate": 9.039914297050505e-06,
+      "loss": 0.8767,
+      "step": 19833
+    },
+    {
+      "epoch": 0.5447254950427068,
+      "grad_norm": 0.0,
+      "learning_rate": 9.039028893949949e-06,
+      "loss": 0.7829,
+      "step": 19834
+    },
+    {
+      "epoch": 0.5447529592705501,
+      "grad_norm": 0.0,
+      "learning_rate": 9.038143498452972e-06,
+      "loss": 0.8411,
+      "step": 19835
+    },
+    {
+      "epoch": 0.5447804234983934,
+      "grad_norm": 0.0,
+      "learning_rate": 9.037258110566578e-06,
+      "loss": 0.9136,
+      "step": 19836
+    },
+    {
+      "epoch": 0.5448078877262366,
+      "grad_norm": 0.0,
+      "learning_rate": 9.036372730297764e-06,
+      "loss": 0.9518,
+      "step": 19837
+    },
+    {
+      "epoch": 0.5448353519540798,
+      "grad_norm": 0.0,
+      "learning_rate": 9.035487357653544e-06,
+      "loss": 0.9679,
+      "step": 19838
+    },
+    {
+      "epoch": 0.544862816181923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.034601992640924e-06,
+      "loss": 0.8807,
+      "step": 19839
+    },
+    {
+      "epoch": 0.5448902804097663,
+      "grad_norm": 0.0,
+      "learning_rate": 9.033716635266906e-06,
+      "loss": 0.937,
+      "step": 19840
+    },
+    {
+      "epoch": 0.5449177446376096,
+      "grad_norm": 0.0,
+      "learning_rate": 9.032831285538499e-06,
+      "loss": 0.8192,
+      "step": 19841
+    },
+    {
+      "epoch": 0.5449452088654527,
+      "grad_norm": 0.0,
+      "learning_rate": 9.0319459434627e-06,
+      "loss": 0.9021,
+      "step": 19842
+    },
+    {
+      "epoch": 0.544972673093296,
+      "grad_norm": 0.0,
+      "learning_rate": 9.031060609046523e-06,
+      "loss": 0.831,
+      "step": 19843
+    },
+    {
+      "epoch": 0.5450001373211392,
+      "grad_norm": 0.0,
+      "learning_rate": 9.030175282296971e-06,
+      "loss": 0.8411,
+      "step": 19844
+    },
+    {
+      "epoch": 0.5450276015489824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.029289963221044e-06,
+      "loss": 0.8466,
+      "step": 19845
+    },
+    {
+      "epoch": 0.5450550657768257,
+      "grad_norm": 0.0,
+      "learning_rate": 9.028404651825752e-06,
+      "loss": 0.822,
+      "step": 19846
+    },
+    {
+      "epoch": 0.5450825300046689,
+      "grad_norm": 0.0,
+      "learning_rate": 9.027519348118102e-06,
+      "loss": 0.9016,
+      "step": 19847
+    },
+    {
+      "epoch": 0.5451099942325122,
+      "grad_norm": 0.0,
+      "learning_rate": 9.026634052105095e-06,
+      "loss": 0.8268,
+      "step": 19848
+    },
+    {
+      "epoch": 0.5451374584603554,
+      "grad_norm": 0.0,
+      "learning_rate": 9.025748763793734e-06,
+      "loss": 0.8811,
+      "step": 19849
+    },
+    {
+      "epoch": 0.5451649226881986,
+      "grad_norm": 0.0,
+      "learning_rate": 9.024863483191025e-06,
+      "loss": 0.931,
+      "step": 19850
+    },
+    {
+      "epoch": 0.5451923869160419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.023978210303976e-06,
+      "loss": 0.819,
+      "step": 19851
+    },
+    {
+      "epoch": 0.545219851143885,
+      "grad_norm": 0.0,
+      "learning_rate": 9.02309294513959e-06,
+      "loss": 0.9219,
+      "step": 19852
+    },
+    {
+      "epoch": 0.5452473153717283,
+      "grad_norm": 0.0,
+      "learning_rate": 9.022207687704871e-06,
+      "loss": 0.9774,
+      "step": 19853
+    },
+    {
+      "epoch": 0.5452747795995716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.021322438006822e-06,
+      "loss": 0.8801,
+      "step": 19854
+    },
+    {
+      "epoch": 0.5453022438274148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.020437196052449e-06,
+      "loss": 0.9088,
+      "step": 19855
+    },
+    {
+      "epoch": 0.545329708055258,
+      "grad_norm": 0.0,
+      "learning_rate": 9.019551961848758e-06,
+      "loss": 0.8875,
+      "step": 19856
+    },
+    {
+      "epoch": 0.5453571722831012,
+      "grad_norm": 0.0,
+      "learning_rate": 9.018666735402749e-06,
+      "loss": 0.8942,
+      "step": 19857
+    },
+    {
+      "epoch": 0.5453846365109445,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01778151672143e-06,
+      "loss": 0.8137,
+      "step": 19858
+    },
+    {
+      "epoch": 0.5454121007387878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.016896305811807e-06,
+      "loss": 0.8562,
+      "step": 19859
+    },
+    {
+      "epoch": 0.5454395649666309,
+      "grad_norm": 0.0,
+      "learning_rate": 9.016011102680881e-06,
+      "loss": 0.8522,
+      "step": 19860
+    },
+    {
+      "epoch": 0.5454670291944742,
+      "grad_norm": 0.0,
+      "learning_rate": 9.015125907335653e-06,
+      "loss": 0.8229,
+      "step": 19861
+    },
+    {
+      "epoch": 0.5454944934223175,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01424071978313e-06,
+      "loss": 0.9447,
+      "step": 19862
+    },
+    {
+      "epoch": 0.5455219576501606,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01335554003032e-06,
+      "loss": 0.8283,
+      "step": 19863
+    },
+    {
+      "epoch": 0.5455494218780039,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01247036808422e-06,
+      "loss": 0.9307,
+      "step": 19864
+    },
+    {
+      "epoch": 0.5455768861058471,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01158520395184e-06,
+      "loss": 0.8044,
+      "step": 19865
+    },
+    {
+      "epoch": 0.5456043503336904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.010700047640179e-06,
+      "loss": 0.8436,
+      "step": 19866
+    },
+    {
+      "epoch": 0.5456318145615336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00981489915624e-06,
+      "loss": 0.8107,
+      "step": 19867
+    },
+    {
+      "epoch": 0.5456592787893768,
+      "grad_norm": 0.0,
+      "learning_rate": 9.008929758507032e-06,
+      "loss": 0.9105,
+      "step": 19868
+    },
+    {
+      "epoch": 0.5456867430172201,
+      "grad_norm": 0.0,
+      "learning_rate": 9.008044625699555e-06,
+      "loss": 0.8806,
+      "step": 19869
+    },
+    {
+      "epoch": 0.5457142072450633,
+      "grad_norm": 0.0,
+      "learning_rate": 9.007159500740812e-06,
+      "loss": 0.8372,
+      "step": 19870
+    },
+    {
+      "epoch": 0.5457416714729065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.006274383637811e-06,
+      "loss": 0.9035,
+      "step": 19871
+    },
+    {
+      "epoch": 0.5457691357007498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.005389274397552e-06,
+      "loss": 0.8224,
+      "step": 19872
+    },
+    {
+      "epoch": 0.545796599928593,
+      "grad_norm": 0.0,
+      "learning_rate": 9.004504173027034e-06,
+      "loss": 0.8777,
+      "step": 19873
+    },
+    {
+      "epoch": 0.5458240641564363,
+      "grad_norm": 0.0,
+      "learning_rate": 9.003619079533266e-06,
+      "loss": 0.8161,
+      "step": 19874
+    },
+    {
+      "epoch": 0.5458515283842795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00273399392325e-06,
+      "loss": 0.8716,
+      "step": 19875
+    },
+    {
+      "epoch": 0.5458789926121227,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00184891620399e-06,
+      "loss": 0.876,
+      "step": 19876
+    },
+    {
+      "epoch": 0.545906456839966,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00096384638249e-06,
+      "loss": 0.8746,
+      "step": 19877
+    },
+    {
+      "epoch": 0.5459339210678091,
+      "grad_norm": 0.0,
+      "learning_rate": 9.000078784465746e-06,
+      "loss": 0.9084,
+      "step": 19878
+    },
+    {
+      "epoch": 0.5459613852956524,
+      "grad_norm": 0.0,
+      "learning_rate": 8.999193730460769e-06,
+      "loss": 0.9974,
+      "step": 19879
+    },
+    {
+      "epoch": 0.5459888495234957,
+      "grad_norm": 0.0,
+      "learning_rate": 8.99830868437456e-06,
+      "loss": 0.8566,
+      "step": 19880
+    },
+    {
+      "epoch": 0.5460163137513389,
+      "grad_norm": 0.0,
+      "learning_rate": 8.997423646214118e-06,
+      "loss": 0.9258,
+      "step": 19881
+    },
+    {
+      "epoch": 0.5460437779791821,
+      "grad_norm": 0.0,
+      "learning_rate": 8.996538615986447e-06,
+      "loss": 0.925,
+      "step": 19882
+    },
+    {
+      "epoch": 0.5460712422070253,
+      "grad_norm": 0.0,
+      "learning_rate": 8.995653593698556e-06,
+      "loss": 0.9716,
+      "step": 19883
+    },
+    {
+      "epoch": 0.5460987064348686,
+      "grad_norm": 0.0,
+      "learning_rate": 8.994768579357442e-06,
+      "loss": 0.7657,
+      "step": 19884
+    },
+    {
+      "epoch": 0.5461261706627119,
+      "grad_norm": 0.0,
+      "learning_rate": 8.993883572970107e-06,
+      "loss": 0.7738,
+      "step": 19885
+    },
+    {
+      "epoch": 0.546153634890555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.992998574543554e-06,
+      "loss": 0.9315,
+      "step": 19886
+    },
+    {
+      "epoch": 0.5461810991183983,
+      "grad_norm": 0.0,
+      "learning_rate": 8.992113584084788e-06,
+      "loss": 0.8265,
+      "step": 19887
+    },
+    {
+      "epoch": 0.5462085633462416,
+      "grad_norm": 0.0,
+      "learning_rate": 8.991228601600808e-06,
+      "loss": 0.8459,
+      "step": 19888
+    },
+    {
+      "epoch": 0.5462360275740847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.990343627098622e-06,
+      "loss": 0.7964,
+      "step": 19889
+    },
+    {
+      "epoch": 0.546263491801928,
+      "grad_norm": 0.0,
+      "learning_rate": 8.989458660585225e-06,
+      "loss": 0.8762,
+      "step": 19890
+    },
+    {
+      "epoch": 0.5462909560297712,
+      "grad_norm": 0.0,
+      "learning_rate": 8.98857370206762e-06,
+      "loss": 0.9253,
+      "step": 19891
+    },
+    {
+      "epoch": 0.5463184202576145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.987688751552816e-06,
+      "loss": 0.8057,
+      "step": 19892
+    },
+    {
+      "epoch": 0.5463458844854577,
+      "grad_norm": 0.0,
+      "learning_rate": 8.986803809047809e-06,
+      "loss": 0.8513,
+      "step": 19893
+    },
+    {
+      "epoch": 0.5463733487133009,
+      "grad_norm": 0.0,
+      "learning_rate": 8.985918874559601e-06,
+      "loss": 0.8971,
+      "step": 19894
+    },
+    {
+      "epoch": 0.5464008129411442,
+      "grad_norm": 0.0,
+      "learning_rate": 8.985033948095201e-06,
+      "loss": 0.9024,
+      "step": 19895
+    },
+    {
+      "epoch": 0.5464282771689873,
+      "grad_norm": 0.0,
+      "learning_rate": 8.984149029661603e-06,
+      "loss": 0.826,
+      "step": 19896
+    },
+    {
+      "epoch": 0.5464557413968306,
+      "grad_norm": 0.0,
+      "learning_rate": 8.983264119265809e-06,
+      "loss": 0.8715,
+      "step": 19897
+    },
+    {
+      "epoch": 0.5464832056246739,
+      "grad_norm": 0.0,
+      "learning_rate": 8.982379216914824e-06,
+      "loss": 0.7555,
+      "step": 19898
+    },
+    {
+      "epoch": 0.5465106698525171,
+      "grad_norm": 0.0,
+      "learning_rate": 8.98149432261565e-06,
+      "loss": 0.8836,
+      "step": 19899
+    },
+    {
+      "epoch": 0.5465381340803603,
+      "grad_norm": 0.0,
+      "learning_rate": 8.980609436375289e-06,
+      "loss": 0.8027,
+      "step": 19900
+    },
+    {
+      "epoch": 0.5465655983082036,
+      "grad_norm": 0.0,
+      "learning_rate": 8.97972455820074e-06,
+      "loss": 0.7718,
+      "step": 19901
+    },
+    {
+      "epoch": 0.5465930625360468,
+      "grad_norm": 0.0,
+      "learning_rate": 8.978839688099003e-06,
+      "loss": 0.8707,
+      "step": 19902
+    },
+    {
+      "epoch": 0.5466205267638901,
+      "grad_norm": 0.0,
+      "learning_rate": 8.977954826077081e-06,
+      "loss": 0.8913,
+      "step": 19903
+    },
+    {
+      "epoch": 0.5466479909917332,
+      "grad_norm": 0.0,
+      "learning_rate": 8.977069972141981e-06,
+      "loss": 0.9684,
+      "step": 19904
+    },
+    {
+      "epoch": 0.5466754552195765,
+      "grad_norm": 0.0,
+      "learning_rate": 8.976185126300696e-06,
+      "loss": 0.7626,
+      "step": 19905
+    },
+    {
+      "epoch": 0.5467029194474198,
+      "grad_norm": 0.0,
+      "learning_rate": 8.975300288560236e-06,
+      "loss": 0.9748,
+      "step": 19906
+    },
+    {
+      "epoch": 0.5467303836752629,
+      "grad_norm": 0.0,
+      "learning_rate": 8.974415458927592e-06,
+      "loss": 0.8717,
+      "step": 19907
+    },
+    {
+      "epoch": 0.5467578479031062,
+      "grad_norm": 0.0,
+      "learning_rate": 8.973530637409772e-06,
+      "loss": 0.9401,
+      "step": 19908
+    },
+    {
+      "epoch": 0.5467853121309494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.972645824013774e-06,
+      "loss": 0.8887,
+      "step": 19909
+    },
+    {
+      "epoch": 0.5468127763587927,
+      "grad_norm": 0.0,
+      "learning_rate": 8.9717610187466e-06,
+      "loss": 0.9022,
+      "step": 19910
+    },
+    {
+      "epoch": 0.5468402405866359,
+      "grad_norm": 0.0,
+      "learning_rate": 8.970876221615253e-06,
+      "loss": 0.896,
+      "step": 19911
+    },
+    {
+      "epoch": 0.5468677048144791,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96999143262673e-06,
+      "loss": 0.8409,
+      "step": 19912
+    },
+    {
+      "epoch": 0.5468951690423224,
+      "grad_norm": 0.0,
+      "learning_rate": 8.969106651788038e-06,
+      "loss": 0.8307,
+      "step": 19913
+    },
+    {
+      "epoch": 0.5469226332701657,
+      "grad_norm": 0.0,
+      "learning_rate": 8.968221879106167e-06,
+      "loss": 0.886,
+      "step": 19914
+    },
+    {
+      "epoch": 0.5469500974980088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.967337114588127e-06,
+      "loss": 0.8909,
+      "step": 19915
+    },
+    {
+      "epoch": 0.5469775617258521,
+      "grad_norm": 0.0,
+      "learning_rate": 8.966452358240918e-06,
+      "loss": 0.8453,
+      "step": 19916
+    },
+    {
+      "epoch": 0.5470050259536953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.965567610071536e-06,
+      "loss": 0.8931,
+      "step": 19917
+    },
+    {
+      "epoch": 0.5470324901815385,
+      "grad_norm": 0.0,
+      "learning_rate": 8.964682870086987e-06,
+      "loss": 0.7796,
+      "step": 19918
+    },
+    {
+      "epoch": 0.5470599544093818,
+      "grad_norm": 0.0,
+      "learning_rate": 8.963798138294265e-06,
+      "loss": 0.8137,
+      "step": 19919
+    },
+    {
+      "epoch": 0.547087418637225,
+      "grad_norm": 0.0,
+      "learning_rate": 8.962913414700374e-06,
+      "loss": 0.8367,
+      "step": 19920
+    },
+    {
+      "epoch": 0.5471148828650683,
+      "grad_norm": 0.0,
+      "learning_rate": 8.962028699312314e-06,
+      "loss": 0.8896,
+      "step": 19921
+    },
+    {
+      "epoch": 0.5471423470929114,
+      "grad_norm": 0.0,
+      "learning_rate": 8.961143992137083e-06,
+      "loss": 0.8732,
+      "step": 19922
+    },
+    {
+      "epoch": 0.5471698113207547,
+      "grad_norm": 0.0,
+      "learning_rate": 8.960259293181685e-06,
+      "loss": 0.897,
+      "step": 19923
+    },
+    {
+      "epoch": 0.547197275548598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.959374602453122e-06,
+      "loss": 0.8597,
+      "step": 19924
+    },
+    {
+      "epoch": 0.5472247397764411,
+      "grad_norm": 0.0,
+      "learning_rate": 8.958489919958383e-06,
+      "loss": 0.8588,
+      "step": 19925
+    },
+    {
+      "epoch": 0.5472522040042844,
+      "grad_norm": 0.0,
+      "learning_rate": 8.957605245704479e-06,
+      "loss": 0.9567,
+      "step": 19926
+    },
+    {
+      "epoch": 0.5472796682321277,
+      "grad_norm": 0.0,
+      "learning_rate": 8.956720579698405e-06,
+      "loss": 0.891,
+      "step": 19927
+    },
+    {
+      "epoch": 0.5473071324599709,
+      "grad_norm": 0.0,
+      "learning_rate": 8.955835921947164e-06,
+      "loss": 0.8441,
+      "step": 19928
+    },
+    {
+      "epoch": 0.5473345966878141,
+      "grad_norm": 0.0,
+      "learning_rate": 8.95495127245775e-06,
+      "loss": 0.7851,
+      "step": 19929
+    },
+    {
+      "epoch": 0.5473620609156573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.954066631237172e-06,
+      "loss": 0.8898,
+      "step": 19930
+    },
+    {
+      "epoch": 0.5473895251435006,
+      "grad_norm": 0.0,
+      "learning_rate": 8.953181998292419e-06,
+      "loss": 0.9141,
+      "step": 19931
+    },
+    {
+      "epoch": 0.5474169893713439,
+      "grad_norm": 0.0,
+      "learning_rate": 8.952297373630498e-06,
+      "loss": 0.913,
+      "step": 19932
+    },
+    {
+      "epoch": 0.547444453599187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.951412757258403e-06,
+      "loss": 0.8778,
+      "step": 19933
+    },
+    {
+      "epoch": 0.5474719178270303,
+      "grad_norm": 0.0,
+      "learning_rate": 8.95052814918314e-06,
+      "loss": 0.9215,
+      "step": 19934
+    },
+    {
+      "epoch": 0.5474993820548735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.949643549411703e-06,
+      "loss": 0.8893,
+      "step": 19935
+    },
+    {
+      "epoch": 0.5475268462827168,
+      "grad_norm": 0.0,
+      "learning_rate": 8.948758957951097e-06,
+      "loss": 0.8432,
+      "step": 19936
+    },
+    {
+      "epoch": 0.54755431051056,
+      "grad_norm": 0.0,
+      "learning_rate": 8.947874374808312e-06,
+      "loss": 0.9146,
+      "step": 19937
+    },
+    {
+      "epoch": 0.5475817747384032,
+      "grad_norm": 0.0,
+      "learning_rate": 8.946989799990353e-06,
+      "loss": 0.8566,
+      "step": 19938
+    },
+    {
+      "epoch": 0.5476092389662465,
+      "grad_norm": 0.0,
+      "learning_rate": 8.946105233504218e-06,
+      "loss": 0.8742,
+      "step": 19939
+    },
+    {
+      "epoch": 0.5476367031940897,
+      "grad_norm": 0.0,
+      "learning_rate": 8.945220675356911e-06,
+      "loss": 0.9085,
+      "step": 19940
+    },
+    {
+      "epoch": 0.5476641674219329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.944336125555423e-06,
+      "loss": 0.8663,
+      "step": 19941
+    },
+    {
+      "epoch": 0.5476916316497762,
+      "grad_norm": 0.0,
+      "learning_rate": 8.94345158410676e-06,
+      "loss": 0.8656,
+      "step": 19942
+    },
+    {
+      "epoch": 0.5477190958776194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.942567051017914e-06,
+      "loss": 0.9539,
+      "step": 19943
+    },
+    {
+      "epoch": 0.5477465601054626,
+      "grad_norm": 0.0,
+      "learning_rate": 8.941682526295889e-06,
+      "loss": 0.9478,
+      "step": 19944
+    },
+    {
+      "epoch": 0.5477740243333059,
+      "grad_norm": 0.0,
+      "learning_rate": 8.94079800994768e-06,
+      "loss": 0.8472,
+      "step": 19945
+    },
+    {
+      "epoch": 0.5478014885611491,
+      "grad_norm": 0.0,
+      "learning_rate": 8.939913501980288e-06,
+      "loss": 0.8706,
+      "step": 19946
+    },
+    {
+      "epoch": 0.5478289527889924,
+      "grad_norm": 0.0,
+      "learning_rate": 8.939029002400711e-06,
+      "loss": 0.858,
+      "step": 19947
+    },
+    {
+      "epoch": 0.5478564170168355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.938144511215951e-06,
+      "loss": 0.908,
+      "step": 19948
+    },
+    {
+      "epoch": 0.5478838812446788,
+      "grad_norm": 0.0,
+      "learning_rate": 8.937260028433e-06,
+      "loss": 0.9129,
+      "step": 19949
+    },
+    {
+      "epoch": 0.5479113454725221,
+      "grad_norm": 0.0,
+      "learning_rate": 8.936375554058857e-06,
+      "loss": 0.8351,
+      "step": 19950
+    },
+    {
+      "epoch": 0.5479388097003652,
+      "grad_norm": 0.0,
+      "learning_rate": 8.935491088100523e-06,
+      "loss": 0.8068,
+      "step": 19951
+    },
+    {
+      "epoch": 0.5479662739282085,
+      "grad_norm": 0.0,
+      "learning_rate": 8.934606630565e-06,
+      "loss": 0.9725,
+      "step": 19952
+    },
+    {
+      "epoch": 0.5479937381560518,
+      "grad_norm": 0.0,
+      "learning_rate": 8.933722181459278e-06,
+      "loss": 0.7633,
+      "step": 19953
+    },
+    {
+      "epoch": 0.548021202383895,
+      "grad_norm": 0.0,
+      "learning_rate": 8.932837740790363e-06,
+      "loss": 0.9456,
+      "step": 19954
+    },
+    {
+      "epoch": 0.5480486666117382,
+      "grad_norm": 0.0,
+      "learning_rate": 8.931953308565246e-06,
+      "loss": 0.8891,
+      "step": 19955
+    },
+    {
+      "epoch": 0.5480761308395814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.93106888479093e-06,
+      "loss": 0.8882,
+      "step": 19956
+    },
+    {
+      "epoch": 0.5481035950674247,
+      "grad_norm": 0.0,
+      "learning_rate": 8.93018446947441e-06,
+      "loss": 0.8931,
+      "step": 19957
+    },
+    {
+      "epoch": 0.548131059295268,
+      "grad_norm": 0.0,
+      "learning_rate": 8.929300062622683e-06,
+      "loss": 0.8479,
+      "step": 19958
+    },
+    {
+      "epoch": 0.5481585235231111,
+      "grad_norm": 0.0,
+      "learning_rate": 8.928415664242755e-06,
+      "loss": 0.8,
+      "step": 19959
+    },
+    {
+      "epoch": 0.5481859877509544,
+      "grad_norm": 0.0,
+      "learning_rate": 8.927531274341614e-06,
+      "loss": 0.8712,
+      "step": 19960
+    },
+    {
+      "epoch": 0.5482134519787976,
+      "grad_norm": 0.0,
+      "learning_rate": 8.926646892926259e-06,
+      "loss": 0.7642,
+      "step": 19961
+    },
+    {
+      "epoch": 0.5482409162066408,
+      "grad_norm": 0.0,
+      "learning_rate": 8.925762520003689e-06,
+      "loss": 0.8025,
+      "step": 19962
+    },
+    {
+      "epoch": 0.5482683804344841,
+      "grad_norm": 0.0,
+      "learning_rate": 8.924878155580904e-06,
+      "loss": 0.879,
+      "step": 19963
+    },
+    {
+      "epoch": 0.5482958446623273,
+      "grad_norm": 0.0,
+      "learning_rate": 8.923993799664898e-06,
+      "loss": 0.8831,
+      "step": 19964
+    },
+    {
+      "epoch": 0.5483233088901706,
+      "grad_norm": 0.0,
+      "learning_rate": 8.923109452262674e-06,
+      "loss": 0.9067,
+      "step": 19965
+    },
+    {
+      "epoch": 0.5483507731180138,
+      "grad_norm": 0.0,
+      "learning_rate": 8.92222511338122e-06,
+      "loss": 0.8487,
+      "step": 19966
+    },
+    {
+      "epoch": 0.548378237345857,
+      "grad_norm": 0.0,
+      "learning_rate": 8.921340783027541e-06,
+      "loss": 0.9113,
+      "step": 19967
+    },
+    {
+      "epoch": 0.5484057015737003,
+      "grad_norm": 0.0,
+      "learning_rate": 8.920456461208632e-06,
+      "loss": 0.8979,
+      "step": 19968
+    },
+    {
+      "epoch": 0.5484331658015434,
+      "grad_norm": 0.0,
+      "learning_rate": 8.919572147931487e-06,
+      "loss": 0.8344,
+      "step": 19969
+    },
+    {
+      "epoch": 0.5484606300293867,
+      "grad_norm": 0.0,
+      "learning_rate": 8.918687843203107e-06,
+      "loss": 0.8028,
+      "step": 19970
+    },
+    {
+      "epoch": 0.54848809425723,
+      "grad_norm": 0.0,
+      "learning_rate": 8.91780354703049e-06,
+      "loss": 0.8733,
+      "step": 19971
+    },
+    {
+      "epoch": 0.5485155584850732,
+      "grad_norm": 0.0,
+      "learning_rate": 8.916919259420631e-06,
+      "loss": 0.8737,
+      "step": 19972
+    },
+    {
+      "epoch": 0.5485430227129164,
+      "grad_norm": 0.0,
+      "learning_rate": 8.916034980380523e-06,
+      "loss": 0.8166,
+      "step": 19973
+    },
+    {
+      "epoch": 0.5485704869407596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.915150709917165e-06,
+      "loss": 0.9072,
+      "step": 19974
+    },
+    {
+      "epoch": 0.5485979511686029,
+      "grad_norm": 0.0,
+      "learning_rate": 8.914266448037558e-06,
+      "loss": 0.8915,
+      "step": 19975
+    },
+    {
+      "epoch": 0.5486254153964462,
+      "grad_norm": 0.0,
+      "learning_rate": 8.913382194748693e-06,
+      "loss": 0.8995,
+      "step": 19976
+    },
+    {
+      "epoch": 0.5486528796242893,
+      "grad_norm": 0.0,
+      "learning_rate": 8.912497950057573e-06,
+      "loss": 0.807,
+      "step": 19977
+    },
+    {
+      "epoch": 0.5486803438521326,
+      "grad_norm": 0.0,
+      "learning_rate": 8.911613713971188e-06,
+      "loss": 0.9807,
+      "step": 19978
+    },
+    {
+      "epoch": 0.5487078080799759,
+      "grad_norm": 0.0,
+      "learning_rate": 8.910729486496535e-06,
+      "loss": 0.7792,
+      "step": 19979
+    },
+    {
+      "epoch": 0.548735272307819,
+      "grad_norm": 0.0,
+      "learning_rate": 8.909845267640616e-06,
+      "loss": 0.9008,
+      "step": 19980
+    },
+    {
+      "epoch": 0.5487627365356623,
+      "grad_norm": 0.0,
+      "learning_rate": 8.90896105741042e-06,
+      "loss": 0.857,
+      "step": 19981
+    },
+    {
+      "epoch": 0.5487902007635055,
+      "grad_norm": 0.0,
+      "learning_rate": 8.908076855812947e-06,
+      "loss": 0.9142,
+      "step": 19982
+    },
+    {
+      "epoch": 0.5488176649913488,
+      "grad_norm": 0.0,
+      "learning_rate": 8.907192662855196e-06,
+      "loss": 0.8092,
+      "step": 19983
+    },
+    {
+      "epoch": 0.548845129219192,
+      "grad_norm": 0.0,
+      "learning_rate": 8.90630847854416e-06,
+      "loss": 0.8675,
+      "step": 19984
+    },
+    {
+      "epoch": 0.5488725934470352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.905424302886832e-06,
+      "loss": 0.8962,
+      "step": 19985
+    },
+    {
+      "epoch": 0.5489000576748785,
+      "grad_norm": 0.0,
+      "learning_rate": 8.904540135890211e-06,
+      "loss": 0.9885,
+      "step": 19986
+    },
+    {
+      "epoch": 0.5489275219027218,
+      "grad_norm": 0.0,
+      "learning_rate": 8.903655977561295e-06,
+      "loss": 0.9993,
+      "step": 19987
+    },
+    {
+      "epoch": 0.5489549861305649,
+      "grad_norm": 0.0,
+      "learning_rate": 8.902771827907075e-06,
+      "loss": 0.8628,
+      "step": 19988
+    },
+    {
+      "epoch": 0.5489824503584082,
+      "grad_norm": 0.0,
+      "learning_rate": 8.901887686934553e-06,
+      "loss": 0.8398,
+      "step": 19989
+    },
+    {
+      "epoch": 0.5490099145862514,
+      "grad_norm": 0.0,
+      "learning_rate": 8.901003554650718e-06,
+      "loss": 0.8901,
+      "step": 19990
+    },
+    {
+      "epoch": 0.5490373788140946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.900119431062568e-06,
+      "loss": 0.8846,
+      "step": 19991
+    },
+    {
+      "epoch": 0.5490648430419379,
+      "grad_norm": 0.0,
+      "learning_rate": 8.899235316177101e-06,
+      "loss": 0.7892,
+      "step": 19992
+    },
+    {
+      "epoch": 0.5490923072697811,
+      "grad_norm": 0.0,
+      "learning_rate": 8.89835121000131e-06,
+      "loss": 0.8469,
+      "step": 19993
+    },
+    {
+      "epoch": 0.5491197714976244,
+      "grad_norm": 0.0,
+      "learning_rate": 8.897467112542188e-06,
+      "loss": 0.8748,
+      "step": 19994
+    },
+    {
+      "epoch": 0.5491472357254675,
+      "grad_norm": 0.0,
+      "learning_rate": 8.896583023806741e-06,
+      "loss": 0.9049,
+      "step": 19995
+    },
+    {
+      "epoch": 0.5491746999533108,
+      "grad_norm": 0.0,
+      "learning_rate": 8.895698943801953e-06,
+      "loss": 0.9996,
+      "step": 19996
+    },
+    {
+      "epoch": 0.5492021641811541,
+      "grad_norm": 0.0,
+      "learning_rate": 8.894814872534819e-06,
+      "loss": 0.9283,
+      "step": 19997
+    },
+    {
+      "epoch": 0.5492296284089973,
+      "grad_norm": 0.0,
+      "learning_rate": 8.893930810012339e-06,
+      "loss": 0.7719,
+      "step": 19998
+    },
+    {
+      "epoch": 0.5492570926368405,
+      "grad_norm": 0.0,
+      "learning_rate": 8.89304675624151e-06,
+      "loss": 0.8667,
+      "step": 19999
+    },
+    {
+      "epoch": 0.5492845568646838,
+      "grad_norm": 0.0,
+      "learning_rate": 8.89216271122932e-06,
+      "loss": 0.8289,
+      "step": 20000
+    },
+    {
+      "epoch": 0.549312021092527,
+      "grad_norm": 0.0,
+      "learning_rate": 8.891278674982774e-06,
+      "loss": 0.8702,
+      "step": 20001
+    },
+    {
+      "epoch": 0.5493394853203702,
+      "grad_norm": 0.0,
+      "learning_rate": 8.890394647508854e-06,
+      "loss": 0.8886,
+      "step": 20002
+    },
+    {
+      "epoch": 0.5493669495482134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.889510628814564e-06,
+      "loss": 0.8391,
+      "step": 20003
+    },
+    {
+      "epoch": 0.5493944137760567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.888626618906896e-06,
+      "loss": 0.7872,
+      "step": 20004
+    },
+    {
+      "epoch": 0.5494218780039,
+      "grad_norm": 0.0,
+      "learning_rate": 8.887742617792844e-06,
+      "loss": 0.9044,
+      "step": 20005
+    },
+    {
+      "epoch": 0.5494493422317431,
+      "grad_norm": 0.0,
+      "learning_rate": 8.886858625479403e-06,
+      "loss": 0.7943,
+      "step": 20006
+    },
+    {
+      "epoch": 0.5494768064595864,
+      "grad_norm": 0.0,
+      "learning_rate": 8.88597464197357e-06,
+      "loss": 0.8823,
+      "step": 20007
+    },
+    {
+      "epoch": 0.5495042706874296,
+      "grad_norm": 0.0,
+      "learning_rate": 8.885090667282337e-06,
+      "loss": 0.8887,
+      "step": 20008
+    },
+    {
+      "epoch": 0.5495317349152729,
+      "grad_norm": 0.0,
+      "learning_rate": 8.884206701412697e-06,
+      "loss": 0.8893,
+      "step": 20009
+    },
+    {
+      "epoch": 0.5495591991431161,
+      "grad_norm": 0.0,
+      "learning_rate": 8.883322744371645e-06,
+      "loss": 0.9889,
+      "step": 20010
+    },
+    {
+      "epoch": 0.5495866633709593,
+      "grad_norm": 0.0,
+      "learning_rate": 8.882438796166178e-06,
+      "loss": 0.8474,
+      "step": 20011
+    },
+    {
+      "epoch": 0.5496141275988026,
+      "grad_norm": 0.0,
+      "learning_rate": 8.881554856803285e-06,
+      "loss": 0.9433,
+      "step": 20012
+    },
+    {
+      "epoch": 0.5496415918266458,
+      "grad_norm": 0.0,
+      "learning_rate": 8.88067092628997e-06,
+      "loss": 0.8489,
+      "step": 20013
+    },
+    {
+      "epoch": 0.549669056054489,
+      "grad_norm": 0.0,
+      "learning_rate": 8.879787004633213e-06,
+      "loss": 0.9068,
+      "step": 20014
+    },
+    {
+      "epoch": 0.5496965202823323,
+      "grad_norm": 0.0,
+      "learning_rate": 8.878903091840017e-06,
+      "loss": 0.9222,
+      "step": 20015
+    },
+    {
+      "epoch": 0.5497239845101755,
+      "grad_norm": 0.0,
+      "learning_rate": 8.878019187917375e-06,
+      "loss": 0.834,
+      "step": 20016
+    },
+    {
+      "epoch": 0.5497514487380187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.87713529287228e-06,
+      "loss": 0.8726,
+      "step": 20017
+    },
+    {
+      "epoch": 0.549778912965862,
+      "grad_norm": 0.0,
+      "learning_rate": 8.876251406711727e-06,
+      "loss": 0.9109,
+      "step": 20018
+    },
+    {
+      "epoch": 0.5498063771937052,
+      "grad_norm": 0.0,
+      "learning_rate": 8.875367529442705e-06,
+      "loss": 0.9128,
+      "step": 20019
+    },
+    {
+      "epoch": 0.5498338414215485,
+      "grad_norm": 0.0,
+      "learning_rate": 8.874483661072213e-06,
+      "loss": 0.8111,
+      "step": 20020
+    },
+    {
+      "epoch": 0.5498613056493916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.87359980160724e-06,
+      "loss": 0.7961,
+      "step": 20021
+    },
+    {
+      "epoch": 0.5498887698772349,
+      "grad_norm": 0.0,
+      "learning_rate": 8.872715951054782e-06,
+      "loss": 0.8716,
+      "step": 20022
+    },
+    {
+      "epoch": 0.5499162341050782,
+      "grad_norm": 0.0,
+      "learning_rate": 8.871832109421833e-06,
+      "loss": 0.8589,
+      "step": 20023
+    },
+    {
+      "epoch": 0.5499436983329213,
+      "grad_norm": 0.0,
+      "learning_rate": 8.870948276715387e-06,
+      "loss": 0.9361,
+      "step": 20024
+    },
+    {
+      "epoch": 0.5499711625607646,
+      "grad_norm": 0.0,
+      "learning_rate": 8.870064452942435e-06,
+      "loss": 0.9749,
+      "step": 20025
+    },
+    {
+      "epoch": 0.5499986267886079,
+      "grad_norm": 0.0,
+      "learning_rate": 8.869180638109969e-06,
+      "loss": 0.9071,
+      "step": 20026
+    },
+    {
+      "epoch": 0.5500260910164511,
+      "grad_norm": 0.0,
+      "learning_rate": 8.868296832224984e-06,
+      "loss": 0.9167,
+      "step": 20027
+    },
+    {
+      "epoch": 0.5500535552442943,
+      "grad_norm": 0.0,
+      "learning_rate": 8.867413035294474e-06,
+      "loss": 0.9362,
+      "step": 20028
+    },
+    {
+      "epoch": 0.5500810194721375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.866529247325428e-06,
+      "loss": 0.9337,
+      "step": 20029
+    },
+    {
+      "epoch": 0.5501084836999808,
+      "grad_norm": 0.0,
+      "learning_rate": 8.865645468324849e-06,
+      "loss": 0.8429,
+      "step": 20030
+    },
+    {
+      "epoch": 0.550135947927824,
+      "grad_norm": 0.0,
+      "learning_rate": 8.864761698299715e-06,
+      "loss": 0.9225,
+      "step": 20031
+    },
+    {
+      "epoch": 0.5501634121556672,
+      "grad_norm": 0.0,
+      "learning_rate": 8.863877937257029e-06,
+      "loss": 0.845,
+      "step": 20032
+    },
+    {
+      "epoch": 0.5501908763835105,
+      "grad_norm": 0.0,
+      "learning_rate": 8.86299418520378e-06,
+      "loss": 0.9115,
+      "step": 20033
+    },
+    {
+      "epoch": 0.5502183406113537,
+      "grad_norm": 0.0,
+      "learning_rate": 8.86211044214696e-06,
+      "loss": 0.8229,
+      "step": 20034
+    },
+    {
+      "epoch": 0.5502458048391969,
+      "grad_norm": 0.0,
+      "learning_rate": 8.861226708093566e-06,
+      "loss": 0.9581,
+      "step": 20035
+    },
+    {
+      "epoch": 0.5502732690670402,
+      "grad_norm": 0.0,
+      "learning_rate": 8.860342983050588e-06,
+      "loss": 0.8029,
+      "step": 20036
+    },
+    {
+      "epoch": 0.5503007332948834,
+      "grad_norm": 0.0,
+      "learning_rate": 8.859459267025016e-06,
+      "loss": 0.76,
+      "step": 20037
+    },
+    {
+      "epoch": 0.5503281975227267,
+      "grad_norm": 0.0,
+      "learning_rate": 8.858575560023843e-06,
+      "loss": 0.8524,
+      "step": 20038
+    },
+    {
+      "epoch": 0.5503556617505699,
+      "grad_norm": 0.0,
+      "learning_rate": 8.857691862054062e-06,
+      "loss": 0.8693,
+      "step": 20039
+    },
+    {
+      "epoch": 0.5503831259784131,
+      "grad_norm": 0.0,
+      "learning_rate": 8.856808173122667e-06,
+      "loss": 0.9401,
+      "step": 20040
+    },
+    {
+      "epoch": 0.5504105902062564,
+      "grad_norm": 0.0,
+      "learning_rate": 8.855924493236647e-06,
+      "loss": 0.8846,
+      "step": 20041
+    },
+    {
+      "epoch": 0.5504380544340995,
+      "grad_norm": 0.0,
+      "learning_rate": 8.855040822403e-06,
+      "loss": 0.8405,
+      "step": 20042
+    },
+    {
+      "epoch": 0.5504655186619428,
+      "grad_norm": 0.0,
+      "learning_rate": 8.85415716062871e-06,
+      "loss": 0.7429,
+      "step": 20043
+    },
+    {
+      "epoch": 0.5504929828897861,
+      "grad_norm": 0.0,
+      "learning_rate": 8.853273507920774e-06,
+      "loss": 0.8686,
+      "step": 20044
+    },
+    {
+      "epoch": 0.5505204471176293,
+      "grad_norm": 0.0,
+      "learning_rate": 8.85238986428618e-06,
+      "loss": 0.938,
+      "step": 20045
+    },
+    {
+      "epoch": 0.5505479113454725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.851506229731922e-06,
+      "loss": 0.8941,
+      "step": 20046
+    },
+    {
+      "epoch": 0.5505753755733157,
+      "grad_norm": 0.0,
+      "learning_rate": 8.850622604264993e-06,
+      "loss": 0.7814,
+      "step": 20047
+    },
+    {
+      "epoch": 0.550602839801159,
+      "grad_norm": 0.0,
+      "learning_rate": 8.849738987892385e-06,
+      "loss": 0.8438,
+      "step": 20048
+    },
+    {
+      "epoch": 0.5506303040290023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.848855380621085e-06,
+      "loss": 0.9373,
+      "step": 20049
+    },
+    {
+      "epoch": 0.5506577682568454,
+      "grad_norm": 0.0,
+      "learning_rate": 8.847971782458087e-06,
+      "loss": 0.8993,
+      "step": 20050
+    },
+    {
+      "epoch": 0.5506852324846887,
+      "grad_norm": 0.0,
+      "learning_rate": 8.847088193410381e-06,
+      "loss": 0.8735,
+      "step": 20051
+    },
+    {
+      "epoch": 0.550712696712532,
+      "grad_norm": 0.0,
+      "learning_rate": 8.846204613484963e-06,
+      "loss": 0.911,
+      "step": 20052
+    },
+    {
+      "epoch": 0.5507401609403751,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84532104268882e-06,
+      "loss": 0.9009,
+      "step": 20053
+    },
+    {
+      "epoch": 0.5507676251682184,
+      "grad_norm": 0.0,
+      "learning_rate": 8.844437481028948e-06,
+      "loss": 0.8488,
+      "step": 20054
+    },
+    {
+      "epoch": 0.5507950893960616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.843553928512329e-06,
+      "loss": 0.8145,
+      "step": 20055
+    },
+    {
+      "epoch": 0.5508225536239049,
+      "grad_norm": 0.0,
+      "learning_rate": 8.842670385145961e-06,
+      "loss": 0.8876,
+      "step": 20056
+    },
+    {
+      "epoch": 0.5508500178517481,
+      "grad_norm": 0.0,
+      "learning_rate": 8.841786850936831e-06,
+      "loss": 0.8766,
+      "step": 20057
+    },
+    {
+      "epoch": 0.5508774820795913,
+      "grad_norm": 0.0,
+      "learning_rate": 8.840903325891934e-06,
+      "loss": 0.8995,
+      "step": 20058
+    },
+    {
+      "epoch": 0.5509049463074346,
+      "grad_norm": 0.0,
+      "learning_rate": 8.840019810018258e-06,
+      "loss": 0.8995,
+      "step": 20059
+    },
+    {
+      "epoch": 0.5509324105352778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.839136303322797e-06,
+      "loss": 0.858,
+      "step": 20060
+    },
+    {
+      "epoch": 0.550959874763121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.83825280581254e-06,
+      "loss": 0.8201,
+      "step": 20061
+    },
+    {
+      "epoch": 0.5509873389909643,
+      "grad_norm": 0.0,
+      "learning_rate": 8.837369317494472e-06,
+      "loss": 0.9008,
+      "step": 20062
+    },
+    {
+      "epoch": 0.5510148032188075,
+      "grad_norm": 0.0,
+      "learning_rate": 8.836485838375591e-06,
+      "loss": 0.9429,
+      "step": 20063
+    },
+    {
+      "epoch": 0.5510422674466507,
+      "grad_norm": 0.0,
+      "learning_rate": 8.835602368462886e-06,
+      "loss": 0.8181,
+      "step": 20064
+    },
+    {
+      "epoch": 0.551069731674494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.834718907763342e-06,
+      "loss": 0.8514,
+      "step": 20065
+    },
+    {
+      "epoch": 0.5510971959023372,
+      "grad_norm": 0.0,
+      "learning_rate": 8.833835456283961e-06,
+      "loss": 0.872,
+      "step": 20066
+    },
+    {
+      "epoch": 0.5511246601301805,
+      "grad_norm": 0.0,
+      "learning_rate": 8.832952014031718e-06,
+      "loss": 0.9219,
+      "step": 20067
+    },
+    {
+      "epoch": 0.5511521243580236,
+      "grad_norm": 0.0,
+      "learning_rate": 8.832068581013616e-06,
+      "loss": 0.8587,
+      "step": 20068
+    },
+    {
+      "epoch": 0.5511795885858669,
+      "grad_norm": 0.0,
+      "learning_rate": 8.831185157236635e-06,
+      "loss": 0.8156,
+      "step": 20069
+    },
+    {
+      "epoch": 0.5512070528137102,
+      "grad_norm": 0.0,
+      "learning_rate": 8.83030174270777e-06,
+      "loss": 0.8566,
+      "step": 20070
+    },
+    {
+      "epoch": 0.5512345170415534,
+      "grad_norm": 0.0,
+      "learning_rate": 8.829418337434014e-06,
+      "loss": 0.9255,
+      "step": 20071
+    },
+    {
+      "epoch": 0.5512619812693966,
+      "grad_norm": 0.0,
+      "learning_rate": 8.828534941422353e-06,
+      "loss": 0.9286,
+      "step": 20072
+    },
+    {
+      "epoch": 0.5512894454972398,
+      "grad_norm": 0.0,
+      "learning_rate": 8.827651554679777e-06,
+      "loss": 0.7726,
+      "step": 20073
+    },
+    {
+      "epoch": 0.5513169097250831,
+      "grad_norm": 0.0,
+      "learning_rate": 8.826768177213276e-06,
+      "loss": 0.8718,
+      "step": 20074
+    },
+    {
+      "epoch": 0.5513443739529263,
+      "grad_norm": 0.0,
+      "learning_rate": 8.825884809029837e-06,
+      "loss": 0.9343,
+      "step": 20075
+    },
+    {
+      "epoch": 0.5513718381807695,
+      "grad_norm": 0.0,
+      "learning_rate": 8.825001450136456e-06,
+      "loss": 0.8581,
+      "step": 20076
+    },
+    {
+      "epoch": 0.5513993024086128,
+      "grad_norm": 0.0,
+      "learning_rate": 8.824118100540117e-06,
+      "loss": 1.0109,
+      "step": 20077
+    },
+    {
+      "epoch": 0.5514267666364561,
+      "grad_norm": 0.0,
+      "learning_rate": 8.823234760247813e-06,
+      "loss": 0.8503,
+      "step": 20078
+    },
+    {
+      "epoch": 0.5514542308642992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.82235142926653e-06,
+      "loss": 0.8898,
+      "step": 20079
+    },
+    {
+      "epoch": 0.5514816950921425,
+      "grad_norm": 0.0,
+      "learning_rate": 8.821468107603258e-06,
+      "loss": 0.8948,
+      "step": 20080
+    },
+    {
+      "epoch": 0.5515091593199857,
+      "grad_norm": 0.0,
+      "learning_rate": 8.820584795264987e-06,
+      "loss": 0.813,
+      "step": 20081
+    },
+    {
+      "epoch": 0.551536623547829,
+      "grad_norm": 0.0,
+      "learning_rate": 8.819701492258704e-06,
+      "loss": 0.8352,
+      "step": 20082
+    },
+    {
+      "epoch": 0.5515640877756722,
+      "grad_norm": 0.0,
+      "learning_rate": 8.818818198591406e-06,
+      "loss": 0.8064,
+      "step": 20083
+    },
+    {
+      "epoch": 0.5515915520035154,
+      "grad_norm": 0.0,
+      "learning_rate": 8.817934914270074e-06,
+      "loss": 0.9446,
+      "step": 20084
+    },
+    {
+      "epoch": 0.5516190162313587,
+      "grad_norm": 0.0,
+      "learning_rate": 8.817051639301695e-06,
+      "loss": 0.7979,
+      "step": 20085
+    },
+    {
+      "epoch": 0.5516464804592018,
+      "grad_norm": 0.0,
+      "learning_rate": 8.816168373693262e-06,
+      "loss": 0.9294,
+      "step": 20086
+    },
+    {
+      "epoch": 0.5516739446870451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.815285117451764e-06,
+      "loss": 0.8451,
+      "step": 20087
+    },
+    {
+      "epoch": 0.5517014089148884,
+      "grad_norm": 0.0,
+      "learning_rate": 8.814401870584191e-06,
+      "loss": 0.8335,
+      "step": 20088
+    },
+    {
+      "epoch": 0.5517288731427316,
+      "grad_norm": 0.0,
+      "learning_rate": 8.813518633097531e-06,
+      "loss": 0.9446,
+      "step": 20089
+    },
+    {
+      "epoch": 0.5517563373705748,
+      "grad_norm": 0.0,
+      "learning_rate": 8.812635404998768e-06,
+      "loss": 0.9161,
+      "step": 20090
+    },
+    {
+      "epoch": 0.5517838015984181,
+      "grad_norm": 0.0,
+      "learning_rate": 8.811752186294895e-06,
+      "loss": 0.8907,
+      "step": 20091
+    },
+    {
+      "epoch": 0.5518112658262613,
+      "grad_norm": 0.0,
+      "learning_rate": 8.810868976992899e-06,
+      "loss": 0.8795,
+      "step": 20092
+    },
+    {
+      "epoch": 0.5518387300541046,
+      "grad_norm": 0.0,
+      "learning_rate": 8.809985777099767e-06,
+      "loss": 0.8504,
+      "step": 20093
+    },
+    {
+      "epoch": 0.5518661942819477,
+      "grad_norm": 0.0,
+      "learning_rate": 8.809102586622487e-06,
+      "loss": 0.8616,
+      "step": 20094
+    },
+    {
+      "epoch": 0.551893658509791,
+      "grad_norm": 0.0,
+      "learning_rate": 8.808219405568054e-06,
+      "loss": 0.818,
+      "step": 20095
+    },
+    {
+      "epoch": 0.5519211227376343,
+      "grad_norm": 0.0,
+      "learning_rate": 8.807336233943449e-06,
+      "loss": 0.8864,
+      "step": 20096
+    },
+    {
+      "epoch": 0.5519485869654774,
+      "grad_norm": 0.0,
+      "learning_rate": 8.80645307175566e-06,
+      "loss": 0.9814,
+      "step": 20097
+    },
+    {
+      "epoch": 0.5519760511933207,
+      "grad_norm": 0.0,
+      "learning_rate": 8.805569919011675e-06,
+      "loss": 0.8533,
+      "step": 20098
+    },
+    {
+      "epoch": 0.5520035154211639,
+      "grad_norm": 0.0,
+      "learning_rate": 8.804686775718484e-06,
+      "loss": 0.8967,
+      "step": 20099
+    },
+    {
+      "epoch": 0.5520309796490072,
+      "grad_norm": 0.0,
+      "learning_rate": 8.803803641883077e-06,
+      "loss": 0.8634,
+      "step": 20100
+    },
+    {
+      "epoch": 0.5520584438768504,
+      "grad_norm": 0.0,
+      "learning_rate": 8.802920517512441e-06,
+      "loss": 0.9271,
+      "step": 20101
+    },
+    {
+      "epoch": 0.5520859081046936,
+      "grad_norm": 0.0,
+      "learning_rate": 8.802037402613556e-06,
+      "loss": 0.7767,
+      "step": 20102
+    },
+    {
+      "epoch": 0.5521133723325369,
+      "grad_norm": 0.0,
+      "learning_rate": 8.801154297193418e-06,
+      "loss": 0.8724,
+      "step": 20103
+    },
+    {
+      "epoch": 0.5521408365603802,
+      "grad_norm": 0.0,
+      "learning_rate": 8.800271201259011e-06,
+      "loss": 0.8648,
+      "step": 20104
+    },
+    {
+      "epoch": 0.5521683007882233,
+      "grad_norm": 0.0,
+      "learning_rate": 8.799388114817321e-06,
+      "loss": 0.876,
+      "step": 20105
+    },
+    {
+      "epoch": 0.5521957650160666,
+      "grad_norm": 0.0,
+      "learning_rate": 8.798505037875338e-06,
+      "loss": 0.942,
+      "step": 20106
+    },
+    {
+      "epoch": 0.5522232292439098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.797621970440053e-06,
+      "loss": 0.9986,
+      "step": 20107
+    },
+    {
+      "epoch": 0.552250693471753,
+      "grad_norm": 0.0,
+      "learning_rate": 8.796738912518447e-06,
+      "loss": 0.9526,
+      "step": 20108
+    },
+    {
+      "epoch": 0.5522781576995963,
+      "grad_norm": 0.0,
+      "learning_rate": 8.795855864117505e-06,
+      "loss": 0.8785,
+      "step": 20109
+    },
+    {
+      "epoch": 0.5523056219274395,
+      "grad_norm": 0.0,
+      "learning_rate": 8.79497282524422e-06,
+      "loss": 0.8355,
+      "step": 20110
+    },
+    {
+      "epoch": 0.5523330861552828,
+      "grad_norm": 0.0,
+      "learning_rate": 8.794089795905575e-06,
+      "loss": 0.8789,
+      "step": 20111
+    },
+    {
+      "epoch": 0.5523605503831259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.793206776108563e-06,
+      "loss": 0.952,
+      "step": 20112
+    },
+    {
+      "epoch": 0.5523880146109692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.792323765860167e-06,
+      "loss": 0.8752,
+      "step": 20113
+    },
+    {
+      "epoch": 0.5524154788388125,
+      "grad_norm": 0.0,
+      "learning_rate": 8.79144076516737e-06,
+      "loss": 0.8348,
+      "step": 20114
+    },
+    {
+      "epoch": 0.5524429430666556,
+      "grad_norm": 0.0,
+      "learning_rate": 8.79055777403716e-06,
+      "loss": 0.8464,
+      "step": 20115
+    },
+    {
+      "epoch": 0.5524704072944989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.78967479247653e-06,
+      "loss": 0.9049,
+      "step": 20116
+    },
+    {
+      "epoch": 0.5524978715223422,
+      "grad_norm": 0.0,
+      "learning_rate": 8.788791820492457e-06,
+      "loss": 0.9656,
+      "step": 20117
+    },
+    {
+      "epoch": 0.5525253357501854,
+      "grad_norm": 0.0,
+      "learning_rate": 8.787908858091934e-06,
+      "loss": 0.8452,
+      "step": 20118
+    },
+    {
+      "epoch": 0.5525527999780286,
+      "grad_norm": 0.0,
+      "learning_rate": 8.787025905281949e-06,
+      "loss": 0.9952,
+      "step": 20119
+    },
+    {
+      "epoch": 0.5525802642058718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.786142962069483e-06,
+      "loss": 0.808,
+      "step": 20120
+    },
+    {
+      "epoch": 0.5526077284337151,
+      "grad_norm": 0.0,
+      "learning_rate": 8.785260028461523e-06,
+      "loss": 0.8407,
+      "step": 20121
+    },
+    {
+      "epoch": 0.5526351926615584,
+      "grad_norm": 0.0,
+      "learning_rate": 8.784377104465053e-06,
+      "loss": 0.778,
+      "step": 20122
+    },
+    {
+      "epoch": 0.5526626568894015,
+      "grad_norm": 0.0,
+      "learning_rate": 8.783494190087067e-06,
+      "loss": 0.9985,
+      "step": 20123
+    },
+    {
+      "epoch": 0.5526901211172448,
+      "grad_norm": 0.0,
+      "learning_rate": 8.782611285334544e-06,
+      "loss": 0.9652,
+      "step": 20124
+    },
+    {
+      "epoch": 0.552717585345088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.781728390214476e-06,
+      "loss": 0.9027,
+      "step": 20125
+    },
+    {
+      "epoch": 0.5527450495729312,
+      "grad_norm": 0.0,
+      "learning_rate": 8.780845504733842e-06,
+      "loss": 0.9024,
+      "step": 20126
+    },
+    {
+      "epoch": 0.5527725138007745,
+      "grad_norm": 0.0,
+      "learning_rate": 8.779962628899629e-06,
+      "loss": 0.9022,
+      "step": 20127
+    },
+    {
+      "epoch": 0.5527999780286177,
+      "grad_norm": 0.0,
+      "learning_rate": 8.779079762718826e-06,
+      "loss": 0.8815,
+      "step": 20128
+    },
+    {
+      "epoch": 0.552827442256461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.778196906198416e-06,
+      "loss": 1.0436,
+      "step": 20129
+    },
+    {
+      "epoch": 0.5528549064843042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.777314059345385e-06,
+      "loss": 0.926,
+      "step": 20130
+    },
+    {
+      "epoch": 0.5528823707121474,
+      "grad_norm": 0.0,
+      "learning_rate": 8.776431222166723e-06,
+      "loss": 0.9393,
+      "step": 20131
+    },
+    {
+      "epoch": 0.5529098349399907,
+      "grad_norm": 0.0,
+      "learning_rate": 8.775548394669409e-06,
+      "loss": 0.8881,
+      "step": 20132
+    },
+    {
+      "epoch": 0.5529372991678339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.774665576860428e-06,
+      "loss": 0.9283,
+      "step": 20133
+    },
+    {
+      "epoch": 0.5529647633956771,
+      "grad_norm": 0.0,
+      "learning_rate": 8.773782768746769e-06,
+      "loss": 0.8849,
+      "step": 20134
+    },
+    {
+      "epoch": 0.5529922276235204,
+      "grad_norm": 0.0,
+      "learning_rate": 8.772899970335416e-06,
+      "loss": 0.9106,
+      "step": 20135
+    },
+    {
+      "epoch": 0.5530196918513636,
+      "grad_norm": 0.0,
+      "learning_rate": 8.77201718163335e-06,
+      "loss": 0.8796,
+      "step": 20136
+    },
+    {
+      "epoch": 0.5530471560792068,
+      "grad_norm": 0.0,
+      "learning_rate": 8.771134402647568e-06,
+      "loss": 0.8779,
+      "step": 20137
+    },
+    {
+      "epoch": 0.55307462030705,
+      "grad_norm": 0.0,
+      "learning_rate": 8.77025163338504e-06,
+      "loss": 0.9483,
+      "step": 20138
+    },
+    {
+      "epoch": 0.5531020845348933,
+      "grad_norm": 0.0,
+      "learning_rate": 8.769368873852758e-06,
+      "loss": 0.8002,
+      "step": 20139
+    },
+    {
+      "epoch": 0.5531295487627366,
+      "grad_norm": 0.0,
+      "learning_rate": 8.768486124057708e-06,
+      "loss": 0.8123,
+      "step": 20140
+    },
+    {
+      "epoch": 0.5531570129905797,
+      "grad_norm": 0.0,
+      "learning_rate": 8.767603384006871e-06,
+      "loss": 0.9327,
+      "step": 20141
+    },
+    {
+      "epoch": 0.553184477218423,
+      "grad_norm": 0.0,
+      "learning_rate": 8.766720653707238e-06,
+      "loss": 0.8442,
+      "step": 20142
+    },
+    {
+      "epoch": 0.5532119414462663,
+      "grad_norm": 0.0,
+      "learning_rate": 8.765837933165786e-06,
+      "loss": 0.9382,
+      "step": 20143
+    },
+    {
+      "epoch": 0.5532394056741095,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764955222389502e-06,
+      "loss": 0.7746,
+      "step": 20144
+    },
+    {
+      "epoch": 0.5532668699019527,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764072521385369e-06,
+      "loss": 0.8982,
+      "step": 20145
+    },
+    {
+      "epoch": 0.5532943341297959,
+      "grad_norm": 0.0,
+      "learning_rate": 8.763189830160374e-06,
+      "loss": 0.8814,
+      "step": 20146
+    },
+    {
+      "epoch": 0.5533217983576392,
+      "grad_norm": 0.0,
+      "learning_rate": 8.762307148721501e-06,
+      "loss": 0.8371,
+      "step": 20147
+    },
+    {
+      "epoch": 0.5533492625854824,
+      "grad_norm": 0.0,
+      "learning_rate": 8.761424477075733e-06,
+      "loss": 0.784,
+      "step": 20148
+    },
+    {
+      "epoch": 0.5533767268133256,
+      "grad_norm": 0.0,
+      "learning_rate": 8.760541815230056e-06,
+      "loss": 0.8611,
+      "step": 20149
+    },
+    {
+      "epoch": 0.5534041910411689,
+      "grad_norm": 0.0,
+      "learning_rate": 8.75965916319145e-06,
+      "loss": 0.8497,
+      "step": 20150
+    },
+    {
+      "epoch": 0.5534316552690121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.7587765209669e-06,
+      "loss": 0.9365,
+      "step": 20151
+    },
+    {
+      "epoch": 0.5534591194968553,
+      "grad_norm": 0.0,
+      "learning_rate": 8.757893888563393e-06,
+      "loss": 0.8808,
+      "step": 20152
+    },
+    {
+      "epoch": 0.5534865837246986,
+      "grad_norm": 0.0,
+      "learning_rate": 8.75701126598791e-06,
+      "loss": 0.8409,
+      "step": 20153
+    },
+    {
+      "epoch": 0.5535140479525418,
+      "grad_norm": 0.0,
+      "learning_rate": 8.756128653247439e-06,
+      "loss": 0.8324,
+      "step": 20154
+    },
+    {
+      "epoch": 0.553541512180385,
+      "grad_norm": 0.0,
+      "learning_rate": 8.755246050348956e-06,
+      "loss": 0.8255,
+      "step": 20155
+    },
+    {
+      "epoch": 0.5535689764082283,
+      "grad_norm": 0.0,
+      "learning_rate": 8.75436345729945e-06,
+      "loss": 0.8344,
+      "step": 20156
+    },
+    {
+      "epoch": 0.5535964406360715,
+      "grad_norm": 0.0,
+      "learning_rate": 8.7534808741059e-06,
+      "loss": 0.8656,
+      "step": 20157
+    },
+    {
+      "epoch": 0.5536239048639148,
+      "grad_norm": 0.0,
+      "learning_rate": 8.752598300775294e-06,
+      "loss": 0.925,
+      "step": 20158
+    },
+    {
+      "epoch": 0.5536513690917579,
+      "grad_norm": 0.0,
+      "learning_rate": 8.751715737314614e-06,
+      "loss": 0.883,
+      "step": 20159
+    },
+    {
+      "epoch": 0.5536788333196012,
+      "grad_norm": 0.0,
+      "learning_rate": 8.750833183730843e-06,
+      "loss": 0.833,
+      "step": 20160
+    },
+    {
+      "epoch": 0.5537062975474445,
+      "grad_norm": 0.0,
+      "learning_rate": 8.749950640030964e-06,
+      "loss": 0.842,
+      "step": 20161
+    },
+    {
+      "epoch": 0.5537337617752877,
+      "grad_norm": 0.0,
+      "learning_rate": 8.749068106221958e-06,
+      "loss": 0.859,
+      "step": 20162
+    },
+    {
+      "epoch": 0.5537612260031309,
+      "grad_norm": 0.0,
+      "learning_rate": 8.748185582310809e-06,
+      "loss": 0.8586,
+      "step": 20163
+    },
+    {
+      "epoch": 0.5537886902309741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.747303068304503e-06,
+      "loss": 0.8739,
+      "step": 20164
+    },
+    {
+      "epoch": 0.5538161544588174,
+      "grad_norm": 0.0,
+      "learning_rate": 8.746420564210018e-06,
+      "loss": 0.894,
+      "step": 20165
+    },
+    {
+      "epoch": 0.5538436186866607,
+      "grad_norm": 0.0,
+      "learning_rate": 8.745538070034343e-06,
+      "loss": 0.8684,
+      "step": 20166
+    },
+    {
+      "epoch": 0.5538710829145038,
+      "grad_norm": 0.0,
+      "learning_rate": 8.744655585784452e-06,
+      "loss": 0.8568,
+      "step": 20167
+    },
+    {
+      "epoch": 0.5538985471423471,
+      "grad_norm": 0.0,
+      "learning_rate": 8.743773111467336e-06,
+      "loss": 0.7866,
+      "step": 20168
+    },
+    {
+      "epoch": 0.5539260113701904,
+      "grad_norm": 0.0,
+      "learning_rate": 8.742890647089971e-06,
+      "loss": 0.8851,
+      "step": 20169
+    },
+    {
+      "epoch": 0.5539534755980335,
+      "grad_norm": 0.0,
+      "learning_rate": 8.742008192659342e-06,
+      "loss": 0.8252,
+      "step": 20170
+    },
+    {
+      "epoch": 0.5539809398258768,
+      "grad_norm": 0.0,
+      "learning_rate": 8.741125748182434e-06,
+      "loss": 0.9668,
+      "step": 20171
+    },
+    {
+      "epoch": 0.55400840405372,
+      "grad_norm": 0.0,
+      "learning_rate": 8.740243313666227e-06,
+      "loss": 0.9244,
+      "step": 20172
+    },
+    {
+      "epoch": 0.5540358682815633,
+      "grad_norm": 0.0,
+      "learning_rate": 8.739360889117702e-06,
+      "loss": 0.9467,
+      "step": 20173
+    },
+    {
+      "epoch": 0.5540633325094065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.73847847454384e-06,
+      "loss": 0.9375,
+      "step": 20174
+    },
+    {
+      "epoch": 0.5540907967372497,
+      "grad_norm": 0.0,
+      "learning_rate": 8.737596069951626e-06,
+      "loss": 1.0272,
+      "step": 20175
+    },
+    {
+      "epoch": 0.554118260965093,
+      "grad_norm": 0.0,
+      "learning_rate": 8.736713675348042e-06,
+      "loss": 0.9161,
+      "step": 20176
+    },
+    {
+      "epoch": 0.5541457251929363,
+      "grad_norm": 0.0,
+      "learning_rate": 8.735831290740066e-06,
+      "loss": 0.8679,
+      "step": 20177
+    },
+    {
+      "epoch": 0.5541731894207794,
+      "grad_norm": 0.0,
+      "learning_rate": 8.734948916134688e-06,
+      "loss": 0.8234,
+      "step": 20178
+    },
+    {
+      "epoch": 0.5542006536486227,
+      "grad_norm": 0.0,
+      "learning_rate": 8.73406655153888e-06,
+      "loss": 0.9161,
+      "step": 20179
+    },
+    {
+      "epoch": 0.5542281178764659,
+      "grad_norm": 0.0,
+      "learning_rate": 8.73318419695963e-06,
+      "loss": 0.8525,
+      "step": 20180
+    },
+    {
+      "epoch": 0.5542555821043091,
+      "grad_norm": 0.0,
+      "learning_rate": 8.732301852403914e-06,
+      "loss": 0.9614,
+      "step": 20181
+    },
+    {
+      "epoch": 0.5542830463321524,
+      "grad_norm": 0.0,
+      "learning_rate": 8.731419517878718e-06,
+      "loss": 0.879,
+      "step": 20182
+    },
+    {
+      "epoch": 0.5543105105599956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.730537193391024e-06,
+      "loss": 0.8847,
+      "step": 20183
+    },
+    {
+      "epoch": 0.5543379747878389,
+      "grad_norm": 0.0,
+      "learning_rate": 8.72965487894781e-06,
+      "loss": 0.9399,
+      "step": 20184
+    },
+    {
+      "epoch": 0.554365439015682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.728772574556062e-06,
+      "loss": 0.8203,
+      "step": 20185
+    },
+    {
+      "epoch": 0.5543929032435253,
+      "grad_norm": 0.0,
+      "learning_rate": 8.727890280222752e-06,
+      "loss": 0.9219,
+      "step": 20186
+    },
+    {
+      "epoch": 0.5544203674713686,
+      "grad_norm": 0.0,
+      "learning_rate": 8.727007995954869e-06,
+      "loss": 0.8155,
+      "step": 20187
+    },
+    {
+      "epoch": 0.5544478316992117,
+      "grad_norm": 0.0,
+      "learning_rate": 8.726125721759393e-06,
+      "loss": 0.8517,
+      "step": 20188
+    },
+    {
+      "epoch": 0.554475295927055,
+      "grad_norm": 0.0,
+      "learning_rate": 8.725243457643302e-06,
+      "loss": 0.9034,
+      "step": 20189
+    },
+    {
+      "epoch": 0.5545027601548983,
+      "grad_norm": 0.0,
+      "learning_rate": 8.724361203613584e-06,
+      "loss": 0.7645,
+      "step": 20190
+    },
+    {
+      "epoch": 0.5545302243827415,
+      "grad_norm": 0.0,
+      "learning_rate": 8.723478959677208e-06,
+      "loss": 0.8681,
+      "step": 20191
+    },
+    {
+      "epoch": 0.5545576886105847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.722596725841165e-06,
+      "loss": 0.9147,
+      "step": 20192
+    },
+    {
+      "epoch": 0.5545851528384279,
+      "grad_norm": 0.0,
+      "learning_rate": 8.721714502112429e-06,
+      "loss": 0.8948,
+      "step": 20193
+    },
+    {
+      "epoch": 0.5546126170662712,
+      "grad_norm": 0.0,
+      "learning_rate": 8.720832288497982e-06,
+      "loss": 0.9712,
+      "step": 20194
+    },
+    {
+      "epoch": 0.5546400812941145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.719950085004807e-06,
+      "loss": 0.8972,
+      "step": 20195
+    },
+    {
+      "epoch": 0.5546675455219576,
+      "grad_norm": 0.0,
+      "learning_rate": 8.719067891639885e-06,
+      "loss": 0.8752,
+      "step": 20196
+    },
+    {
+      "epoch": 0.5546950097498009,
+      "grad_norm": 0.0,
+      "learning_rate": 8.718185708410193e-06,
+      "loss": 0.869,
+      "step": 20197
+    },
+    {
+      "epoch": 0.5547224739776441,
+      "grad_norm": 0.0,
+      "learning_rate": 8.71730353532271e-06,
+      "loss": 0.9243,
+      "step": 20198
+    },
+    {
+      "epoch": 0.5547499382054873,
+      "grad_norm": 0.0,
+      "learning_rate": 8.716421372384418e-06,
+      "loss": 0.8195,
+      "step": 20199
+    },
+    {
+      "epoch": 0.5547774024333306,
+      "grad_norm": 0.0,
+      "learning_rate": 8.715539219602299e-06,
+      "loss": 0.7315,
+      "step": 20200
+    },
+    {
+      "epoch": 0.5548048666611738,
+      "grad_norm": 0.0,
+      "learning_rate": 8.714657076983335e-06,
+      "loss": 0.8269,
+      "step": 20201
+    },
+    {
+      "epoch": 0.5548323308890171,
+      "grad_norm": 0.0,
+      "learning_rate": 8.713774944534496e-06,
+      "loss": 0.8615,
+      "step": 20202
+    },
+    {
+      "epoch": 0.5548597951168603,
+      "grad_norm": 0.0,
+      "learning_rate": 8.712892822262769e-06,
+      "loss": 0.918,
+      "step": 20203
+    },
+    {
+      "epoch": 0.5548872593447035,
+      "grad_norm": 0.0,
+      "learning_rate": 8.712010710175133e-06,
+      "loss": 0.742,
+      "step": 20204
+    },
+    {
+      "epoch": 0.5549147235725468,
+      "grad_norm": 0.0,
+      "learning_rate": 8.711128608278568e-06,
+      "loss": 0.885,
+      "step": 20205
+    },
+    {
+      "epoch": 0.55494218780039,
+      "grad_norm": 0.0,
+      "learning_rate": 8.71024651658005e-06,
+      "loss": 0.9091,
+      "step": 20206
+    },
+    {
+      "epoch": 0.5549696520282332,
+      "grad_norm": 0.0,
+      "learning_rate": 8.709364435086567e-06,
+      "loss": 0.8936,
+      "step": 20207
+    },
+    {
+      "epoch": 0.5549971162560765,
+      "grad_norm": 0.0,
+      "learning_rate": 8.708482363805088e-06,
+      "loss": 0.9269,
+      "step": 20208
+    },
+    {
+      "epoch": 0.5550245804839197,
+      "grad_norm": 0.0,
+      "learning_rate": 8.707600302742599e-06,
+      "loss": 0.8738,
+      "step": 20209
+    },
+    {
+      "epoch": 0.555052044711763,
+      "grad_norm": 0.0,
+      "learning_rate": 8.706718251906076e-06,
+      "loss": 0.8845,
+      "step": 20210
+    },
+    {
+      "epoch": 0.5550795089396061,
+      "grad_norm": 0.0,
+      "learning_rate": 8.705836211302498e-06,
+      "loss": 0.9292,
+      "step": 20211
+    },
+    {
+      "epoch": 0.5551069731674494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.704954180938848e-06,
+      "loss": 0.8262,
+      "step": 20212
+    },
+    {
+      "epoch": 0.5551344373952927,
+      "grad_norm": 0.0,
+      "learning_rate": 8.704072160822102e-06,
+      "loss": 0.8502,
+      "step": 20213
+    },
+    {
+      "epoch": 0.5551619016231358,
+      "grad_norm": 0.0,
+      "learning_rate": 8.703190150959236e-06,
+      "loss": 0.8311,
+      "step": 20214
+    },
+    {
+      "epoch": 0.5551893658509791,
+      "grad_norm": 0.0,
+      "learning_rate": 8.702308151357232e-06,
+      "loss": 0.9308,
+      "step": 20215
+    },
+    {
+      "epoch": 0.5552168300788224,
+      "grad_norm": 0.0,
+      "learning_rate": 8.701426162023072e-06,
+      "loss": 0.8676,
+      "step": 20216
+    },
+    {
+      "epoch": 0.5552442943066656,
+      "grad_norm": 0.0,
+      "learning_rate": 8.700544182963729e-06,
+      "loss": 0.9304,
+      "step": 20217
+    },
+    {
+      "epoch": 0.5552717585345088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.699662214186182e-06,
+      "loss": 0.957,
+      "step": 20218
+    },
+    {
+      "epoch": 0.555299222762352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.698780255697416e-06,
+      "loss": 0.9916,
+      "step": 20219
+    },
+    {
+      "epoch": 0.5553266869901953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.697898307504401e-06,
+      "loss": 0.8908,
+      "step": 20220
+    },
+    {
+      "epoch": 0.5553541512180385,
+      "grad_norm": 0.0,
+      "learning_rate": 8.697016369614121e-06,
+      "loss": 0.923,
+      "step": 20221
+    },
+    {
+      "epoch": 0.5553816154458817,
+      "grad_norm": 0.0,
+      "learning_rate": 8.69613444203355e-06,
+      "loss": 0.8585,
+      "step": 20222
+    },
+    {
+      "epoch": 0.555409079673725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.695252524769669e-06,
+      "loss": 0.9098,
+      "step": 20223
+    },
+    {
+      "epoch": 0.5554365439015682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.694370617829456e-06,
+      "loss": 0.9224,
+      "step": 20224
+    },
+    {
+      "epoch": 0.5554640081294114,
+      "grad_norm": 0.0,
+      "learning_rate": 8.693488721219892e-06,
+      "loss": 0.7892,
+      "step": 20225
+    },
+    {
+      "epoch": 0.5554914723572547,
+      "grad_norm": 0.0,
+      "learning_rate": 8.692606834947946e-06,
+      "loss": 0.8755,
+      "step": 20226
+    },
+    {
+      "epoch": 0.5555189365850979,
+      "grad_norm": 0.0,
+      "learning_rate": 8.6917249590206e-06,
+      "loss": 0.9316,
+      "step": 20227
+    },
+    {
+      "epoch": 0.5555464008129412,
+      "grad_norm": 0.0,
+      "learning_rate": 8.690843093444838e-06,
+      "loss": 0.8078,
+      "step": 20228
+    },
+    {
+      "epoch": 0.5555738650407844,
+      "grad_norm": 0.0,
+      "learning_rate": 8.689961238227628e-06,
+      "loss": 0.8572,
+      "step": 20229
+    },
+    {
+      "epoch": 0.5556013292686276,
+      "grad_norm": 0.0,
+      "learning_rate": 8.689079393375954e-06,
+      "loss": 0.8486,
+      "step": 20230
+    },
+    {
+      "epoch": 0.5556287934964709,
+      "grad_norm": 0.0,
+      "learning_rate": 8.688197558896795e-06,
+      "loss": 0.8709,
+      "step": 20231
+    },
+    {
+      "epoch": 0.555656257724314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.687315734797121e-06,
+      "loss": 0.9302,
+      "step": 20232
+    },
+    {
+      "epoch": 0.5556837219521573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.686433921083917e-06,
+      "loss": 0.9104,
+      "step": 20233
+    },
+    {
+      "epoch": 0.5557111861800006,
+      "grad_norm": 0.0,
+      "learning_rate": 8.685552117764152e-06,
+      "loss": 0.9198,
+      "step": 20234
+    },
+    {
+      "epoch": 0.5557386504078438,
+      "grad_norm": 0.0,
+      "learning_rate": 8.684670324844809e-06,
+      "loss": 0.8953,
+      "step": 20235
+    },
+    {
+      "epoch": 0.555766114635687,
+      "grad_norm": 0.0,
+      "learning_rate": 8.683788542332865e-06,
+      "loss": 0.8524,
+      "step": 20236
+    },
+    {
+      "epoch": 0.5557935788635302,
+      "grad_norm": 0.0,
+      "learning_rate": 8.6829067702353e-06,
+      "loss": 0.9534,
+      "step": 20237
+    },
+    {
+      "epoch": 0.5558210430913735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.68202500855908e-06,
+      "loss": 0.9181,
+      "step": 20238
+    },
+    {
+      "epoch": 0.5558485073192168,
+      "grad_norm": 0.0,
+      "learning_rate": 8.681143257311191e-06,
+      "loss": 0.9103,
+      "step": 20239
+    },
+    {
+      "epoch": 0.5558759715470599,
+      "grad_norm": 0.0,
+      "learning_rate": 8.68026151649861e-06,
+      "loss": 0.8743,
+      "step": 20240
+    },
+    {
+      "epoch": 0.5559034357749032,
+      "grad_norm": 0.0,
+      "learning_rate": 8.679379786128306e-06,
+      "loss": 0.9658,
+      "step": 20241
+    },
+    {
+      "epoch": 0.5559309000027465,
+      "grad_norm": 0.0,
+      "learning_rate": 8.678498066207263e-06,
+      "loss": 0.798,
+      "step": 20242
+    },
+    {
+      "epoch": 0.5559583642305896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67761635674246e-06,
+      "loss": 0.9757,
+      "step": 20243
+    },
+    {
+      "epoch": 0.5559858284584329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.676734657740864e-06,
+      "loss": 0.8261,
+      "step": 20244
+    },
+    {
+      "epoch": 0.5560132926862761,
+      "grad_norm": 0.0,
+      "learning_rate": 8.675852969209455e-06,
+      "loss": 0.904,
+      "step": 20245
+    },
+    {
+      "epoch": 0.5560407569141194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67497129115521e-06,
+      "loss": 0.8697,
+      "step": 20246
+    },
+    {
+      "epoch": 0.5560682211419626,
+      "grad_norm": 0.0,
+      "learning_rate": 8.674089623585106e-06,
+      "loss": 0.8874,
+      "step": 20247
+    },
+    {
+      "epoch": 0.5560956853698058,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67320796650612e-06,
+      "loss": 0.7731,
+      "step": 20248
+    },
+    {
+      "epoch": 0.5561231495976491,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67232631992523e-06,
+      "loss": 0.8783,
+      "step": 20249
+    },
+    {
+      "epoch": 0.5561506138254922,
+      "grad_norm": 0.0,
+      "learning_rate": 8.671444683849402e-06,
+      "loss": 0.8926,
+      "step": 20250
+    },
+    {
+      "epoch": 0.5561780780533355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67056305828562e-06,
+      "loss": 0.9379,
+      "step": 20251
+    },
+    {
+      "epoch": 0.5562055422811788,
+      "grad_norm": 0.0,
+      "learning_rate": 8.66968144324086e-06,
+      "loss": 0.8701,
+      "step": 20252
+    },
+    {
+      "epoch": 0.556233006509022,
+      "grad_norm": 0.0,
+      "learning_rate": 8.668799838722093e-06,
+      "loss": 0.9338,
+      "step": 20253
+    },
+    {
+      "epoch": 0.5562604707368652,
+      "grad_norm": 0.0,
+      "learning_rate": 8.667918244736297e-06,
+      "loss": 0.9798,
+      "step": 20254
+    },
+    {
+      "epoch": 0.5562879349647085,
+      "grad_norm": 0.0,
+      "learning_rate": 8.667036661290454e-06,
+      "loss": 0.8625,
+      "step": 20255
+    },
+    {
+      "epoch": 0.5563153991925517,
+      "grad_norm": 0.0,
+      "learning_rate": 8.66615508839153e-06,
+      "loss": 0.7834,
+      "step": 20256
+    },
+    {
+      "epoch": 0.556342863420395,
+      "grad_norm": 0.0,
+      "learning_rate": 8.665273526046501e-06,
+      "loss": 0.886,
+      "step": 20257
+    },
+    {
+      "epoch": 0.5563703276482381,
+      "grad_norm": 0.0,
+      "learning_rate": 8.664391974262344e-06,
+      "loss": 0.8835,
+      "step": 20258
+    },
+    {
+      "epoch": 0.5563977918760814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.663510433046038e-06,
+      "loss": 0.9394,
+      "step": 20259
+    },
+    {
+      "epoch": 0.5564252561039247,
+      "grad_norm": 0.0,
+      "learning_rate": 8.662628902404557e-06,
+      "loss": 0.9421,
+      "step": 20260
+    },
+    {
+      "epoch": 0.5564527203317678,
+      "grad_norm": 0.0,
+      "learning_rate": 8.661747382344874e-06,
+      "loss": 0.8137,
+      "step": 20261
+    },
+    {
+      "epoch": 0.5564801845596111,
+      "grad_norm": 0.0,
+      "learning_rate": 8.66086587287396e-06,
+      "loss": 0.827,
+      "step": 20262
+    },
+    {
+      "epoch": 0.5565076487874543,
+      "grad_norm": 0.0,
+      "learning_rate": 8.659984373998795e-06,
+      "loss": 0.8574,
+      "step": 20263
+    },
+    {
+      "epoch": 0.5565351130152976,
+      "grad_norm": 0.0,
+      "learning_rate": 8.659102885726354e-06,
+      "loss": 0.8172,
+      "step": 20264
+    },
+    {
+      "epoch": 0.5565625772431408,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65822140806361e-06,
+      "loss": 0.8662,
+      "step": 20265
+    },
+    {
+      "epoch": 0.556590041470984,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65733994101754e-06,
+      "loss": 0.972,
+      "step": 20266
+    },
+    {
+      "epoch": 0.5566175056988273,
+      "grad_norm": 0.0,
+      "learning_rate": 8.656458484595114e-06,
+      "loss": 0.8948,
+      "step": 20267
+    },
+    {
+      "epoch": 0.5566449699266706,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65557703880331e-06,
+      "loss": 0.9507,
+      "step": 20268
+    },
+    {
+      "epoch": 0.5566724341545137,
+      "grad_norm": 0.0,
+      "learning_rate": 8.654695603649099e-06,
+      "loss": 0.872,
+      "step": 20269
+    },
+    {
+      "epoch": 0.556699898382357,
+      "grad_norm": 0.0,
+      "learning_rate": 8.653814179139459e-06,
+      "loss": 0.9381,
+      "step": 20270
+    },
+    {
+      "epoch": 0.5567273626102002,
+      "grad_norm": 0.0,
+      "learning_rate": 8.652932765281361e-06,
+      "loss": 0.8834,
+      "step": 20271
+    },
+    {
+      "epoch": 0.5567548268380434,
+      "grad_norm": 0.0,
+      "learning_rate": 8.652051362081786e-06,
+      "loss": 0.8755,
+      "step": 20272
+    },
+    {
+      "epoch": 0.5567822910658867,
+      "grad_norm": 0.0,
+      "learning_rate": 8.651169969547701e-06,
+      "loss": 0.8819,
+      "step": 20273
+    },
+    {
+      "epoch": 0.5568097552937299,
+      "grad_norm": 0.0,
+      "learning_rate": 8.650288587686078e-06,
+      "loss": 0.9381,
+      "step": 20274
+    },
+    {
+      "epoch": 0.5568372195215732,
+      "grad_norm": 0.0,
+      "learning_rate": 8.649407216503896e-06,
+      "loss": 1.0286,
+      "step": 20275
+    },
+    {
+      "epoch": 0.5568646837494163,
+      "grad_norm": 0.0,
+      "learning_rate": 8.648525856008127e-06,
+      "loss": 0.8443,
+      "step": 20276
+    },
+    {
+      "epoch": 0.5568921479772596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.647644506205746e-06,
+      "loss": 0.8744,
+      "step": 20277
+    },
+    {
+      "epoch": 0.5569196122051029,
+      "grad_norm": 0.0,
+      "learning_rate": 8.646763167103728e-06,
+      "loss": 0.874,
+      "step": 20278
+    },
+    {
+      "epoch": 0.556947076432946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.64588183870904e-06,
+      "loss": 0.8896,
+      "step": 20279
+    },
+    {
+      "epoch": 0.5569745406607893,
+      "grad_norm": 0.0,
+      "learning_rate": 8.645000521028661e-06,
+      "loss": 1.0294,
+      "step": 20280
+    },
+    {
+      "epoch": 0.5570020048886326,
+      "grad_norm": 0.0,
+      "learning_rate": 8.644119214069562e-06,
+      "loss": 0.8921,
+      "step": 20281
+    },
+    {
+      "epoch": 0.5570294691164758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.643237917838717e-06,
+      "loss": 0.8862,
+      "step": 20282
+    },
+    {
+      "epoch": 0.557056933344319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.642356632343099e-06,
+      "loss": 0.8611,
+      "step": 20283
+    },
+    {
+      "epoch": 0.5570843975721622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.641475357589684e-06,
+      "loss": 0.7673,
+      "step": 20284
+    },
+    {
+      "epoch": 0.5571118618000055,
+      "grad_norm": 0.0,
+      "learning_rate": 8.640594093585442e-06,
+      "loss": 0.8744,
+      "step": 20285
+    },
+    {
+      "epoch": 0.5571393260278488,
+      "grad_norm": 0.0,
+      "learning_rate": 8.639712840337342e-06,
+      "loss": 0.8691,
+      "step": 20286
+    },
+    {
+      "epoch": 0.5571667902556919,
+      "grad_norm": 0.0,
+      "learning_rate": 8.638831597852362e-06,
+      "loss": 0.8235,
+      "step": 20287
+    },
+    {
+      "epoch": 0.5571942544835352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.637950366137475e-06,
+      "loss": 0.9266,
+      "step": 20288
+    },
+    {
+      "epoch": 0.5572217187113784,
+      "grad_norm": 0.0,
+      "learning_rate": 8.63706914519965e-06,
+      "loss": 0.8866,
+      "step": 20289
+    },
+    {
+      "epoch": 0.5572491829392217,
+      "grad_norm": 0.0,
+      "learning_rate": 8.636187935045869e-06,
+      "loss": 0.8878,
+      "step": 20290
+    },
+    {
+      "epoch": 0.5572766471670649,
+      "grad_norm": 0.0,
+      "learning_rate": 8.63530673568309e-06,
+      "loss": 0.8968,
+      "step": 20291
+    },
+    {
+      "epoch": 0.5573041113949081,
+      "grad_norm": 0.0,
+      "learning_rate": 8.634425547118298e-06,
+      "loss": 0.908,
+      "step": 20292
+    },
+    {
+      "epoch": 0.5573315756227514,
+      "grad_norm": 0.0,
+      "learning_rate": 8.633544369358454e-06,
+      "loss": 0.8468,
+      "step": 20293
+    },
+    {
+      "epoch": 0.5573590398505947,
+      "grad_norm": 0.0,
+      "learning_rate": 8.632663202410539e-06,
+      "loss": 0.8745,
+      "step": 20294
+    },
+    {
+      "epoch": 0.5573865040784378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.631782046281523e-06,
+      "loss": 0.8732,
+      "step": 20295
+    },
+    {
+      "epoch": 0.5574139683062811,
+      "grad_norm": 0.0,
+      "learning_rate": 8.630900900978379e-06,
+      "loss": 0.9025,
+      "step": 20296
+    },
+    {
+      "epoch": 0.5574414325341243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.630019766508077e-06,
+      "loss": 0.7732,
+      "step": 20297
+    },
+    {
+      "epoch": 0.5574688967619675,
+      "grad_norm": 0.0,
+      "learning_rate": 8.629138642877586e-06,
+      "loss": 0.8692,
+      "step": 20298
+    },
+    {
+      "epoch": 0.5574963609898108,
+      "grad_norm": 0.0,
+      "learning_rate": 8.628257530093882e-06,
+      "loss": 0.8645,
+      "step": 20299
+    },
+    {
+      "epoch": 0.557523825217654,
+      "grad_norm": 0.0,
+      "learning_rate": 8.627376428163938e-06,
+      "loss": 0.8655,
+      "step": 20300
+    },
+    {
+      "epoch": 0.5575512894454973,
+      "grad_norm": 0.0,
+      "learning_rate": 8.626495337094721e-06,
+      "loss": 0.9089,
+      "step": 20301
+    },
+    {
+      "epoch": 0.5575787536733404,
+      "grad_norm": 0.0,
+      "learning_rate": 8.625614256893208e-06,
+      "loss": 0.8522,
+      "step": 20302
+    },
+    {
+      "epoch": 0.5576062179011837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.624733187566365e-06,
+      "loss": 0.7632,
+      "step": 20303
+    },
+    {
+      "epoch": 0.557633682129027,
+      "grad_norm": 0.0,
+      "learning_rate": 8.623852129121168e-06,
+      "loss": 0.8147,
+      "step": 20304
+    },
+    {
+      "epoch": 0.5576611463568701,
+      "grad_norm": 0.0,
+      "learning_rate": 8.622971081564581e-06,
+      "loss": 0.9971,
+      "step": 20305
+    },
+    {
+      "epoch": 0.5576886105847134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.622090044903584e-06,
+      "loss": 0.8591,
+      "step": 20306
+    },
+    {
+      "epoch": 0.5577160748125567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.621209019145145e-06,
+      "loss": 0.9422,
+      "step": 20307
+    },
+    {
+      "epoch": 0.5577435390403999,
+      "grad_norm": 0.0,
+      "learning_rate": 8.620328004296233e-06,
+      "loss": 0.8517,
+      "step": 20308
+    },
+    {
+      "epoch": 0.5577710032682431,
+      "grad_norm": 0.0,
+      "learning_rate": 8.61944700036382e-06,
+      "loss": 0.7397,
+      "step": 20309
+    },
+    {
+      "epoch": 0.5577984674960863,
+      "grad_norm": 0.0,
+      "learning_rate": 8.618566007354878e-06,
+      "loss": 0.9028,
+      "step": 20310
+    },
+    {
+      "epoch": 0.5578259317239296,
+      "grad_norm": 0.0,
+      "learning_rate": 8.617685025276374e-06,
+      "loss": 0.8305,
+      "step": 20311
+    },
+    {
+      "epoch": 0.5578533959517729,
+      "grad_norm": 0.0,
+      "learning_rate": 8.616804054135286e-06,
+      "loss": 0.8374,
+      "step": 20312
+    },
+    {
+      "epoch": 0.557880860179616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.615923093938576e-06,
+      "loss": 0.8562,
+      "step": 20313
+    },
+    {
+      "epoch": 0.5579083244074593,
+      "grad_norm": 0.0,
+      "learning_rate": 8.615042144693224e-06,
+      "loss": 0.9744,
+      "step": 20314
+    },
+    {
+      "epoch": 0.5579357886353025,
+      "grad_norm": 0.0,
+      "learning_rate": 8.614161206406191e-06,
+      "loss": 0.8712,
+      "step": 20315
+    },
+    {
+      "epoch": 0.5579632528631457,
+      "grad_norm": 0.0,
+      "learning_rate": 8.613280279084452e-06,
+      "loss": 0.8709,
+      "step": 20316
+    },
+    {
+      "epoch": 0.557990717090989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.612399362734976e-06,
+      "loss": 0.7549,
+      "step": 20317
+    },
+    {
+      "epoch": 0.5580181813188322,
+      "grad_norm": 0.0,
+      "learning_rate": 8.611518457364733e-06,
+      "loss": 0.8632,
+      "step": 20318
+    },
+    {
+      "epoch": 0.5580456455466755,
+      "grad_norm": 0.0,
+      "learning_rate": 8.610637562980697e-06,
+      "loss": 0.8408,
+      "step": 20319
+    },
+    {
+      "epoch": 0.5580731097745187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.609756679589834e-06,
+      "loss": 0.8976,
+      "step": 20320
+    },
+    {
+      "epoch": 0.5581005740023619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.608875807199114e-06,
+      "loss": 0.9225,
+      "step": 20321
+    },
+    {
+      "epoch": 0.5581280382302052,
+      "grad_norm": 0.0,
+      "learning_rate": 8.607994945815505e-06,
+      "loss": 0.8784,
+      "step": 20322
+    },
+    {
+      "epoch": 0.5581555024580483,
+      "grad_norm": 0.0,
+      "learning_rate": 8.60711409544598e-06,
+      "loss": 0.9296,
+      "step": 20323
+    },
+    {
+      "epoch": 0.5581829666858916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.60623325609751e-06,
+      "loss": 0.8205,
+      "step": 20324
+    },
+    {
+      "epoch": 0.5582104309137349,
+      "grad_norm": 0.0,
+      "learning_rate": 8.605352427777066e-06,
+      "loss": 0.8085,
+      "step": 20325
+    },
+    {
+      "epoch": 0.5582378951415781,
+      "grad_norm": 0.0,
+      "learning_rate": 8.604471610491608e-06,
+      "loss": 0.8861,
+      "step": 20326
+    },
+    {
+      "epoch": 0.5582653593694213,
+      "grad_norm": 0.0,
+      "learning_rate": 8.603590804248111e-06,
+      "loss": 0.8445,
+      "step": 20327
+    },
+    {
+      "epoch": 0.5582928235972645,
+      "grad_norm": 0.0,
+      "learning_rate": 8.602710009053548e-06,
+      "loss": 0.937,
+      "step": 20328
+    },
+    {
+      "epoch": 0.5583202878251078,
+      "grad_norm": 0.0,
+      "learning_rate": 8.601829224914882e-06,
+      "loss": 0.9193,
+      "step": 20329
+    },
+    {
+      "epoch": 0.5583477520529511,
+      "grad_norm": 0.0,
+      "learning_rate": 8.600948451839084e-06,
+      "loss": 0.8712,
+      "step": 20330
+    },
+    {
+      "epoch": 0.5583752162807942,
+      "grad_norm": 0.0,
+      "learning_rate": 8.600067689833128e-06,
+      "loss": 0.853,
+      "step": 20331
+    },
+    {
+      "epoch": 0.5584026805086375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.599186938903974e-06,
+      "loss": 0.9659,
+      "step": 20332
+    },
+    {
+      "epoch": 0.5584301447364808,
+      "grad_norm": 0.0,
+      "learning_rate": 8.598306199058599e-06,
+      "loss": 0.8854,
+      "step": 20333
+    },
+    {
+      "epoch": 0.558457608964324,
+      "grad_norm": 0.0,
+      "learning_rate": 8.597425470303967e-06,
+      "loss": 1.037,
+      "step": 20334
+    },
+    {
+      "epoch": 0.5584850731921672,
+      "grad_norm": 0.0,
+      "learning_rate": 8.596544752647045e-06,
+      "loss": 0.7976,
+      "step": 20335
+    },
+    {
+      "epoch": 0.5585125374200104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.59566404609481e-06,
+      "loss": 0.8817,
+      "step": 20336
+    },
+    {
+      "epoch": 0.5585400016478537,
+      "grad_norm": 0.0,
+      "learning_rate": 8.594783350654225e-06,
+      "loss": 0.913,
+      "step": 20337
+    },
+    {
+      "epoch": 0.558567465875697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.593902666332254e-06,
+      "loss": 0.8375,
+      "step": 20338
+    },
+    {
+      "epoch": 0.5585949301035401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.593021993135871e-06,
+      "loss": 0.8967,
+      "step": 20339
+    },
+    {
+      "epoch": 0.5586223943313834,
+      "grad_norm": 0.0,
+      "learning_rate": 8.592141331072044e-06,
+      "loss": 0.8222,
+      "step": 20340
+    },
+    {
+      "epoch": 0.5586498585592266,
+      "grad_norm": 0.0,
+      "learning_rate": 8.59126068014774e-06,
+      "loss": 0.9261,
+      "step": 20341
+    },
+    {
+      "epoch": 0.5586773227870698,
+      "grad_norm": 0.0,
+      "learning_rate": 8.590380040369924e-06,
+      "loss": 0.8374,
+      "step": 20342
+    },
+    {
+      "epoch": 0.5587047870149131,
+      "grad_norm": 0.0,
+      "learning_rate": 8.589499411745574e-06,
+      "loss": 0.8141,
+      "step": 20343
+    },
+    {
+      "epoch": 0.5587322512427563,
+      "grad_norm": 0.0,
+      "learning_rate": 8.588618794281647e-06,
+      "loss": 0.919,
+      "step": 20344
+    },
+    {
+      "epoch": 0.5587597154705995,
+      "grad_norm": 0.0,
+      "learning_rate": 8.587738187985115e-06,
+      "loss": 0.9476,
+      "step": 20345
+    },
+    {
+      "epoch": 0.5587871796984428,
+      "grad_norm": 0.0,
+      "learning_rate": 8.586857592862946e-06,
+      "loss": 0.9334,
+      "step": 20346
+    },
+    {
+      "epoch": 0.558814643926286,
+      "grad_norm": 0.0,
+      "learning_rate": 8.585977008922104e-06,
+      "loss": 0.9194,
+      "step": 20347
+    },
+    {
+      "epoch": 0.5588421081541293,
+      "grad_norm": 0.0,
+      "learning_rate": 8.585096436169563e-06,
+      "loss": 0.8764,
+      "step": 20348
+    },
+    {
+      "epoch": 0.5588695723819724,
+      "grad_norm": 0.0,
+      "learning_rate": 8.58421587461229e-06,
+      "loss": 0.9312,
+      "step": 20349
+    },
+    {
+      "epoch": 0.5588970366098157,
+      "grad_norm": 0.0,
+      "learning_rate": 8.583335324257245e-06,
+      "loss": 0.9256,
+      "step": 20350
+    },
+    {
+      "epoch": 0.558924500837659,
+      "grad_norm": 0.0,
+      "learning_rate": 8.582454785111399e-06,
+      "loss": 0.8355,
+      "step": 20351
+    },
+    {
+      "epoch": 0.5589519650655022,
+      "grad_norm": 0.0,
+      "learning_rate": 8.581574257181722e-06,
+      "loss": 0.808,
+      "step": 20352
+    },
+    {
+      "epoch": 0.5589794292933454,
+      "grad_norm": 0.0,
+      "learning_rate": 8.580693740475177e-06,
+      "loss": 0.8944,
+      "step": 20353
+    },
+    {
+      "epoch": 0.5590068935211886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.579813234998733e-06,
+      "loss": 0.8729,
+      "step": 20354
+    },
+    {
+      "epoch": 0.5590343577490319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.57893274075936e-06,
+      "loss": 0.9368,
+      "step": 20355
+    },
+    {
+      "epoch": 0.5590618219768752,
+      "grad_norm": 0.0,
+      "learning_rate": 8.578052257764019e-06,
+      "loss": 0.8817,
+      "step": 20356
+    },
+    {
+      "epoch": 0.5590892862047183,
+      "grad_norm": 0.0,
+      "learning_rate": 8.57717178601968e-06,
+      "loss": 0.8857,
+      "step": 20357
+    },
+    {
+      "epoch": 0.5591167504325616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.57629132553331e-06,
+      "loss": 0.8535,
+      "step": 20358
+    },
+    {
+      "epoch": 0.5591442146604049,
+      "grad_norm": 0.0,
+      "learning_rate": 8.575410876311871e-06,
+      "loss": 0.7841,
+      "step": 20359
+    },
+    {
+      "epoch": 0.559171678888248,
+      "grad_norm": 0.0,
+      "learning_rate": 8.574530438362336e-06,
+      "loss": 0.9105,
+      "step": 20360
+    },
+    {
+      "epoch": 0.5591991431160913,
+      "grad_norm": 0.0,
+      "learning_rate": 8.573650011691672e-06,
+      "loss": 0.8382,
+      "step": 20361
+    },
+    {
+      "epoch": 0.5592266073439345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.572769596306837e-06,
+      "loss": 0.8707,
+      "step": 20362
+    },
+    {
+      "epoch": 0.5592540715717778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5718891922148e-06,
+      "loss": 0.9506,
+      "step": 20363
+    },
+    {
+      "epoch": 0.559281535799621,
+      "grad_norm": 0.0,
+      "learning_rate": 8.571008799422531e-06,
+      "loss": 0.8991,
+      "step": 20364
+    },
+    {
+      "epoch": 0.5593090000274642,
+      "grad_norm": 0.0,
+      "learning_rate": 8.570128417936994e-06,
+      "loss": 0.8747,
+      "step": 20365
+    },
+    {
+      "epoch": 0.5593364642553075,
+      "grad_norm": 0.0,
+      "learning_rate": 8.569248047765152e-06,
+      "loss": 0.9371,
+      "step": 20366
+    },
+    {
+      "epoch": 0.5593639284831506,
+      "grad_norm": 0.0,
+      "learning_rate": 8.56836768891398e-06,
+      "loss": 0.8043,
+      "step": 20367
+    },
+    {
+      "epoch": 0.5593913927109939,
+      "grad_norm": 0.0,
+      "learning_rate": 8.567487341390434e-06,
+      "loss": 0.9298,
+      "step": 20368
+    },
+    {
+      "epoch": 0.5594188569388372,
+      "grad_norm": 0.0,
+      "learning_rate": 8.566607005201483e-06,
+      "loss": 0.8737,
+      "step": 20369
+    },
+    {
+      "epoch": 0.5594463211666804,
+      "grad_norm": 0.0,
+      "learning_rate": 8.565726680354091e-06,
+      "loss": 0.9115,
+      "step": 20370
+    },
+    {
+      "epoch": 0.5594737853945236,
+      "grad_norm": 0.0,
+      "learning_rate": 8.564846366855227e-06,
+      "loss": 0.8946,
+      "step": 20371
+    },
+    {
+      "epoch": 0.5595012496223669,
+      "grad_norm": 0.0,
+      "learning_rate": 8.563966064711855e-06,
+      "loss": 0.8847,
+      "step": 20372
+    },
+    {
+      "epoch": 0.5595287138502101,
+      "grad_norm": 0.0,
+      "learning_rate": 8.563085773930941e-06,
+      "loss": 0.8918,
+      "step": 20373
+    },
+    {
+      "epoch": 0.5595561780780534,
+      "grad_norm": 0.0,
+      "learning_rate": 8.562205494519447e-06,
+      "loss": 0.8808,
+      "step": 20374
+    },
+    {
+      "epoch": 0.5595836423058965,
+      "grad_norm": 0.0,
+      "learning_rate": 8.561325226484339e-06,
+      "loss": 0.9282,
+      "step": 20375
+    },
+    {
+      "epoch": 0.5596111065337398,
+      "grad_norm": 0.0,
+      "learning_rate": 8.560444969832584e-06,
+      "loss": 0.8189,
+      "step": 20376
+    },
+    {
+      "epoch": 0.5596385707615831,
+      "grad_norm": 0.0,
+      "learning_rate": 8.559564724571145e-06,
+      "loss": 0.8043,
+      "step": 20377
+    },
+    {
+      "epoch": 0.5596660349894262,
+      "grad_norm": 0.0,
+      "learning_rate": 8.558684490706987e-06,
+      "loss": 0.9786,
+      "step": 20378
+    },
+    {
+      "epoch": 0.5596934992172695,
+      "grad_norm": 0.0,
+      "learning_rate": 8.557804268247081e-06,
+      "loss": 1.0464,
+      "step": 20379
+    },
+    {
+      "epoch": 0.5597209634451128,
+      "grad_norm": 0.0,
+      "learning_rate": 8.556924057198383e-06,
+      "loss": 0.9471,
+      "step": 20380
+    },
+    {
+      "epoch": 0.559748427672956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.556043857567862e-06,
+      "loss": 0.9295,
+      "step": 20381
+    },
+    {
+      "epoch": 0.5597758919007992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.555163669362479e-06,
+      "loss": 0.8099,
+      "step": 20382
+    },
+    {
+      "epoch": 0.5598033561286424,
+      "grad_norm": 0.0,
+      "learning_rate": 8.554283492589199e-06,
+      "loss": 0.8955,
+      "step": 20383
+    },
+    {
+      "epoch": 0.5598308203564857,
+      "grad_norm": 0.0,
+      "learning_rate": 8.553403327254994e-06,
+      "loss": 0.9385,
+      "step": 20384
+    },
+    {
+      "epoch": 0.559858284584329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.552523173366822e-06,
+      "loss": 0.9146,
+      "step": 20385
+    },
+    {
+      "epoch": 0.5598857488121721,
+      "grad_norm": 0.0,
+      "learning_rate": 8.551643030931643e-06,
+      "loss": 0.9138,
+      "step": 20386
+    },
+    {
+      "epoch": 0.5599132130400154,
+      "grad_norm": 0.0,
+      "learning_rate": 8.550762899956424e-06,
+      "loss": 0.8775,
+      "step": 20387
+    },
+    {
+      "epoch": 0.5599406772678586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.549882780448136e-06,
+      "loss": 0.9161,
+      "step": 20388
+    },
+    {
+      "epoch": 0.5599681414957018,
+      "grad_norm": 0.0,
+      "learning_rate": 8.549002672413732e-06,
+      "loss": 0.8793,
+      "step": 20389
+    },
+    {
+      "epoch": 0.5599956057235451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.548122575860186e-06,
+      "loss": 0.8581,
+      "step": 20390
+    },
+    {
+      "epoch": 0.5600230699513883,
+      "grad_norm": 0.0,
+      "learning_rate": 8.547242490794452e-06,
+      "loss": 0.8358,
+      "step": 20391
+    },
+    {
+      "epoch": 0.5600505341792316,
+      "grad_norm": 0.0,
+      "learning_rate": 8.546362417223501e-06,
+      "loss": 0.8842,
+      "step": 20392
+    },
+    {
+      "epoch": 0.5600779984070748,
+      "grad_norm": 0.0,
+      "learning_rate": 8.545482355154292e-06,
+      "loss": 0.8809,
+      "step": 20393
+    },
+    {
+      "epoch": 0.560105462634918,
+      "grad_norm": 0.0,
+      "learning_rate": 8.54460230459379e-06,
+      "loss": 0.8545,
+      "step": 20394
+    },
+    {
+      "epoch": 0.5601329268627613,
+      "grad_norm": 0.0,
+      "learning_rate": 8.543722265548957e-06,
+      "loss": 0.8768,
+      "step": 20395
+    },
+    {
+      "epoch": 0.5601603910906044,
+      "grad_norm": 0.0,
+      "learning_rate": 8.542842238026764e-06,
+      "loss": 0.8873,
+      "step": 20396
+    },
+    {
+      "epoch": 0.5601878553184477,
+      "grad_norm": 0.0,
+      "learning_rate": 8.541962222034163e-06,
+      "loss": 0.9611,
+      "step": 20397
+    },
+    {
+      "epoch": 0.560215319546291,
+      "grad_norm": 0.0,
+      "learning_rate": 8.54108221757812e-06,
+      "loss": 0.9441,
+      "step": 20398
+    },
+    {
+      "epoch": 0.5602427837741342,
+      "grad_norm": 0.0,
+      "learning_rate": 8.540202224665601e-06,
+      "loss": 0.9245,
+      "step": 20399
+    },
+    {
+      "epoch": 0.5602702480019774,
+      "grad_norm": 0.0,
+      "learning_rate": 8.539322243303568e-06,
+      "loss": 0.8698,
+      "step": 20400
+    },
+    {
+      "epoch": 0.5602977122298206,
+      "grad_norm": 0.0,
+      "learning_rate": 8.538442273498983e-06,
+      "loss": 0.8447,
+      "step": 20401
+    },
+    {
+      "epoch": 0.5603251764576639,
+      "grad_norm": 0.0,
+      "learning_rate": 8.537562315258811e-06,
+      "loss": 0.8125,
+      "step": 20402
+    },
+    {
+      "epoch": 0.5603526406855072,
+      "grad_norm": 0.0,
+      "learning_rate": 8.53668236859001e-06,
+      "loss": 0.877,
+      "step": 20403
+    },
+    {
+      "epoch": 0.5603801049133503,
+      "grad_norm": 0.0,
+      "learning_rate": 8.535802433499547e-06,
+      "loss": 0.8672,
+      "step": 20404
+    },
+    {
+      "epoch": 0.5604075691411936,
+      "grad_norm": 0.0,
+      "learning_rate": 8.534922509994379e-06,
+      "loss": 0.8424,
+      "step": 20405
+    },
+    {
+      "epoch": 0.5604350333690369,
+      "grad_norm": 0.0,
+      "learning_rate": 8.534042598081472e-06,
+      "loss": 0.8914,
+      "step": 20406
+    },
+    {
+      "epoch": 0.56046249759688,
+      "grad_norm": 0.0,
+      "learning_rate": 8.53316269776779e-06,
+      "loss": 0.9998,
+      "step": 20407
+    },
+    {
+      "epoch": 0.5604899618247233,
+      "grad_norm": 0.0,
+      "learning_rate": 8.532282809060293e-06,
+      "loss": 0.7693,
+      "step": 20408
+    },
+    {
+      "epoch": 0.5605174260525665,
+      "grad_norm": 0.0,
+      "learning_rate": 8.531402931965944e-06,
+      "loss": 0.8091,
+      "step": 20409
+    },
+    {
+      "epoch": 0.5605448902804098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5305230664917e-06,
+      "loss": 0.8747,
+      "step": 20410
+    },
+    {
+      "epoch": 0.560572354508253,
+      "grad_norm": 0.0,
+      "learning_rate": 8.529643212644528e-06,
+      "loss": 0.866,
+      "step": 20411
+    },
+    {
+      "epoch": 0.5605998187360962,
+      "grad_norm": 0.0,
+      "learning_rate": 8.52876337043139e-06,
+      "loss": 0.9595,
+      "step": 20412
+    },
+    {
+      "epoch": 0.5606272829639395,
+      "grad_norm": 0.0,
+      "learning_rate": 8.527883539859244e-06,
+      "loss": 0.8593,
+      "step": 20413
+    },
+    {
+      "epoch": 0.5606547471917827,
+      "grad_norm": 0.0,
+      "learning_rate": 8.527003720935057e-06,
+      "loss": 0.8815,
+      "step": 20414
+    },
+    {
+      "epoch": 0.5606822114196259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.526123913665784e-06,
+      "loss": 0.769,
+      "step": 20415
+    },
+    {
+      "epoch": 0.5607096756474692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.525244118058393e-06,
+      "loss": 0.8971,
+      "step": 20416
+    },
+    {
+      "epoch": 0.5607371398753124,
+      "grad_norm": 0.0,
+      "learning_rate": 8.524364334119838e-06,
+      "loss": 0.9322,
+      "step": 20417
+    },
+    {
+      "epoch": 0.5607646041031557,
+      "grad_norm": 0.0,
+      "learning_rate": 8.523484561857083e-06,
+      "loss": 0.8386,
+      "step": 20418
+    },
+    {
+      "epoch": 0.5607920683309989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.522604801277093e-06,
+      "loss": 0.9342,
+      "step": 20419
+    },
+    {
+      "epoch": 0.5608195325588421,
+      "grad_norm": 0.0,
+      "learning_rate": 8.521725052386828e-06,
+      "loss": 0.9176,
+      "step": 20420
+    },
+    {
+      "epoch": 0.5608469967866854,
+      "grad_norm": 0.0,
+      "learning_rate": 8.520845315193245e-06,
+      "loss": 0.8489,
+      "step": 20421
+    },
+    {
+      "epoch": 0.5608744610145285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.519965589703305e-06,
+      "loss": 0.7703,
+      "step": 20422
+    },
+    {
+      "epoch": 0.5609019252423718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.51908587592397e-06,
+      "loss": 0.9344,
+      "step": 20423
+    },
+    {
+      "epoch": 0.5609293894702151,
+      "grad_norm": 0.0,
+      "learning_rate": 8.518206173862205e-06,
+      "loss": 0.7566,
+      "step": 20424
+    },
+    {
+      "epoch": 0.5609568536980583,
+      "grad_norm": 0.0,
+      "learning_rate": 8.517326483524963e-06,
+      "loss": 0.8664,
+      "step": 20425
+    },
+    {
+      "epoch": 0.5609843179259015,
+      "grad_norm": 0.0,
+      "learning_rate": 8.516446804919214e-06,
+      "loss": 0.8534,
+      "step": 20426
+    },
+    {
+      "epoch": 0.5610117821537447,
+      "grad_norm": 0.0,
+      "learning_rate": 8.515567138051908e-06,
+      "loss": 0.8025,
+      "step": 20427
+    },
+    {
+      "epoch": 0.561039246381588,
+      "grad_norm": 0.0,
+      "learning_rate": 8.51468748293001e-06,
+      "loss": 0.9619,
+      "step": 20428
+    },
+    {
+      "epoch": 0.5610667106094313,
+      "grad_norm": 0.0,
+      "learning_rate": 8.51380783956048e-06,
+      "loss": 0.9389,
+      "step": 20429
+    },
+    {
+      "epoch": 0.5610941748372744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.512928207950278e-06,
+      "loss": 0.8353,
+      "step": 20430
+    },
+    {
+      "epoch": 0.5611216390651177,
+      "grad_norm": 0.0,
+      "learning_rate": 8.512048588106365e-06,
+      "loss": 0.8705,
+      "step": 20431
+    },
+    {
+      "epoch": 0.561149103292961,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5111689800357e-06,
+      "loss": 0.9271,
+      "step": 20432
+    },
+    {
+      "epoch": 0.5611765675208041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.510289383745245e-06,
+      "loss": 0.9225,
+      "step": 20433
+    },
+    {
+      "epoch": 0.5612040317486474,
+      "grad_norm": 0.0,
+      "learning_rate": 8.509409799241954e-06,
+      "loss": 0.8953,
+      "step": 20434
+    },
+    {
+      "epoch": 0.5612314959764906,
+      "grad_norm": 0.0,
+      "learning_rate": 8.508530226532789e-06,
+      "loss": 0.9189,
+      "step": 20435
+    },
+    {
+      "epoch": 0.5612589602043339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.507650665624714e-06,
+      "loss": 0.8739,
+      "step": 20436
+    },
+    {
+      "epoch": 0.5612864244321771,
+      "grad_norm": 0.0,
+      "learning_rate": 8.506771116524681e-06,
+      "loss": 0.8172,
+      "step": 20437
+    },
+    {
+      "epoch": 0.5613138886600203,
+      "grad_norm": 0.0,
+      "learning_rate": 8.505891579239659e-06,
+      "loss": 0.8067,
+      "step": 20438
+    },
+    {
+      "epoch": 0.5613413528878636,
+      "grad_norm": 0.0,
+      "learning_rate": 8.505012053776597e-06,
+      "loss": 0.9349,
+      "step": 20439
+    },
+    {
+      "epoch": 0.5613688171157067,
+      "grad_norm": 0.0,
+      "learning_rate": 8.504132540142461e-06,
+      "loss": 0.8594,
+      "step": 20440
+    },
+    {
+      "epoch": 0.56139628134355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.503253038344206e-06,
+      "loss": 0.8461,
+      "step": 20441
+    },
+    {
+      "epoch": 0.5614237455713933,
+      "grad_norm": 0.0,
+      "learning_rate": 8.502373548388793e-06,
+      "loss": 0.8725,
+      "step": 20442
+    },
+    {
+      "epoch": 0.5614512097992365,
+      "grad_norm": 0.0,
+      "learning_rate": 8.501494070283184e-06,
+      "loss": 0.9149,
+      "step": 20443
+    },
+    {
+      "epoch": 0.5614786740270797,
+      "grad_norm": 0.0,
+      "learning_rate": 8.50061460403433e-06,
+      "loss": 0.9347,
+      "step": 20444
+    },
+    {
+      "epoch": 0.561506138254923,
+      "grad_norm": 0.0,
+      "learning_rate": 8.499735149649197e-06,
+      "loss": 0.8804,
+      "step": 20445
+    },
+    {
+      "epoch": 0.5615336024827662,
+      "grad_norm": 0.0,
+      "learning_rate": 8.498855707134739e-06,
+      "loss": 0.8777,
+      "step": 20446
+    },
+    {
+      "epoch": 0.5615610667106095,
+      "grad_norm": 0.0,
+      "learning_rate": 8.497976276497915e-06,
+      "loss": 1.0012,
+      "step": 20447
+    },
+    {
+      "epoch": 0.5615885309384526,
+      "grad_norm": 0.0,
+      "learning_rate": 8.497096857745687e-06,
+      "loss": 0.9006,
+      "step": 20448
+    },
+    {
+      "epoch": 0.5616159951662959,
+      "grad_norm": 0.0,
+      "learning_rate": 8.496217450885013e-06,
+      "loss": 0.9342,
+      "step": 20449
+    },
+    {
+      "epoch": 0.5616434593941392,
+      "grad_norm": 0.0,
+      "learning_rate": 8.495338055922844e-06,
+      "loss": 0.7864,
+      "step": 20450
+    },
+    {
+      "epoch": 0.5616709236219823,
+      "grad_norm": 0.0,
+      "learning_rate": 8.494458672866144e-06,
+      "loss": 0.9594,
+      "step": 20451
+    },
+    {
+      "epoch": 0.5616983878498256,
+      "grad_norm": 0.0,
+      "learning_rate": 8.493579301721872e-06,
+      "loss": 0.8077,
+      "step": 20452
+    },
+    {
+      "epoch": 0.5617258520776688,
+      "grad_norm": 0.0,
+      "learning_rate": 8.492699942496982e-06,
+      "loss": 0.9229,
+      "step": 20453
+    },
+    {
+      "epoch": 0.5617533163055121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.491820595198434e-06,
+      "loss": 0.8447,
+      "step": 20454
+    },
+    {
+      "epoch": 0.5617807805333553,
+      "grad_norm": 0.0,
+      "learning_rate": 8.49094125983319e-06,
+      "loss": 0.8646,
+      "step": 20455
+    },
+    {
+      "epoch": 0.5618082447611985,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4900619364082e-06,
+      "loss": 0.7455,
+      "step": 20456
+    },
+    {
+      "epoch": 0.5618357089890418,
+      "grad_norm": 0.0,
+      "learning_rate": 8.489182624930425e-06,
+      "loss": 0.9176,
+      "step": 20457
+    },
+    {
+      "epoch": 0.5618631732168851,
+      "grad_norm": 0.0,
+      "learning_rate": 8.488303325406822e-06,
+      "loss": 0.8582,
+      "step": 20458
+    },
+    {
+      "epoch": 0.5618906374447282,
+      "grad_norm": 0.0,
+      "learning_rate": 8.487424037844348e-06,
+      "loss": 0.856,
+      "step": 20459
+    },
+    {
+      "epoch": 0.5619181016725715,
+      "grad_norm": 0.0,
+      "learning_rate": 8.486544762249961e-06,
+      "loss": 0.8994,
+      "step": 20460
+    },
+    {
+      "epoch": 0.5619455659004147,
+      "grad_norm": 0.0,
+      "learning_rate": 8.485665498630624e-06,
+      "loss": 0.87,
+      "step": 20461
+    },
+    {
+      "epoch": 0.5619730301282579,
+      "grad_norm": 0.0,
+      "learning_rate": 8.484786246993282e-06,
+      "loss": 0.9897,
+      "step": 20462
+    },
+    {
+      "epoch": 0.5620004943561012,
+      "grad_norm": 0.0,
+      "learning_rate": 8.483907007344898e-06,
+      "loss": 0.8366,
+      "step": 20463
+    },
+    {
+      "epoch": 0.5620279585839444,
+      "grad_norm": 0.0,
+      "learning_rate": 8.48302777969243e-06,
+      "loss": 0.7926,
+      "step": 20464
+    },
+    {
+      "epoch": 0.5620554228117877,
+      "grad_norm": 0.0,
+      "learning_rate": 8.482148564042834e-06,
+      "loss": 0.9693,
+      "step": 20465
+    },
+    {
+      "epoch": 0.5620828870396308,
+      "grad_norm": 0.0,
+      "learning_rate": 8.481269360403066e-06,
+      "loss": 0.9459,
+      "step": 20466
+    },
+    {
+      "epoch": 0.5621103512674741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.480390168780085e-06,
+      "loss": 0.8555,
+      "step": 20467
+    },
+    {
+      "epoch": 0.5621378154953174,
+      "grad_norm": 0.0,
+      "learning_rate": 8.479510989180844e-06,
+      "loss": 0.9141,
+      "step": 20468
+    },
+    {
+      "epoch": 0.5621652797231605,
+      "grad_norm": 0.0,
+      "learning_rate": 8.478631821612303e-06,
+      "loss": 0.9269,
+      "step": 20469
+    },
+    {
+      "epoch": 0.5621927439510038,
+      "grad_norm": 0.0,
+      "learning_rate": 8.477752666081413e-06,
+      "loss": 0.7881,
+      "step": 20470
+    },
+    {
+      "epoch": 0.5622202081788471,
+      "grad_norm": 0.0,
+      "learning_rate": 8.476873522595135e-06,
+      "loss": 0.7919,
+      "step": 20471
+    },
+    {
+      "epoch": 0.5622476724066903,
+      "grad_norm": 0.0,
+      "learning_rate": 8.475994391160425e-06,
+      "loss": 0.843,
+      "step": 20472
+    },
+    {
+      "epoch": 0.5622751366345335,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47511527178424e-06,
+      "loss": 0.9104,
+      "step": 20473
+    },
+    {
+      "epoch": 0.5623026008623767,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47423616447353e-06,
+      "loss": 0.9526,
+      "step": 20474
+    },
+    {
+      "epoch": 0.56233006509022,
+      "grad_norm": 0.0,
+      "learning_rate": 8.473357069235253e-06,
+      "loss": 0.8379,
+      "step": 20475
+    },
+    {
+      "epoch": 0.5623575293180633,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47247798607637e-06,
+      "loss": 0.9299,
+      "step": 20476
+    },
+    {
+      "epoch": 0.5623849935459064,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47159891500383e-06,
+      "loss": 0.9811,
+      "step": 20477
+    },
+    {
+      "epoch": 0.5624124577737497,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47071985602459e-06,
+      "loss": 0.8443,
+      "step": 20478
+    },
+    {
+      "epoch": 0.5624399220015929,
+      "grad_norm": 0.0,
+      "learning_rate": 8.469840809145613e-06,
+      "loss": 0.8371,
+      "step": 20479
+    },
+    {
+      "epoch": 0.5624673862294362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.468961774373844e-06,
+      "loss": 0.8277,
+      "step": 20480
+    },
+    {
+      "epoch": 0.5624948504572794,
+      "grad_norm": 0.0,
+      "learning_rate": 8.468082751716245e-06,
+      "loss": 0.7647,
+      "step": 20481
+    },
+    {
+      "epoch": 0.5625223146851226,
+      "grad_norm": 0.0,
+      "learning_rate": 8.467203741179764e-06,
+      "loss": 0.9504,
+      "step": 20482
+    },
+    {
+      "epoch": 0.5625497789129659,
+      "grad_norm": 0.0,
+      "learning_rate": 8.466324742771363e-06,
+      "loss": 0.8864,
+      "step": 20483
+    },
+    {
+      "epoch": 0.5625772431408091,
+      "grad_norm": 0.0,
+      "learning_rate": 8.465445756497997e-06,
+      "loss": 0.9092,
+      "step": 20484
+    },
+    {
+      "epoch": 0.5626047073686523,
+      "grad_norm": 0.0,
+      "learning_rate": 8.464566782366619e-06,
+      "loss": 0.9202,
+      "step": 20485
+    },
+    {
+      "epoch": 0.5626321715964956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.46368782038418e-06,
+      "loss": 0.9399,
+      "step": 20486
+    },
+    {
+      "epoch": 0.5626596358243388,
+      "grad_norm": 0.0,
+      "learning_rate": 8.462808870557639e-06,
+      "loss": 0.9311,
+      "step": 20487
+    },
+    {
+      "epoch": 0.562687100052182,
+      "grad_norm": 0.0,
+      "learning_rate": 8.46192993289395e-06,
+      "loss": 0.8804,
+      "step": 20488
+    },
+    {
+      "epoch": 0.5627145642800253,
+      "grad_norm": 0.0,
+      "learning_rate": 8.461051007400068e-06,
+      "loss": 0.8071,
+      "step": 20489
+    },
+    {
+      "epoch": 0.5627420285078685,
+      "grad_norm": 0.0,
+      "learning_rate": 8.460172094082944e-06,
+      "loss": 0.8585,
+      "step": 20490
+    },
+    {
+      "epoch": 0.5627694927357118,
+      "grad_norm": 0.0,
+      "learning_rate": 8.45929319294954e-06,
+      "loss": 0.9376,
+      "step": 20491
+    },
+    {
+      "epoch": 0.5627969569635549,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4584143040068e-06,
+      "loss": 0.9562,
+      "step": 20492
+    },
+    {
+      "epoch": 0.5628244211913982,
+      "grad_norm": 0.0,
+      "learning_rate": 8.457535427261685e-06,
+      "loss": 0.7669,
+      "step": 20493
+    },
+    {
+      "epoch": 0.5628518854192415,
+      "grad_norm": 0.0,
+      "learning_rate": 8.456656562721146e-06,
+      "loss": 0.9443,
+      "step": 20494
+    },
+    {
+      "epoch": 0.5628793496470846,
+      "grad_norm": 0.0,
+      "learning_rate": 8.455777710392138e-06,
+      "loss": 0.9089,
+      "step": 20495
+    },
+    {
+      "epoch": 0.5629068138749279,
+      "grad_norm": 0.0,
+      "learning_rate": 8.454898870281619e-06,
+      "loss": 0.8464,
+      "step": 20496
+    },
+    {
+      "epoch": 0.5629342781027712,
+      "grad_norm": 0.0,
+      "learning_rate": 8.454020042396537e-06,
+      "loss": 0.8195,
+      "step": 20497
+    },
+    {
+      "epoch": 0.5629617423306144,
+      "grad_norm": 0.0,
+      "learning_rate": 8.453141226743845e-06,
+      "loss": 0.9633,
+      "step": 20498
+    },
+    {
+      "epoch": 0.5629892065584576,
+      "grad_norm": 0.0,
+      "learning_rate": 8.452262423330499e-06,
+      "loss": 0.8173,
+      "step": 20499
+    },
+    {
+      "epoch": 0.5630166707863008,
+      "grad_norm": 0.0,
+      "learning_rate": 8.451383632163455e-06,
+      "loss": 0.9425,
+      "step": 20500
+    },
+    {
+      "epoch": 0.5630441350141441,
+      "grad_norm": 0.0,
+      "learning_rate": 8.45050485324966e-06,
+      "loss": 0.9839,
+      "step": 20501
+    },
+    {
+      "epoch": 0.5630715992419874,
+      "grad_norm": 0.0,
+      "learning_rate": 8.449626086596075e-06,
+      "loss": 0.7944,
+      "step": 20502
+    },
+    {
+      "epoch": 0.5630990634698305,
+      "grad_norm": 0.0,
+      "learning_rate": 8.448747332209645e-06,
+      "loss": 0.8195,
+      "step": 20503
+    },
+    {
+      "epoch": 0.5631265276976738,
+      "grad_norm": 0.0,
+      "learning_rate": 8.447868590097327e-06,
+      "loss": 0.8942,
+      "step": 20504
+    },
+    {
+      "epoch": 0.563153991925517,
+      "grad_norm": 0.0,
+      "learning_rate": 8.446989860266077e-06,
+      "loss": 0.8756,
+      "step": 20505
+    },
+    {
+      "epoch": 0.5631814561533602,
+      "grad_norm": 0.0,
+      "learning_rate": 8.446111142722843e-06,
+      "loss": 0.9027,
+      "step": 20506
+    },
+    {
+      "epoch": 0.5632089203812035,
+      "grad_norm": 0.0,
+      "learning_rate": 8.445232437474578e-06,
+      "loss": 0.8773,
+      "step": 20507
+    },
+    {
+      "epoch": 0.5632363846090467,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44435374452824e-06,
+      "loss": 0.8358,
+      "step": 20508
+    },
+    {
+      "epoch": 0.56326384883689,
+      "grad_norm": 0.0,
+      "learning_rate": 8.443475063890778e-06,
+      "loss": 0.9459,
+      "step": 20509
+    },
+    {
+      "epoch": 0.5632913130647332,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44259639556914e-06,
+      "loss": 0.7925,
+      "step": 20510
+    },
+    {
+      "epoch": 0.5633187772925764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.441717739570284e-06,
+      "loss": 0.8129,
+      "step": 20511
+    },
+    {
+      "epoch": 0.5633462415204197,
+      "grad_norm": 0.0,
+      "learning_rate": 8.440839095901162e-06,
+      "loss": 0.876,
+      "step": 20512
+    },
+    {
+      "epoch": 0.5633737057482628,
+      "grad_norm": 0.0,
+      "learning_rate": 8.439960464568724e-06,
+      "loss": 0.9604,
+      "step": 20513
+    },
+    {
+      "epoch": 0.5634011699761061,
+      "grad_norm": 0.0,
+      "learning_rate": 8.439081845579927e-06,
+      "loss": 0.944,
+      "step": 20514
+    },
+    {
+      "epoch": 0.5634286342039494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.438203238941713e-06,
+      "loss": 0.8704,
+      "step": 20515
+    },
+    {
+      "epoch": 0.5634560984317926,
+      "grad_norm": 0.0,
+      "learning_rate": 8.437324644661042e-06,
+      "loss": 0.7933,
+      "step": 20516
+    },
+    {
+      "epoch": 0.5634835626596358,
+      "grad_norm": 0.0,
+      "learning_rate": 8.436446062744865e-06,
+      "loss": 0.858,
+      "step": 20517
+    },
+    {
+      "epoch": 0.563511026887479,
+      "grad_norm": 0.0,
+      "learning_rate": 8.435567493200132e-06,
+      "loss": 0.8065,
+      "step": 20518
+    },
+    {
+      "epoch": 0.5635384911153223,
+      "grad_norm": 0.0,
+      "learning_rate": 8.434688936033792e-06,
+      "loss": 0.8468,
+      "step": 20519
+    },
+    {
+      "epoch": 0.5635659553431656,
+      "grad_norm": 0.0,
+      "learning_rate": 8.433810391252807e-06,
+      "loss": 0.9127,
+      "step": 20520
+    },
+    {
+      "epoch": 0.5635934195710087,
+      "grad_norm": 0.0,
+      "learning_rate": 8.432931858864115e-06,
+      "loss": 0.877,
+      "step": 20521
+    },
+    {
+      "epoch": 0.563620883798852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.432053338874675e-06,
+      "loss": 0.908,
+      "step": 20522
+    },
+    {
+      "epoch": 0.5636483480266953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.431174831291435e-06,
+      "loss": 0.9269,
+      "step": 20523
+    },
+    {
+      "epoch": 0.5636758122545384,
+      "grad_norm": 0.0,
+      "learning_rate": 8.43029633612135e-06,
+      "loss": 0.9555,
+      "step": 20524
+    },
+    {
+      "epoch": 0.5637032764823817,
+      "grad_norm": 0.0,
+      "learning_rate": 8.429417853371363e-06,
+      "loss": 0.9263,
+      "step": 20525
+    },
+    {
+      "epoch": 0.5637307407102249,
+      "grad_norm": 0.0,
+      "learning_rate": 8.428539383048437e-06,
+      "loss": 0.8929,
+      "step": 20526
+    },
+    {
+      "epoch": 0.5637582049380682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.427660925159512e-06,
+      "loss": 0.9095,
+      "step": 20527
+    },
+    {
+      "epoch": 0.5637856691659114,
+      "grad_norm": 0.0,
+      "learning_rate": 8.426782479711542e-06,
+      "loss": 0.8958,
+      "step": 20528
+    },
+    {
+      "epoch": 0.5638131333937546,
+      "grad_norm": 0.0,
+      "learning_rate": 8.425904046711481e-06,
+      "loss": 0.9457,
+      "step": 20529
+    },
+    {
+      "epoch": 0.5638405976215979,
+      "grad_norm": 0.0,
+      "learning_rate": 8.425025626166277e-06,
+      "loss": 0.9247,
+      "step": 20530
+    },
+    {
+      "epoch": 0.563868061849441,
+      "grad_norm": 0.0,
+      "learning_rate": 8.424147218082877e-06,
+      "loss": 0.8601,
+      "step": 20531
+    },
+    {
+      "epoch": 0.5638955260772843,
+      "grad_norm": 0.0,
+      "learning_rate": 8.42326882246824e-06,
+      "loss": 0.9178,
+      "step": 20532
+    },
+    {
+      "epoch": 0.5639229903051276,
+      "grad_norm": 0.0,
+      "learning_rate": 8.42239043932931e-06,
+      "loss": 0.8838,
+      "step": 20533
+    },
+    {
+      "epoch": 0.5639504545329708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.421512068673035e-06,
+      "loss": 0.8474,
+      "step": 20534
+    },
+    {
+      "epoch": 0.563977918760814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.420633710506368e-06,
+      "loss": 0.8728,
+      "step": 20535
+    },
+    {
+      "epoch": 0.5640053829886573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.41975536483626e-06,
+      "loss": 0.986,
+      "step": 20536
+    },
+    {
+      "epoch": 0.5640328472165005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.418877031669658e-06,
+      "loss": 0.8155,
+      "step": 20537
+    },
+    {
+      "epoch": 0.5640603114443438,
+      "grad_norm": 0.0,
+      "learning_rate": 8.417998711013518e-06,
+      "loss": 0.8073,
+      "step": 20538
+    },
+    {
+      "epoch": 0.5640877756721869,
+      "grad_norm": 0.0,
+      "learning_rate": 8.417120402874781e-06,
+      "loss": 0.8518,
+      "step": 20539
+    },
+    {
+      "epoch": 0.5641152399000302,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4162421072604e-06,
+      "loss": 0.8602,
+      "step": 20540
+    },
+    {
+      "epoch": 0.5641427041278735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.415363824177328e-06,
+      "loss": 0.8605,
+      "step": 20541
+    },
+    {
+      "epoch": 0.5641701683557167,
+      "grad_norm": 0.0,
+      "learning_rate": 8.414485553632508e-06,
+      "loss": 0.8683,
+      "step": 20542
+    },
+    {
+      "epoch": 0.5641976325835599,
+      "grad_norm": 0.0,
+      "learning_rate": 8.413607295632892e-06,
+      "loss": 0.9018,
+      "step": 20543
+    },
+    {
+      "epoch": 0.5642250968114031,
+      "grad_norm": 0.0,
+      "learning_rate": 8.412729050185436e-06,
+      "loss": 0.875,
+      "step": 20544
+    },
+    {
+      "epoch": 0.5642525610392464,
+      "grad_norm": 0.0,
+      "learning_rate": 8.411850817297078e-06,
+      "loss": 0.9799,
+      "step": 20545
+    },
+    {
+      "epoch": 0.5642800252670896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.410972596974772e-06,
+      "loss": 0.8623,
+      "step": 20546
+    },
+    {
+      "epoch": 0.5643074894949328,
+      "grad_norm": 0.0,
+      "learning_rate": 8.410094389225464e-06,
+      "loss": 0.7933,
+      "step": 20547
+    },
+    {
+      "epoch": 0.5643349537227761,
+      "grad_norm": 0.0,
+      "learning_rate": 8.409216194056108e-06,
+      "loss": 0.9238,
+      "step": 20548
+    },
+    {
+      "epoch": 0.5643624179506194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.408338011473647e-06,
+      "loss": 0.9113,
+      "step": 20549
+    },
+    {
+      "epoch": 0.5643898821784625,
+      "grad_norm": 0.0,
+      "learning_rate": 8.407459841485037e-06,
+      "loss": 0.903,
+      "step": 20550
+    },
+    {
+      "epoch": 0.5644173464063058,
+      "grad_norm": 0.0,
+      "learning_rate": 8.406581684097218e-06,
+      "loss": 0.9221,
+      "step": 20551
+    },
+    {
+      "epoch": 0.564444810634149,
+      "grad_norm": 0.0,
+      "learning_rate": 8.405703539317144e-06,
+      "loss": 0.942,
+      "step": 20552
+    },
+    {
+      "epoch": 0.5644722748619923,
+      "grad_norm": 0.0,
+      "learning_rate": 8.404825407151758e-06,
+      "loss": 0.8543,
+      "step": 20553
+    },
+    {
+      "epoch": 0.5644997390898355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.403947287608012e-06,
+      "loss": 0.8915,
+      "step": 20554
+    },
+    {
+      "epoch": 0.5645272033176787,
+      "grad_norm": 0.0,
+      "learning_rate": 8.403069180692854e-06,
+      "loss": 0.7875,
+      "step": 20555
+    },
+    {
+      "epoch": 0.564554667545522,
+      "grad_norm": 0.0,
+      "learning_rate": 8.402191086413233e-06,
+      "loss": 0.7614,
+      "step": 20556
+    },
+    {
+      "epoch": 0.5645821317733651,
+      "grad_norm": 0.0,
+      "learning_rate": 8.401313004776096e-06,
+      "loss": 0.7808,
+      "step": 20557
+    },
+    {
+      "epoch": 0.5646095960012084,
+      "grad_norm": 0.0,
+      "learning_rate": 8.400434935788386e-06,
+      "loss": 0.8148,
+      "step": 20558
+    },
+    {
+      "epoch": 0.5646370602290517,
+      "grad_norm": 0.0,
+      "learning_rate": 8.399556879457055e-06,
+      "loss": 0.7735,
+      "step": 20559
+    },
+    {
+      "epoch": 0.5646645244568949,
+      "grad_norm": 0.0,
+      "learning_rate": 8.398678835789052e-06,
+      "loss": 0.8163,
+      "step": 20560
+    },
+    {
+      "epoch": 0.5646919886847381,
+      "grad_norm": 0.0,
+      "learning_rate": 8.397800804791324e-06,
+      "loss": 0.7957,
+      "step": 20561
+    },
+    {
+      "epoch": 0.5647194529125814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.396922786470813e-06,
+      "loss": 0.9039,
+      "step": 20562
+    },
+    {
+      "epoch": 0.5647469171404246,
+      "grad_norm": 0.0,
+      "learning_rate": 8.396044780834469e-06,
+      "loss": 0.8417,
+      "step": 20563
+    },
+    {
+      "epoch": 0.5647743813682679,
+      "grad_norm": 0.0,
+      "learning_rate": 8.395166787889242e-06,
+      "loss": 0.8549,
+      "step": 20564
+    },
+    {
+      "epoch": 0.564801845596111,
+      "grad_norm": 0.0,
+      "learning_rate": 8.394288807642076e-06,
+      "loss": 0.9396,
+      "step": 20565
+    },
+    {
+      "epoch": 0.5648293098239543,
+      "grad_norm": 0.0,
+      "learning_rate": 8.393410840099919e-06,
+      "loss": 0.8105,
+      "step": 20566
+    },
+    {
+      "epoch": 0.5648567740517976,
+      "grad_norm": 0.0,
+      "learning_rate": 8.392532885269721e-06,
+      "loss": 0.9719,
+      "step": 20567
+    },
+    {
+      "epoch": 0.5648842382796407,
+      "grad_norm": 0.0,
+      "learning_rate": 8.39165494315842e-06,
+      "loss": 0.8575,
+      "step": 20568
+    },
+    {
+      "epoch": 0.564911702507484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.390777013772974e-06,
+      "loss": 0.8684,
+      "step": 20569
+    },
+    {
+      "epoch": 0.5649391667353273,
+      "grad_norm": 0.0,
+      "learning_rate": 8.389899097120318e-06,
+      "loss": 0.894,
+      "step": 20570
+    },
+    {
+      "epoch": 0.5649666309631705,
+      "grad_norm": 0.0,
+      "learning_rate": 8.389021193207407e-06,
+      "loss": 0.9093,
+      "step": 20571
+    },
+    {
+      "epoch": 0.5649940951910137,
+      "grad_norm": 0.0,
+      "learning_rate": 8.388143302041184e-06,
+      "loss": 0.8457,
+      "step": 20572
+    },
+    {
+      "epoch": 0.5650215594188569,
+      "grad_norm": 0.0,
+      "learning_rate": 8.387265423628598e-06,
+      "loss": 0.9039,
+      "step": 20573
+    },
+    {
+      "epoch": 0.5650490236467002,
+      "grad_norm": 0.0,
+      "learning_rate": 8.38638755797659e-06,
+      "loss": 0.858,
+      "step": 20574
+    },
+    {
+      "epoch": 0.5650764878745435,
+      "grad_norm": 0.0,
+      "learning_rate": 8.385509705092107e-06,
+      "loss": 0.8387,
+      "step": 20575
+    },
+    {
+      "epoch": 0.5651039521023866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.3846318649821e-06,
+      "loss": 0.8076,
+      "step": 20576
+    },
+    {
+      "epoch": 0.5651314163302299,
+      "grad_norm": 0.0,
+      "learning_rate": 8.383754037653508e-06,
+      "loss": 0.8814,
+      "step": 20577
+    },
+    {
+      "epoch": 0.5651588805580731,
+      "grad_norm": 0.0,
+      "learning_rate": 8.382876223113283e-06,
+      "loss": 0.9072,
+      "step": 20578
+    },
+    {
+      "epoch": 0.5651863447859163,
+      "grad_norm": 0.0,
+      "learning_rate": 8.38199842136837e-06,
+      "loss": 0.7965,
+      "step": 20579
+    },
+    {
+      "epoch": 0.5652138090137596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.381120632425706e-06,
+      "loss": 0.8386,
+      "step": 20580
+    },
+    {
+      "epoch": 0.5652412732416028,
+      "grad_norm": 0.0,
+      "learning_rate": 8.380242856292247e-06,
+      "loss": 0.9487,
+      "step": 20581
+    },
+    {
+      "epoch": 0.5652687374694461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.379365092974931e-06,
+      "loss": 0.9017,
+      "step": 20582
+    },
+    {
+      "epoch": 0.5652962016972893,
+      "grad_norm": 0.0,
+      "learning_rate": 8.378487342480709e-06,
+      "loss": 0.8114,
+      "step": 20583
+    },
+    {
+      "epoch": 0.5653236659251325,
+      "grad_norm": 0.0,
+      "learning_rate": 8.377609604816521e-06,
+      "loss": 0.8742,
+      "step": 20584
+    },
+    {
+      "epoch": 0.5653511301529758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.376731879989319e-06,
+      "loss": 0.7623,
+      "step": 20585
+    },
+    {
+      "epoch": 0.5653785943808189,
+      "grad_norm": 0.0,
+      "learning_rate": 8.375854168006039e-06,
+      "loss": 0.8638,
+      "step": 20586
+    },
+    {
+      "epoch": 0.5654060586086622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.374976468873629e-06,
+      "loss": 0.8841,
+      "step": 20587
+    },
+    {
+      "epoch": 0.5654335228365055,
+      "grad_norm": 0.0,
+      "learning_rate": 8.374098782599037e-06,
+      "loss": 0.9895,
+      "step": 20588
+    },
+    {
+      "epoch": 0.5654609870643487,
+      "grad_norm": 0.0,
+      "learning_rate": 8.373221109189203e-06,
+      "loss": 0.9023,
+      "step": 20589
+    },
+    {
+      "epoch": 0.5654884512921919,
+      "grad_norm": 0.0,
+      "learning_rate": 8.372343448651074e-06,
+      "loss": 0.817,
+      "step": 20590
+    },
+    {
+      "epoch": 0.5655159155200351,
+      "grad_norm": 0.0,
+      "learning_rate": 8.371465800991598e-06,
+      "loss": 0.8849,
+      "step": 20591
+    },
+    {
+      "epoch": 0.5655433797478784,
+      "grad_norm": 0.0,
+      "learning_rate": 8.370588166217712e-06,
+      "loss": 0.8801,
+      "step": 20592
+    },
+    {
+      "epoch": 0.5655708439757217,
+      "grad_norm": 0.0,
+      "learning_rate": 8.369710544336365e-06,
+      "loss": 0.8702,
+      "step": 20593
+    },
+    {
+      "epoch": 0.5655983082035648,
+      "grad_norm": 0.0,
+      "learning_rate": 8.368832935354498e-06,
+      "loss": 0.8438,
+      "step": 20594
+    },
+    {
+      "epoch": 0.5656257724314081,
+      "grad_norm": 0.0,
+      "learning_rate": 8.367955339279058e-06,
+      "loss": 0.8472,
+      "step": 20595
+    },
+    {
+      "epoch": 0.5656532366592514,
+      "grad_norm": 0.0,
+      "learning_rate": 8.367077756116986e-06,
+      "loss": 0.7681,
+      "step": 20596
+    },
+    {
+      "epoch": 0.5656807008870945,
+      "grad_norm": 0.0,
+      "learning_rate": 8.366200185875232e-06,
+      "loss": 0.8429,
+      "step": 20597
+    },
+    {
+      "epoch": 0.5657081651149378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.36532262856073e-06,
+      "loss": 0.9189,
+      "step": 20598
+    },
+    {
+      "epoch": 0.565735629342781,
+      "grad_norm": 0.0,
+      "learning_rate": 8.364445084180429e-06,
+      "loss": 0.9912,
+      "step": 20599
+    },
+    {
+      "epoch": 0.5657630935706243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.363567552741274e-06,
+      "loss": 0.8982,
+      "step": 20600
+    },
+    {
+      "epoch": 0.5657905577984675,
+      "grad_norm": 0.0,
+      "learning_rate": 8.362690034250206e-06,
+      "loss": 0.9484,
+      "step": 20601
+    },
+    {
+      "epoch": 0.5658180220263107,
+      "grad_norm": 0.0,
+      "learning_rate": 8.361812528714166e-06,
+      "loss": 0.8835,
+      "step": 20602
+    },
+    {
+      "epoch": 0.565845486254154,
+      "grad_norm": 0.0,
+      "learning_rate": 8.360935036140105e-06,
+      "loss": 0.9541,
+      "step": 20603
+    },
+    {
+      "epoch": 0.5658729504819972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.360057556534957e-06,
+      "loss": 0.9,
+      "step": 20604
+    },
+    {
+      "epoch": 0.5659004147098404,
+      "grad_norm": 0.0,
+      "learning_rate": 8.359180089905671e-06,
+      "loss": 0.8328,
+      "step": 20605
+    },
+    {
+      "epoch": 0.5659278789376837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.358302636259186e-06,
+      "loss": 0.7901,
+      "step": 20606
+    },
+    {
+      "epoch": 0.5659553431655269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.357425195602444e-06,
+      "loss": 0.7805,
+      "step": 20607
+    },
+    {
+      "epoch": 0.5659828073933701,
+      "grad_norm": 0.0,
+      "learning_rate": 8.356547767942395e-06,
+      "loss": 0.8492,
+      "step": 20608
+    },
+    {
+      "epoch": 0.5660102716212134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.355670353285978e-06,
+      "loss": 0.8883,
+      "step": 20609
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.0,
+      "learning_rate": 8.354792951640129e-06,
+      "loss": 0.9079,
+      "step": 20610
+    },
+    {
+      "epoch": 0.5660652000768999,
+      "grad_norm": 0.0,
+      "learning_rate": 8.353915563011798e-06,
+      "loss": 0.8992,
+      "step": 20611
+    },
+    {
+      "epoch": 0.566092664304743,
+      "grad_norm": 0.0,
+      "learning_rate": 8.353038187407925e-06,
+      "loss": 0.8738,
+      "step": 20612
+    },
+    {
+      "epoch": 0.5661201285325863,
+      "grad_norm": 0.0,
+      "learning_rate": 8.35216082483545e-06,
+      "loss": 0.8112,
+      "step": 20613
+    },
+    {
+      "epoch": 0.5661475927604296,
+      "grad_norm": 0.0,
+      "learning_rate": 8.351283475301314e-06,
+      "loss": 0.8102,
+      "step": 20614
+    },
+    {
+      "epoch": 0.5661750569882728,
+      "grad_norm": 0.0,
+      "learning_rate": 8.35040613881247e-06,
+      "loss": 0.8971,
+      "step": 20615
+    },
+    {
+      "epoch": 0.566202521216116,
+      "grad_norm": 0.0,
+      "learning_rate": 8.349528815375847e-06,
+      "loss": 0.8791,
+      "step": 20616
+    },
+    {
+      "epoch": 0.5662299854439592,
+      "grad_norm": 0.0,
+      "learning_rate": 8.348651504998391e-06,
+      "loss": 0.8141,
+      "step": 20617
+    },
+    {
+      "epoch": 0.5662574496718025,
+      "grad_norm": 0.0,
+      "learning_rate": 8.347774207687045e-06,
+      "loss": 0.8332,
+      "step": 20618
+    },
+    {
+      "epoch": 0.5662849138996457,
+      "grad_norm": 0.0,
+      "learning_rate": 8.346896923448747e-06,
+      "loss": 0.9175,
+      "step": 20619
+    },
+    {
+      "epoch": 0.5663123781274889,
+      "grad_norm": 0.0,
+      "learning_rate": 8.346019652290443e-06,
+      "loss": 0.9509,
+      "step": 20620
+    },
+    {
+      "epoch": 0.5663398423553322,
+      "grad_norm": 0.0,
+      "learning_rate": 8.345142394219077e-06,
+      "loss": 0.8903,
+      "step": 20621
+    },
+    {
+      "epoch": 0.5663673065831755,
+      "grad_norm": 0.0,
+      "learning_rate": 8.344265149241578e-06,
+      "loss": 0.7326,
+      "step": 20622
+    },
+    {
+      "epoch": 0.5663947708110186,
+      "grad_norm": 0.0,
+      "learning_rate": 8.343387917364896e-06,
+      "loss": 0.8512,
+      "step": 20623
+    },
+    {
+      "epoch": 0.5664222350388619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.342510698595972e-06,
+      "loss": 0.859,
+      "step": 20624
+    },
+    {
+      "epoch": 0.5664496992667051,
+      "grad_norm": 0.0,
+      "learning_rate": 8.341633492941744e-06,
+      "loss": 0.8493,
+      "step": 20625
+    },
+    {
+      "epoch": 0.5664771634945484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.340756300409159e-06,
+      "loss": 0.7738,
+      "step": 20626
+    },
+    {
+      "epoch": 0.5665046277223916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.339879121005147e-06,
+      "loss": 0.8496,
+      "step": 20627
+    },
+    {
+      "epoch": 0.5665320919502348,
+      "grad_norm": 0.0,
+      "learning_rate": 8.339001954736656e-06,
+      "loss": 0.9021,
+      "step": 20628
+    },
+    {
+      "epoch": 0.5665595561780781,
+      "grad_norm": 0.0,
+      "learning_rate": 8.338124801610627e-06,
+      "loss": 0.9241,
+      "step": 20629
+    },
+    {
+      "epoch": 0.5665870204059212,
+      "grad_norm": 0.0,
+      "learning_rate": 8.337247661633995e-06,
+      "loss": 0.9291,
+      "step": 20630
+    },
+    {
+      "epoch": 0.5666144846337645,
+      "grad_norm": 0.0,
+      "learning_rate": 8.336370534813704e-06,
+      "loss": 0.8485,
+      "step": 20631
+    },
+    {
+      "epoch": 0.5666419488616078,
+      "grad_norm": 0.0,
+      "learning_rate": 8.335493421156698e-06,
+      "loss": 0.8527,
+      "step": 20632
+    },
+    {
+      "epoch": 0.566669413089451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.334616320669911e-06,
+      "loss": 0.8864,
+      "step": 20633
+    },
+    {
+      "epoch": 0.5666968773172942,
+      "grad_norm": 0.0,
+      "learning_rate": 8.333739233360284e-06,
+      "loss": 1.0225,
+      "step": 20634
+    },
+    {
+      "epoch": 0.5667243415451375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.332862159234754e-06,
+      "loss": 0.8631,
+      "step": 20635
+    },
+    {
+      "epoch": 0.5667518057729807,
+      "grad_norm": 0.0,
+      "learning_rate": 8.331985098300269e-06,
+      "loss": 0.9063,
+      "step": 20636
+    },
+    {
+      "epoch": 0.566779270000824,
+      "grad_norm": 0.0,
+      "learning_rate": 8.331108050563761e-06,
+      "loss": 0.8223,
+      "step": 20637
+    },
+    {
+      "epoch": 0.5668067342286671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.330231016032179e-06,
+      "loss": 0.9003,
+      "step": 20638
+    },
+    {
+      "epoch": 0.5668341984565104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.329353994712449e-06,
+      "loss": 0.9507,
+      "step": 20639
+    },
+    {
+      "epoch": 0.5668616626843537,
+      "grad_norm": 0.0,
+      "learning_rate": 8.328476986611518e-06,
+      "loss": 0.869,
+      "step": 20640
+    },
+    {
+      "epoch": 0.5668891269121968,
+      "grad_norm": 0.0,
+      "learning_rate": 8.327599991736328e-06,
+      "loss": 0.857,
+      "step": 20641
+    },
+    {
+      "epoch": 0.5669165911400401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.326723010093811e-06,
+      "loss": 0.8516,
+      "step": 20642
+    },
+    {
+      "epoch": 0.5669440553678833,
+      "grad_norm": 0.0,
+      "learning_rate": 8.32584604169091e-06,
+      "loss": 0.9224,
+      "step": 20643
+    },
+    {
+      "epoch": 0.5669715195957266,
+      "grad_norm": 0.0,
+      "learning_rate": 8.32496908653457e-06,
+      "loss": 0.9525,
+      "step": 20644
+    },
+    {
+      "epoch": 0.5669989838235698,
+      "grad_norm": 0.0,
+      "learning_rate": 8.32409214463172e-06,
+      "loss": 0.8852,
+      "step": 20645
+    },
+    {
+      "epoch": 0.567026448051413,
+      "grad_norm": 0.0,
+      "learning_rate": 8.323215215989298e-06,
+      "loss": 0.8907,
+      "step": 20646
+    },
+    {
+      "epoch": 0.5670539122792563,
+      "grad_norm": 0.0,
+      "learning_rate": 8.322338300614248e-06,
+      "loss": 0.8797,
+      "step": 20647
+    },
+    {
+      "epoch": 0.5670813765070996,
+      "grad_norm": 0.0,
+      "learning_rate": 8.321461398513511e-06,
+      "loss": 0.8524,
+      "step": 20648
+    },
+    {
+      "epoch": 0.5671088407349427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.320584509694018e-06,
+      "loss": 0.9656,
+      "step": 20649
+    },
+    {
+      "epoch": 0.567136304962786,
+      "grad_norm": 0.0,
+      "learning_rate": 8.319707634162716e-06,
+      "loss": 1.0564,
+      "step": 20650
+    },
+    {
+      "epoch": 0.5671637691906292,
+      "grad_norm": 0.0,
+      "learning_rate": 8.318830771926532e-06,
+      "loss": 0.9907,
+      "step": 20651
+    },
+    {
+      "epoch": 0.5671912334184724,
+      "grad_norm": 0.0,
+      "learning_rate": 8.317953922992413e-06,
+      "loss": 0.8405,
+      "step": 20652
+    },
+    {
+      "epoch": 0.5672186976463157,
+      "grad_norm": 0.0,
+      "learning_rate": 8.317077087367294e-06,
+      "loss": 1.0048,
+      "step": 20653
+    },
+    {
+      "epoch": 0.5672461618741589,
+      "grad_norm": 0.0,
+      "learning_rate": 8.31620026505811e-06,
+      "loss": 0.8041,
+      "step": 20654
+    },
+    {
+      "epoch": 0.5672736261020022,
+      "grad_norm": 0.0,
+      "learning_rate": 8.315323456071805e-06,
+      "loss": 0.8717,
+      "step": 20655
+    },
+    {
+      "epoch": 0.5673010903298453,
+      "grad_norm": 0.0,
+      "learning_rate": 8.314446660415314e-06,
+      "loss": 0.9148,
+      "step": 20656
+    },
+    {
+      "epoch": 0.5673285545576886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.313569878095575e-06,
+      "loss": 0.7775,
+      "step": 20657
+    },
+    {
+      "epoch": 0.5673560187855319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.312693109119521e-06,
+      "loss": 0.8426,
+      "step": 20658
+    },
+    {
+      "epoch": 0.567383483013375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.311816353494094e-06,
+      "loss": 0.8818,
+      "step": 20659
+    },
+    {
+      "epoch": 0.5674109472412183,
+      "grad_norm": 0.0,
+      "learning_rate": 8.310939611226229e-06,
+      "loss": 0.8782,
+      "step": 20660
+    },
+    {
+      "epoch": 0.5674384114690616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.310062882322865e-06,
+      "loss": 0.879,
+      "step": 20661
+    },
+    {
+      "epoch": 0.5674658756969048,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30918616679094e-06,
+      "loss": 0.9246,
+      "step": 20662
+    },
+    {
+      "epoch": 0.567493339924748,
+      "grad_norm": 0.0,
+      "learning_rate": 8.308309464637385e-06,
+      "loss": 0.8768,
+      "step": 20663
+    },
+    {
+      "epoch": 0.5675208041525912,
+      "grad_norm": 0.0,
+      "learning_rate": 8.307432775869142e-06,
+      "loss": 0.834,
+      "step": 20664
+    },
+    {
+      "epoch": 0.5675482683804345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.306556100493148e-06,
+      "loss": 0.9346,
+      "step": 20665
+    },
+    {
+      "epoch": 0.5675757326082778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.305679438516336e-06,
+      "loss": 0.875,
+      "step": 20666
+    },
+    {
+      "epoch": 0.5676031968361209,
+      "grad_norm": 0.0,
+      "learning_rate": 8.304802789945645e-06,
+      "loss": 0.9104,
+      "step": 20667
+    },
+    {
+      "epoch": 0.5676306610639642,
+      "grad_norm": 0.0,
+      "learning_rate": 8.303926154788015e-06,
+      "loss": 0.8145,
+      "step": 20668
+    },
+    {
+      "epoch": 0.5676581252918074,
+      "grad_norm": 0.0,
+      "learning_rate": 8.303049533050377e-06,
+      "loss": 0.8554,
+      "step": 20669
+    },
+    {
+      "epoch": 0.5676855895196506,
+      "grad_norm": 0.0,
+      "learning_rate": 8.302172924739665e-06,
+      "loss": 0.8486,
+      "step": 20670
+    },
+    {
+      "epoch": 0.5677130537474939,
+      "grad_norm": 0.0,
+      "learning_rate": 8.301296329862821e-06,
+      "loss": 0.8319,
+      "step": 20671
+    },
+    {
+      "epoch": 0.5677405179753371,
+      "grad_norm": 0.0,
+      "learning_rate": 8.300419748426777e-06,
+      "loss": 0.9078,
+      "step": 20672
+    },
+    {
+      "epoch": 0.5677679822031804,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29954318043847e-06,
+      "loss": 0.8418,
+      "step": 20673
+    },
+    {
+      "epoch": 0.5677954464310236,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29866662590484e-06,
+      "loss": 0.897,
+      "step": 20674
+    },
+    {
+      "epoch": 0.5678229106588668,
+      "grad_norm": 0.0,
+      "learning_rate": 8.297790084832814e-06,
+      "loss": 0.9007,
+      "step": 20675
+    },
+    {
+      "epoch": 0.5678503748867101,
+      "grad_norm": 0.0,
+      "learning_rate": 8.296913557229334e-06,
+      "loss": 0.8838,
+      "step": 20676
+    },
+    {
+      "epoch": 0.5678778391145533,
+      "grad_norm": 0.0,
+      "learning_rate": 8.296037043101334e-06,
+      "loss": 0.9456,
+      "step": 20677
+    },
+    {
+      "epoch": 0.5679053033423965,
+      "grad_norm": 0.0,
+      "learning_rate": 8.295160542455746e-06,
+      "loss": 0.9318,
+      "step": 20678
+    },
+    {
+      "epoch": 0.5679327675702398,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29428405529951e-06,
+      "loss": 0.9222,
+      "step": 20679
+    },
+    {
+      "epoch": 0.567960231798083,
+      "grad_norm": 0.0,
+      "learning_rate": 8.293407581639563e-06,
+      "loss": 0.828,
+      "step": 20680
+    },
+    {
+      "epoch": 0.5679876960259262,
+      "grad_norm": 0.0,
+      "learning_rate": 8.292531121482835e-06,
+      "loss": 0.8626,
+      "step": 20681
+    },
+    {
+      "epoch": 0.5680151602537694,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29165467483626e-06,
+      "loss": 0.8039,
+      "step": 20682
+    },
+    {
+      "epoch": 0.5680426244816127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.290778241706772e-06,
+      "loss": 0.838,
+      "step": 20683
+    },
+    {
+      "epoch": 0.568070088709456,
+      "grad_norm": 0.0,
+      "learning_rate": 8.289901822101315e-06,
+      "loss": 0.8973,
+      "step": 20684
+    },
+    {
+      "epoch": 0.5680975529372991,
+      "grad_norm": 0.0,
+      "learning_rate": 8.289025416026817e-06,
+      "loss": 0.7977,
+      "step": 20685
+    },
+    {
+      "epoch": 0.5681250171651424,
+      "grad_norm": 0.0,
+      "learning_rate": 8.28814902349021e-06,
+      "loss": 0.827,
+      "step": 20686
+    },
+    {
+      "epoch": 0.5681524813929857,
+      "grad_norm": 0.0,
+      "learning_rate": 8.287272644498428e-06,
+      "loss": 0.7719,
+      "step": 20687
+    },
+    {
+      "epoch": 0.5681799456208289,
+      "grad_norm": 0.0,
+      "learning_rate": 8.286396279058411e-06,
+      "loss": 0.877,
+      "step": 20688
+    },
+    {
+      "epoch": 0.5682074098486721,
+      "grad_norm": 0.0,
+      "learning_rate": 8.28551992717709e-06,
+      "loss": 0.7612,
+      "step": 20689
+    },
+    {
+      "epoch": 0.5682348740765153,
+      "grad_norm": 0.0,
+      "learning_rate": 8.284643588861399e-06,
+      "loss": 0.8575,
+      "step": 20690
+    },
+    {
+      "epoch": 0.5682623383043586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.283767264118278e-06,
+      "loss": 0.8668,
+      "step": 20691
+    },
+    {
+      "epoch": 0.5682898025322018,
+      "grad_norm": 0.0,
+      "learning_rate": 8.282890952954649e-06,
+      "loss": 0.8268,
+      "step": 20692
+    },
+    {
+      "epoch": 0.568317266760045,
+      "grad_norm": 0.0,
+      "learning_rate": 8.282014655377455e-06,
+      "loss": 0.791,
+      "step": 20693
+    },
+    {
+      "epoch": 0.5683447309878883,
+      "grad_norm": 0.0,
+      "learning_rate": 8.281138371393624e-06,
+      "loss": 0.9145,
+      "step": 20694
+    },
+    {
+      "epoch": 0.5683721952157315,
+      "grad_norm": 0.0,
+      "learning_rate": 8.280262101010092e-06,
+      "loss": 0.8114,
+      "step": 20695
+    },
+    {
+      "epoch": 0.5683996594435747,
+      "grad_norm": 0.0,
+      "learning_rate": 8.279385844233794e-06,
+      "loss": 0.8322,
+      "step": 20696
+    },
+    {
+      "epoch": 0.568427123671418,
+      "grad_norm": 0.0,
+      "learning_rate": 8.278509601071664e-06,
+      "loss": 0.799,
+      "step": 20697
+    },
+    {
+      "epoch": 0.5684545878992612,
+      "grad_norm": 0.0,
+      "learning_rate": 8.277633371530628e-06,
+      "loss": 0.9018,
+      "step": 20698
+    },
+    {
+      "epoch": 0.5684820521271045,
+      "grad_norm": 0.0,
+      "learning_rate": 8.276757155617626e-06,
+      "loss": 0.8985,
+      "step": 20699
+    },
+    {
+      "epoch": 0.5685095163549477,
+      "grad_norm": 0.0,
+      "learning_rate": 8.275880953339587e-06,
+      "loss": 0.8665,
+      "step": 20700
+    },
+    {
+      "epoch": 0.5685369805827909,
+      "grad_norm": 0.0,
+      "learning_rate": 8.275004764703451e-06,
+      "loss": 0.936,
+      "step": 20701
+    },
+    {
+      "epoch": 0.5685644448106342,
+      "grad_norm": 0.0,
+      "learning_rate": 8.274128589716141e-06,
+      "loss": 0.8543,
+      "step": 20702
+    },
+    {
+      "epoch": 0.5685919090384773,
+      "grad_norm": 0.0,
+      "learning_rate": 8.273252428384599e-06,
+      "loss": 0.7081,
+      "step": 20703
+    },
+    {
+      "epoch": 0.5686193732663206,
+      "grad_norm": 0.0,
+      "learning_rate": 8.272376280715746e-06,
+      "loss": 0.8334,
+      "step": 20704
+    },
+    {
+      "epoch": 0.5686468374941639,
+      "grad_norm": 0.0,
+      "learning_rate": 8.271500146716525e-06,
+      "loss": 0.8967,
+      "step": 20705
+    },
+    {
+      "epoch": 0.5686743017220071,
+      "grad_norm": 0.0,
+      "learning_rate": 8.270624026393862e-06,
+      "loss": 0.7998,
+      "step": 20706
+    },
+    {
+      "epoch": 0.5687017659498503,
+      "grad_norm": 0.0,
+      "learning_rate": 8.269747919754693e-06,
+      "loss": 0.9005,
+      "step": 20707
+    },
+    {
+      "epoch": 0.5687292301776935,
+      "grad_norm": 0.0,
+      "learning_rate": 8.268871826805952e-06,
+      "loss": 0.832,
+      "step": 20708
+    },
+    {
+      "epoch": 0.5687566944055368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.267995747554566e-06,
+      "loss": 0.9281,
+      "step": 20709
+    },
+    {
+      "epoch": 0.56878415863338,
+      "grad_norm": 0.0,
+      "learning_rate": 8.267119682007466e-06,
+      "loss": 0.839,
+      "step": 20710
+    },
+    {
+      "epoch": 0.5688116228612232,
+      "grad_norm": 0.0,
+      "learning_rate": 8.266243630171586e-06,
+      "loss": 0.8857,
+      "step": 20711
+    },
+    {
+      "epoch": 0.5688390870890665,
+      "grad_norm": 0.0,
+      "learning_rate": 8.265367592053862e-06,
+      "loss": 0.8611,
+      "step": 20712
+    },
+    {
+      "epoch": 0.5688665513169098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.264491567661217e-06,
+      "loss": 0.9769,
+      "step": 20713
+    },
+    {
+      "epoch": 0.5688940155447529,
+      "grad_norm": 0.0,
+      "learning_rate": 8.263615557000587e-06,
+      "loss": 0.8716,
+      "step": 20714
+    },
+    {
+      "epoch": 0.5689214797725962,
+      "grad_norm": 0.0,
+      "learning_rate": 8.262739560078908e-06,
+      "loss": 0.9249,
+      "step": 20715
+    },
+    {
+      "epoch": 0.5689489440004394,
+      "grad_norm": 0.0,
+      "learning_rate": 8.261863576903104e-06,
+      "loss": 0.8838,
+      "step": 20716
+    },
+    {
+      "epoch": 0.5689764082282827,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260987607480109e-06,
+      "loss": 0.8503,
+      "step": 20717
+    },
+    {
+      "epoch": 0.5690038724561259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260111651816853e-06,
+      "loss": 0.8832,
+      "step": 20718
+    },
+    {
+      "epoch": 0.5690313366839691,
+      "grad_norm": 0.0,
+      "learning_rate": 8.259235709920266e-06,
+      "loss": 0.8645,
+      "step": 20719
+    },
+    {
+      "epoch": 0.5690588009118124,
+      "grad_norm": 0.0,
+      "learning_rate": 8.258359781797283e-06,
+      "loss": 0.8541,
+      "step": 20720
+    },
+    {
+      "epoch": 0.5690862651396555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.257483867454834e-06,
+      "loss": 0.9125,
+      "step": 20721
+    },
+    {
+      "epoch": 0.5691137293674988,
+      "grad_norm": 0.0,
+      "learning_rate": 8.256607966899843e-06,
+      "loss": 0.9313,
+      "step": 20722
+    },
+    {
+      "epoch": 0.5691411935953421,
+      "grad_norm": 0.0,
+      "learning_rate": 8.255732080139248e-06,
+      "loss": 0.7949,
+      "step": 20723
+    },
+    {
+      "epoch": 0.5691686578231853,
+      "grad_norm": 0.0,
+      "learning_rate": 8.254856207179975e-06,
+      "loss": 0.7619,
+      "step": 20724
+    },
+    {
+      "epoch": 0.5691961220510285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.253980348028955e-06,
+      "loss": 0.8132,
+      "step": 20725
+    },
+    {
+      "epoch": 0.5692235862788718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.253104502693119e-06,
+      "loss": 0.8825,
+      "step": 20726
+    },
+    {
+      "epoch": 0.569251050506715,
+      "grad_norm": 0.0,
+      "learning_rate": 8.2522286711794e-06,
+      "loss": 0.9207,
+      "step": 20727
+    },
+    {
+      "epoch": 0.5692785147345583,
+      "grad_norm": 0.0,
+      "learning_rate": 8.251352853494722e-06,
+      "loss": 0.879,
+      "step": 20728
+    },
+    {
+      "epoch": 0.5693059789624014,
+      "grad_norm": 0.0,
+      "learning_rate": 8.250477049646018e-06,
+      "loss": 0.8978,
+      "step": 20729
+    },
+    {
+      "epoch": 0.5693334431902447,
+      "grad_norm": 0.0,
+      "learning_rate": 8.249601259640216e-06,
+      "loss": 0.8133,
+      "step": 20730
+    },
+    {
+      "epoch": 0.569360907418088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.248725483484248e-06,
+      "loss": 0.835,
+      "step": 20731
+    },
+    {
+      "epoch": 0.5693883716459311,
+      "grad_norm": 0.0,
+      "learning_rate": 8.247849721185044e-06,
+      "loss": 0.78,
+      "step": 20732
+    },
+    {
+      "epoch": 0.5694158358737744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.246973972749534e-06,
+      "loss": 0.8329,
+      "step": 20733
+    },
+    {
+      "epoch": 0.5694433001016176,
+      "grad_norm": 0.0,
+      "learning_rate": 8.246098238184638e-06,
+      "loss": 0.9347,
+      "step": 20734
+    },
+    {
+      "epoch": 0.5694707643294609,
+      "grad_norm": 0.0,
+      "learning_rate": 8.245222517497295e-06,
+      "loss": 0.7427,
+      "step": 20735
+    },
+    {
+      "epoch": 0.5694982285573041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.244346810694433e-06,
+      "loss": 0.8724,
+      "step": 20736
+    },
+    {
+      "epoch": 0.5695256927851473,
+      "grad_norm": 0.0,
+      "learning_rate": 8.243471117782978e-06,
+      "loss": 0.7946,
+      "step": 20737
+    },
+    {
+      "epoch": 0.5695531570129906,
+      "grad_norm": 0.0,
+      "learning_rate": 8.242595438769857e-06,
+      "loss": 0.8222,
+      "step": 20738
+    },
+    {
+      "epoch": 0.5695806212408339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.241719773662008e-06,
+      "loss": 0.9179,
+      "step": 20739
+    },
+    {
+      "epoch": 0.569608085468677,
+      "grad_norm": 0.0,
+      "learning_rate": 8.240844122466349e-06,
+      "loss": 0.7861,
+      "step": 20740
+    },
+    {
+      "epoch": 0.5696355496965203,
+      "grad_norm": 0.0,
+      "learning_rate": 8.239968485189813e-06,
+      "loss": 0.8238,
+      "step": 20741
+    },
+    {
+      "epoch": 0.5696630139243635,
+      "grad_norm": 0.0,
+      "learning_rate": 8.239092861839327e-06,
+      "loss": 0.898,
+      "step": 20742
+    },
+    {
+      "epoch": 0.5696904781522067,
+      "grad_norm": 0.0,
+      "learning_rate": 8.23821725242182e-06,
+      "loss": 0.8325,
+      "step": 20743
+    },
+    {
+      "epoch": 0.56971794238005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.237341656944226e-06,
+      "loss": 0.9467,
+      "step": 20744
+    },
+    {
+      "epoch": 0.5697454066078932,
+      "grad_norm": 0.0,
+      "learning_rate": 8.236466075413464e-06,
+      "loss": 0.8174,
+      "step": 20745
+    },
+    {
+      "epoch": 0.5697728708357365,
+      "grad_norm": 0.0,
+      "learning_rate": 8.235590507836466e-06,
+      "loss": 0.833,
+      "step": 20746
+    },
+    {
+      "epoch": 0.5698003350635796,
+      "grad_norm": 0.0,
+      "learning_rate": 8.234714954220158e-06,
+      "loss": 0.7692,
+      "step": 20747
+    },
+    {
+      "epoch": 0.5698277992914229,
+      "grad_norm": 0.0,
+      "learning_rate": 8.233839414571471e-06,
+      "loss": 0.8609,
+      "step": 20748
+    },
+    {
+      "epoch": 0.5698552635192662,
+      "grad_norm": 0.0,
+      "learning_rate": 8.232963888897328e-06,
+      "loss": 0.8341,
+      "step": 20749
+    },
+    {
+      "epoch": 0.5698827277471094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.232088377204665e-06,
+      "loss": 0.821,
+      "step": 20750
+    },
+    {
+      "epoch": 0.5699101919749526,
+      "grad_norm": 0.0,
+      "learning_rate": 8.231212879500397e-06,
+      "loss": 0.8228,
+      "step": 20751
+    },
+    {
+      "epoch": 0.5699376562027959,
+      "grad_norm": 0.0,
+      "learning_rate": 8.230337395791458e-06,
+      "loss": 0.8208,
+      "step": 20752
+    },
+    {
+      "epoch": 0.5699651204306391,
+      "grad_norm": 0.0,
+      "learning_rate": 8.22946192608478e-06,
+      "loss": 0.808,
+      "step": 20753
+    },
+    {
+      "epoch": 0.5699925846584823,
+      "grad_norm": 0.0,
+      "learning_rate": 8.22858647038728e-06,
+      "loss": 0.8845,
+      "step": 20754
+    },
+    {
+      "epoch": 0.5700200488863255,
+      "grad_norm": 0.0,
+      "learning_rate": 8.227711028705892e-06,
+      "loss": 0.9153,
+      "step": 20755
+    },
+    {
+      "epoch": 0.5700475131141688,
+      "grad_norm": 0.0,
+      "learning_rate": 8.226835601047545e-06,
+      "loss": 0.8878,
+      "step": 20756
+    },
+    {
+      "epoch": 0.5700749773420121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.225960187419158e-06,
+      "loss": 0.7847,
+      "step": 20757
+    },
+    {
+      "epoch": 0.5701024415698552,
+      "grad_norm": 0.0,
+      "learning_rate": 8.22508478782766e-06,
+      "loss": 0.8324,
+      "step": 20758
+    },
+    {
+      "epoch": 0.5701299057976985,
+      "grad_norm": 0.0,
+      "learning_rate": 8.224209402279978e-06,
+      "loss": 0.8266,
+      "step": 20759
+    },
+    {
+      "epoch": 0.5701573700255417,
+      "grad_norm": 0.0,
+      "learning_rate": 8.223334030783042e-06,
+      "loss": 0.9157,
+      "step": 20760
+    },
+    {
+      "epoch": 0.570184834253385,
+      "grad_norm": 0.0,
+      "learning_rate": 8.222458673343774e-06,
+      "loss": 0.8121,
+      "step": 20761
+    },
+    {
+      "epoch": 0.5702122984812282,
+      "grad_norm": 0.0,
+      "learning_rate": 8.221583329969106e-06,
+      "loss": 0.8486,
+      "step": 20762
+    },
+    {
+      "epoch": 0.5702397627090714,
+      "grad_norm": 0.0,
+      "learning_rate": 8.220708000665952e-06,
+      "loss": 0.7916,
+      "step": 20763
+    },
+    {
+      "epoch": 0.5702672269369147,
+      "grad_norm": 0.0,
+      "learning_rate": 8.219832685441248e-06,
+      "loss": 0.8436,
+      "step": 20764
+    },
+    {
+      "epoch": 0.570294691164758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.21895738430192e-06,
+      "loss": 0.8522,
+      "step": 20765
+    },
+    {
+      "epoch": 0.5703221553926011,
+      "grad_norm": 0.0,
+      "learning_rate": 8.218082097254887e-06,
+      "loss": 0.8604,
+      "step": 20766
+    },
+    {
+      "epoch": 0.5703496196204444,
+      "grad_norm": 0.0,
+      "learning_rate": 8.21720682430708e-06,
+      "loss": 0.9364,
+      "step": 20767
+    },
+    {
+      "epoch": 0.5703770838482876,
+      "grad_norm": 0.0,
+      "learning_rate": 8.216331565465427e-06,
+      "loss": 0.9323,
+      "step": 20768
+    },
+    {
+      "epoch": 0.5704045480761308,
+      "grad_norm": 0.0,
+      "learning_rate": 8.21545632073685e-06,
+      "loss": 0.8499,
+      "step": 20769
+    },
+    {
+      "epoch": 0.5704320123039741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.214581090128269e-06,
+      "loss": 1.0087,
+      "step": 20770
+    },
+    {
+      "epoch": 0.5704594765318173,
+      "grad_norm": 0.0,
+      "learning_rate": 8.213705873646613e-06,
+      "loss": 0.7585,
+      "step": 20771
+    },
+    {
+      "epoch": 0.5704869407596606,
+      "grad_norm": 0.0,
+      "learning_rate": 8.212830671298812e-06,
+      "loss": 0.8024,
+      "step": 20772
+    },
+    {
+      "epoch": 0.5705144049875038,
+      "grad_norm": 0.0,
+      "learning_rate": 8.211955483091784e-06,
+      "loss": 0.9322,
+      "step": 20773
+    },
+    {
+      "epoch": 0.570541869215347,
+      "grad_norm": 0.0,
+      "learning_rate": 8.211080309032461e-06,
+      "loss": 0.831,
+      "step": 20774
+    },
+    {
+      "epoch": 0.5705693334431903,
+      "grad_norm": 0.0,
+      "learning_rate": 8.210205149127762e-06,
+      "loss": 0.8458,
+      "step": 20775
+    },
+    {
+      "epoch": 0.5705967976710334,
+      "grad_norm": 0.0,
+      "learning_rate": 8.20933000338461e-06,
+      "loss": 0.843,
+      "step": 20776
+    },
+    {
+      "epoch": 0.5706242618988767,
+      "grad_norm": 0.0,
+      "learning_rate": 8.208454871809937e-06,
+      "loss": 0.932,
+      "step": 20777
+    },
+    {
+      "epoch": 0.57065172612672,
+      "grad_norm": 0.0,
+      "learning_rate": 8.20757975441066e-06,
+      "loss": 1.0029,
+      "step": 20778
+    },
+    {
+      "epoch": 0.5706791903545632,
+      "grad_norm": 0.0,
+      "learning_rate": 8.206704651193705e-06,
+      "loss": 0.9165,
+      "step": 20779
+    },
+    {
+      "epoch": 0.5707066545824064,
+      "grad_norm": 0.0,
+      "learning_rate": 8.205829562166004e-06,
+      "loss": 1.0023,
+      "step": 20780
+    },
+    {
+      "epoch": 0.5707341188102496,
+      "grad_norm": 0.0,
+      "learning_rate": 8.20495448733447e-06,
+      "loss": 0.7441,
+      "step": 20781
+    },
+    {
+      "epoch": 0.5707615830380929,
+      "grad_norm": 0.0,
+      "learning_rate": 8.204079426706031e-06,
+      "loss": 0.8525,
+      "step": 20782
+    },
+    {
+      "epoch": 0.5707890472659362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.20320438028761e-06,
+      "loss": 0.8297,
+      "step": 20783
+    },
+    {
+      "epoch": 0.5708165114937793,
+      "grad_norm": 0.0,
+      "learning_rate": 8.202329348086135e-06,
+      "loss": 0.8911,
+      "step": 20784
+    },
+    {
+      "epoch": 0.5708439757216226,
+      "grad_norm": 0.0,
+      "learning_rate": 8.201454330108525e-06,
+      "loss": 0.7997,
+      "step": 20785
+    },
+    {
+      "epoch": 0.5708714399494659,
+      "grad_norm": 0.0,
+      "learning_rate": 8.200579326361708e-06,
+      "loss": 0.8332,
+      "step": 20786
+    },
+    {
+      "epoch": 0.570898904177309,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1997043368526e-06,
+      "loss": 0.8576,
+      "step": 20787
+    },
+    {
+      "epoch": 0.5709263684051523,
+      "grad_norm": 0.0,
+      "learning_rate": 8.198829361588129e-06,
+      "loss": 0.7566,
+      "step": 20788
+    },
+    {
+      "epoch": 0.5709538326329955,
+      "grad_norm": 0.0,
+      "learning_rate": 8.19795440057522e-06,
+      "loss": 0.8659,
+      "step": 20789
+    },
+    {
+      "epoch": 0.5709812968608388,
+      "grad_norm": 0.0,
+      "learning_rate": 8.197079453820792e-06,
+      "loss": 0.8279,
+      "step": 20790
+    },
+    {
+      "epoch": 0.571008761088682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.196204521331771e-06,
+      "loss": 0.8572,
+      "step": 20791
+    },
+    {
+      "epoch": 0.5710362253165252,
+      "grad_norm": 0.0,
+      "learning_rate": 8.195329603115082e-06,
+      "loss": 0.8185,
+      "step": 20792
+    },
+    {
+      "epoch": 0.5710636895443685,
+      "grad_norm": 0.0,
+      "learning_rate": 8.19445469917764e-06,
+      "loss": 0.8668,
+      "step": 20793
+    },
+    {
+      "epoch": 0.5710911537722116,
+      "grad_norm": 0.0,
+      "learning_rate": 8.193579809526372e-06,
+      "loss": 0.828,
+      "step": 20794
+    },
+    {
+      "epoch": 0.5711186180000549,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1927049341682e-06,
+      "loss": 0.8198,
+      "step": 20795
+    },
+    {
+      "epoch": 0.5711460822278982,
+      "grad_norm": 0.0,
+      "learning_rate": 8.191830073110048e-06,
+      "loss": 0.8311,
+      "step": 20796
+    },
+    {
+      "epoch": 0.5711735464557414,
+      "grad_norm": 0.0,
+      "learning_rate": 8.190955226358835e-06,
+      "loss": 0.8393,
+      "step": 20797
+    },
+    {
+      "epoch": 0.5712010106835846,
+      "grad_norm": 0.0,
+      "learning_rate": 8.19008039392149e-06,
+      "loss": 0.8362,
+      "step": 20798
+    },
+    {
+      "epoch": 0.5712284749114279,
+      "grad_norm": 0.0,
+      "learning_rate": 8.189205575804925e-06,
+      "loss": 0.7627,
+      "step": 20799
+    },
+    {
+      "epoch": 0.5712559391392711,
+      "grad_norm": 0.0,
+      "learning_rate": 8.188330772016067e-06,
+      "loss": 0.805,
+      "step": 20800
+    },
+    {
+      "epoch": 0.5712834033671144,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18745598256184e-06,
+      "loss": 0.9187,
+      "step": 20801
+    },
+    {
+      "epoch": 0.5713108675949575,
+      "grad_norm": 0.0,
+      "learning_rate": 8.186581207449162e-06,
+      "loss": 0.8043,
+      "step": 20802
+    },
+    {
+      "epoch": 0.5713383318228008,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18570644668496e-06,
+      "loss": 0.8394,
+      "step": 20803
+    },
+    {
+      "epoch": 0.5713657960506441,
+      "grad_norm": 0.0,
+      "learning_rate": 8.184831700276146e-06,
+      "loss": 0.7984,
+      "step": 20804
+    },
+    {
+      "epoch": 0.5713932602784872,
+      "grad_norm": 0.0,
+      "learning_rate": 8.183956968229648e-06,
+      "loss": 0.7907,
+      "step": 20805
+    },
+    {
+      "epoch": 0.5714207245063305,
+      "grad_norm": 0.0,
+      "learning_rate": 8.183082250552385e-06,
+      "loss": 0.9002,
+      "step": 20806
+    },
+    {
+      "epoch": 0.5714481887341737,
+      "grad_norm": 0.0,
+      "learning_rate": 8.182207547251279e-06,
+      "loss": 0.7784,
+      "step": 20807
+    },
+    {
+      "epoch": 0.571475652962017,
+      "grad_norm": 0.0,
+      "learning_rate": 8.181332858333254e-06,
+      "loss": 0.9156,
+      "step": 20808
+    },
+    {
+      "epoch": 0.5715031171898602,
+      "grad_norm": 0.0,
+      "learning_rate": 8.180458183805227e-06,
+      "loss": 0.8922,
+      "step": 20809
+    },
+    {
+      "epoch": 0.5715305814177034,
+      "grad_norm": 0.0,
+      "learning_rate": 8.17958352367412e-06,
+      "loss": 0.8033,
+      "step": 20810
+    },
+    {
+      "epoch": 0.5715580456455467,
+      "grad_norm": 0.0,
+      "learning_rate": 8.178708877946852e-06,
+      "loss": 0.8784,
+      "step": 20811
+    },
+    {
+      "epoch": 0.57158550987339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.177834246630342e-06,
+      "loss": 0.8338,
+      "step": 20812
+    },
+    {
+      "epoch": 0.5716129741012331,
+      "grad_norm": 0.0,
+      "learning_rate": 8.176959629731518e-06,
+      "loss": 0.9655,
+      "step": 20813
+    },
+    {
+      "epoch": 0.5716404383290764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.176085027257293e-06,
+      "loss": 0.8208,
+      "step": 20814
+    },
+    {
+      "epoch": 0.5716679025569196,
+      "grad_norm": 0.0,
+      "learning_rate": 8.175210439214594e-06,
+      "loss": 0.845,
+      "step": 20815
+    },
+    {
+      "epoch": 0.5716953667847628,
+      "grad_norm": 0.0,
+      "learning_rate": 8.174335865610333e-06,
+      "loss": 0.9032,
+      "step": 20816
+    },
+    {
+      "epoch": 0.5717228310126061,
+      "grad_norm": 0.0,
+      "learning_rate": 8.173461306451435e-06,
+      "loss": 0.8718,
+      "step": 20817
+    },
+    {
+      "epoch": 0.5717502952404493,
+      "grad_norm": 0.0,
+      "learning_rate": 8.172586761744817e-06,
+      "loss": 0.9029,
+      "step": 20818
+    },
+    {
+      "epoch": 0.5717777594682926,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1717122314974e-06,
+      "loss": 0.8563,
+      "step": 20819
+    },
+    {
+      "epoch": 0.5718052236961357,
+      "grad_norm": 0.0,
+      "learning_rate": 8.170837715716107e-06,
+      "loss": 0.923,
+      "step": 20820
+    },
+    {
+      "epoch": 0.571832687923979,
+      "grad_norm": 0.0,
+      "learning_rate": 8.169963214407856e-06,
+      "loss": 0.7494,
+      "step": 20821
+    },
+    {
+      "epoch": 0.5718601521518223,
+      "grad_norm": 0.0,
+      "learning_rate": 8.169088727579562e-06,
+      "loss": 0.8624,
+      "step": 20822
+    },
+    {
+      "epoch": 0.5718876163796655,
+      "grad_norm": 0.0,
+      "learning_rate": 8.168214255238145e-06,
+      "loss": 0.9192,
+      "step": 20823
+    },
+    {
+      "epoch": 0.5719150806075087,
+      "grad_norm": 0.0,
+      "learning_rate": 8.167339797390527e-06,
+      "loss": 0.8779,
+      "step": 20824
+    },
+    {
+      "epoch": 0.571942544835352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.16646535404363e-06,
+      "loss": 0.9041,
+      "step": 20825
+    },
+    {
+      "epoch": 0.5719700090631952,
+      "grad_norm": 0.0,
+      "learning_rate": 8.165590925204366e-06,
+      "loss": 0.828,
+      "step": 20826
+    },
+    {
+      "epoch": 0.5719974732910384,
+      "grad_norm": 0.0,
+      "learning_rate": 8.164716510879662e-06,
+      "loss": 0.7876,
+      "step": 20827
+    },
+    {
+      "epoch": 0.5720249375188816,
+      "grad_norm": 0.0,
+      "learning_rate": 8.163842111076427e-06,
+      "loss": 0.7823,
+      "step": 20828
+    },
+    {
+      "epoch": 0.5720524017467249,
+      "grad_norm": 0.0,
+      "learning_rate": 8.162967725801586e-06,
+      "loss": 0.892,
+      "step": 20829
+    },
+    {
+      "epoch": 0.5720798659745682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.162093355062056e-06,
+      "loss": 0.9136,
+      "step": 20830
+    },
+    {
+      "epoch": 0.5721073302024113,
+      "grad_norm": 0.0,
+      "learning_rate": 8.161218998864753e-06,
+      "loss": 0.8356,
+      "step": 20831
+    },
+    {
+      "epoch": 0.5721347944302546,
+      "grad_norm": 0.0,
+      "learning_rate": 8.160344657216602e-06,
+      "loss": 0.9124,
+      "step": 20832
+    },
+    {
+      "epoch": 0.5721622586580978,
+      "grad_norm": 0.0,
+      "learning_rate": 8.159470330124517e-06,
+      "loss": 0.9213,
+      "step": 20833
+    },
+    {
+      "epoch": 0.572189722885941,
+      "grad_norm": 0.0,
+      "learning_rate": 8.158596017595413e-06,
+      "loss": 0.8236,
+      "step": 20834
+    },
+    {
+      "epoch": 0.5722171871137843,
+      "grad_norm": 0.0,
+      "learning_rate": 8.15772171963621e-06,
+      "loss": 0.8178,
+      "step": 20835
+    },
+    {
+      "epoch": 0.5722446513416275,
+      "grad_norm": 0.0,
+      "learning_rate": 8.156847436253825e-06,
+      "loss": 0.8802,
+      "step": 20836
+    },
+    {
+      "epoch": 0.5722721155694708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.15597316745518e-06,
+      "loss": 0.8197,
+      "step": 20837
+    },
+    {
+      "epoch": 0.572299579797314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.155098913247189e-06,
+      "loss": 0.8861,
+      "step": 20838
+    },
+    {
+      "epoch": 0.5723270440251572,
+      "grad_norm": 0.0,
+      "learning_rate": 8.154224673636773e-06,
+      "loss": 0.8866,
+      "step": 20839
+    },
+    {
+      "epoch": 0.5723545082530005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.153350448630842e-06,
+      "loss": 0.8752,
+      "step": 20840
+    },
+    {
+      "epoch": 0.5723819724808437,
+      "grad_norm": 0.0,
+      "learning_rate": 8.152476238236319e-06,
+      "loss": 0.8979,
+      "step": 20841
+    },
+    {
+      "epoch": 0.5724094367086869,
+      "grad_norm": 0.0,
+      "learning_rate": 8.151602042460119e-06,
+      "loss": 0.8575,
+      "step": 20842
+    },
+    {
+      "epoch": 0.5724369009365302,
+      "grad_norm": 0.0,
+      "learning_rate": 8.150727861309158e-06,
+      "loss": 0.8938,
+      "step": 20843
+    },
+    {
+      "epoch": 0.5724643651643734,
+      "grad_norm": 0.0,
+      "learning_rate": 8.149853694790357e-06,
+      "loss": 0.7605,
+      "step": 20844
+    },
+    {
+      "epoch": 0.5724918293922167,
+      "grad_norm": 0.0,
+      "learning_rate": 8.148979542910633e-06,
+      "loss": 0.8693,
+      "step": 20845
+    },
+    {
+      "epoch": 0.5725192936200598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.148105405676895e-06,
+      "loss": 0.8883,
+      "step": 20846
+    },
+    {
+      "epoch": 0.5725467578479031,
+      "grad_norm": 0.0,
+      "learning_rate": 8.147231283096067e-06,
+      "loss": 0.8935,
+      "step": 20847
+    },
+    {
+      "epoch": 0.5725742220757464,
+      "grad_norm": 0.0,
+      "learning_rate": 8.14635717517506e-06,
+      "loss": 0.8318,
+      "step": 20848
+    },
+    {
+      "epoch": 0.5726016863035895,
+      "grad_norm": 0.0,
+      "learning_rate": 8.145483081920795e-06,
+      "loss": 0.947,
+      "step": 20849
+    },
+    {
+      "epoch": 0.5726291505314328,
+      "grad_norm": 0.0,
+      "learning_rate": 8.144609003340186e-06,
+      "loss": 0.9033,
+      "step": 20850
+    },
+    {
+      "epoch": 0.5726566147592761,
+      "grad_norm": 0.0,
+      "learning_rate": 8.143734939440153e-06,
+      "loss": 0.792,
+      "step": 20851
+    },
+    {
+      "epoch": 0.5726840789871193,
+      "grad_norm": 0.0,
+      "learning_rate": 8.142860890227603e-06,
+      "loss": 0.8198,
+      "step": 20852
+    },
+    {
+      "epoch": 0.5727115432149625,
+      "grad_norm": 0.0,
+      "learning_rate": 8.141986855709459e-06,
+      "loss": 0.9365,
+      "step": 20853
+    },
+    {
+      "epoch": 0.5727390074428057,
+      "grad_norm": 0.0,
+      "learning_rate": 8.141112835892635e-06,
+      "loss": 0.8947,
+      "step": 20854
+    },
+    {
+      "epoch": 0.572766471670649,
+      "grad_norm": 0.0,
+      "learning_rate": 8.140238830784044e-06,
+      "loss": 0.8602,
+      "step": 20855
+    },
+    {
+      "epoch": 0.5727939358984923,
+      "grad_norm": 0.0,
+      "learning_rate": 8.139364840390607e-06,
+      "loss": 0.9052,
+      "step": 20856
+    },
+    {
+      "epoch": 0.5728214001263354,
+      "grad_norm": 0.0,
+      "learning_rate": 8.138490864719237e-06,
+      "loss": 0.8352,
+      "step": 20857
+    },
+    {
+      "epoch": 0.5728488643541787,
+      "grad_norm": 0.0,
+      "learning_rate": 8.137616903776845e-06,
+      "loss": 0.9249,
+      "step": 20858
+    },
+    {
+      "epoch": 0.5728763285820219,
+      "grad_norm": 0.0,
+      "learning_rate": 8.13674295757035e-06,
+      "loss": 0.7963,
+      "step": 20859
+    },
+    {
+      "epoch": 0.5729037928098651,
+      "grad_norm": 0.0,
+      "learning_rate": 8.135869026106666e-06,
+      "loss": 0.8164,
+      "step": 20860
+    },
+    {
+      "epoch": 0.5729312570377084,
+      "grad_norm": 0.0,
+      "learning_rate": 8.13499510939271e-06,
+      "loss": 0.9078,
+      "step": 20861
+    },
+    {
+      "epoch": 0.5729587212655516,
+      "grad_norm": 0.0,
+      "learning_rate": 8.134121207435397e-06,
+      "loss": 0.9167,
+      "step": 20862
+    },
+    {
+      "epoch": 0.5729861854933949,
+      "grad_norm": 0.0,
+      "learning_rate": 8.133247320241635e-06,
+      "loss": 0.9295,
+      "step": 20863
+    },
+    {
+      "epoch": 0.5730136497212381,
+      "grad_norm": 0.0,
+      "learning_rate": 8.132373447818345e-06,
+      "loss": 0.8257,
+      "step": 20864
+    },
+    {
+      "epoch": 0.5730411139490813,
+      "grad_norm": 0.0,
+      "learning_rate": 8.131499590172441e-06,
+      "loss": 0.8593,
+      "step": 20865
+    },
+    {
+      "epoch": 0.5730685781769246,
+      "grad_norm": 0.0,
+      "learning_rate": 8.130625747310835e-06,
+      "loss": 0.866,
+      "step": 20866
+    },
+    {
+      "epoch": 0.5730960424047677,
+      "grad_norm": 0.0,
+      "learning_rate": 8.12975191924044e-06,
+      "loss": 0.8805,
+      "step": 20867
+    },
+    {
+      "epoch": 0.573123506632611,
+      "grad_norm": 0.0,
+      "learning_rate": 8.128878105968178e-06,
+      "loss": 0.9576,
+      "step": 20868
+    },
+    {
+      "epoch": 0.5731509708604543,
+      "grad_norm": 0.0,
+      "learning_rate": 8.128004307500956e-06,
+      "loss": 0.7935,
+      "step": 20869
+    },
+    {
+      "epoch": 0.5731784350882975,
+      "grad_norm": 0.0,
+      "learning_rate": 8.127130523845686e-06,
+      "loss": 0.8401,
+      "step": 20870
+    },
+    {
+      "epoch": 0.5732058993161407,
+      "grad_norm": 0.0,
+      "learning_rate": 8.126256755009284e-06,
+      "loss": 0.8425,
+      "step": 20871
+    },
+    {
+      "epoch": 0.5732333635439839,
+      "grad_norm": 0.0,
+      "learning_rate": 8.125383000998667e-06,
+      "loss": 0.8493,
+      "step": 20872
+    },
+    {
+      "epoch": 0.5732608277718272,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124509261820745e-06,
+      "loss": 0.8392,
+      "step": 20873
+    },
+    {
+      "epoch": 0.5732882919996705,
+      "grad_norm": 0.0,
+      "learning_rate": 8.123635537482434e-06,
+      "loss": 0.7961,
+      "step": 20874
+    },
+    {
+      "epoch": 0.5733157562275136,
+      "grad_norm": 0.0,
+      "learning_rate": 8.122761827990643e-06,
+      "loss": 0.8388,
+      "step": 20875
+    },
+    {
+      "epoch": 0.5733432204553569,
+      "grad_norm": 0.0,
+      "learning_rate": 8.121888133352286e-06,
+      "loss": 0.8614,
+      "step": 20876
+    },
+    {
+      "epoch": 0.5733706846832002,
+      "grad_norm": 0.0,
+      "learning_rate": 8.121014453574284e-06,
+      "loss": 0.8613,
+      "step": 20877
+    },
+    {
+      "epoch": 0.5733981489110433,
+      "grad_norm": 0.0,
+      "learning_rate": 8.120140788663536e-06,
+      "loss": 0.8481,
+      "step": 20878
+    },
+    {
+      "epoch": 0.5734256131388866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.119267138626966e-06,
+      "loss": 0.8445,
+      "step": 20879
+    },
+    {
+      "epoch": 0.5734530773667298,
+      "grad_norm": 0.0,
+      "learning_rate": 8.118393503471484e-06,
+      "loss": 0.8578,
+      "step": 20880
+    },
+    {
+      "epoch": 0.5734805415945731,
+      "grad_norm": 0.0,
+      "learning_rate": 8.117519883204002e-06,
+      "loss": 0.9713,
+      "step": 20881
+    },
+    {
+      "epoch": 0.5735080058224163,
+      "grad_norm": 0.0,
+      "learning_rate": 8.116646277831428e-06,
+      "loss": 0.824,
+      "step": 20882
+    },
+    {
+      "epoch": 0.5735354700502595,
+      "grad_norm": 0.0,
+      "learning_rate": 8.11577268736068e-06,
+      "loss": 0.904,
+      "step": 20883
+    },
+    {
+      "epoch": 0.5735629342781028,
+      "grad_norm": 0.0,
+      "learning_rate": 8.114899111798669e-06,
+      "loss": 0.993,
+      "step": 20884
+    },
+    {
+      "epoch": 0.573590398505946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.114025551152305e-06,
+      "loss": 0.8868,
+      "step": 20885
+    },
+    {
+      "epoch": 0.5736178627337892,
+      "grad_norm": 0.0,
+      "learning_rate": 8.113152005428506e-06,
+      "loss": 0.8417,
+      "step": 20886
+    },
+    {
+      "epoch": 0.5736453269616325,
+      "grad_norm": 0.0,
+      "learning_rate": 8.112278474634173e-06,
+      "loss": 0.8195,
+      "step": 20887
+    },
+    {
+      "epoch": 0.5736727911894757,
+      "grad_norm": 0.0,
+      "learning_rate": 8.111404958776226e-06,
+      "loss": 0.8779,
+      "step": 20888
+    },
+    {
+      "epoch": 0.573700255417319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.110531457861577e-06,
+      "loss": 0.8979,
+      "step": 20889
+    },
+    {
+      "epoch": 0.5737277196451622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.109657971897132e-06,
+      "loss": 0.8724,
+      "step": 20890
+    },
+    {
+      "epoch": 0.5737551838730054,
+      "grad_norm": 0.0,
+      "learning_rate": 8.108784500889805e-06,
+      "loss": 0.8164,
+      "step": 20891
+    },
+    {
+      "epoch": 0.5737826481008487,
+      "grad_norm": 0.0,
+      "learning_rate": 8.107911044846513e-06,
+      "loss": 0.8148,
+      "step": 20892
+    },
+    {
+      "epoch": 0.5738101123286918,
+      "grad_norm": 0.0,
+      "learning_rate": 8.107037603774158e-06,
+      "loss": 0.8353,
+      "step": 20893
+    },
+    {
+      "epoch": 0.5738375765565351,
+      "grad_norm": 0.0,
+      "learning_rate": 8.106164177679654e-06,
+      "loss": 0.9297,
+      "step": 20894
+    },
+    {
+      "epoch": 0.5738650407843784,
+      "grad_norm": 0.0,
+      "learning_rate": 8.10529076656991e-06,
+      "loss": 0.8769,
+      "step": 20895
+    },
+    {
+      "epoch": 0.5738925050122216,
+      "grad_norm": 0.0,
+      "learning_rate": 8.104417370451845e-06,
+      "loss": 0.8554,
+      "step": 20896
+    },
+    {
+      "epoch": 0.5739199692400648,
+      "grad_norm": 0.0,
+      "learning_rate": 8.10354398933236e-06,
+      "loss": 0.8376,
+      "step": 20897
+    },
+    {
+      "epoch": 0.573947433467908,
+      "grad_norm": 0.0,
+      "learning_rate": 8.102670623218375e-06,
+      "loss": 0.9036,
+      "step": 20898
+    },
+    {
+      "epoch": 0.5739748976957513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.101797272116791e-06,
+      "loss": 0.8663,
+      "step": 20899
+    },
+    {
+      "epoch": 0.5740023619235946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.100923936034521e-06,
+      "loss": 0.8969,
+      "step": 20900
+    },
+    {
+      "epoch": 0.5740298261514377,
+      "grad_norm": 0.0,
+      "learning_rate": 8.100050614978478e-06,
+      "loss": 0.8706,
+      "step": 20901
+    },
+    {
+      "epoch": 0.574057290379281,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09917730895557e-06,
+      "loss": 0.9951,
+      "step": 20902
+    },
+    {
+      "epoch": 0.5740847546071243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.098304017972707e-06,
+      "loss": 0.8662,
+      "step": 20903
+    },
+    {
+      "epoch": 0.5741122188349674,
+      "grad_norm": 0.0,
+      "learning_rate": 8.097430742036803e-06,
+      "loss": 0.9214,
+      "step": 20904
+    },
+    {
+      "epoch": 0.5741396830628107,
+      "grad_norm": 0.0,
+      "learning_rate": 8.096557481154762e-06,
+      "loss": 0.8107,
+      "step": 20905
+    },
+    {
+      "epoch": 0.5741671472906539,
+      "grad_norm": 0.0,
+      "learning_rate": 8.095684235333494e-06,
+      "loss": 0.8475,
+      "step": 20906
+    },
+    {
+      "epoch": 0.5741946115184972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.094811004579908e-06,
+      "loss": 0.922,
+      "step": 20907
+    },
+    {
+      "epoch": 0.5742220757463404,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09393778890092e-06,
+      "loss": 0.7901,
+      "step": 20908
+    },
+    {
+      "epoch": 0.5742495399741836,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09306458830343e-06,
+      "loss": 0.9067,
+      "step": 20909
+    },
+    {
+      "epoch": 0.5742770042020269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.092191402794359e-06,
+      "loss": 0.9395,
+      "step": 20910
+    },
+    {
+      "epoch": 0.57430446842987,
+      "grad_norm": 0.0,
+      "learning_rate": 8.091318232380602e-06,
+      "loss": 0.7838,
+      "step": 20911
+    },
+    {
+      "epoch": 0.5743319326577133,
+      "grad_norm": 0.0,
+      "learning_rate": 8.090445077069074e-06,
+      "loss": 0.8534,
+      "step": 20912
+    },
+    {
+      "epoch": 0.5743593968855566,
+      "grad_norm": 0.0,
+      "learning_rate": 8.089571936866689e-06,
+      "loss": 0.9196,
+      "step": 20913
+    },
+    {
+      "epoch": 0.5743868611133998,
+      "grad_norm": 0.0,
+      "learning_rate": 8.088698811780346e-06,
+      "loss": 0.9445,
+      "step": 20914
+    },
+    {
+      "epoch": 0.574414325341243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.08782570181696e-06,
+      "loss": 0.9644,
+      "step": 20915
+    },
+    {
+      "epoch": 0.5744417895690863,
+      "grad_norm": 0.0,
+      "learning_rate": 8.086952606983442e-06,
+      "loss": 0.8798,
+      "step": 20916
+    },
+    {
+      "epoch": 0.5744692537969295,
+      "grad_norm": 0.0,
+      "learning_rate": 8.086079527286693e-06,
+      "loss": 0.8435,
+      "step": 20917
+    },
+    {
+      "epoch": 0.5744967180247728,
+      "grad_norm": 0.0,
+      "learning_rate": 8.085206462733623e-06,
+      "loss": 0.8196,
+      "step": 20918
+    },
+    {
+      "epoch": 0.5745241822526159,
+      "grad_norm": 0.0,
+      "learning_rate": 8.08433341333114e-06,
+      "loss": 0.9268,
+      "step": 20919
+    },
+    {
+      "epoch": 0.5745516464804592,
+      "grad_norm": 0.0,
+      "learning_rate": 8.083460379086157e-06,
+      "loss": 0.8976,
+      "step": 20920
+    },
+    {
+      "epoch": 0.5745791107083025,
+      "grad_norm": 0.0,
+      "learning_rate": 8.082587360005575e-06,
+      "loss": 0.9008,
+      "step": 20921
+    },
+    {
+      "epoch": 0.5746065749361456,
+      "grad_norm": 0.0,
+      "learning_rate": 8.081714356096308e-06,
+      "loss": 0.8696,
+      "step": 20922
+    },
+    {
+      "epoch": 0.5746340391639889,
+      "grad_norm": 0.0,
+      "learning_rate": 8.080841367365256e-06,
+      "loss": 0.8867,
+      "step": 20923
+    },
+    {
+      "epoch": 0.5746615033918321,
+      "grad_norm": 0.0,
+      "learning_rate": 8.07996839381933e-06,
+      "loss": 0.8415,
+      "step": 20924
+    },
+    {
+      "epoch": 0.5746889676196754,
+      "grad_norm": 0.0,
+      "learning_rate": 8.079095435465442e-06,
+      "loss": 0.8372,
+      "step": 20925
+    },
+    {
+      "epoch": 0.5747164318475186,
+      "grad_norm": 0.0,
+      "learning_rate": 8.07822249231049e-06,
+      "loss": 0.7796,
+      "step": 20926
+    },
+    {
+      "epoch": 0.5747438960753618,
+      "grad_norm": 0.0,
+      "learning_rate": 8.077349564361394e-06,
+      "loss": 0.7063,
+      "step": 20927
+    },
+    {
+      "epoch": 0.5747713603032051,
+      "grad_norm": 0.0,
+      "learning_rate": 8.076476651625046e-06,
+      "loss": 0.8455,
+      "step": 20928
+    },
+    {
+      "epoch": 0.5747988245310484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.075603754108364e-06,
+      "loss": 0.9026,
+      "step": 20929
+    },
+    {
+      "epoch": 0.5748262887588915,
+      "grad_norm": 0.0,
+      "learning_rate": 8.074730871818246e-06,
+      "loss": 0.9105,
+      "step": 20930
+    },
+    {
+      "epoch": 0.5748537529867348,
+      "grad_norm": 0.0,
+      "learning_rate": 8.073858004761603e-06,
+      "loss": 0.8492,
+      "step": 20931
+    },
+    {
+      "epoch": 0.574881217214578,
+      "grad_norm": 0.0,
+      "learning_rate": 8.072985152945345e-06,
+      "loss": 0.8715,
+      "step": 20932
+    },
+    {
+      "epoch": 0.5749086814424212,
+      "grad_norm": 0.0,
+      "learning_rate": 8.072112316376375e-06,
+      "loss": 0.9055,
+      "step": 20933
+    },
+    {
+      "epoch": 0.5749361456702645,
+      "grad_norm": 0.0,
+      "learning_rate": 8.071239495061601e-06,
+      "loss": 0.9434,
+      "step": 20934
+    },
+    {
+      "epoch": 0.5749636098981077,
+      "grad_norm": 0.0,
+      "learning_rate": 8.070366689007923e-06,
+      "loss": 0.8925,
+      "step": 20935
+    },
+    {
+      "epoch": 0.574991074125951,
+      "grad_norm": 0.0,
+      "learning_rate": 8.06949389822225e-06,
+      "loss": 0.8625,
+      "step": 20936
+    },
+    {
+      "epoch": 0.5750185383537941,
+      "grad_norm": 0.0,
+      "learning_rate": 8.068621122711493e-06,
+      "loss": 0.8445,
+      "step": 20937
+    },
+    {
+      "epoch": 0.5750460025816374,
+      "grad_norm": 0.0,
+      "learning_rate": 8.067748362482551e-06,
+      "loss": 0.808,
+      "step": 20938
+    },
+    {
+      "epoch": 0.5750734668094807,
+      "grad_norm": 0.0,
+      "learning_rate": 8.066875617542337e-06,
+      "loss": 0.7761,
+      "step": 20939
+    },
+    {
+      "epoch": 0.5751009310373238,
+      "grad_norm": 0.0,
+      "learning_rate": 8.066002887897748e-06,
+      "loss": 0.8448,
+      "step": 20940
+    },
+    {
+      "epoch": 0.5751283952651671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.065130173555694e-06,
+      "loss": 0.8496,
+      "step": 20941
+    },
+    {
+      "epoch": 0.5751558594930104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.064257474523079e-06,
+      "loss": 0.9176,
+      "step": 20942
+    },
+    {
+      "epoch": 0.5751833237208536,
+      "grad_norm": 0.0,
+      "learning_rate": 8.063384790806808e-06,
+      "loss": 0.9314,
+      "step": 20943
+    },
+    {
+      "epoch": 0.5752107879486968,
+      "grad_norm": 0.0,
+      "learning_rate": 8.06251212241379e-06,
+      "loss": 0.9669,
+      "step": 20944
+    },
+    {
+      "epoch": 0.57523825217654,
+      "grad_norm": 0.0,
+      "learning_rate": 8.061639469350925e-06,
+      "loss": 0.8771,
+      "step": 20945
+    },
+    {
+      "epoch": 0.5752657164043833,
+      "grad_norm": 0.0,
+      "learning_rate": 8.06076683162512e-06,
+      "loss": 0.8903,
+      "step": 20946
+    },
+    {
+      "epoch": 0.5752931806322266,
+      "grad_norm": 0.0,
+      "learning_rate": 8.059894209243278e-06,
+      "loss": 0.9157,
+      "step": 20947
+    },
+    {
+      "epoch": 0.5753206448600697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.059021602212302e-06,
+      "loss": 0.7844,
+      "step": 20948
+    },
+    {
+      "epoch": 0.575348109087913,
+      "grad_norm": 0.0,
+      "learning_rate": 8.058149010539103e-06,
+      "loss": 0.9412,
+      "step": 20949
+    },
+    {
+      "epoch": 0.5753755733157562,
+      "grad_norm": 0.0,
+      "learning_rate": 8.057276434230579e-06,
+      "loss": 0.8166,
+      "step": 20950
+    },
+    {
+      "epoch": 0.5754030375435994,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05640387329364e-06,
+      "loss": 0.9065,
+      "step": 20951
+    },
+    {
+      "epoch": 0.5754305017714427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.055531327735181e-06,
+      "loss": 0.8728,
+      "step": 20952
+    },
+    {
+      "epoch": 0.5754579659992859,
+      "grad_norm": 0.0,
+      "learning_rate": 8.054658797562116e-06,
+      "loss": 0.8493,
+      "step": 20953
+    },
+    {
+      "epoch": 0.5754854302271292,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05378628278134e-06,
+      "loss": 0.8845,
+      "step": 20954
+    },
+    {
+      "epoch": 0.5755128944549724,
+      "grad_norm": 0.0,
+      "learning_rate": 8.052913783399762e-06,
+      "loss": 0.9027,
+      "step": 20955
+    },
+    {
+      "epoch": 0.5755403586828156,
+      "grad_norm": 0.0,
+      "learning_rate": 8.052041299424287e-06,
+      "loss": 0.8002,
+      "step": 20956
+    },
+    {
+      "epoch": 0.5755678229106589,
+      "grad_norm": 0.0,
+      "learning_rate": 8.051168830861815e-06,
+      "loss": 0.8337,
+      "step": 20957
+    },
+    {
+      "epoch": 0.575595287138502,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05029637771925e-06,
+      "loss": 0.9159,
+      "step": 20958
+    },
+    {
+      "epoch": 0.5756227513663453,
+      "grad_norm": 0.0,
+      "learning_rate": 8.049423940003493e-06,
+      "loss": 0.856,
+      "step": 20959
+    },
+    {
+      "epoch": 0.5756502155941886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.04855151772145e-06,
+      "loss": 0.8416,
+      "step": 20960
+    },
+    {
+      "epoch": 0.5756776798220318,
+      "grad_norm": 0.0,
+      "learning_rate": 8.047679110880026e-06,
+      "loss": 0.9034,
+      "step": 20961
+    },
+    {
+      "epoch": 0.575705144049875,
+      "grad_norm": 0.0,
+      "learning_rate": 8.046806719486118e-06,
+      "loss": 0.8595,
+      "step": 20962
+    },
+    {
+      "epoch": 0.5757326082777183,
+      "grad_norm": 0.0,
+      "learning_rate": 8.045934343546634e-06,
+      "loss": 0.8815,
+      "step": 20963
+    },
+    {
+      "epoch": 0.5757600725055615,
+      "grad_norm": 0.0,
+      "learning_rate": 8.045061983068473e-06,
+      "loss": 0.8015,
+      "step": 20964
+    },
+    {
+      "epoch": 0.5757875367334048,
+      "grad_norm": 0.0,
+      "learning_rate": 8.044189638058542e-06,
+      "loss": 0.8029,
+      "step": 20965
+    },
+    {
+      "epoch": 0.5758150009612479,
+      "grad_norm": 0.0,
+      "learning_rate": 8.043317308523736e-06,
+      "loss": 0.9281,
+      "step": 20966
+    },
+    {
+      "epoch": 0.5758424651890912,
+      "grad_norm": 0.0,
+      "learning_rate": 8.042444994470961e-06,
+      "loss": 0.8008,
+      "step": 20967
+    },
+    {
+      "epoch": 0.5758699294169345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.041572695907122e-06,
+      "loss": 0.8863,
+      "step": 20968
+    },
+    {
+      "epoch": 0.5758973936447777,
+      "grad_norm": 0.0,
+      "learning_rate": 8.04070041283912e-06,
+      "loss": 0.8423,
+      "step": 20969
+    },
+    {
+      "epoch": 0.5759248578726209,
+      "grad_norm": 0.0,
+      "learning_rate": 8.039828145273856e-06,
+      "loss": 0.9192,
+      "step": 20970
+    },
+    {
+      "epoch": 0.5759523221004641,
+      "grad_norm": 0.0,
+      "learning_rate": 8.038955893218226e-06,
+      "loss": 0.8863,
+      "step": 20971
+    },
+    {
+      "epoch": 0.5759797863283074,
+      "grad_norm": 0.0,
+      "learning_rate": 8.038083656679137e-06,
+      "loss": 0.8293,
+      "step": 20972
+    },
+    {
+      "epoch": 0.5760072505561507,
+      "grad_norm": 0.0,
+      "learning_rate": 8.037211435663492e-06,
+      "loss": 0.8834,
+      "step": 20973
+    },
+    {
+      "epoch": 0.5760347147839938,
+      "grad_norm": 0.0,
+      "learning_rate": 8.03633923017819e-06,
+      "loss": 0.9614,
+      "step": 20974
+    },
+    {
+      "epoch": 0.5760621790118371,
+      "grad_norm": 0.0,
+      "learning_rate": 8.035467040230136e-06,
+      "loss": 0.8989,
+      "step": 20975
+    },
+    {
+      "epoch": 0.5760896432396804,
+      "grad_norm": 0.0,
+      "learning_rate": 8.034594865826224e-06,
+      "loss": 0.875,
+      "step": 20976
+    },
+    {
+      "epoch": 0.5761171074675235,
+      "grad_norm": 0.0,
+      "learning_rate": 8.03372270697336e-06,
+      "loss": 0.9291,
+      "step": 20977
+    },
+    {
+      "epoch": 0.5761445716953668,
+      "grad_norm": 0.0,
+      "learning_rate": 8.032850563678443e-06,
+      "loss": 0.7727,
+      "step": 20978
+    },
+    {
+      "epoch": 0.57617203592321,
+      "grad_norm": 0.0,
+      "learning_rate": 8.031978435948373e-06,
+      "loss": 0.8815,
+      "step": 20979
+    },
+    {
+      "epoch": 0.5761995001510533,
+      "grad_norm": 0.0,
+      "learning_rate": 8.031106323790055e-06,
+      "loss": 0.8683,
+      "step": 20980
+    },
+    {
+      "epoch": 0.5762269643788965,
+      "grad_norm": 0.0,
+      "learning_rate": 8.030234227210387e-06,
+      "loss": 0.823,
+      "step": 20981
+    },
+    {
+      "epoch": 0.5762544286067397,
+      "grad_norm": 0.0,
+      "learning_rate": 8.029362146216266e-06,
+      "loss": 0.8834,
+      "step": 20982
+    },
+    {
+      "epoch": 0.576281892834583,
+      "grad_norm": 0.0,
+      "learning_rate": 8.028490080814596e-06,
+      "loss": 1.0203,
+      "step": 20983
+    },
+    {
+      "epoch": 0.5763093570624261,
+      "grad_norm": 0.0,
+      "learning_rate": 8.027618031012273e-06,
+      "loss": 0.8916,
+      "step": 20984
+    },
+    {
+      "epoch": 0.5763368212902694,
+      "grad_norm": 0.0,
+      "learning_rate": 8.026745996816204e-06,
+      "loss": 0.8357,
+      "step": 20985
+    },
+    {
+      "epoch": 0.5763642855181127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.025873978233287e-06,
+      "loss": 0.8188,
+      "step": 20986
+    },
+    {
+      "epoch": 0.5763917497459559,
+      "grad_norm": 0.0,
+      "learning_rate": 8.025001975270416e-06,
+      "loss": 0.9067,
+      "step": 20987
+    },
+    {
+      "epoch": 0.5764192139737991,
+      "grad_norm": 0.0,
+      "learning_rate": 8.024129987934495e-06,
+      "loss": 1.0186,
+      "step": 20988
+    },
+    {
+      "epoch": 0.5764466782016424,
+      "grad_norm": 0.0,
+      "learning_rate": 8.023258016232424e-06,
+      "loss": 0.8464,
+      "step": 20989
+    },
+    {
+      "epoch": 0.5764741424294856,
+      "grad_norm": 0.0,
+      "learning_rate": 8.0223860601711e-06,
+      "loss": 0.8886,
+      "step": 20990
+    },
+    {
+      "epoch": 0.5765016066573289,
+      "grad_norm": 0.0,
+      "learning_rate": 8.021514119757422e-06,
+      "loss": 0.7943,
+      "step": 20991
+    },
+    {
+      "epoch": 0.576529070885172,
+      "grad_norm": 0.0,
+      "learning_rate": 8.020642194998296e-06,
+      "loss": 0.7972,
+      "step": 20992
+    },
+    {
+      "epoch": 0.5765565351130153,
+      "grad_norm": 0.0,
+      "learning_rate": 8.019770285900613e-06,
+      "loss": 0.8239,
+      "step": 20993
+    },
+    {
+      "epoch": 0.5765839993408586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.018898392471273e-06,
+      "loss": 0.8469,
+      "step": 20994
+    },
+    {
+      "epoch": 0.5766114635687017,
+      "grad_norm": 0.0,
+      "learning_rate": 8.018026514717176e-06,
+      "loss": 0.8511,
+      "step": 20995
+    },
+    {
+      "epoch": 0.576638927796545,
+      "grad_norm": 0.0,
+      "learning_rate": 8.017154652645221e-06,
+      "loss": 0.8433,
+      "step": 20996
+    },
+    {
+      "epoch": 0.5766663920243882,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01628280626231e-06,
+      "loss": 0.8239,
+      "step": 20997
+    },
+    {
+      "epoch": 0.5766938562522315,
+      "grad_norm": 0.0,
+      "learning_rate": 8.015410975575337e-06,
+      "loss": 0.8578,
+      "step": 20998
+    },
+    {
+      "epoch": 0.5767213204800747,
+      "grad_norm": 0.0,
+      "learning_rate": 8.0145391605912e-06,
+      "loss": 0.8472,
+      "step": 20999
+    },
+    {
+      "epoch": 0.5767487847079179,
+      "grad_norm": 0.0,
+      "learning_rate": 8.013667361316795e-06,
+      "loss": 0.9069,
+      "step": 21000
+    },
+    {
+      "epoch": 0.5767762489357612,
+      "grad_norm": 0.0,
+      "learning_rate": 8.012795577759029e-06,
+      "loss": 0.7951,
+      "step": 21001
+    },
+    {
+      "epoch": 0.5768037131636045,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01192380992479e-06,
+      "loss": 0.8991,
+      "step": 21002
+    },
+    {
+      "epoch": 0.5768311773914476,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01105205782098e-06,
+      "loss": 0.8979,
+      "step": 21003
+    },
+    {
+      "epoch": 0.5768586416192909,
+      "grad_norm": 0.0,
+      "learning_rate": 8.010180321454501e-06,
+      "loss": 0.898,
+      "step": 21004
+    },
+    {
+      "epoch": 0.5768861058471341,
+      "grad_norm": 0.0,
+      "learning_rate": 8.009308600832243e-06,
+      "loss": 0.9678,
+      "step": 21005
+    },
+    {
+      "epoch": 0.5769135700749773,
+      "grad_norm": 0.0,
+      "learning_rate": 8.008436895961105e-06,
+      "loss": 0.8493,
+      "step": 21006
+    },
+    {
+      "epoch": 0.5769410343028206,
+      "grad_norm": 0.0,
+      "learning_rate": 8.007565206847987e-06,
+      "loss": 0.8952,
+      "step": 21007
+    },
+    {
+      "epoch": 0.5769684985306638,
+      "grad_norm": 0.0,
+      "learning_rate": 8.006693533499783e-06,
+      "loss": 0.8189,
+      "step": 21008
+    },
+    {
+      "epoch": 0.5769959627585071,
+      "grad_norm": 0.0,
+      "learning_rate": 8.005821875923396e-06,
+      "loss": 0.9624,
+      "step": 21009
+    },
+    {
+      "epoch": 0.5770234269863502,
+      "grad_norm": 0.0,
+      "learning_rate": 8.00495023412572e-06,
+      "loss": 0.9554,
+      "step": 21010
+    },
+    {
+      "epoch": 0.5770508912141935,
+      "grad_norm": 0.0,
+      "learning_rate": 8.004078608113645e-06,
+      "loss": 0.8512,
+      "step": 21011
+    },
+    {
+      "epoch": 0.5770783554420368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.003206997894075e-06,
+      "loss": 0.9465,
+      "step": 21012
+    },
+    {
+      "epoch": 0.57710581966988,
+      "grad_norm": 0.0,
+      "learning_rate": 8.002335403473906e-06,
+      "loss": 0.8144,
+      "step": 21013
+    },
+    {
+      "epoch": 0.5771332838977232,
+      "grad_norm": 0.0,
+      "learning_rate": 8.001463824860032e-06,
+      "loss": 0.9333,
+      "step": 21014
+    },
+    {
+      "epoch": 0.5771607481255665,
+      "grad_norm": 0.0,
+      "learning_rate": 8.000592262059348e-06,
+      "loss": 0.8604,
+      "step": 21015
+    },
+    {
+      "epoch": 0.5771882123534097,
+      "grad_norm": 0.0,
+      "learning_rate": 7.999720715078759e-06,
+      "loss": 0.9179,
+      "step": 21016
+    },
+    {
+      "epoch": 0.577215676581253,
+      "grad_norm": 0.0,
+      "learning_rate": 7.998849183925152e-06,
+      "loss": 0.8665,
+      "step": 21017
+    },
+    {
+      "epoch": 0.5772431408090961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.997977668605424e-06,
+      "loss": 0.9007,
+      "step": 21018
+    },
+    {
+      "epoch": 0.5772706050369394,
+      "grad_norm": 0.0,
+      "learning_rate": 7.99710616912647e-06,
+      "loss": 0.9068,
+      "step": 21019
+    },
+    {
+      "epoch": 0.5772980692647827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.996234685495192e-06,
+      "loss": 0.8333,
+      "step": 21020
+    },
+    {
+      "epoch": 0.5773255334926258,
+      "grad_norm": 0.0,
+      "learning_rate": 7.995363217718477e-06,
+      "loss": 0.7995,
+      "step": 21021
+    },
+    {
+      "epoch": 0.5773529977204691,
+      "grad_norm": 0.0,
+      "learning_rate": 7.994491765803231e-06,
+      "loss": 0.9476,
+      "step": 21022
+    },
+    {
+      "epoch": 0.5773804619483123,
+      "grad_norm": 0.0,
+      "learning_rate": 7.993620329756337e-06,
+      "loss": 0.9471,
+      "step": 21023
+    },
+    {
+      "epoch": 0.5774079261761556,
+      "grad_norm": 0.0,
+      "learning_rate": 7.992748909584698e-06,
+      "loss": 0.8825,
+      "step": 21024
+    },
+    {
+      "epoch": 0.5774353904039988,
+      "grad_norm": 0.0,
+      "learning_rate": 7.991877505295208e-06,
+      "loss": 0.7496,
+      "step": 21025
+    },
+    {
+      "epoch": 0.577462854631842,
+      "grad_norm": 0.0,
+      "learning_rate": 7.99100611689476e-06,
+      "loss": 0.9076,
+      "step": 21026
+    },
+    {
+      "epoch": 0.5774903188596853,
+      "grad_norm": 0.0,
+      "learning_rate": 7.99013474439025e-06,
+      "loss": 0.8192,
+      "step": 21027
+    },
+    {
+      "epoch": 0.5775177830875285,
+      "grad_norm": 0.0,
+      "learning_rate": 7.989263387788577e-06,
+      "loss": 0.8665,
+      "step": 21028
+    },
+    {
+      "epoch": 0.5775452473153717,
+      "grad_norm": 0.0,
+      "learning_rate": 7.988392047096627e-06,
+      "loss": 0.8785,
+      "step": 21029
+    },
+    {
+      "epoch": 0.577572711543215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.987520722321297e-06,
+      "loss": 0.9542,
+      "step": 21030
+    },
+    {
+      "epoch": 0.5776001757710582,
+      "grad_norm": 0.0,
+      "learning_rate": 7.986649413469484e-06,
+      "loss": 0.7817,
+      "step": 21031
+    },
+    {
+      "epoch": 0.5776276399989014,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98577812054808e-06,
+      "loss": 0.8123,
+      "step": 21032
+    },
+    {
+      "epoch": 0.5776551042267447,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98490684356398e-06,
+      "loss": 0.8856,
+      "step": 21033
+    },
+    {
+      "epoch": 0.5776825684545879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98403558252408e-06,
+      "loss": 0.8408,
+      "step": 21034
+    },
+    {
+      "epoch": 0.5777100326824312,
+      "grad_norm": 0.0,
+      "learning_rate": 7.983164337435268e-06,
+      "loss": 0.9686,
+      "step": 21035
+    },
+    {
+      "epoch": 0.5777374969102743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.982293108304442e-06,
+      "loss": 0.8323,
+      "step": 21036
+    },
+    {
+      "epoch": 0.5777649611381176,
+      "grad_norm": 0.0,
+      "learning_rate": 7.981421895138497e-06,
+      "loss": 0.9299,
+      "step": 21037
+    },
+    {
+      "epoch": 0.5777924253659609,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98055069794432e-06,
+      "loss": 0.943,
+      "step": 21038
+    },
+    {
+      "epoch": 0.577819889593804,
+      "grad_norm": 0.0,
+      "learning_rate": 7.979679516728807e-06,
+      "loss": 0.8827,
+      "step": 21039
+    },
+    {
+      "epoch": 0.5778473538216473,
+      "grad_norm": 0.0,
+      "learning_rate": 7.978808351498858e-06,
+      "loss": 0.9191,
+      "step": 21040
+    },
+    {
+      "epoch": 0.5778748180494906,
+      "grad_norm": 0.0,
+      "learning_rate": 7.97793720226136e-06,
+      "loss": 0.8412,
+      "step": 21041
+    },
+    {
+      "epoch": 0.5779022822773338,
+      "grad_norm": 0.0,
+      "learning_rate": 7.977066069023201e-06,
+      "loss": 0.8974,
+      "step": 21042
+    },
+    {
+      "epoch": 0.577929746505177,
+      "grad_norm": 0.0,
+      "learning_rate": 7.97619495179128e-06,
+      "loss": 0.8918,
+      "step": 21043
+    },
+    {
+      "epoch": 0.5779572107330202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.975323850572492e-06,
+      "loss": 0.8405,
+      "step": 21044
+    },
+    {
+      "epoch": 0.5779846749608635,
+      "grad_norm": 0.0,
+      "learning_rate": 7.974452765373725e-06,
+      "loss": 0.8773,
+      "step": 21045
+    },
+    {
+      "epoch": 0.5780121391887068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.973581696201872e-06,
+      "loss": 0.9664,
+      "step": 21046
+    },
+    {
+      "epoch": 0.5780396034165499,
+      "grad_norm": 0.0,
+      "learning_rate": 7.972710643063824e-06,
+      "loss": 0.9458,
+      "step": 21047
+    },
+    {
+      "epoch": 0.5780670676443932,
+      "grad_norm": 0.0,
+      "learning_rate": 7.971839605966473e-06,
+      "loss": 0.9229,
+      "step": 21048
+    },
+    {
+      "epoch": 0.5780945318722364,
+      "grad_norm": 0.0,
+      "learning_rate": 7.970968584916717e-06,
+      "loss": 0.9563,
+      "step": 21049
+    },
+    {
+      "epoch": 0.5781219961000796,
+      "grad_norm": 0.0,
+      "learning_rate": 7.970097579921441e-06,
+      "loss": 0.8504,
+      "step": 21050
+    },
+    {
+      "epoch": 0.5781494603279229,
+      "grad_norm": 0.0,
+      "learning_rate": 7.969226590987543e-06,
+      "loss": 0.805,
+      "step": 21051
+    },
+    {
+      "epoch": 0.5781769245557661,
+      "grad_norm": 0.0,
+      "learning_rate": 7.968355618121905e-06,
+      "loss": 0.894,
+      "step": 21052
+    },
+    {
+      "epoch": 0.5782043887836094,
+      "grad_norm": 0.0,
+      "learning_rate": 7.96748466133143e-06,
+      "loss": 0.9268,
+      "step": 21053
+    },
+    {
+      "epoch": 0.5782318530114526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.966613720622999e-06,
+      "loss": 0.8191,
+      "step": 21054
+    },
+    {
+      "epoch": 0.5782593172392958,
+      "grad_norm": 0.0,
+      "learning_rate": 7.965742796003509e-06,
+      "loss": 0.8594,
+      "step": 21055
+    },
+    {
+      "epoch": 0.5782867814671391,
+      "grad_norm": 0.0,
+      "learning_rate": 7.964871887479851e-06,
+      "loss": 0.958,
+      "step": 21056
+    },
+    {
+      "epoch": 0.5783142456949822,
+      "grad_norm": 0.0,
+      "learning_rate": 7.964000995058916e-06,
+      "loss": 0.8718,
+      "step": 21057
+    },
+    {
+      "epoch": 0.5783417099228255,
+      "grad_norm": 0.0,
+      "learning_rate": 7.963130118747592e-06,
+      "loss": 0.8319,
+      "step": 21058
+    },
+    {
+      "epoch": 0.5783691741506688,
+      "grad_norm": 0.0,
+      "learning_rate": 7.962259258552773e-06,
+      "loss": 0.9573,
+      "step": 21059
+    },
+    {
+      "epoch": 0.578396638378512,
+      "grad_norm": 0.0,
+      "learning_rate": 7.961388414481344e-06,
+      "loss": 0.8304,
+      "step": 21060
+    },
+    {
+      "epoch": 0.5784241026063552,
+      "grad_norm": 0.0,
+      "learning_rate": 7.960517586540205e-06,
+      "loss": 0.7943,
+      "step": 21061
+    },
+    {
+      "epoch": 0.5784515668341984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.959646774736237e-06,
+      "loss": 0.8797,
+      "step": 21062
+    },
+    {
+      "epoch": 0.5784790310620417,
+      "grad_norm": 0.0,
+      "learning_rate": 7.958775979076338e-06,
+      "loss": 0.749,
+      "step": 21063
+    },
+    {
+      "epoch": 0.578506495289885,
+      "grad_norm": 0.0,
+      "learning_rate": 7.95790519956739e-06,
+      "loss": 0.8694,
+      "step": 21064
+    },
+    {
+      "epoch": 0.5785339595177281,
+      "grad_norm": 0.0,
+      "learning_rate": 7.957034436216291e-06,
+      "loss": 0.8839,
+      "step": 21065
+    },
+    {
+      "epoch": 0.5785614237455714,
+      "grad_norm": 0.0,
+      "learning_rate": 7.956163689029924e-06,
+      "loss": 0.8536,
+      "step": 21066
+    },
+    {
+      "epoch": 0.5785888879734147,
+      "grad_norm": 0.0,
+      "learning_rate": 7.955292958015181e-06,
+      "loss": 0.8353,
+      "step": 21067
+    },
+    {
+      "epoch": 0.5786163522012578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.954422243178954e-06,
+      "loss": 0.8572,
+      "step": 21068
+    },
+    {
+      "epoch": 0.5786438164291011,
+      "grad_norm": 0.0,
+      "learning_rate": 7.953551544528131e-06,
+      "loss": 0.8344,
+      "step": 21069
+    },
+    {
+      "epoch": 0.5786712806569443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.952680862069601e-06,
+      "loss": 0.8472,
+      "step": 21070
+    },
+    {
+      "epoch": 0.5786987448847876,
+      "grad_norm": 0.0,
+      "learning_rate": 7.951810195810251e-06,
+      "loss": 0.93,
+      "step": 21071
+    },
+    {
+      "epoch": 0.5787262091126308,
+      "grad_norm": 0.0,
+      "learning_rate": 7.950939545756972e-06,
+      "loss": 0.8758,
+      "step": 21072
+    },
+    {
+      "epoch": 0.578753673340474,
+      "grad_norm": 0.0,
+      "learning_rate": 7.950068911916655e-06,
+      "loss": 0.8571,
+      "step": 21073
+    },
+    {
+      "epoch": 0.5787811375683173,
+      "grad_norm": 0.0,
+      "learning_rate": 7.949198294296185e-06,
+      "loss": 0.925,
+      "step": 21074
+    },
+    {
+      "epoch": 0.5788086017961604,
+      "grad_norm": 0.0,
+      "learning_rate": 7.948327692902457e-06,
+      "loss": 0.8959,
+      "step": 21075
+    },
+    {
+      "epoch": 0.5788360660240037,
+      "grad_norm": 0.0,
+      "learning_rate": 7.94745710774235e-06,
+      "loss": 0.8493,
+      "step": 21076
+    },
+    {
+      "epoch": 0.578863530251847,
+      "grad_norm": 0.0,
+      "learning_rate": 7.94658653882276e-06,
+      "loss": 0.9822,
+      "step": 21077
+    },
+    {
+      "epoch": 0.5788909944796902,
+      "grad_norm": 0.0,
+      "learning_rate": 7.94571598615057e-06,
+      "loss": 0.932,
+      "step": 21078
+    },
+    {
+      "epoch": 0.5789184587075334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.944845449732672e-06,
+      "loss": 0.8884,
+      "step": 21079
+    },
+    {
+      "epoch": 0.5789459229353767,
+      "grad_norm": 0.0,
+      "learning_rate": 7.943974929575955e-06,
+      "loss": 0.9898,
+      "step": 21080
+    },
+    {
+      "epoch": 0.5789733871632199,
+      "grad_norm": 0.0,
+      "learning_rate": 7.943104425687303e-06,
+      "loss": 0.8576,
+      "step": 21081
+    },
+    {
+      "epoch": 0.5790008513910632,
+      "grad_norm": 0.0,
+      "learning_rate": 7.942233938073607e-06,
+      "loss": 0.9111,
+      "step": 21082
+    },
+    {
+      "epoch": 0.5790283156189063,
+      "grad_norm": 0.0,
+      "learning_rate": 7.941363466741751e-06,
+      "loss": 0.8932,
+      "step": 21083
+    },
+    {
+      "epoch": 0.5790557798467496,
+      "grad_norm": 0.0,
+      "learning_rate": 7.940493011698625e-06,
+      "loss": 0.8889,
+      "step": 21084
+    },
+    {
+      "epoch": 0.5790832440745929,
+      "grad_norm": 0.0,
+      "learning_rate": 7.939622572951118e-06,
+      "loss": 0.8402,
+      "step": 21085
+    },
+    {
+      "epoch": 0.579110708302436,
+      "grad_norm": 0.0,
+      "learning_rate": 7.938752150506113e-06,
+      "loss": 0.9321,
+      "step": 21086
+    },
+    {
+      "epoch": 0.5791381725302793,
+      "grad_norm": 0.0,
+      "learning_rate": 7.937881744370503e-06,
+      "loss": 0.7483,
+      "step": 21087
+    },
+    {
+      "epoch": 0.5791656367581225,
+      "grad_norm": 0.0,
+      "learning_rate": 7.937011354551168e-06,
+      "loss": 0.8189,
+      "step": 21088
+    },
+    {
+      "epoch": 0.5791931009859658,
+      "grad_norm": 0.0,
+      "learning_rate": 7.936140981055e-06,
+      "loss": 0.9883,
+      "step": 21089
+    },
+    {
+      "epoch": 0.579220565213809,
+      "grad_norm": 0.0,
+      "learning_rate": 7.935270623888881e-06,
+      "loss": 0.7929,
+      "step": 21090
+    },
+    {
+      "epoch": 0.5792480294416522,
+      "grad_norm": 0.0,
+      "learning_rate": 7.934400283059703e-06,
+      "loss": 0.9504,
+      "step": 21091
+    },
+    {
+      "epoch": 0.5792754936694955,
+      "grad_norm": 0.0,
+      "learning_rate": 7.933529958574351e-06,
+      "loss": 0.9417,
+      "step": 21092
+    },
+    {
+      "epoch": 0.5793029578973388,
+      "grad_norm": 0.0,
+      "learning_rate": 7.932659650439711e-06,
+      "loss": 0.8059,
+      "step": 21093
+    },
+    {
+      "epoch": 0.5793304221251819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.931789358662668e-06,
+      "loss": 0.8217,
+      "step": 21094
+    },
+    {
+      "epoch": 0.5793578863530252,
+      "grad_norm": 0.0,
+      "learning_rate": 7.930919083250106e-06,
+      "loss": 0.8419,
+      "step": 21095
+    },
+    {
+      "epoch": 0.5793853505808684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.930048824208914e-06,
+      "loss": 0.811,
+      "step": 21096
+    },
+    {
+      "epoch": 0.5794128148087117,
+      "grad_norm": 0.0,
+      "learning_rate": 7.929178581545981e-06,
+      "loss": 0.9336,
+      "step": 21097
+    },
+    {
+      "epoch": 0.5794402790365549,
+      "grad_norm": 0.0,
+      "learning_rate": 7.928308355268187e-06,
+      "loss": 0.8483,
+      "step": 21098
+    },
+    {
+      "epoch": 0.5794677432643981,
+      "grad_norm": 0.0,
+      "learning_rate": 7.927438145382423e-06,
+      "loss": 0.8917,
+      "step": 21099
+    },
+    {
+      "epoch": 0.5794952074922414,
+      "grad_norm": 0.0,
+      "learning_rate": 7.926567951895566e-06,
+      "loss": 0.8546,
+      "step": 21100
+    },
+    {
+      "epoch": 0.5795226717200845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.925697774814511e-06,
+      "loss": 0.8507,
+      "step": 21101
+    },
+    {
+      "epoch": 0.5795501359479278,
+      "grad_norm": 0.0,
+      "learning_rate": 7.924827614146134e-06,
+      "loss": 0.7957,
+      "step": 21102
+    },
+    {
+      "epoch": 0.5795776001757711,
+      "grad_norm": 0.0,
+      "learning_rate": 7.923957469897326e-06,
+      "loss": 0.7943,
+      "step": 21103
+    },
+    {
+      "epoch": 0.5796050644036143,
+      "grad_norm": 0.0,
+      "learning_rate": 7.923087342074976e-06,
+      "loss": 0.9506,
+      "step": 21104
+    },
+    {
+      "epoch": 0.5796325286314575,
+      "grad_norm": 0.0,
+      "learning_rate": 7.922217230685957e-06,
+      "loss": 0.8557,
+      "step": 21105
+    },
+    {
+      "epoch": 0.5796599928593008,
+      "grad_norm": 0.0,
+      "learning_rate": 7.921347135737165e-06,
+      "loss": 0.7327,
+      "step": 21106
+    },
+    {
+      "epoch": 0.579687457087144,
+      "grad_norm": 0.0,
+      "learning_rate": 7.920477057235477e-06,
+      "loss": 0.9567,
+      "step": 21107
+    },
+    {
+      "epoch": 0.5797149213149873,
+      "grad_norm": 0.0,
+      "learning_rate": 7.919606995187777e-06,
+      "loss": 0.8228,
+      "step": 21108
+    },
+    {
+      "epoch": 0.5797423855428304,
+      "grad_norm": 0.0,
+      "learning_rate": 7.918736949600957e-06,
+      "loss": 0.8709,
+      "step": 21109
+    },
+    {
+      "epoch": 0.5797698497706737,
+      "grad_norm": 0.0,
+      "learning_rate": 7.917866920481898e-06,
+      "loss": 0.8355,
+      "step": 21110
+    },
+    {
+      "epoch": 0.579797313998517,
+      "grad_norm": 0.0,
+      "learning_rate": 7.916996907837478e-06,
+      "loss": 0.8961,
+      "step": 21111
+    },
+    {
+      "epoch": 0.5798247782263601,
+      "grad_norm": 0.0,
+      "learning_rate": 7.916126911674587e-06,
+      "loss": 0.9746,
+      "step": 21112
+    },
+    {
+      "epoch": 0.5798522424542034,
+      "grad_norm": 0.0,
+      "learning_rate": 7.915256932000108e-06,
+      "loss": 0.9246,
+      "step": 21113
+    },
+    {
+      "epoch": 0.5798797066820466,
+      "grad_norm": 0.0,
+      "learning_rate": 7.91438696882092e-06,
+      "loss": 0.8919,
+      "step": 21114
+    },
+    {
+      "epoch": 0.5799071709098899,
+      "grad_norm": 0.0,
+      "learning_rate": 7.913517022143912e-06,
+      "loss": 0.8604,
+      "step": 21115
+    },
+    {
+      "epoch": 0.5799346351377331,
+      "grad_norm": 0.0,
+      "learning_rate": 7.91264709197597e-06,
+      "loss": 0.986,
+      "step": 21116
+    },
+    {
+      "epoch": 0.5799620993655763,
+      "grad_norm": 0.0,
+      "learning_rate": 7.911777178323967e-06,
+      "loss": 0.8289,
+      "step": 21117
+    },
+    {
+      "epoch": 0.5799895635934196,
+      "grad_norm": 0.0,
+      "learning_rate": 7.910907281194796e-06,
+      "loss": 0.7758,
+      "step": 21118
+    },
+    {
+      "epoch": 0.5800170278212629,
+      "grad_norm": 0.0,
+      "learning_rate": 7.910037400595334e-06,
+      "loss": 0.8154,
+      "step": 21119
+    },
+    {
+      "epoch": 0.580044492049106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.909167536532463e-06,
+      "loss": 0.8854,
+      "step": 21120
+    },
+    {
+      "epoch": 0.5800719562769493,
+      "grad_norm": 0.0,
+      "learning_rate": 7.908297689013073e-06,
+      "loss": 0.9073,
+      "step": 21121
+    },
+    {
+      "epoch": 0.5800994205047925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.907427858044042e-06,
+      "loss": 0.8994,
+      "step": 21122
+    },
+    {
+      "epoch": 0.5801268847326357,
+      "grad_norm": 0.0,
+      "learning_rate": 7.90655804363225e-06,
+      "loss": 0.8555,
+      "step": 21123
+    },
+    {
+      "epoch": 0.580154348960479,
+      "grad_norm": 0.0,
+      "learning_rate": 7.90568824578458e-06,
+      "loss": 0.8326,
+      "step": 21124
+    },
+    {
+      "epoch": 0.5801818131883222,
+      "grad_norm": 0.0,
+      "learning_rate": 7.904818464507918e-06,
+      "loss": 0.8125,
+      "step": 21125
+    },
+    {
+      "epoch": 0.5802092774161655,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903948699809143e-06,
+      "loss": 0.9173,
+      "step": 21126
+    },
+    {
+      "epoch": 0.5802367416440086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903078951695137e-06,
+      "loss": 0.8678,
+      "step": 21127
+    },
+    {
+      "epoch": 0.5802642058718519,
+      "grad_norm": 0.0,
+      "learning_rate": 7.902209220172787e-06,
+      "loss": 0.9103,
+      "step": 21128
+    },
+    {
+      "epoch": 0.5802916700996952,
+      "grad_norm": 0.0,
+      "learning_rate": 7.901339505248966e-06,
+      "loss": 0.8063,
+      "step": 21129
+    },
+    {
+      "epoch": 0.5803191343275383,
+      "grad_norm": 0.0,
+      "learning_rate": 7.900469806930562e-06,
+      "loss": 0.8204,
+      "step": 21130
+    },
+    {
+      "epoch": 0.5803465985553816,
+      "grad_norm": 0.0,
+      "learning_rate": 7.899600125224453e-06,
+      "loss": 0.8395,
+      "step": 21131
+    },
+    {
+      "epoch": 0.5803740627832249,
+      "grad_norm": 0.0,
+      "learning_rate": 7.898730460137521e-06,
+      "loss": 0.9411,
+      "step": 21132
+    },
+    {
+      "epoch": 0.5804015270110681,
+      "grad_norm": 0.0,
+      "learning_rate": 7.897860811676649e-06,
+      "loss": 0.8168,
+      "step": 21133
+    },
+    {
+      "epoch": 0.5804289912389113,
+      "grad_norm": 0.0,
+      "learning_rate": 7.896991179848718e-06,
+      "loss": 0.8776,
+      "step": 21134
+    },
+    {
+      "epoch": 0.5804564554667545,
+      "grad_norm": 0.0,
+      "learning_rate": 7.896121564660606e-06,
+      "loss": 0.7916,
+      "step": 21135
+    },
+    {
+      "epoch": 0.5804839196945978,
+      "grad_norm": 0.0,
+      "learning_rate": 7.895251966119191e-06,
+      "loss": 0.8518,
+      "step": 21136
+    },
+    {
+      "epoch": 0.5805113839224411,
+      "grad_norm": 0.0,
+      "learning_rate": 7.894382384231363e-06,
+      "loss": 0.824,
+      "step": 21137
+    },
+    {
+      "epoch": 0.5805388481502842,
+      "grad_norm": 0.0,
+      "learning_rate": 7.893512819003995e-06,
+      "loss": 0.785,
+      "step": 21138
+    },
+    {
+      "epoch": 0.5805663123781275,
+      "grad_norm": 0.0,
+      "learning_rate": 7.892643270443968e-06,
+      "loss": 0.8728,
+      "step": 21139
+    },
+    {
+      "epoch": 0.5805937766059707,
+      "grad_norm": 0.0,
+      "learning_rate": 7.891773738558168e-06,
+      "loss": 0.8966,
+      "step": 21140
+    },
+    {
+      "epoch": 0.580621240833814,
+      "grad_norm": 0.0,
+      "learning_rate": 7.890904223353471e-06,
+      "loss": 0.8889,
+      "step": 21141
+    },
+    {
+      "epoch": 0.5806487050616572,
+      "grad_norm": 0.0,
+      "learning_rate": 7.890034724836753e-06,
+      "loss": 0.7498,
+      "step": 21142
+    },
+    {
+      "epoch": 0.5806761692895004,
+      "grad_norm": 0.0,
+      "learning_rate": 7.889165243014898e-06,
+      "loss": 0.8944,
+      "step": 21143
+    },
+    {
+      "epoch": 0.5807036335173437,
+      "grad_norm": 0.0,
+      "learning_rate": 7.888295777894782e-06,
+      "loss": 0.8671,
+      "step": 21144
+    },
+    {
+      "epoch": 0.5807310977451869,
+      "grad_norm": 0.0,
+      "learning_rate": 7.887426329483293e-06,
+      "loss": 0.897,
+      "step": 21145
+    },
+    {
+      "epoch": 0.5807585619730301,
+      "grad_norm": 0.0,
+      "learning_rate": 7.886556897787308e-06,
+      "loss": 0.8547,
+      "step": 21146
+    },
+    {
+      "epoch": 0.5807860262008734,
+      "grad_norm": 0.0,
+      "learning_rate": 7.8856874828137e-06,
+      "loss": 0.8119,
+      "step": 21147
+    },
+    {
+      "epoch": 0.5808134904287166,
+      "grad_norm": 0.0,
+      "learning_rate": 7.88481808456935e-06,
+      "loss": 0.9725,
+      "step": 21148
+    },
+    {
+      "epoch": 0.5808409546565598,
+      "grad_norm": 0.0,
+      "learning_rate": 7.88394870306114e-06,
+      "loss": 0.8714,
+      "step": 21149
+    },
+    {
+      "epoch": 0.5808684188844031,
+      "grad_norm": 0.0,
+      "learning_rate": 7.883079338295947e-06,
+      "loss": 0.8635,
+      "step": 21150
+    },
+    {
+      "epoch": 0.5808958831122463,
+      "grad_norm": 0.0,
+      "learning_rate": 7.882209990280649e-06,
+      "loss": 0.876,
+      "step": 21151
+    },
+    {
+      "epoch": 0.5809233473400895,
+      "grad_norm": 0.0,
+      "learning_rate": 7.881340659022131e-06,
+      "loss": 0.7836,
+      "step": 21152
+    },
+    {
+      "epoch": 0.5809508115679327,
+      "grad_norm": 0.0,
+      "learning_rate": 7.880471344527264e-06,
+      "loss": 0.9196,
+      "step": 21153
+    },
+    {
+      "epoch": 0.580978275795776,
+      "grad_norm": 0.0,
+      "learning_rate": 7.879602046802926e-06,
+      "loss": 0.8895,
+      "step": 21154
+    },
+    {
+      "epoch": 0.5810057400236193,
+      "grad_norm": 0.0,
+      "learning_rate": 7.878732765855999e-06,
+      "loss": 0.8688,
+      "step": 21155
+    },
+    {
+      "epoch": 0.5810332042514624,
+      "grad_norm": 0.0,
+      "learning_rate": 7.87786350169336e-06,
+      "loss": 0.8584,
+      "step": 21156
+    },
+    {
+      "epoch": 0.5810606684793057,
+      "grad_norm": 0.0,
+      "learning_rate": 7.876994254321888e-06,
+      "loss": 0.95,
+      "step": 21157
+    },
+    {
+      "epoch": 0.581088132707149,
+      "grad_norm": 0.0,
+      "learning_rate": 7.876125023748463e-06,
+      "loss": 0.8809,
+      "step": 21158
+    },
+    {
+      "epoch": 0.5811155969349922,
+      "grad_norm": 0.0,
+      "learning_rate": 7.875255809979954e-06,
+      "loss": 0.7998,
+      "step": 21159
+    },
+    {
+      "epoch": 0.5811430611628354,
+      "grad_norm": 0.0,
+      "learning_rate": 7.874386613023247e-06,
+      "loss": 0.8397,
+      "step": 21160
+    },
+    {
+      "epoch": 0.5811705253906786,
+      "grad_norm": 0.0,
+      "learning_rate": 7.873517432885215e-06,
+      "loss": 0.8217,
+      "step": 21161
+    },
+    {
+      "epoch": 0.5811979896185219,
+      "grad_norm": 0.0,
+      "learning_rate": 7.872648269572736e-06,
+      "loss": 0.8029,
+      "step": 21162
+    },
+    {
+      "epoch": 0.5812254538463651,
+      "grad_norm": 0.0,
+      "learning_rate": 7.871779123092694e-06,
+      "loss": 0.9493,
+      "step": 21163
+    },
+    {
+      "epoch": 0.5812529180742083,
+      "grad_norm": 0.0,
+      "learning_rate": 7.870909993451954e-06,
+      "loss": 0.8237,
+      "step": 21164
+    },
+    {
+      "epoch": 0.5812803823020516,
+      "grad_norm": 0.0,
+      "learning_rate": 7.870040880657401e-06,
+      "loss": 0.8764,
+      "step": 21165
+    },
+    {
+      "epoch": 0.5813078465298949,
+      "grad_norm": 0.0,
+      "learning_rate": 7.869171784715908e-06,
+      "loss": 0.8826,
+      "step": 21166
+    },
+    {
+      "epoch": 0.581335310757738,
+      "grad_norm": 0.0,
+      "learning_rate": 7.868302705634354e-06,
+      "loss": 0.9099,
+      "step": 21167
+    },
+    {
+      "epoch": 0.5813627749855813,
+      "grad_norm": 0.0,
+      "learning_rate": 7.867433643419614e-06,
+      "loss": 0.9192,
+      "step": 21168
+    },
+    {
+      "epoch": 0.5813902392134245,
+      "grad_norm": 0.0,
+      "learning_rate": 7.866564598078568e-06,
+      "loss": 0.8242,
+      "step": 21169
+    },
+    {
+      "epoch": 0.5814177034412678,
+      "grad_norm": 0.0,
+      "learning_rate": 7.865695569618088e-06,
+      "loss": 0.9388,
+      "step": 21170
+    },
+    {
+      "epoch": 0.581445167669111,
+      "grad_norm": 0.0,
+      "learning_rate": 7.86482655804505e-06,
+      "loss": 0.8191,
+      "step": 21171
+    },
+    {
+      "epoch": 0.5814726318969542,
+      "grad_norm": 0.0,
+      "learning_rate": 7.863957563366331e-06,
+      "loss": 0.8416,
+      "step": 21172
+    },
+    {
+      "epoch": 0.5815000961247975,
+      "grad_norm": 0.0,
+      "learning_rate": 7.86308858558881e-06,
+      "loss": 0.8456,
+      "step": 21173
+    },
+    {
+      "epoch": 0.5815275603526406,
+      "grad_norm": 0.0,
+      "learning_rate": 7.862219624719356e-06,
+      "loss": 0.8598,
+      "step": 21174
+    },
+    {
+      "epoch": 0.5815550245804839,
+      "grad_norm": 0.0,
+      "learning_rate": 7.861350680764854e-06,
+      "loss": 0.8541,
+      "step": 21175
+    },
+    {
+      "epoch": 0.5815824888083272,
+      "grad_norm": 0.0,
+      "learning_rate": 7.860481753732168e-06,
+      "loss": 0.9017,
+      "step": 21176
+    },
+    {
+      "epoch": 0.5816099530361704,
+      "grad_norm": 0.0,
+      "learning_rate": 7.859612843628184e-06,
+      "loss": 0.8906,
+      "step": 21177
+    },
+    {
+      "epoch": 0.5816374172640136,
+      "grad_norm": 0.0,
+      "learning_rate": 7.858743950459769e-06,
+      "loss": 0.7912,
+      "step": 21178
+    },
+    {
+      "epoch": 0.5816648814918569,
+      "grad_norm": 0.0,
+      "learning_rate": 7.857875074233801e-06,
+      "loss": 0.8907,
+      "step": 21179
+    },
+    {
+      "epoch": 0.5816923457197001,
+      "grad_norm": 0.0,
+      "learning_rate": 7.857006214957158e-06,
+      "loss": 0.8386,
+      "step": 21180
+    },
+    {
+      "epoch": 0.5817198099475434,
+      "grad_norm": 0.0,
+      "learning_rate": 7.85613737263671e-06,
+      "loss": 0.8173,
+      "step": 21181
+    },
+    {
+      "epoch": 0.5817472741753865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.855268547279334e-06,
+      "loss": 0.8755,
+      "step": 21182
+    },
+    {
+      "epoch": 0.5817747384032298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.854399738891903e-06,
+      "loss": 0.9026,
+      "step": 21183
+    },
+    {
+      "epoch": 0.5818022026310731,
+      "grad_norm": 0.0,
+      "learning_rate": 7.853530947481293e-06,
+      "loss": 0.9321,
+      "step": 21184
+    },
+    {
+      "epoch": 0.5818296668589162,
+      "grad_norm": 0.0,
+      "learning_rate": 7.852662173054379e-06,
+      "loss": 0.909,
+      "step": 21185
+    },
+    {
+      "epoch": 0.5818571310867595,
+      "grad_norm": 0.0,
+      "learning_rate": 7.851793415618031e-06,
+      "loss": 0.8106,
+      "step": 21186
+    },
+    {
+      "epoch": 0.5818845953146027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.850924675179131e-06,
+      "loss": 0.8494,
+      "step": 21187
+    },
+    {
+      "epoch": 0.581912059542446,
+      "grad_norm": 0.0,
+      "learning_rate": 7.850055951744542e-06,
+      "loss": 0.8524,
+      "step": 21188
+    },
+    {
+      "epoch": 0.5819395237702892,
+      "grad_norm": 0.0,
+      "learning_rate": 7.849187245321145e-06,
+      "loss": 0.7433,
+      "step": 21189
+    },
+    {
+      "epoch": 0.5819669879981324,
+      "grad_norm": 0.0,
+      "learning_rate": 7.84831855591581e-06,
+      "loss": 0.836,
+      "step": 21190
+    },
+    {
+      "epoch": 0.5819944522259757,
+      "grad_norm": 0.0,
+      "learning_rate": 7.847449883535413e-06,
+      "loss": 0.8514,
+      "step": 21191
+    },
+    {
+      "epoch": 0.582021916453819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.846581228186827e-06,
+      "loss": 0.813,
+      "step": 21192
+    },
+    {
+      "epoch": 0.5820493806816621,
+      "grad_norm": 0.0,
+      "learning_rate": 7.845712589876926e-06,
+      "loss": 0.9254,
+      "step": 21193
+    },
+    {
+      "epoch": 0.5820768449095054,
+      "grad_norm": 0.0,
+      "learning_rate": 7.844843968612581e-06,
+      "loss": 0.8528,
+      "step": 21194
+    },
+    {
+      "epoch": 0.5821043091373486,
+      "grad_norm": 0.0,
+      "learning_rate": 7.843975364400662e-06,
+      "loss": 0.9192,
+      "step": 21195
+    },
+    {
+      "epoch": 0.5821317733651918,
+      "grad_norm": 0.0,
+      "learning_rate": 7.843106777248047e-06,
+      "loss": 0.9572,
+      "step": 21196
+    },
+    {
+      "epoch": 0.5821592375930351,
+      "grad_norm": 0.0,
+      "learning_rate": 7.842238207161607e-06,
+      "loss": 0.8546,
+      "step": 21197
+    },
+    {
+      "epoch": 0.5821867018208783,
+      "grad_norm": 0.0,
+      "learning_rate": 7.841369654148213e-06,
+      "loss": 0.858,
+      "step": 21198
+    },
+    {
+      "epoch": 0.5822141660487216,
+      "grad_norm": 0.0,
+      "learning_rate": 7.840501118214742e-06,
+      "loss": 0.8514,
+      "step": 21199
+    },
+    {
+      "epoch": 0.5822416302765647,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83963259936806e-06,
+      "loss": 0.8549,
+      "step": 21200
+    },
+    {
+      "epoch": 0.582269094504408,
+      "grad_norm": 0.0,
+      "learning_rate": 7.838764097615043e-06,
+      "loss": 0.778,
+      "step": 21201
+    },
+    {
+      "epoch": 0.5822965587322513,
+      "grad_norm": 0.0,
+      "learning_rate": 7.837895612962558e-06,
+      "loss": 0.8603,
+      "step": 21202
+    },
+    {
+      "epoch": 0.5823240229600944,
+      "grad_norm": 0.0,
+      "learning_rate": 7.837027145417483e-06,
+      "loss": 0.9149,
+      "step": 21203
+    },
+    {
+      "epoch": 0.5823514871879377,
+      "grad_norm": 0.0,
+      "learning_rate": 7.836158694986687e-06,
+      "loss": 0.8223,
+      "step": 21204
+    },
+    {
+      "epoch": 0.582378951415781,
+      "grad_norm": 0.0,
+      "learning_rate": 7.835290261677043e-06,
+      "loss": 0.9022,
+      "step": 21205
+    },
+    {
+      "epoch": 0.5824064156436242,
+      "grad_norm": 0.0,
+      "learning_rate": 7.834421845495421e-06,
+      "loss": 0.891,
+      "step": 21206
+    },
+    {
+      "epoch": 0.5824338798714674,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83355344644869e-06,
+      "loss": 0.8255,
+      "step": 21207
+    },
+    {
+      "epoch": 0.5824613440993106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.832685064543722e-06,
+      "loss": 0.8637,
+      "step": 21208
+    },
+    {
+      "epoch": 0.5824888083271539,
+      "grad_norm": 0.0,
+      "learning_rate": 7.831816699787393e-06,
+      "loss": 0.8935,
+      "step": 21209
+    },
+    {
+      "epoch": 0.5825162725549972,
+      "grad_norm": 0.0,
+      "learning_rate": 7.830948352186567e-06,
+      "loss": 0.76,
+      "step": 21210
+    },
+    {
+      "epoch": 0.5825437367828403,
+      "grad_norm": 0.0,
+      "learning_rate": 7.830080021748123e-06,
+      "loss": 0.8542,
+      "step": 21211
+    },
+    {
+      "epoch": 0.5825712010106836,
+      "grad_norm": 0.0,
+      "learning_rate": 7.829211708478923e-06,
+      "loss": 0.8397,
+      "step": 21212
+    },
+    {
+      "epoch": 0.5825986652385268,
+      "grad_norm": 0.0,
+      "learning_rate": 7.828343412385841e-06,
+      "loss": 0.8635,
+      "step": 21213
+    },
+    {
+      "epoch": 0.58262612946637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.827475133475744e-06,
+      "loss": 0.8222,
+      "step": 21214
+    },
+    {
+      "epoch": 0.5826535936942133,
+      "grad_norm": 0.0,
+      "learning_rate": 7.826606871755508e-06,
+      "loss": 0.8061,
+      "step": 21215
+    },
+    {
+      "epoch": 0.5826810579220565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.825738627232001e-06,
+      "loss": 0.8326,
+      "step": 21216
+    },
+    {
+      "epoch": 0.5827085221498998,
+      "grad_norm": 0.0,
+      "learning_rate": 7.824870399912093e-06,
+      "loss": 0.915,
+      "step": 21217
+    },
+    {
+      "epoch": 0.582735986377743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.824002189802654e-06,
+      "loss": 0.8944,
+      "step": 21218
+    },
+    {
+      "epoch": 0.5827634506055862,
+      "grad_norm": 0.0,
+      "learning_rate": 7.823133996910548e-06,
+      "loss": 0.9168,
+      "step": 21219
+    },
+    {
+      "epoch": 0.5827909148334295,
+      "grad_norm": 0.0,
+      "learning_rate": 7.82226582124265e-06,
+      "loss": 0.8588,
+      "step": 21220
+    },
+    {
+      "epoch": 0.5828183790612727,
+      "grad_norm": 0.0,
+      "learning_rate": 7.82139766280583e-06,
+      "loss": 0.9384,
+      "step": 21221
+    },
+    {
+      "epoch": 0.5828458432891159,
+      "grad_norm": 0.0,
+      "learning_rate": 7.820529521606955e-06,
+      "loss": 0.8997,
+      "step": 21222
+    },
+    {
+      "epoch": 0.5828733075169592,
+      "grad_norm": 0.0,
+      "learning_rate": 7.819661397652898e-06,
+      "loss": 0.8683,
+      "step": 21223
+    },
+    {
+      "epoch": 0.5829007717448024,
+      "grad_norm": 0.0,
+      "learning_rate": 7.81879329095052e-06,
+      "loss": 0.8171,
+      "step": 21224
+    },
+    {
+      "epoch": 0.5829282359726456,
+      "grad_norm": 0.0,
+      "learning_rate": 7.817925201506697e-06,
+      "loss": 0.868,
+      "step": 21225
+    },
+    {
+      "epoch": 0.5829557002004888,
+      "grad_norm": 0.0,
+      "learning_rate": 7.817057129328293e-06,
+      "loss": 0.8359,
+      "step": 21226
+    },
+    {
+      "epoch": 0.5829831644283321,
+      "grad_norm": 0.0,
+      "learning_rate": 7.816189074422178e-06,
+      "loss": 0.9217,
+      "step": 21227
+    },
+    {
+      "epoch": 0.5830106286561754,
+      "grad_norm": 0.0,
+      "learning_rate": 7.815321036795227e-06,
+      "loss": 0.9662,
+      "step": 21228
+    },
+    {
+      "epoch": 0.5830380928840185,
+      "grad_norm": 0.0,
+      "learning_rate": 7.814453016454295e-06,
+      "loss": 0.9215,
+      "step": 21229
+    },
+    {
+      "epoch": 0.5830655571118618,
+      "grad_norm": 0.0,
+      "learning_rate": 7.813585013406261e-06,
+      "loss": 0.8606,
+      "step": 21230
+    },
+    {
+      "epoch": 0.5830930213397051,
+      "grad_norm": 0.0,
+      "learning_rate": 7.812717027657988e-06,
+      "loss": 0.9276,
+      "step": 21231
+    },
+    {
+      "epoch": 0.5831204855675483,
+      "grad_norm": 0.0,
+      "learning_rate": 7.811849059216344e-06,
+      "loss": 0.8651,
+      "step": 21232
+    },
+    {
+      "epoch": 0.5831479497953915,
+      "grad_norm": 0.0,
+      "learning_rate": 7.8109811080882e-06,
+      "loss": 0.8998,
+      "step": 21233
+    },
+    {
+      "epoch": 0.5831754140232347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.810113174280424e-06,
+      "loss": 0.8206,
+      "step": 21234
+    },
+    {
+      "epoch": 0.583202878251078,
+      "grad_norm": 0.0,
+      "learning_rate": 7.809245257799878e-06,
+      "loss": 0.9512,
+      "step": 21235
+    },
+    {
+      "epoch": 0.5832303424789212,
+      "grad_norm": 0.0,
+      "learning_rate": 7.80837735865343e-06,
+      "loss": 0.8695,
+      "step": 21236
+    },
+    {
+      "epoch": 0.5832578067067644,
+      "grad_norm": 0.0,
+      "learning_rate": 7.807509476847952e-06,
+      "loss": 0.9099,
+      "step": 21237
+    },
+    {
+      "epoch": 0.5832852709346077,
+      "grad_norm": 0.0,
+      "learning_rate": 7.806641612390307e-06,
+      "loss": 0.7909,
+      "step": 21238
+    },
+    {
+      "epoch": 0.5833127351624509,
+      "grad_norm": 0.0,
+      "learning_rate": 7.805773765287363e-06,
+      "loss": 0.9024,
+      "step": 21239
+    },
+    {
+      "epoch": 0.5833401993902941,
+      "grad_norm": 0.0,
+      "learning_rate": 7.80490593554599e-06,
+      "loss": 0.8125,
+      "step": 21240
+    },
+    {
+      "epoch": 0.5833676636181374,
+      "grad_norm": 0.0,
+      "learning_rate": 7.804038123173048e-06,
+      "loss": 0.8571,
+      "step": 21241
+    },
+    {
+      "epoch": 0.5833951278459806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.803170328175409e-06,
+      "loss": 0.7478,
+      "step": 21242
+    },
+    {
+      "epoch": 0.5834225920738239,
+      "grad_norm": 0.0,
+      "learning_rate": 7.802302550559935e-06,
+      "loss": 0.9029,
+      "step": 21243
+    },
+    {
+      "epoch": 0.5834500563016671,
+      "grad_norm": 0.0,
+      "learning_rate": 7.801434790333496e-06,
+      "loss": 0.8174,
+      "step": 21244
+    },
+    {
+      "epoch": 0.5834775205295103,
+      "grad_norm": 0.0,
+      "learning_rate": 7.800567047502955e-06,
+      "loss": 0.8091,
+      "step": 21245
+    },
+    {
+      "epoch": 0.5835049847573536,
+      "grad_norm": 0.0,
+      "learning_rate": 7.799699322075184e-06,
+      "loss": 0.8719,
+      "step": 21246
+    },
+    {
+      "epoch": 0.5835324489851967,
+      "grad_norm": 0.0,
+      "learning_rate": 7.798831614057042e-06,
+      "loss": 0.8292,
+      "step": 21247
+    },
+    {
+      "epoch": 0.58355991321304,
+      "grad_norm": 0.0,
+      "learning_rate": 7.797963923455395e-06,
+      "loss": 0.8571,
+      "step": 21248
+    },
+    {
+      "epoch": 0.5835873774408833,
+      "grad_norm": 0.0,
+      "learning_rate": 7.797096250277112e-06,
+      "loss": 0.9225,
+      "step": 21249
+    },
+    {
+      "epoch": 0.5836148416687265,
+      "grad_norm": 0.0,
+      "learning_rate": 7.796228594529056e-06,
+      "loss": 0.7944,
+      "step": 21250
+    },
+    {
+      "epoch": 0.5836423058965697,
+      "grad_norm": 0.0,
+      "learning_rate": 7.795360956218092e-06,
+      "loss": 0.838,
+      "step": 21251
+    },
+    {
+      "epoch": 0.5836697701244129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.794493335351091e-06,
+      "loss": 0.9704,
+      "step": 21252
+    },
+    {
+      "epoch": 0.5836972343522562,
+      "grad_norm": 0.0,
+      "learning_rate": 7.793625731934908e-06,
+      "loss": 0.8144,
+      "step": 21253
+    },
+    {
+      "epoch": 0.5837246985800995,
+      "grad_norm": 0.0,
+      "learning_rate": 7.792758145976416e-06,
+      "loss": 0.9196,
+      "step": 21254
+    },
+    {
+      "epoch": 0.5837521628079426,
+      "grad_norm": 0.0,
+      "learning_rate": 7.791890577482474e-06,
+      "loss": 0.8589,
+      "step": 21255
+    },
+    {
+      "epoch": 0.5837796270357859,
+      "grad_norm": 0.0,
+      "learning_rate": 7.791023026459949e-06,
+      "loss": 0.8536,
+      "step": 21256
+    },
+    {
+      "epoch": 0.5838070912636292,
+      "grad_norm": 0.0,
+      "learning_rate": 7.790155492915708e-06,
+      "loss": 0.8132,
+      "step": 21257
+    },
+    {
+      "epoch": 0.5838345554914723,
+      "grad_norm": 0.0,
+      "learning_rate": 7.789287976856614e-06,
+      "loss": 0.9304,
+      "step": 21258
+    },
+    {
+      "epoch": 0.5838620197193156,
+      "grad_norm": 0.0,
+      "learning_rate": 7.788420478289525e-06,
+      "loss": 0.8486,
+      "step": 21259
+    },
+    {
+      "epoch": 0.5838894839471588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.787552997221311e-06,
+      "loss": 0.9148,
+      "step": 21260
+    },
+    {
+      "epoch": 0.5839169481750021,
+      "grad_norm": 0.0,
+      "learning_rate": 7.786685533658838e-06,
+      "loss": 0.8052,
+      "step": 21261
+    },
+    {
+      "epoch": 0.5839444124028453,
+      "grad_norm": 0.0,
+      "learning_rate": 7.785818087608963e-06,
+      "loss": 0.8756,
+      "step": 21262
+    },
+    {
+      "epoch": 0.5839718766306885,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784950659078553e-06,
+      "loss": 0.8594,
+      "step": 21263
+    },
+    {
+      "epoch": 0.5839993408585318,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784083248074475e-06,
+      "loss": 0.8479,
+      "step": 21264
+    },
+    {
+      "epoch": 0.584026805086375,
+      "grad_norm": 0.0,
+      "learning_rate": 7.783215854603585e-06,
+      "loss": 0.9134,
+      "step": 21265
+    },
+    {
+      "epoch": 0.5840542693142182,
+      "grad_norm": 0.0,
+      "learning_rate": 7.782348478672753e-06,
+      "loss": 0.8531,
+      "step": 21266
+    },
+    {
+      "epoch": 0.5840817335420615,
+      "grad_norm": 0.0,
+      "learning_rate": 7.781481120288836e-06,
+      "loss": 0.752,
+      "step": 21267
+    },
+    {
+      "epoch": 0.5841091977699047,
+      "grad_norm": 0.0,
+      "learning_rate": 7.780613779458699e-06,
+      "loss": 0.9847,
+      "step": 21268
+    },
+    {
+      "epoch": 0.5841366619977479,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77974645618921e-06,
+      "loss": 0.8805,
+      "step": 21269
+    },
+    {
+      "epoch": 0.5841641262255912,
+      "grad_norm": 0.0,
+      "learning_rate": 7.778879150487226e-06,
+      "loss": 0.8322,
+      "step": 21270
+    },
+    {
+      "epoch": 0.5841915904534344,
+      "grad_norm": 0.0,
+      "learning_rate": 7.778011862359608e-06,
+      "loss": 0.9277,
+      "step": 21271
+    },
+    {
+      "epoch": 0.5842190546812777,
+      "grad_norm": 0.0,
+      "learning_rate": 7.777144591813222e-06,
+      "loss": 0.9399,
+      "step": 21272
+    },
+    {
+      "epoch": 0.5842465189091208,
+      "grad_norm": 0.0,
+      "learning_rate": 7.776277338854928e-06,
+      "loss": 0.9562,
+      "step": 21273
+    },
+    {
+      "epoch": 0.5842739831369641,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77541010349159e-06,
+      "loss": 0.8395,
+      "step": 21274
+    },
+    {
+      "epoch": 0.5843014473648074,
+      "grad_norm": 0.0,
+      "learning_rate": 7.774542885730067e-06,
+      "loss": 0.8882,
+      "step": 21275
+    },
+    {
+      "epoch": 0.5843289115926505,
+      "grad_norm": 0.0,
+      "learning_rate": 7.773675685577227e-06,
+      "loss": 0.8381,
+      "step": 21276
+    },
+    {
+      "epoch": 0.5843563758204938,
+      "grad_norm": 0.0,
+      "learning_rate": 7.772808503039924e-06,
+      "loss": 0.7442,
+      "step": 21277
+    },
+    {
+      "epoch": 0.584383840048337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.771941338125027e-06,
+      "loss": 0.8098,
+      "step": 21278
+    },
+    {
+      "epoch": 0.5844113042761803,
+      "grad_norm": 0.0,
+      "learning_rate": 7.771074190839387e-06,
+      "loss": 0.8644,
+      "step": 21279
+    },
+    {
+      "epoch": 0.5844387685040235,
+      "grad_norm": 0.0,
+      "learning_rate": 7.770207061189873e-06,
+      "loss": 0.9178,
+      "step": 21280
+    },
+    {
+      "epoch": 0.5844662327318667,
+      "grad_norm": 0.0,
+      "learning_rate": 7.769339949183347e-06,
+      "loss": 0.8571,
+      "step": 21281
+    },
+    {
+      "epoch": 0.58449369695971,
+      "grad_norm": 0.0,
+      "learning_rate": 7.76847285482667e-06,
+      "loss": 0.7951,
+      "step": 21282
+    },
+    {
+      "epoch": 0.5845211611875533,
+      "grad_norm": 0.0,
+      "learning_rate": 7.767605778126695e-06,
+      "loss": 0.922,
+      "step": 21283
+    },
+    {
+      "epoch": 0.5845486254153964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.76673871909029e-06,
+      "loss": 0.8975,
+      "step": 21284
+    },
+    {
+      "epoch": 0.5845760896432397,
+      "grad_norm": 0.0,
+      "learning_rate": 7.765871677724312e-06,
+      "loss": 0.9065,
+      "step": 21285
+    },
+    {
+      "epoch": 0.5846035538710829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.765004654035623e-06,
+      "loss": 0.9218,
+      "step": 21286
+    },
+    {
+      "epoch": 0.5846310180989261,
+      "grad_norm": 0.0,
+      "learning_rate": 7.764137648031086e-06,
+      "loss": 0.867,
+      "step": 21287
+    },
+    {
+      "epoch": 0.5846584823267694,
+      "grad_norm": 0.0,
+      "learning_rate": 7.763270659717554e-06,
+      "loss": 0.7829,
+      "step": 21288
+    },
+    {
+      "epoch": 0.5846859465546126,
+      "grad_norm": 0.0,
+      "learning_rate": 7.762403689101894e-06,
+      "loss": 0.9084,
+      "step": 21289
+    },
+    {
+      "epoch": 0.5847134107824559,
+      "grad_norm": 0.0,
+      "learning_rate": 7.76153673619096e-06,
+      "loss": 0.8012,
+      "step": 21290
+    },
+    {
+      "epoch": 0.584740875010299,
+      "grad_norm": 0.0,
+      "learning_rate": 7.760669800991614e-06,
+      "loss": 0.9004,
+      "step": 21291
+    },
+    {
+      "epoch": 0.5847683392381423,
+      "grad_norm": 0.0,
+      "learning_rate": 7.759802883510716e-06,
+      "loss": 0.8596,
+      "step": 21292
+    },
+    {
+      "epoch": 0.5847958034659856,
+      "grad_norm": 0.0,
+      "learning_rate": 7.758935983755131e-06,
+      "loss": 0.7279,
+      "step": 21293
+    },
+    {
+      "epoch": 0.5848232676938288,
+      "grad_norm": 0.0,
+      "learning_rate": 7.75806910173171e-06,
+      "loss": 0.9088,
+      "step": 21294
+    },
+    {
+      "epoch": 0.584850731921672,
+      "grad_norm": 0.0,
+      "learning_rate": 7.757202237447311e-06,
+      "loss": 0.8834,
+      "step": 21295
+    },
+    {
+      "epoch": 0.5848781961495153,
+      "grad_norm": 0.0,
+      "learning_rate": 7.756335390908798e-06,
+      "loss": 0.8842,
+      "step": 21296
+    },
+    {
+      "epoch": 0.5849056603773585,
+      "grad_norm": 0.0,
+      "learning_rate": 7.75546856212303e-06,
+      "loss": 0.8814,
+      "step": 21297
+    },
+    {
+      "epoch": 0.5849331246052017,
+      "grad_norm": 0.0,
+      "learning_rate": 7.754601751096864e-06,
+      "loss": 0.8518,
+      "step": 21298
+    },
+    {
+      "epoch": 0.5849605888330449,
+      "grad_norm": 0.0,
+      "learning_rate": 7.75373495783716e-06,
+      "loss": 0.803,
+      "step": 21299
+    },
+    {
+      "epoch": 0.5849880530608882,
+      "grad_norm": 0.0,
+      "learning_rate": 7.752868182350771e-06,
+      "loss": 0.8712,
+      "step": 21300
+    },
+    {
+      "epoch": 0.5850155172887315,
+      "grad_norm": 0.0,
+      "learning_rate": 7.752001424644565e-06,
+      "loss": 0.8152,
+      "step": 21301
+    },
+    {
+      "epoch": 0.5850429815165746,
+      "grad_norm": 0.0,
+      "learning_rate": 7.751134684725389e-06,
+      "loss": 0.8196,
+      "step": 21302
+    },
+    {
+      "epoch": 0.5850704457444179,
+      "grad_norm": 0.0,
+      "learning_rate": 7.750267962600107e-06,
+      "loss": 0.8353,
+      "step": 21303
+    },
+    {
+      "epoch": 0.5850979099722611,
+      "grad_norm": 0.0,
+      "learning_rate": 7.749401258275577e-06,
+      "loss": 0.9043,
+      "step": 21304
+    },
+    {
+      "epoch": 0.5851253742001044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.748534571758661e-06,
+      "loss": 0.8652,
+      "step": 21305
+    },
+    {
+      "epoch": 0.5851528384279476,
+      "grad_norm": 0.0,
+      "learning_rate": 7.747667903056208e-06,
+      "loss": 0.8391,
+      "step": 21306
+    },
+    {
+      "epoch": 0.5851803026557908,
+      "grad_norm": 0.0,
+      "learning_rate": 7.746801252175079e-06,
+      "loss": 0.7832,
+      "step": 21307
+    },
+    {
+      "epoch": 0.5852077668836341,
+      "grad_norm": 0.0,
+      "learning_rate": 7.74593461912213e-06,
+      "loss": 0.767,
+      "step": 21308
+    },
+    {
+      "epoch": 0.5852352311114773,
+      "grad_norm": 0.0,
+      "learning_rate": 7.745068003904221e-06,
+      "loss": 0.8508,
+      "step": 21309
+    },
+    {
+      "epoch": 0.5852626953393205,
+      "grad_norm": 0.0,
+      "learning_rate": 7.744201406528207e-06,
+      "loss": 0.852,
+      "step": 21310
+    },
+    {
+      "epoch": 0.5852901595671638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.743334827000947e-06,
+      "loss": 0.9281,
+      "step": 21311
+    },
+    {
+      "epoch": 0.585317623795007,
+      "grad_norm": 0.0,
+      "learning_rate": 7.742468265329293e-06,
+      "loss": 0.8774,
+      "step": 21312
+    },
+    {
+      "epoch": 0.5853450880228502,
+      "grad_norm": 0.0,
+      "learning_rate": 7.741601721520108e-06,
+      "loss": 0.8134,
+      "step": 21313
+    },
+    {
+      "epoch": 0.5853725522506935,
+      "grad_norm": 0.0,
+      "learning_rate": 7.74073519558024e-06,
+      "loss": 0.6863,
+      "step": 21314
+    },
+    {
+      "epoch": 0.5854000164785367,
+      "grad_norm": 0.0,
+      "learning_rate": 7.739868687516553e-06,
+      "loss": 0.8176,
+      "step": 21315
+    },
+    {
+      "epoch": 0.58542748070638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.7390021973359e-06,
+      "loss": 1.0087,
+      "step": 21316
+    },
+    {
+      "epoch": 0.5854549449342231,
+      "grad_norm": 0.0,
+      "learning_rate": 7.738135725045141e-06,
+      "loss": 0.9542,
+      "step": 21317
+    },
+    {
+      "epoch": 0.5854824091620664,
+      "grad_norm": 0.0,
+      "learning_rate": 7.737269270651126e-06,
+      "loss": 0.8196,
+      "step": 21318
+    },
+    {
+      "epoch": 0.5855098733899097,
+      "grad_norm": 0.0,
+      "learning_rate": 7.736402834160711e-06,
+      "loss": 0.8371,
+      "step": 21319
+    },
+    {
+      "epoch": 0.5855373376177528,
+      "grad_norm": 0.0,
+      "learning_rate": 7.735536415580755e-06,
+      "loss": 0.8986,
+      "step": 21320
+    },
+    {
+      "epoch": 0.5855648018455961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.734670014918112e-06,
+      "loss": 0.8221,
+      "step": 21321
+    },
+    {
+      "epoch": 0.5855922660734394,
+      "grad_norm": 0.0,
+      "learning_rate": 7.733803632179635e-06,
+      "loss": 0.9156,
+      "step": 21322
+    },
+    {
+      "epoch": 0.5856197303012826,
+      "grad_norm": 0.0,
+      "learning_rate": 7.732937267372189e-06,
+      "loss": 0.8216,
+      "step": 21323
+    },
+    {
+      "epoch": 0.5856471945291258,
+      "grad_norm": 0.0,
+      "learning_rate": 7.732070920502615e-06,
+      "loss": 0.8002,
+      "step": 21324
+    },
+    {
+      "epoch": 0.585674658756969,
+      "grad_norm": 0.0,
+      "learning_rate": 7.731204591577777e-06,
+      "loss": 0.8593,
+      "step": 21325
+    },
+    {
+      "epoch": 0.5857021229848123,
+      "grad_norm": 0.0,
+      "learning_rate": 7.730338280604524e-06,
+      "loss": 0.8905,
+      "step": 21326
+    },
+    {
+      "epoch": 0.5857295872126556,
+      "grad_norm": 0.0,
+      "learning_rate": 7.729471987589713e-06,
+      "loss": 0.8433,
+      "step": 21327
+    },
+    {
+      "epoch": 0.5857570514404987,
+      "grad_norm": 0.0,
+      "learning_rate": 7.728605712540205e-06,
+      "loss": 0.8075,
+      "step": 21328
+    },
+    {
+      "epoch": 0.585784515668342,
+      "grad_norm": 0.0,
+      "learning_rate": 7.727739455462845e-06,
+      "loss": 0.8453,
+      "step": 21329
+    },
+    {
+      "epoch": 0.5858119798961852,
+      "grad_norm": 0.0,
+      "learning_rate": 7.726873216364492e-06,
+      "loss": 0.8804,
+      "step": 21330
+    },
+    {
+      "epoch": 0.5858394441240284,
+      "grad_norm": 0.0,
+      "learning_rate": 7.726006995251997e-06,
+      "loss": 0.9076,
+      "step": 21331
+    },
+    {
+      "epoch": 0.5858669083518717,
+      "grad_norm": 0.0,
+      "learning_rate": 7.725140792132215e-06,
+      "loss": 0.8039,
+      "step": 21332
+    },
+    {
+      "epoch": 0.5858943725797149,
+      "grad_norm": 0.0,
+      "learning_rate": 7.724274607012001e-06,
+      "loss": 0.8793,
+      "step": 21333
+    },
+    {
+      "epoch": 0.5859218368075582,
+      "grad_norm": 0.0,
+      "learning_rate": 7.723408439898208e-06,
+      "loss": 0.8081,
+      "step": 21334
+    },
+    {
+      "epoch": 0.5859493010354014,
+      "grad_norm": 0.0,
+      "learning_rate": 7.722542290797692e-06,
+      "loss": 0.9476,
+      "step": 21335
+    },
+    {
+      "epoch": 0.5859767652632446,
+      "grad_norm": 0.0,
+      "learning_rate": 7.7216761597173e-06,
+      "loss": 0.828,
+      "step": 21336
+    },
+    {
+      "epoch": 0.5860042294910879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.72081004666389e-06,
+      "loss": 0.8465,
+      "step": 21337
+    },
+    {
+      "epoch": 0.586031693718931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.719943951644313e-06,
+      "loss": 0.8661,
+      "step": 21338
+    },
+    {
+      "epoch": 0.5860591579467743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.719077874665421e-06,
+      "loss": 0.8766,
+      "step": 21339
+    },
+    {
+      "epoch": 0.5860866221746176,
+      "grad_norm": 0.0,
+      "learning_rate": 7.718211815734072e-06,
+      "loss": 0.8746,
+      "step": 21340
+    },
+    {
+      "epoch": 0.5861140864024608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.717345774857115e-06,
+      "loss": 0.8369,
+      "step": 21341
+    },
+    {
+      "epoch": 0.586141550630304,
+      "grad_norm": 0.0,
+      "learning_rate": 7.716479752041401e-06,
+      "loss": 0.8167,
+      "step": 21342
+    },
+    {
+      "epoch": 0.5861690148581472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.715613747293783e-06,
+      "loss": 0.8412,
+      "step": 21343
+    },
+    {
+      "epoch": 0.5861964790859905,
+      "grad_norm": 0.0,
+      "learning_rate": 7.714747760621114e-06,
+      "loss": 0.8354,
+      "step": 21344
+    },
+    {
+      "epoch": 0.5862239433138338,
+      "grad_norm": 0.0,
+      "learning_rate": 7.713881792030247e-06,
+      "loss": 0.9127,
+      "step": 21345
+    },
+    {
+      "epoch": 0.5862514075416769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.713015841528036e-06,
+      "loss": 0.8688,
+      "step": 21346
+    },
+    {
+      "epoch": 0.5862788717695202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.712149909121326e-06,
+      "loss": 0.936,
+      "step": 21347
+    },
+    {
+      "epoch": 0.5863063359973635,
+      "grad_norm": 0.0,
+      "learning_rate": 7.71128399481697e-06,
+      "loss": 0.8198,
+      "step": 21348
+    },
+    {
+      "epoch": 0.5863338002252066,
+      "grad_norm": 0.0,
+      "learning_rate": 7.710418098621824e-06,
+      "loss": 0.9064,
+      "step": 21349
+    },
+    {
+      "epoch": 0.5863612644530499,
+      "grad_norm": 0.0,
+      "learning_rate": 7.709552220542736e-06,
+      "loss": 0.8248,
+      "step": 21350
+    },
+    {
+      "epoch": 0.5863887286808931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.708686360586557e-06,
+      "loss": 0.7991,
+      "step": 21351
+    },
+    {
+      "epoch": 0.5864161929087364,
+      "grad_norm": 0.0,
+      "learning_rate": 7.707820518760144e-06,
+      "loss": 0.9034,
+      "step": 21352
+    },
+    {
+      "epoch": 0.5864436571365796,
+      "grad_norm": 0.0,
+      "learning_rate": 7.706954695070339e-06,
+      "loss": 0.9173,
+      "step": 21353
+    },
+    {
+      "epoch": 0.5864711213644228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.706088889523999e-06,
+      "loss": 0.8794,
+      "step": 21354
+    },
+    {
+      "epoch": 0.5864985855922661,
+      "grad_norm": 0.0,
+      "learning_rate": 7.705223102127969e-06,
+      "loss": 0.8145,
+      "step": 21355
+    },
+    {
+      "epoch": 0.5865260498201094,
+      "grad_norm": 0.0,
+      "learning_rate": 7.704357332889103e-06,
+      "loss": 0.8991,
+      "step": 21356
+    },
+    {
+      "epoch": 0.5865535140479525,
+      "grad_norm": 0.0,
+      "learning_rate": 7.703491581814253e-06,
+      "loss": 0.8563,
+      "step": 21357
+    },
+    {
+      "epoch": 0.5865809782757958,
+      "grad_norm": 0.0,
+      "learning_rate": 7.70262584891027e-06,
+      "loss": 0.8266,
+      "step": 21358
+    },
+    {
+      "epoch": 0.586608442503639,
+      "grad_norm": 0.0,
+      "learning_rate": 7.701760134183997e-06,
+      "loss": 0.9081,
+      "step": 21359
+    },
+    {
+      "epoch": 0.5866359067314822,
+      "grad_norm": 0.0,
+      "learning_rate": 7.700894437642288e-06,
+      "loss": 0.764,
+      "step": 21360
+    },
+    {
+      "epoch": 0.5866633709593255,
+      "grad_norm": 0.0,
+      "learning_rate": 7.700028759291995e-06,
+      "loss": 0.8091,
+      "step": 21361
+    },
+    {
+      "epoch": 0.5866908351871687,
+      "grad_norm": 0.0,
+      "learning_rate": 7.699163099139963e-06,
+      "loss": 0.9391,
+      "step": 21362
+    },
+    {
+      "epoch": 0.586718299415012,
+      "grad_norm": 0.0,
+      "learning_rate": 7.698297457193044e-06,
+      "loss": 0.9377,
+      "step": 21363
+    },
+    {
+      "epoch": 0.5867457636428551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.69743183345809e-06,
+      "loss": 0.8882,
+      "step": 21364
+    },
+    {
+      "epoch": 0.5867732278706984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.696566227941944e-06,
+      "loss": 0.8495,
+      "step": 21365
+    },
+    {
+      "epoch": 0.5868006920985417,
+      "grad_norm": 0.0,
+      "learning_rate": 7.695700640651462e-06,
+      "loss": 0.8628,
+      "step": 21366
+    },
+    {
+      "epoch": 0.5868281563263849,
+      "grad_norm": 0.0,
+      "learning_rate": 7.694835071593485e-06,
+      "loss": 0.9472,
+      "step": 21367
+    },
+    {
+      "epoch": 0.5868556205542281,
+      "grad_norm": 0.0,
+      "learning_rate": 7.693969520774867e-06,
+      "loss": 0.8994,
+      "step": 21368
+    },
+    {
+      "epoch": 0.5868830847820714,
+      "grad_norm": 0.0,
+      "learning_rate": 7.693103988202456e-06,
+      "loss": 0.8092,
+      "step": 21369
+    },
+    {
+      "epoch": 0.5869105490099146,
+      "grad_norm": 0.0,
+      "learning_rate": 7.692238473883102e-06,
+      "loss": 0.8328,
+      "step": 21370
+    },
+    {
+      "epoch": 0.5869380132377578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.69137297782365e-06,
+      "loss": 0.9312,
+      "step": 21371
+    },
+    {
+      "epoch": 0.586965477465601,
+      "grad_norm": 0.0,
+      "learning_rate": 7.690507500030946e-06,
+      "loss": 0.8625,
+      "step": 21372
+    },
+    {
+      "epoch": 0.5869929416934443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.689642040511844e-06,
+      "loss": 0.8616,
+      "step": 21373
+    },
+    {
+      "epoch": 0.5870204059212876,
+      "grad_norm": 0.0,
+      "learning_rate": 7.688776599273187e-06,
+      "loss": 0.927,
+      "step": 21374
+    },
+    {
+      "epoch": 0.5870478701491307,
+      "grad_norm": 0.0,
+      "learning_rate": 7.687911176321826e-06,
+      "loss": 0.8136,
+      "step": 21375
+    },
+    {
+      "epoch": 0.587075334376974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68704577166461e-06,
+      "loss": 0.8817,
+      "step": 21376
+    },
+    {
+      "epoch": 0.5871027986048172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.686180385308382e-06,
+      "loss": 0.8931,
+      "step": 21377
+    },
+    {
+      "epoch": 0.5871302628326605,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68531501725999e-06,
+      "loss": 0.9476,
+      "step": 21378
+    },
+    {
+      "epoch": 0.5871577270605037,
+      "grad_norm": 0.0,
+      "learning_rate": 7.684449667526284e-06,
+      "loss": 0.8946,
+      "step": 21379
+    },
+    {
+      "epoch": 0.5871851912883469,
+      "grad_norm": 0.0,
+      "learning_rate": 7.683584336114107e-06,
+      "loss": 0.7722,
+      "step": 21380
+    },
+    {
+      "epoch": 0.5872126555161902,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68271902303031e-06,
+      "loss": 0.798,
+      "step": 21381
+    },
+    {
+      "epoch": 0.5872401197440335,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68185372828174e-06,
+      "loss": 0.9102,
+      "step": 21382
+    },
+    {
+      "epoch": 0.5872675839718766,
+      "grad_norm": 0.0,
+      "learning_rate": 7.680988451875237e-06,
+      "loss": 0.9232,
+      "step": 21383
+    },
+    {
+      "epoch": 0.5872950481997199,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68012319381765e-06,
+      "loss": 0.7945,
+      "step": 21384
+    },
+    {
+      "epoch": 0.5873225124275631,
+      "grad_norm": 0.0,
+      "learning_rate": 7.679257954115832e-06,
+      "loss": 0.8736,
+      "step": 21385
+    },
+    {
+      "epoch": 0.5873499766554063,
+      "grad_norm": 0.0,
+      "learning_rate": 7.678392732776621e-06,
+      "loss": 0.983,
+      "step": 21386
+    },
+    {
+      "epoch": 0.5873774408832496,
+      "grad_norm": 0.0,
+      "learning_rate": 7.677527529806865e-06,
+      "loss": 0.9283,
+      "step": 21387
+    },
+    {
+      "epoch": 0.5874049051110928,
+      "grad_norm": 0.0,
+      "learning_rate": 7.676662345213416e-06,
+      "loss": 0.8587,
+      "step": 21388
+    },
+    {
+      "epoch": 0.5874323693389361,
+      "grad_norm": 0.0,
+      "learning_rate": 7.675797179003109e-06,
+      "loss": 0.8284,
+      "step": 21389
+    },
+    {
+      "epoch": 0.5874598335667792,
+      "grad_norm": 0.0,
+      "learning_rate": 7.674932031182797e-06,
+      "loss": 0.8617,
+      "step": 21390
+    },
+    {
+      "epoch": 0.5874872977946225,
+      "grad_norm": 0.0,
+      "learning_rate": 7.674066901759324e-06,
+      "loss": 0.8229,
+      "step": 21391
+    },
+    {
+      "epoch": 0.5875147620224658,
+      "grad_norm": 0.0,
+      "learning_rate": 7.673201790739533e-06,
+      "loss": 0.8831,
+      "step": 21392
+    },
+    {
+      "epoch": 0.5875422262503089,
+      "grad_norm": 0.0,
+      "learning_rate": 7.672336698130272e-06,
+      "loss": 0.8905,
+      "step": 21393
+    },
+    {
+      "epoch": 0.5875696904781522,
+      "grad_norm": 0.0,
+      "learning_rate": 7.671471623938389e-06,
+      "loss": 0.8594,
+      "step": 21394
+    },
+    {
+      "epoch": 0.5875971547059955,
+      "grad_norm": 0.0,
+      "learning_rate": 7.67060656817072e-06,
+      "loss": 0.8988,
+      "step": 21395
+    },
+    {
+      "epoch": 0.5876246189338387,
+      "grad_norm": 0.0,
+      "learning_rate": 7.669741530834112e-06,
+      "loss": 0.9568,
+      "step": 21396
+    },
+    {
+      "epoch": 0.5876520831616819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.668876511935415e-06,
+      "loss": 0.8929,
+      "step": 21397
+    },
+    {
+      "epoch": 0.5876795473895251,
+      "grad_norm": 0.0,
+      "learning_rate": 7.668011511481467e-06,
+      "loss": 0.8885,
+      "step": 21398
+    },
+    {
+      "epoch": 0.5877070116173684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.667146529479117e-06,
+      "loss": 0.8444,
+      "step": 21399
+    },
+    {
+      "epoch": 0.5877344758452117,
+      "grad_norm": 0.0,
+      "learning_rate": 7.66628156593521e-06,
+      "loss": 0.8639,
+      "step": 21400
+    },
+    {
+      "epoch": 0.5877619400730548,
+      "grad_norm": 0.0,
+      "learning_rate": 7.665416620856583e-06,
+      "loss": 0.9165,
+      "step": 21401
+    },
+    {
+      "epoch": 0.5877894043008981,
+      "grad_norm": 0.0,
+      "learning_rate": 7.664551694250088e-06,
+      "loss": 0.9578,
+      "step": 21402
+    },
+    {
+      "epoch": 0.5878168685287413,
+      "grad_norm": 0.0,
+      "learning_rate": 7.663686786122562e-06,
+      "loss": 0.8217,
+      "step": 21403
+    },
+    {
+      "epoch": 0.5878443327565845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.66282189648085e-06,
+      "loss": 0.7212,
+      "step": 21404
+    },
+    {
+      "epoch": 0.5878717969844278,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6619570253318e-06,
+      "loss": 0.8606,
+      "step": 21405
+    },
+    {
+      "epoch": 0.587899261212271,
+      "grad_norm": 0.0,
+      "learning_rate": 7.661092172682252e-06,
+      "loss": 0.8163,
+      "step": 21406
+    },
+    {
+      "epoch": 0.5879267254401143,
+      "grad_norm": 0.0,
+      "learning_rate": 7.660227338539044e-06,
+      "loss": 0.9824,
+      "step": 21407
+    },
+    {
+      "epoch": 0.5879541896679575,
+      "grad_norm": 0.0,
+      "learning_rate": 7.659362522909027e-06,
+      "loss": 0.865,
+      "step": 21408
+    },
+    {
+      "epoch": 0.5879816538958007,
+      "grad_norm": 0.0,
+      "learning_rate": 7.658497725799039e-06,
+      "loss": 0.8414,
+      "step": 21409
+    },
+    {
+      "epoch": 0.588009118123644,
+      "grad_norm": 0.0,
+      "learning_rate": 7.657632947215923e-06,
+      "loss": 0.8501,
+      "step": 21410
+    },
+    {
+      "epoch": 0.5880365823514871,
+      "grad_norm": 0.0,
+      "learning_rate": 7.656768187166528e-06,
+      "loss": 0.8275,
+      "step": 21411
+    },
+    {
+      "epoch": 0.5880640465793304,
+      "grad_norm": 0.0,
+      "learning_rate": 7.655903445657687e-06,
+      "loss": 0.888,
+      "step": 21412
+    },
+    {
+      "epoch": 0.5880915108071737,
+      "grad_norm": 0.0,
+      "learning_rate": 7.655038722696243e-06,
+      "loss": 0.9042,
+      "step": 21413
+    },
+    {
+      "epoch": 0.5881189750350169,
+      "grad_norm": 0.0,
+      "learning_rate": 7.654174018289045e-06,
+      "loss": 0.9823,
+      "step": 21414
+    },
+    {
+      "epoch": 0.5881464392628601,
+      "grad_norm": 0.0,
+      "learning_rate": 7.653309332442928e-06,
+      "loss": 0.9469,
+      "step": 21415
+    },
+    {
+      "epoch": 0.5881739034907033,
+      "grad_norm": 0.0,
+      "learning_rate": 7.652444665164738e-06,
+      "loss": 0.7904,
+      "step": 21416
+    },
+    {
+      "epoch": 0.5882013677185466,
+      "grad_norm": 0.0,
+      "learning_rate": 7.651580016461318e-06,
+      "loss": 0.8396,
+      "step": 21417
+    },
+    {
+      "epoch": 0.5882288319463899,
+      "grad_norm": 0.0,
+      "learning_rate": 7.650715386339505e-06,
+      "loss": 0.9994,
+      "step": 21418
+    },
+    {
+      "epoch": 0.588256296174233,
+      "grad_norm": 0.0,
+      "learning_rate": 7.64985077480614e-06,
+      "loss": 0.8334,
+      "step": 21419
+    },
+    {
+      "epoch": 0.5882837604020763,
+      "grad_norm": 0.0,
+      "learning_rate": 7.648986181868066e-06,
+      "loss": 0.8702,
+      "step": 21420
+    },
+    {
+      "epoch": 0.5883112246299196,
+      "grad_norm": 0.0,
+      "learning_rate": 7.648121607532125e-06,
+      "loss": 0.8864,
+      "step": 21421
+    },
+    {
+      "epoch": 0.5883386888577627,
+      "grad_norm": 0.0,
+      "learning_rate": 7.647257051805156e-06,
+      "loss": 0.812,
+      "step": 21422
+    },
+    {
+      "epoch": 0.588366153085606,
+      "grad_norm": 0.0,
+      "learning_rate": 7.646392514694004e-06,
+      "loss": 0.8382,
+      "step": 21423
+    },
+    {
+      "epoch": 0.5883936173134492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.645527996205502e-06,
+      "loss": 0.7509,
+      "step": 21424
+    },
+    {
+      "epoch": 0.5884210815412925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.644663496346496e-06,
+      "loss": 0.8483,
+      "step": 21425
+    },
+    {
+      "epoch": 0.5884485457691357,
+      "grad_norm": 0.0,
+      "learning_rate": 7.643799015123824e-06,
+      "loss": 0.7674,
+      "step": 21426
+    },
+    {
+      "epoch": 0.5884760099969789,
+      "grad_norm": 0.0,
+      "learning_rate": 7.642934552544327e-06,
+      "loss": 0.8345,
+      "step": 21427
+    },
+    {
+      "epoch": 0.5885034742248222,
+      "grad_norm": 0.0,
+      "learning_rate": 7.642070108614846e-06,
+      "loss": 0.8111,
+      "step": 21428
+    },
+    {
+      "epoch": 0.5885309384526654,
+      "grad_norm": 0.0,
+      "learning_rate": 7.641205683342222e-06,
+      "loss": 0.9527,
+      "step": 21429
+    },
+    {
+      "epoch": 0.5885584026805086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.640341276733291e-06,
+      "loss": 0.7858,
+      "step": 21430
+    },
+    {
+      "epoch": 0.5885858669083519,
+      "grad_norm": 0.0,
+      "learning_rate": 7.639476888794892e-06,
+      "loss": 0.8508,
+      "step": 21431
+    },
+    {
+      "epoch": 0.5886133311361951,
+      "grad_norm": 0.0,
+      "learning_rate": 7.638612519533865e-06,
+      "loss": 0.8171,
+      "step": 21432
+    },
+    {
+      "epoch": 0.5886407953640383,
+      "grad_norm": 0.0,
+      "learning_rate": 7.637748168957051e-06,
+      "loss": 0.8254,
+      "step": 21433
+    },
+    {
+      "epoch": 0.5886682595918816,
+      "grad_norm": 0.0,
+      "learning_rate": 7.636883837071289e-06,
+      "loss": 0.8979,
+      "step": 21434
+    },
+    {
+      "epoch": 0.5886957238197248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.636019523883422e-06,
+      "loss": 0.8656,
+      "step": 21435
+    },
+    {
+      "epoch": 0.5887231880475681,
+      "grad_norm": 0.0,
+      "learning_rate": 7.63515522940028e-06,
+      "loss": 0.8967,
+      "step": 21436
+    },
+    {
+      "epoch": 0.5887506522754112,
+      "grad_norm": 0.0,
+      "learning_rate": 7.634290953628705e-06,
+      "loss": 0.8699,
+      "step": 21437
+    },
+    {
+      "epoch": 0.5887781165032545,
+      "grad_norm": 0.0,
+      "learning_rate": 7.633426696575538e-06,
+      "loss": 0.8328,
+      "step": 21438
+    },
+    {
+      "epoch": 0.5888055807310978,
+      "grad_norm": 0.0,
+      "learning_rate": 7.632562458247614e-06,
+      "loss": 0.8081,
+      "step": 21439
+    },
+    {
+      "epoch": 0.588833044958941,
+      "grad_norm": 0.0,
+      "learning_rate": 7.631698238651773e-06,
+      "loss": 0.8044,
+      "step": 21440
+    },
+    {
+      "epoch": 0.5888605091867842,
+      "grad_norm": 0.0,
+      "learning_rate": 7.630834037794857e-06,
+      "loss": 0.8818,
+      "step": 21441
+    },
+    {
+      "epoch": 0.5888879734146274,
+      "grad_norm": 0.0,
+      "learning_rate": 7.629969855683698e-06,
+      "loss": 0.807,
+      "step": 21442
+    },
+    {
+      "epoch": 0.5889154376424707,
+      "grad_norm": 0.0,
+      "learning_rate": 7.629105692325134e-06,
+      "loss": 0.9457,
+      "step": 21443
+    },
+    {
+      "epoch": 0.588942901870314,
+      "grad_norm": 0.0,
+      "learning_rate": 7.628241547726003e-06,
+      "loss": 0.9588,
+      "step": 21444
+    },
+    {
+      "epoch": 0.5889703660981571,
+      "grad_norm": 0.0,
+      "learning_rate": 7.627377421893147e-06,
+      "loss": 0.7324,
+      "step": 21445
+    },
+    {
+      "epoch": 0.5889978303260004,
+      "grad_norm": 0.0,
+      "learning_rate": 7.626513314833397e-06,
+      "loss": 0.7631,
+      "step": 21446
+    },
+    {
+      "epoch": 0.5890252945538437,
+      "grad_norm": 0.0,
+      "learning_rate": 7.625649226553598e-06,
+      "loss": 0.8631,
+      "step": 21447
+    },
+    {
+      "epoch": 0.5890527587816868,
+      "grad_norm": 0.0,
+      "learning_rate": 7.624785157060577e-06,
+      "loss": 0.8424,
+      "step": 21448
+    },
+    {
+      "epoch": 0.5890802230095301,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6239211063611775e-06,
+      "loss": 0.75,
+      "step": 21449
+    },
+    {
+      "epoch": 0.5891076872373733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.623057074462233e-06,
+      "loss": 0.827,
+      "step": 21450
+    },
+    {
+      "epoch": 0.5891351514652166,
+      "grad_norm": 0.0,
+      "learning_rate": 7.622193061370582e-06,
+      "loss": 0.7895,
+      "step": 21451
+    },
+    {
+      "epoch": 0.5891626156930598,
+      "grad_norm": 0.0,
+      "learning_rate": 7.621329067093061e-06,
+      "loss": 0.889,
+      "step": 21452
+    },
+    {
+      "epoch": 0.589190079920903,
+      "grad_norm": 0.0,
+      "learning_rate": 7.620465091636509e-06,
+      "loss": 0.844,
+      "step": 21453
+    },
+    {
+      "epoch": 0.5892175441487463,
+      "grad_norm": 0.0,
+      "learning_rate": 7.619601135007757e-06,
+      "loss": 0.9015,
+      "step": 21454
+    },
+    {
+      "epoch": 0.5892450083765894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.618737197213642e-06,
+      "loss": 0.8594,
+      "step": 21455
+    },
+    {
+      "epoch": 0.5892724726044327,
+      "grad_norm": 0.0,
+      "learning_rate": 7.617873278260999e-06,
+      "loss": 0.8249,
+      "step": 21456
+    },
+    {
+      "epoch": 0.589299936832276,
+      "grad_norm": 0.0,
+      "learning_rate": 7.617009378156668e-06,
+      "loss": 0.7459,
+      "step": 21457
+    },
+    {
+      "epoch": 0.5893274010601192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.61614549690748e-06,
+      "loss": 0.9071,
+      "step": 21458
+    },
+    {
+      "epoch": 0.5893548652879624,
+      "grad_norm": 0.0,
+      "learning_rate": 7.615281634520277e-06,
+      "loss": 0.8758,
+      "step": 21459
+    },
+    {
+      "epoch": 0.5893823295158057,
+      "grad_norm": 0.0,
+      "learning_rate": 7.614417791001885e-06,
+      "loss": 0.917,
+      "step": 21460
+    },
+    {
+      "epoch": 0.5894097937436489,
+      "grad_norm": 0.0,
+      "learning_rate": 7.613553966359145e-06,
+      "loss": 0.8334,
+      "step": 21461
+    },
+    {
+      "epoch": 0.5894372579714922,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6126901605988886e-06,
+      "loss": 0.8624,
+      "step": 21462
+    },
+    {
+      "epoch": 0.5894647221993353,
+      "grad_norm": 0.0,
+      "learning_rate": 7.611826373727953e-06,
+      "loss": 0.8904,
+      "step": 21463
+    },
+    {
+      "epoch": 0.5894921864271786,
+      "grad_norm": 0.0,
+      "learning_rate": 7.610962605753172e-06,
+      "loss": 0.9137,
+      "step": 21464
+    },
+    {
+      "epoch": 0.5895196506550219,
+      "grad_norm": 0.0,
+      "learning_rate": 7.610098856681384e-06,
+      "loss": 0.8276,
+      "step": 21465
+    },
+    {
+      "epoch": 0.589547114882865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.609235126519417e-06,
+      "loss": 0.8833,
+      "step": 21466
+    },
+    {
+      "epoch": 0.5895745791107083,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6083714152741075e-06,
+      "loss": 0.7495,
+      "step": 21467
+    },
+    {
+      "epoch": 0.5896020433385515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.607507722952287e-06,
+      "loss": 0.8628,
+      "step": 21468
+    },
+    {
+      "epoch": 0.5896295075663948,
+      "grad_norm": 0.0,
+      "learning_rate": 7.606644049560796e-06,
+      "loss": 0.7857,
+      "step": 21469
+    },
+    {
+      "epoch": 0.589656971794238,
+      "grad_norm": 0.0,
+      "learning_rate": 7.605780395106468e-06,
+      "loss": 0.8143,
+      "step": 21470
+    },
+    {
+      "epoch": 0.5896844360220812,
+      "grad_norm": 0.0,
+      "learning_rate": 7.604916759596126e-06,
+      "loss": 0.725,
+      "step": 21471
+    },
+    {
+      "epoch": 0.5897119002499245,
+      "grad_norm": 0.0,
+      "learning_rate": 7.604053143036612e-06,
+      "loss": 0.9244,
+      "step": 21472
+    },
+    {
+      "epoch": 0.5897393644777678,
+      "grad_norm": 0.0,
+      "learning_rate": 7.603189545434759e-06,
+      "loss": 0.805,
+      "step": 21473
+    },
+    {
+      "epoch": 0.5897668287056109,
+      "grad_norm": 0.0,
+      "learning_rate": 7.602325966797397e-06,
+      "loss": 0.9078,
+      "step": 21474
+    },
+    {
+      "epoch": 0.5897942929334542,
+      "grad_norm": 0.0,
+      "learning_rate": 7.601462407131362e-06,
+      "loss": 0.949,
+      "step": 21475
+    },
+    {
+      "epoch": 0.5898217571612974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6005988664434874e-06,
+      "loss": 0.8504,
+      "step": 21476
+    },
+    {
+      "epoch": 0.5898492213891406,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5997353447406004e-06,
+      "loss": 0.8066,
+      "step": 21477
+    },
+    {
+      "epoch": 0.5898766856169839,
+      "grad_norm": 0.0,
+      "learning_rate": 7.598871842029541e-06,
+      "loss": 0.7198,
+      "step": 21478
+    },
+    {
+      "epoch": 0.5899041498448271,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5980083583171325e-06,
+      "loss": 0.9074,
+      "step": 21479
+    },
+    {
+      "epoch": 0.5899316140726704,
+      "grad_norm": 0.0,
+      "learning_rate": 7.597144893610214e-06,
+      "loss": 0.8163,
+      "step": 21480
+    },
+    {
+      "epoch": 0.5899590783005135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.596281447915619e-06,
+      "loss": 0.8434,
+      "step": 21481
+    },
+    {
+      "epoch": 0.5899865425283568,
+      "grad_norm": 0.0,
+      "learning_rate": 7.595418021240177e-06,
+      "loss": 0.881,
+      "step": 21482
+    },
+    {
+      "epoch": 0.5900140067562001,
+      "grad_norm": 0.0,
+      "learning_rate": 7.594554613590715e-06,
+      "loss": 0.9305,
+      "step": 21483
+    },
+    {
+      "epoch": 0.5900414709840432,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5936912249740694e-06,
+      "loss": 0.9943,
+      "step": 21484
+    },
+    {
+      "epoch": 0.5900689352118865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.592827855397072e-06,
+      "loss": 0.8367,
+      "step": 21485
+    },
+    {
+      "epoch": 0.5900963994397298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.591964504866552e-06,
+      "loss": 0.8683,
+      "step": 21486
+    },
+    {
+      "epoch": 0.590123863667573,
+      "grad_norm": 0.0,
+      "learning_rate": 7.591101173389341e-06,
+      "loss": 0.8332,
+      "step": 21487
+    },
+    {
+      "epoch": 0.5901513278954162,
+      "grad_norm": 0.0,
+      "learning_rate": 7.590237860972275e-06,
+      "loss": 0.8793,
+      "step": 21488
+    },
+    {
+      "epoch": 0.5901787921232594,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5893745676221765e-06,
+      "loss": 0.8329,
+      "step": 21489
+    },
+    {
+      "epoch": 0.5902062563511027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.588511293345881e-06,
+      "loss": 0.8225,
+      "step": 21490
+    },
+    {
+      "epoch": 0.590233720578946,
+      "grad_norm": 0.0,
+      "learning_rate": 7.587648038150218e-06,
+      "loss": 0.9604,
+      "step": 21491
+    },
+    {
+      "epoch": 0.5902611848067891,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5867848020420176e-06,
+      "loss": 0.7927,
+      "step": 21492
+    },
+    {
+      "epoch": 0.5902886490346324,
+      "grad_norm": 0.0,
+      "learning_rate": 7.585921585028113e-06,
+      "loss": 0.9047,
+      "step": 21493
+    },
+    {
+      "epoch": 0.5903161132624756,
+      "grad_norm": 0.0,
+      "learning_rate": 7.585058387115334e-06,
+      "loss": 0.8365,
+      "step": 21494
+    },
+    {
+      "epoch": 0.5903435774903188,
+      "grad_norm": 0.0,
+      "learning_rate": 7.584195208310503e-06,
+      "loss": 0.8374,
+      "step": 21495
+    },
+    {
+      "epoch": 0.5903710417181621,
+      "grad_norm": 0.0,
+      "learning_rate": 7.583332048620457e-06,
+      "loss": 0.8799,
+      "step": 21496
+    },
+    {
+      "epoch": 0.5903985059460053,
+      "grad_norm": 0.0,
+      "learning_rate": 7.582468908052025e-06,
+      "loss": 0.8486,
+      "step": 21497
+    },
+    {
+      "epoch": 0.5904259701738486,
+      "grad_norm": 0.0,
+      "learning_rate": 7.581605786612034e-06,
+      "loss": 0.7468,
+      "step": 21498
+    },
+    {
+      "epoch": 0.5904534344016918,
+      "grad_norm": 0.0,
+      "learning_rate": 7.580742684307314e-06,
+      "loss": 0.8654,
+      "step": 21499
+    },
+    {
+      "epoch": 0.590480898629535,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5798796011446995e-06,
+      "loss": 0.8925,
+      "step": 21500
+    },
+    {
+      "epoch": 0.5905083628573783,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5790165371310105e-06,
+      "loss": 0.8239,
+      "step": 21501
+    },
+    {
+      "epoch": 0.5905358270852215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.578153492273083e-06,
+      "loss": 0.9044,
+      "step": 21502
+    },
+    {
+      "epoch": 0.5905632913130647,
+      "grad_norm": 0.0,
+      "learning_rate": 7.577290466577743e-06,
+      "loss": 0.7638,
+      "step": 21503
+    },
+    {
+      "epoch": 0.590590755540908,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5764274600518165e-06,
+      "loss": 0.8556,
+      "step": 21504
+    },
+    {
+      "epoch": 0.5906182197687512,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5755644727021385e-06,
+      "loss": 0.8418,
+      "step": 21505
+    },
+    {
+      "epoch": 0.5906456839965945,
+      "grad_norm": 0.0,
+      "learning_rate": 7.574701504535534e-06,
+      "loss": 0.945,
+      "step": 21506
+    },
+    {
+      "epoch": 0.5906731482244376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.573838555558828e-06,
+      "loss": 0.9699,
+      "step": 21507
+    },
+    {
+      "epoch": 0.5907006124522809,
+      "grad_norm": 0.0,
+      "learning_rate": 7.57297562577885e-06,
+      "loss": 0.8134,
+      "step": 21508
+    },
+    {
+      "epoch": 0.5907280766801242,
+      "grad_norm": 0.0,
+      "learning_rate": 7.572112715202432e-06,
+      "loss": 0.8626,
+      "step": 21509
+    },
+    {
+      "epoch": 0.5907555409079673,
+      "grad_norm": 0.0,
+      "learning_rate": 7.571249823836396e-06,
+      "loss": 0.8169,
+      "step": 21510
+    },
+    {
+      "epoch": 0.5907830051358106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.570386951687572e-06,
+      "loss": 0.8312,
+      "step": 21511
+    },
+    {
+      "epoch": 0.5908104693636539,
+      "grad_norm": 0.0,
+      "learning_rate": 7.569524098762793e-06,
+      "loss": 0.9213,
+      "step": 21512
+    },
+    {
+      "epoch": 0.590837933591497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.568661265068876e-06,
+      "loss": 0.9164,
+      "step": 21513
+    },
+    {
+      "epoch": 0.5908653978193403,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5677984506126546e-06,
+      "loss": 0.9114,
+      "step": 21514
+    },
+    {
+      "epoch": 0.5908928620471835,
+      "grad_norm": 0.0,
+      "learning_rate": 7.566935655400952e-06,
+      "loss": 0.8582,
+      "step": 21515
+    },
+    {
+      "epoch": 0.5909203262750268,
+      "grad_norm": 0.0,
+      "learning_rate": 7.566072879440597e-06,
+      "loss": 0.9244,
+      "step": 21516
+    },
+    {
+      "epoch": 0.59094779050287,
+      "grad_norm": 0.0,
+      "learning_rate": 7.565210122738419e-06,
+      "loss": 0.9671,
+      "step": 21517
+    },
+    {
+      "epoch": 0.5909752547307132,
+      "grad_norm": 0.0,
+      "learning_rate": 7.564347385301243e-06,
+      "loss": 0.8983,
+      "step": 21518
+    },
+    {
+      "epoch": 0.5910027189585565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5634846671358896e-06,
+      "loss": 0.9391,
+      "step": 21519
+    },
+    {
+      "epoch": 0.5910301831863997,
+      "grad_norm": 0.0,
+      "learning_rate": 7.56262196824919e-06,
+      "loss": 0.7658,
+      "step": 21520
+    },
+    {
+      "epoch": 0.5910576474142429,
+      "grad_norm": 0.0,
+      "learning_rate": 7.561759288647971e-06,
+      "loss": 0.8286,
+      "step": 21521
+    },
+    {
+      "epoch": 0.5910851116420862,
+      "grad_norm": 0.0,
+      "learning_rate": 7.560896628339055e-06,
+      "loss": 0.9361,
+      "step": 21522
+    },
+    {
+      "epoch": 0.5911125758699294,
+      "grad_norm": 0.0,
+      "learning_rate": 7.560033987329269e-06,
+      "loss": 0.8256,
+      "step": 21523
+    },
+    {
+      "epoch": 0.5911400400977727,
+      "grad_norm": 0.0,
+      "learning_rate": 7.559171365625445e-06,
+      "loss": 0.8406,
+      "step": 21524
+    },
+    {
+      "epoch": 0.5911675043256159,
+      "grad_norm": 0.0,
+      "learning_rate": 7.558308763234398e-06,
+      "loss": 0.8152,
+      "step": 21525
+    },
+    {
+      "epoch": 0.5911949685534591,
+      "grad_norm": 0.0,
+      "learning_rate": 7.55744618016296e-06,
+      "loss": 0.7983,
+      "step": 21526
+    },
+    {
+      "epoch": 0.5912224327813024,
+      "grad_norm": 0.0,
+      "learning_rate": 7.556583616417952e-06,
+      "loss": 0.9562,
+      "step": 21527
+    },
+    {
+      "epoch": 0.5912498970091455,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5557210720062e-06,
+      "loss": 0.8568,
+      "step": 21528
+    },
+    {
+      "epoch": 0.5912773612369888,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5548585469345345e-06,
+      "loss": 0.8763,
+      "step": 21529
+    },
+    {
+      "epoch": 0.5913048254648321,
+      "grad_norm": 0.0,
+      "learning_rate": 7.553996041209773e-06,
+      "loss": 0.7453,
+      "step": 21530
+    },
+    {
+      "epoch": 0.5913322896926753,
+      "grad_norm": 0.0,
+      "learning_rate": 7.553133554838741e-06,
+      "loss": 0.8549,
+      "step": 21531
+    },
+    {
+      "epoch": 0.5913597539205185,
+      "grad_norm": 0.0,
+      "learning_rate": 7.552271087828263e-06,
+      "loss": 0.7931,
+      "step": 21532
+    },
+    {
+      "epoch": 0.5913872181483617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5514086401851676e-06,
+      "loss": 0.8348,
+      "step": 21533
+    },
+    {
+      "epoch": 0.591414682376205,
+      "grad_norm": 0.0,
+      "learning_rate": 7.550546211916272e-06,
+      "loss": 0.7502,
+      "step": 21534
+    },
+    {
+      "epoch": 0.5914421466040483,
+      "grad_norm": 0.0,
+      "learning_rate": 7.549683803028409e-06,
+      "loss": 1.0038,
+      "step": 21535
+    },
+    {
+      "epoch": 0.5914696108318914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.548821413528392e-06,
+      "loss": 0.8432,
+      "step": 21536
+    },
+    {
+      "epoch": 0.5914970750597347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.54795904342305e-06,
+      "loss": 0.8274,
+      "step": 21537
+    },
+    {
+      "epoch": 0.591524539287578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.547096692719207e-06,
+      "loss": 0.8382,
+      "step": 21538
+    },
+    {
+      "epoch": 0.5915520035154211,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5462343614236845e-06,
+      "loss": 0.8222,
+      "step": 21539
+    },
+    {
+      "epoch": 0.5915794677432644,
+      "grad_norm": 0.0,
+      "learning_rate": 7.545372049543305e-06,
+      "loss": 0.8115,
+      "step": 21540
+    },
+    {
+      "epoch": 0.5916069319711076,
+      "grad_norm": 0.0,
+      "learning_rate": 7.544509757084898e-06,
+      "loss": 0.8568,
+      "step": 21541
+    },
+    {
+      "epoch": 0.5916343961989509,
+      "grad_norm": 0.0,
+      "learning_rate": 7.543647484055277e-06,
+      "loss": 0.8289,
+      "step": 21542
+    },
+    {
+      "epoch": 0.5916618604267941,
+      "grad_norm": 0.0,
+      "learning_rate": 7.54278523046127e-06,
+      "loss": 0.9802,
+      "step": 21543
+    },
+    {
+      "epoch": 0.5916893246546373,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5419229963096945e-06,
+      "loss": 0.8484,
+      "step": 21544
+    },
+    {
+      "epoch": 0.5917167888824806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.54106078160738e-06,
+      "loss": 0.9372,
+      "step": 21545
+    },
+    {
+      "epoch": 0.5917442531103237,
+      "grad_norm": 0.0,
+      "learning_rate": 7.540198586361143e-06,
+      "loss": 0.7878,
+      "step": 21546
+    },
+    {
+      "epoch": 0.591771717338167,
+      "grad_norm": 0.0,
+      "learning_rate": 7.539336410577811e-06,
+      "loss": 0.7527,
+      "step": 21547
+    },
+    {
+      "epoch": 0.5917991815660103,
+      "grad_norm": 0.0,
+      "learning_rate": 7.538474254264199e-06,
+      "loss": 0.7976,
+      "step": 21548
+    },
+    {
+      "epoch": 0.5918266457938535,
+      "grad_norm": 0.0,
+      "learning_rate": 7.537612117427132e-06,
+      "loss": 0.9505,
+      "step": 21549
+    },
+    {
+      "epoch": 0.5918541100216967,
+      "grad_norm": 0.0,
+      "learning_rate": 7.536750000073434e-06,
+      "loss": 0.8989,
+      "step": 21550
+    },
+    {
+      "epoch": 0.59188157424954,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5358879022099215e-06,
+      "loss": 0.847,
+      "step": 21551
+    },
+    {
+      "epoch": 0.5919090384773832,
+      "grad_norm": 0.0,
+      "learning_rate": 7.53502582384342e-06,
+      "loss": 0.7327,
+      "step": 21552
+    },
+    {
+      "epoch": 0.5919365027052265,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5341637649807516e-06,
+      "loss": 0.8345,
+      "step": 21553
+    },
+    {
+      "epoch": 0.5919639669330696,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5333017256287345e-06,
+      "loss": 0.8203,
+      "step": 21554
+    },
+    {
+      "epoch": 0.5919914311609129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.532439705794185e-06,
+      "loss": 0.9042,
+      "step": 21555
+    },
+    {
+      "epoch": 0.5920188953887562,
+      "grad_norm": 0.0,
+      "learning_rate": 7.53157770548393e-06,
+      "loss": 0.7477,
+      "step": 21556
+    },
+    {
+      "epoch": 0.5920463596165993,
+      "grad_norm": 0.0,
+      "learning_rate": 7.53071572470479e-06,
+      "loss": 0.9894,
+      "step": 21557
+    },
+    {
+      "epoch": 0.5920738238444426,
+      "grad_norm": 0.0,
+      "learning_rate": 7.529853763463582e-06,
+      "loss": 0.822,
+      "step": 21558
+    },
+    {
+      "epoch": 0.5921012880722859,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5289918217671325e-06,
+      "loss": 0.9204,
+      "step": 21559
+    },
+    {
+      "epoch": 0.5921287523001291,
+      "grad_norm": 0.0,
+      "learning_rate": 7.528129899622252e-06,
+      "loss": 0.8363,
+      "step": 21560
+    },
+    {
+      "epoch": 0.5921562165279723,
+      "grad_norm": 0.0,
+      "learning_rate": 7.527267997035764e-06,
+      "loss": 0.8497,
+      "step": 21561
+    },
+    {
+      "epoch": 0.5921836807558155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.526406114014493e-06,
+      "loss": 0.8294,
+      "step": 21562
+    },
+    {
+      "epoch": 0.5922111449836588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5255442505652535e-06,
+      "loss": 0.9096,
+      "step": 21563
+    },
+    {
+      "epoch": 0.5922386092115021,
+      "grad_norm": 0.0,
+      "learning_rate": 7.524682406694865e-06,
+      "loss": 0.8734,
+      "step": 21564
+    },
+    {
+      "epoch": 0.5922660734393452,
+      "grad_norm": 0.0,
+      "learning_rate": 7.523820582410153e-06,
+      "loss": 0.9491,
+      "step": 21565
+    },
+    {
+      "epoch": 0.5922935376671885,
+      "grad_norm": 0.0,
+      "learning_rate": 7.52295877771793e-06,
+      "loss": 0.7925,
+      "step": 21566
+    },
+    {
+      "epoch": 0.5923210018950317,
+      "grad_norm": 0.0,
+      "learning_rate": 7.522096992625014e-06,
+      "loss": 0.8971,
+      "step": 21567
+    },
+    {
+      "epoch": 0.592348466122875,
+      "grad_norm": 0.0,
+      "learning_rate": 7.521235227138226e-06,
+      "loss": 0.9165,
+      "step": 21568
+    },
+    {
+      "epoch": 0.5923759303507182,
+      "grad_norm": 0.0,
+      "learning_rate": 7.520373481264388e-06,
+      "loss": 0.8268,
+      "step": 21569
+    },
+    {
+      "epoch": 0.5924033945785614,
+      "grad_norm": 0.0,
+      "learning_rate": 7.519511755010314e-06,
+      "loss": 0.98,
+      "step": 21570
+    },
+    {
+      "epoch": 0.5924308588064047,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5186500483828265e-06,
+      "loss": 0.796,
+      "step": 21571
+    },
+    {
+      "epoch": 0.592458323034248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.517788361388737e-06,
+      "loss": 0.8503,
+      "step": 21572
+    },
+    {
+      "epoch": 0.5924857872620911,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5169266940348686e-06,
+      "loss": 0.909,
+      "step": 21573
+    },
+    {
+      "epoch": 0.5925132514899344,
+      "grad_norm": 0.0,
+      "learning_rate": 7.516065046328039e-06,
+      "loss": 0.888,
+      "step": 21574
+    },
+    {
+      "epoch": 0.5925407157177776,
+      "grad_norm": 0.0,
+      "learning_rate": 7.515203418275063e-06,
+      "loss": 0.9298,
+      "step": 21575
+    },
+    {
+      "epoch": 0.5925681799456208,
+      "grad_norm": 0.0,
+      "learning_rate": 7.514341809882761e-06,
+      "loss": 0.8601,
+      "step": 21576
+    },
+    {
+      "epoch": 0.5925956441734641,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5134802211579525e-06,
+      "loss": 0.9503,
+      "step": 21577
+    },
+    {
+      "epoch": 0.5926231084013073,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5126186521074506e-06,
+      "loss": 0.8701,
+      "step": 21578
+    },
+    {
+      "epoch": 0.5926505726291506,
+      "grad_norm": 0.0,
+      "learning_rate": 7.511757102738071e-06,
+      "loss": 0.9031,
+      "step": 21579
+    },
+    {
+      "epoch": 0.5926780368569937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.510895573056634e-06,
+      "loss": 0.7894,
+      "step": 21580
+    },
+    {
+      "epoch": 0.592705501084837,
+      "grad_norm": 0.0,
+      "learning_rate": 7.510034063069955e-06,
+      "loss": 0.9397,
+      "step": 21581
+    },
+    {
+      "epoch": 0.5927329653126803,
+      "grad_norm": 0.0,
+      "learning_rate": 7.509172572784851e-06,
+      "loss": 0.9303,
+      "step": 21582
+    },
+    {
+      "epoch": 0.5927604295405234,
+      "grad_norm": 0.0,
+      "learning_rate": 7.508311102208143e-06,
+      "loss": 0.9263,
+      "step": 21583
+    },
+    {
+      "epoch": 0.5927878937683667,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5074496513466365e-06,
+      "loss": 0.8676,
+      "step": 21584
+    },
+    {
+      "epoch": 0.59281535799621,
+      "grad_norm": 0.0,
+      "learning_rate": 7.506588220207155e-06,
+      "loss": 0.9461,
+      "step": 21585
+    },
+    {
+      "epoch": 0.5928428222240532,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5057268087965166e-06,
+      "loss": 0.9414,
+      "step": 21586
+    },
+    {
+      "epoch": 0.5928702864518964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.50486541712153e-06,
+      "loss": 1.002,
+      "step": 21587
+    },
+    {
+      "epoch": 0.5928977506797396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5040040451890194e-06,
+      "loss": 0.8409,
+      "step": 21588
+    },
+    {
+      "epoch": 0.5929252149075829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.503142693005792e-06,
+      "loss": 0.8465,
+      "step": 21589
+    },
+    {
+      "epoch": 0.5929526791354262,
+      "grad_norm": 0.0,
+      "learning_rate": 7.502281360578668e-06,
+      "loss": 0.8542,
+      "step": 21590
+    },
+    {
+      "epoch": 0.5929801433632693,
+      "grad_norm": 0.0,
+      "learning_rate": 7.50142004791446e-06,
+      "loss": 0.9059,
+      "step": 21591
+    },
+    {
+      "epoch": 0.5930076075911126,
+      "grad_norm": 0.0,
+      "learning_rate": 7.500558755019984e-06,
+      "loss": 0.8442,
+      "step": 21592
+    },
+    {
+      "epoch": 0.5930350718189558,
+      "grad_norm": 0.0,
+      "learning_rate": 7.499697481902057e-06,
+      "loss": 0.8236,
+      "step": 21593
+    },
+    {
+      "epoch": 0.593062536046799,
+      "grad_norm": 0.0,
+      "learning_rate": 7.498836228567496e-06,
+      "loss": 0.8749,
+      "step": 21594
+    },
+    {
+      "epoch": 0.5930900002746423,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4979749950231054e-06,
+      "loss": 0.84,
+      "step": 21595
+    },
+    {
+      "epoch": 0.5931174645024855,
+      "grad_norm": 0.0,
+      "learning_rate": 7.497113781275706e-06,
+      "loss": 0.8836,
+      "step": 21596
+    },
+    {
+      "epoch": 0.5931449287303288,
+      "grad_norm": 0.0,
+      "learning_rate": 7.496252587332114e-06,
+      "loss": 0.843,
+      "step": 21597
+    },
+    {
+      "epoch": 0.593172392958172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.495391413199142e-06,
+      "loss": 0.831,
+      "step": 21598
+    },
+    {
+      "epoch": 0.5931998571860152,
+      "grad_norm": 0.0,
+      "learning_rate": 7.494530258883601e-06,
+      "loss": 0.9205,
+      "step": 21599
+    },
+    {
+      "epoch": 0.5932273214138585,
+      "grad_norm": 0.0,
+      "learning_rate": 7.493669124392311e-06,
+      "loss": 0.9108,
+      "step": 21600
+    },
+    {
+      "epoch": 0.5932547856417016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.492808009732079e-06,
+      "loss": 0.9317,
+      "step": 21601
+    },
+    {
+      "epoch": 0.5932822498695449,
+      "grad_norm": 0.0,
+      "learning_rate": 7.491946914909723e-06,
+      "loss": 0.7736,
+      "step": 21602
+    },
+    {
+      "epoch": 0.5933097140973882,
+      "grad_norm": 0.0,
+      "learning_rate": 7.491085839932051e-06,
+      "loss": 0.8746,
+      "step": 21603
+    },
+    {
+      "epoch": 0.5933371783252314,
+      "grad_norm": 0.0,
+      "learning_rate": 7.490224784805881e-06,
+      "loss": 0.8778,
+      "step": 21604
+    },
+    {
+      "epoch": 0.5933646425530746,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4893637495380255e-06,
+      "loss": 0.876,
+      "step": 21605
+    },
+    {
+      "epoch": 0.5933921067809178,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4885027341353e-06,
+      "loss": 0.9247,
+      "step": 21606
+    },
+    {
+      "epoch": 0.5934195710087611,
+      "grad_norm": 0.0,
+      "learning_rate": 7.487641738604509e-06,
+      "loss": 0.7653,
+      "step": 21607
+    },
+    {
+      "epoch": 0.5934470352366044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.48678076295247e-06,
+      "loss": 0.7871,
+      "step": 21608
+    },
+    {
+      "epoch": 0.5934744994644475,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4859198071859976e-06,
+      "loss": 0.8651,
+      "step": 21609
+    },
+    {
+      "epoch": 0.5935019636922908,
+      "grad_norm": 0.0,
+      "learning_rate": 7.485058871311899e-06,
+      "loss": 0.8882,
+      "step": 21610
+    },
+    {
+      "epoch": 0.5935294279201341,
+      "grad_norm": 0.0,
+      "learning_rate": 7.484197955336988e-06,
+      "loss": 0.8619,
+      "step": 21611
+    },
+    {
+      "epoch": 0.5935568921479772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.483337059268083e-06,
+      "loss": 0.865,
+      "step": 21612
+    },
+    {
+      "epoch": 0.5935843563758205,
+      "grad_norm": 0.0,
+      "learning_rate": 7.482476183111984e-06,
+      "loss": 0.8377,
+      "step": 21613
+    },
+    {
+      "epoch": 0.5936118206036637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.481615326875512e-06,
+      "loss": 0.862,
+      "step": 21614
+    },
+    {
+      "epoch": 0.593639284831507,
+      "grad_norm": 0.0,
+      "learning_rate": 7.480754490565473e-06,
+      "loss": 0.8884,
+      "step": 21615
+    },
+    {
+      "epoch": 0.5936667490593502,
+      "grad_norm": 0.0,
+      "learning_rate": 7.479893674188681e-06,
+      "loss": 0.8024,
+      "step": 21616
+    },
+    {
+      "epoch": 0.5936942132871934,
+      "grad_norm": 0.0,
+      "learning_rate": 7.479032877751946e-06,
+      "loss": 0.8582,
+      "step": 21617
+    },
+    {
+      "epoch": 0.5937216775150367,
+      "grad_norm": 0.0,
+      "learning_rate": 7.478172101262085e-06,
+      "loss": 0.8044,
+      "step": 21618
+    },
+    {
+      "epoch": 0.5937491417428798,
+      "grad_norm": 0.0,
+      "learning_rate": 7.477311344725896e-06,
+      "loss": 0.7898,
+      "step": 21619
+    },
+    {
+      "epoch": 0.5937766059707231,
+      "grad_norm": 0.0,
+      "learning_rate": 7.476450608150198e-06,
+      "loss": 0.8818,
+      "step": 21620
+    },
+    {
+      "epoch": 0.5938040701985664,
+      "grad_norm": 0.0,
+      "learning_rate": 7.475589891541802e-06,
+      "loss": 0.7678,
+      "step": 21621
+    },
+    {
+      "epoch": 0.5938315344264096,
+      "grad_norm": 0.0,
+      "learning_rate": 7.474729194907515e-06,
+      "loss": 0.8531,
+      "step": 21622
+    },
+    {
+      "epoch": 0.5938589986542528,
+      "grad_norm": 0.0,
+      "learning_rate": 7.473868518254149e-06,
+      "loss": 0.8849,
+      "step": 21623
+    },
+    {
+      "epoch": 0.5938864628820961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.473007861588517e-06,
+      "loss": 0.8688,
+      "step": 21624
+    },
+    {
+      "epoch": 0.5939139271099393,
+      "grad_norm": 0.0,
+      "learning_rate": 7.472147224917423e-06,
+      "loss": 0.8953,
+      "step": 21625
+    },
+    {
+      "epoch": 0.5939413913377826,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4712866082476805e-06,
+      "loss": 0.8807,
+      "step": 21626
+    },
+    {
+      "epoch": 0.5939688555656257,
+      "grad_norm": 0.0,
+      "learning_rate": 7.470426011586096e-06,
+      "loss": 0.9172,
+      "step": 21627
+    },
+    {
+      "epoch": 0.593996319793469,
+      "grad_norm": 0.0,
+      "learning_rate": 7.469565434939481e-06,
+      "loss": 0.8842,
+      "step": 21628
+    },
+    {
+      "epoch": 0.5940237840213123,
+      "grad_norm": 0.0,
+      "learning_rate": 7.468704878314645e-06,
+      "loss": 0.9143,
+      "step": 21629
+    },
+    {
+      "epoch": 0.5940512482491555,
+      "grad_norm": 0.0,
+      "learning_rate": 7.467844341718401e-06,
+      "loss": 0.8491,
+      "step": 21630
+    },
+    {
+      "epoch": 0.5940787124769987,
+      "grad_norm": 0.0,
+      "learning_rate": 7.466983825157548e-06,
+      "loss": 0.8321,
+      "step": 21631
+    },
+    {
+      "epoch": 0.5941061767048419,
+      "grad_norm": 0.0,
+      "learning_rate": 7.466123328638901e-06,
+      "loss": 0.8179,
+      "step": 21632
+    },
+    {
+      "epoch": 0.5941336409326852,
+      "grad_norm": 0.0,
+      "learning_rate": 7.465262852169268e-06,
+      "loss": 0.8551,
+      "step": 21633
+    },
+    {
+      "epoch": 0.5941611051605284,
+      "grad_norm": 0.0,
+      "learning_rate": 7.464402395755456e-06,
+      "loss": 0.8136,
+      "step": 21634
+    },
+    {
+      "epoch": 0.5941885693883716,
+      "grad_norm": 0.0,
+      "learning_rate": 7.463541959404275e-06,
+      "loss": 0.865,
+      "step": 21635
+    },
+    {
+      "epoch": 0.5942160336162149,
+      "grad_norm": 0.0,
+      "learning_rate": 7.462681543122537e-06,
+      "loss": 0.801,
+      "step": 21636
+    },
+    {
+      "epoch": 0.5942434978440582,
+      "grad_norm": 0.0,
+      "learning_rate": 7.461821146917041e-06,
+      "loss": 0.836,
+      "step": 21637
+    },
+    {
+      "epoch": 0.5942709620719013,
+      "grad_norm": 0.0,
+      "learning_rate": 7.460960770794601e-06,
+      "loss": 0.7894,
+      "step": 21638
+    },
+    {
+      "epoch": 0.5942984262997446,
+      "grad_norm": 0.0,
+      "learning_rate": 7.460100414762021e-06,
+      "loss": 0.8012,
+      "step": 21639
+    },
+    {
+      "epoch": 0.5943258905275878,
+      "grad_norm": 0.0,
+      "learning_rate": 7.459240078826111e-06,
+      "loss": 0.8949,
+      "step": 21640
+    },
+    {
+      "epoch": 0.594353354755431,
+      "grad_norm": 0.0,
+      "learning_rate": 7.458379762993679e-06,
+      "loss": 0.8616,
+      "step": 21641
+    },
+    {
+      "epoch": 0.5943808189832743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.457519467271532e-06,
+      "loss": 0.8663,
+      "step": 21642
+    },
+    {
+      "epoch": 0.5944082832111175,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4566591916664735e-06,
+      "loss": 0.8967,
+      "step": 21643
+    },
+    {
+      "epoch": 0.5944357474389608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.455798936185312e-06,
+      "loss": 0.8909,
+      "step": 21644
+    },
+    {
+      "epoch": 0.5944632116668039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.454938700834856e-06,
+      "loss": 0.9391,
+      "step": 21645
+    },
+    {
+      "epoch": 0.5944906758946472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4540784856219095e-06,
+      "loss": 0.8585,
+      "step": 21646
+    },
+    {
+      "epoch": 0.5945181401224905,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4532182905532854e-06,
+      "loss": 0.8457,
+      "step": 21647
+    },
+    {
+      "epoch": 0.5945456043503337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.452358115635779e-06,
+      "loss": 0.9231,
+      "step": 21648
+    },
+    {
+      "epoch": 0.5945730685781769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.451497960876203e-06,
+      "loss": 0.9063,
+      "step": 21649
+    },
+    {
+      "epoch": 0.5946005328060202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450637826281364e-06,
+      "loss": 0.8855,
+      "step": 21650
+    },
+    {
+      "epoch": 0.5946279970338634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.449777711858064e-06,
+      "loss": 0.7113,
+      "step": 21651
+    },
+    {
+      "epoch": 0.5946554612617067,
+      "grad_norm": 0.0,
+      "learning_rate": 7.44891761761311e-06,
+      "loss": 0.8214,
+      "step": 21652
+    },
+    {
+      "epoch": 0.5946829254895498,
+      "grad_norm": 0.0,
+      "learning_rate": 7.448057543553313e-06,
+      "loss": 0.9732,
+      "step": 21653
+    },
+    {
+      "epoch": 0.5947103897173931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.447197489685472e-06,
+      "loss": 0.8608,
+      "step": 21654
+    },
+    {
+      "epoch": 0.5947378539452364,
+      "grad_norm": 0.0,
+      "learning_rate": 7.446337456016391e-06,
+      "loss": 0.8247,
+      "step": 21655
+    },
+    {
+      "epoch": 0.5947653181730795,
+      "grad_norm": 0.0,
+      "learning_rate": 7.445477442552876e-06,
+      "loss": 0.7974,
+      "step": 21656
+    },
+    {
+      "epoch": 0.5947927824009228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.444617449301734e-06,
+      "loss": 0.8276,
+      "step": 21657
+    },
+    {
+      "epoch": 0.594820246628766,
+      "grad_norm": 0.0,
+      "learning_rate": 7.443757476269768e-06,
+      "loss": 0.877,
+      "step": 21658
+    },
+    {
+      "epoch": 0.5948477108566093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.442897523463788e-06,
+      "loss": 0.8557,
+      "step": 21659
+    },
+    {
+      "epoch": 0.5948751750844525,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4420375908905885e-06,
+      "loss": 0.895,
+      "step": 21660
+    },
+    {
+      "epoch": 0.5949026393122957,
+      "grad_norm": 0.0,
+      "learning_rate": 7.441177678556979e-06,
+      "loss": 0.8194,
+      "step": 21661
+    },
+    {
+      "epoch": 0.594930103540139,
+      "grad_norm": 0.0,
+      "learning_rate": 7.440317786469764e-06,
+      "loss": 0.8747,
+      "step": 21662
+    },
+    {
+      "epoch": 0.5949575677679823,
+      "grad_norm": 0.0,
+      "learning_rate": 7.439457914635744e-06,
+      "loss": 0.8346,
+      "step": 21663
+    },
+    {
+      "epoch": 0.5949850319958254,
+      "grad_norm": 0.0,
+      "learning_rate": 7.438598063061725e-06,
+      "loss": 0.8262,
+      "step": 21664
+    },
+    {
+      "epoch": 0.5950124962236687,
+      "grad_norm": 0.0,
+      "learning_rate": 7.437738231754514e-06,
+      "loss": 0.8405,
+      "step": 21665
+    },
+    {
+      "epoch": 0.5950399604515119,
+      "grad_norm": 0.0,
+      "learning_rate": 7.43687842072091e-06,
+      "loss": 0.8446,
+      "step": 21666
+    },
+    {
+      "epoch": 0.5950674246793551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.436018629967714e-06,
+      "loss": 0.8569,
+      "step": 21667
+    },
+    {
+      "epoch": 0.5950948889071984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.435158859501732e-06,
+      "loss": 0.8905,
+      "step": 21668
+    },
+    {
+      "epoch": 0.5951223531350416,
+      "grad_norm": 0.0,
+      "learning_rate": 7.434299109329769e-06,
+      "loss": 0.8749,
+      "step": 21669
+    },
+    {
+      "epoch": 0.5951498173628849,
+      "grad_norm": 0.0,
+      "learning_rate": 7.433439379458622e-06,
+      "loss": 0.8854,
+      "step": 21670
+    },
+    {
+      "epoch": 0.595177281590728,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4325796698951025e-06,
+      "loss": 0.9041,
+      "step": 21671
+    },
+    {
+      "epoch": 0.5952047458185713,
+      "grad_norm": 0.0,
+      "learning_rate": 7.431719980646002e-06,
+      "loss": 0.7613,
+      "step": 21672
+    },
+    {
+      "epoch": 0.5952322100464146,
+      "grad_norm": 0.0,
+      "learning_rate": 7.430860311718129e-06,
+      "loss": 0.8812,
+      "step": 21673
+    },
+    {
+      "epoch": 0.5952596742742577,
+      "grad_norm": 0.0,
+      "learning_rate": 7.430000663118284e-06,
+      "loss": 0.8798,
+      "step": 21674
+    },
+    {
+      "epoch": 0.595287138502101,
+      "grad_norm": 0.0,
+      "learning_rate": 7.429141034853269e-06,
+      "loss": 0.7902,
+      "step": 21675
+    },
+    {
+      "epoch": 0.5953146027299443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.428281426929886e-06,
+      "loss": 0.8398,
+      "step": 21676
+    },
+    {
+      "epoch": 0.5953420669577875,
+      "grad_norm": 0.0,
+      "learning_rate": 7.42742183935494e-06,
+      "loss": 0.8916,
+      "step": 21677
+    },
+    {
+      "epoch": 0.5953695311856307,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4265622721352285e-06,
+      "loss": 0.85,
+      "step": 21678
+    },
+    {
+      "epoch": 0.5953969954134739,
+      "grad_norm": 0.0,
+      "learning_rate": 7.42570272527755e-06,
+      "loss": 0.8246,
+      "step": 21679
+    },
+    {
+      "epoch": 0.5954244596413172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.424843198788708e-06,
+      "loss": 0.7031,
+      "step": 21680
+    },
+    {
+      "epoch": 0.5954519238691605,
+      "grad_norm": 0.0,
+      "learning_rate": 7.423983692675507e-06,
+      "loss": 0.8846,
+      "step": 21681
+    },
+    {
+      "epoch": 0.5954793880970036,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4231242069447425e-06,
+      "loss": 0.9056,
+      "step": 21682
+    },
+    {
+      "epoch": 0.5955068523248469,
+      "grad_norm": 0.0,
+      "learning_rate": 7.422264741603222e-06,
+      "loss": 0.8077,
+      "step": 21683
+    },
+    {
+      "epoch": 0.5955343165526901,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4214052966577375e-06,
+      "loss": 0.8113,
+      "step": 21684
+    },
+    {
+      "epoch": 0.5955617807805333,
+      "grad_norm": 0.0,
+      "learning_rate": 7.420545872115092e-06,
+      "loss": 0.8981,
+      "step": 21685
+    },
+    {
+      "epoch": 0.5955892450083766,
+      "grad_norm": 0.0,
+      "learning_rate": 7.41968646798209e-06,
+      "loss": 0.8829,
+      "step": 21686
+    },
+    {
+      "epoch": 0.5956167092362198,
+      "grad_norm": 0.0,
+      "learning_rate": 7.418827084265525e-06,
+      "loss": 0.8904,
+      "step": 21687
+    },
+    {
+      "epoch": 0.5956441734640631,
+      "grad_norm": 0.0,
+      "learning_rate": 7.417967720972201e-06,
+      "loss": 0.8208,
+      "step": 21688
+    },
+    {
+      "epoch": 0.5956716376919063,
+      "grad_norm": 0.0,
+      "learning_rate": 7.41710837810892e-06,
+      "loss": 0.8788,
+      "step": 21689
+    },
+    {
+      "epoch": 0.5956991019197495,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4162490556824765e-06,
+      "loss": 0.8839,
+      "step": 21690
+    },
+    {
+      "epoch": 0.5957265661475928,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4153897536996685e-06,
+      "loss": 0.9352,
+      "step": 21691
+    },
+    {
+      "epoch": 0.595754030375436,
+      "grad_norm": 0.0,
+      "learning_rate": 7.414530472167299e-06,
+      "loss": 0.8483,
+      "step": 21692
+    },
+    {
+      "epoch": 0.5957814946032792,
+      "grad_norm": 0.0,
+      "learning_rate": 7.413671211092167e-06,
+      "loss": 0.8414,
+      "step": 21693
+    },
+    {
+      "epoch": 0.5958089588311225,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4128119704810686e-06,
+      "loss": 0.8598,
+      "step": 21694
+    },
+    {
+      "epoch": 0.5958364230589657,
+      "grad_norm": 0.0,
+      "learning_rate": 7.411952750340808e-06,
+      "loss": 0.7682,
+      "step": 21695
+    },
+    {
+      "epoch": 0.595863887286809,
+      "grad_norm": 0.0,
+      "learning_rate": 7.411093550678177e-06,
+      "loss": 0.7401,
+      "step": 21696
+    },
+    {
+      "epoch": 0.5958913515146521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.410234371499975e-06,
+      "loss": 0.9384,
+      "step": 21697
+    },
+    {
+      "epoch": 0.5959188157424954,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4093752128130034e-06,
+      "loss": 0.867,
+      "step": 21698
+    },
+    {
+      "epoch": 0.5959462799703387,
+      "grad_norm": 0.0,
+      "learning_rate": 7.408516074624058e-06,
+      "loss": 0.9786,
+      "step": 21699
+    },
+    {
+      "epoch": 0.5959737441981818,
+      "grad_norm": 0.0,
+      "learning_rate": 7.407656956939937e-06,
+      "loss": 0.8096,
+      "step": 21700
+    },
+    {
+      "epoch": 0.5960012084260251,
+      "grad_norm": 0.0,
+      "learning_rate": 7.406797859767443e-06,
+      "loss": 0.8147,
+      "step": 21701
+    },
+    {
+      "epoch": 0.5960286726538684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4059387831133655e-06,
+      "loss": 0.8049,
+      "step": 21702
+    },
+    {
+      "epoch": 0.5960561368817116,
+      "grad_norm": 0.0,
+      "learning_rate": 7.405079726984503e-06,
+      "loss": 0.9214,
+      "step": 21703
+    },
+    {
+      "epoch": 0.5960836011095548,
+      "grad_norm": 0.0,
+      "learning_rate": 7.404220691387656e-06,
+      "loss": 0.8555,
+      "step": 21704
+    },
+    {
+      "epoch": 0.596111065337398,
+      "grad_norm": 0.0,
+      "learning_rate": 7.403361676329622e-06,
+      "loss": 0.8372,
+      "step": 21705
+    },
+    {
+      "epoch": 0.5961385295652413,
+      "grad_norm": 0.0,
+      "learning_rate": 7.402502681817196e-06,
+      "loss": 0.8408,
+      "step": 21706
+    },
+    {
+      "epoch": 0.5961659937930845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.401643707857176e-06,
+      "loss": 0.8689,
+      "step": 21707
+    },
+    {
+      "epoch": 0.5961934580209277,
+      "grad_norm": 0.0,
+      "learning_rate": 7.400784754456354e-06,
+      "loss": 0.791,
+      "step": 21708
+    },
+    {
+      "epoch": 0.596220922248771,
+      "grad_norm": 0.0,
+      "learning_rate": 7.399925821621531e-06,
+      "loss": 0.8689,
+      "step": 21709
+    },
+    {
+      "epoch": 0.5962483864766142,
+      "grad_norm": 0.0,
+      "learning_rate": 7.399066909359502e-06,
+      "loss": 0.9154,
+      "step": 21710
+    },
+    {
+      "epoch": 0.5962758507044574,
+      "grad_norm": 0.0,
+      "learning_rate": 7.398208017677064e-06,
+      "loss": 0.8642,
+      "step": 21711
+    },
+    {
+      "epoch": 0.5963033149323007,
+      "grad_norm": 0.0,
+      "learning_rate": 7.397349146581014e-06,
+      "loss": 0.8628,
+      "step": 21712
+    },
+    {
+      "epoch": 0.5963307791601439,
+      "grad_norm": 0.0,
+      "learning_rate": 7.396490296078143e-06,
+      "loss": 0.8701,
+      "step": 21713
+    },
+    {
+      "epoch": 0.5963582433879872,
+      "grad_norm": 0.0,
+      "learning_rate": 7.395631466175251e-06,
+      "loss": 0.8677,
+      "step": 21714
+    },
+    {
+      "epoch": 0.5963857076158304,
+      "grad_norm": 0.0,
+      "learning_rate": 7.39477265687913e-06,
+      "loss": 0.882,
+      "step": 21715
+    },
+    {
+      "epoch": 0.5964131718436736,
+      "grad_norm": 0.0,
+      "learning_rate": 7.393913868196576e-06,
+      "loss": 0.856,
+      "step": 21716
+    },
+    {
+      "epoch": 0.5964406360715169,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3930551001343875e-06,
+      "loss": 0.9504,
+      "step": 21717
+    },
+    {
+      "epoch": 0.59646810029936,
+      "grad_norm": 0.0,
+      "learning_rate": 7.392196352699359e-06,
+      "loss": 0.8862,
+      "step": 21718
+    },
+    {
+      "epoch": 0.5964955645272033,
+      "grad_norm": 0.0,
+      "learning_rate": 7.391337625898278e-06,
+      "loss": 0.9205,
+      "step": 21719
+    },
+    {
+      "epoch": 0.5965230287550466,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3904789197379445e-06,
+      "loss": 0.911,
+      "step": 21720
+    },
+    {
+      "epoch": 0.5965504929828898,
+      "grad_norm": 0.0,
+      "learning_rate": 7.389620234225153e-06,
+      "loss": 0.8414,
+      "step": 21721
+    },
+    {
+      "epoch": 0.596577957210733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3887615693667e-06,
+      "loss": 0.7775,
+      "step": 21722
+    },
+    {
+      "epoch": 0.5966054214385762,
+      "grad_norm": 0.0,
+      "learning_rate": 7.387902925169375e-06,
+      "loss": 0.8439,
+      "step": 21723
+    },
+    {
+      "epoch": 0.5966328856664195,
+      "grad_norm": 0.0,
+      "learning_rate": 7.387044301639976e-06,
+      "loss": 0.8696,
+      "step": 21724
+    },
+    {
+      "epoch": 0.5966603498942628,
+      "grad_norm": 0.0,
+      "learning_rate": 7.386185698785293e-06,
+      "loss": 0.8449,
+      "step": 21725
+    },
+    {
+      "epoch": 0.5966878141221059,
+      "grad_norm": 0.0,
+      "learning_rate": 7.385327116612122e-06,
+      "loss": 0.8242,
+      "step": 21726
+    },
+    {
+      "epoch": 0.5967152783499492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.384468555127253e-06,
+      "loss": 0.8413,
+      "step": 21727
+    },
+    {
+      "epoch": 0.5967427425777925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.383610014337483e-06,
+      "loss": 0.7758,
+      "step": 21728
+    },
+    {
+      "epoch": 0.5967702068056356,
+      "grad_norm": 0.0,
+      "learning_rate": 7.382751494249607e-06,
+      "loss": 0.882,
+      "step": 21729
+    },
+    {
+      "epoch": 0.5967976710334789,
+      "grad_norm": 0.0,
+      "learning_rate": 7.381892994870416e-06,
+      "loss": 0.9027,
+      "step": 21730
+    },
+    {
+      "epoch": 0.5968251352613221,
+      "grad_norm": 0.0,
+      "learning_rate": 7.381034516206698e-06,
+      "loss": 0.805,
+      "step": 21731
+    },
+    {
+      "epoch": 0.5968525994891654,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3801760582652495e-06,
+      "loss": 0.7994,
+      "step": 21732
+    },
+    {
+      "epoch": 0.5968800637170086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.379317621052862e-06,
+      "loss": 0.7791,
+      "step": 21733
+    },
+    {
+      "epoch": 0.5969075279448518,
+      "grad_norm": 0.0,
+      "learning_rate": 7.37845920457633e-06,
+      "loss": 0.9189,
+      "step": 21734
+    },
+    {
+      "epoch": 0.5969349921726951,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3776008088424444e-06,
+      "loss": 0.9236,
+      "step": 21735
+    },
+    {
+      "epoch": 0.5969624564005382,
+      "grad_norm": 0.0,
+      "learning_rate": 7.376742433858e-06,
+      "loss": 0.9192,
+      "step": 21736
+    },
+    {
+      "epoch": 0.5969899206283815,
+      "grad_norm": 0.0,
+      "learning_rate": 7.37588407962978e-06,
+      "loss": 0.7687,
+      "step": 21737
+    },
+    {
+      "epoch": 0.5970173848562248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.375025746164586e-06,
+      "loss": 0.8554,
+      "step": 21738
+    },
+    {
+      "epoch": 0.597044849084068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3741674334692024e-06,
+      "loss": 0.8559,
+      "step": 21739
+    },
+    {
+      "epoch": 0.5970723133119112,
+      "grad_norm": 0.0,
+      "learning_rate": 7.373309141550423e-06,
+      "loss": 0.86,
+      "step": 21740
+    },
+    {
+      "epoch": 0.5970997775397545,
+      "grad_norm": 0.0,
+      "learning_rate": 7.372450870415042e-06,
+      "loss": 0.9033,
+      "step": 21741
+    },
+    {
+      "epoch": 0.5971272417675977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3715926200698495e-06,
+      "loss": 0.8429,
+      "step": 21742
+    },
+    {
+      "epoch": 0.597154705995441,
+      "grad_norm": 0.0,
+      "learning_rate": 7.370734390521629e-06,
+      "loss": 0.9072,
+      "step": 21743
+    },
+    {
+      "epoch": 0.5971821702232841,
+      "grad_norm": 0.0,
+      "learning_rate": 7.369876181777177e-06,
+      "loss": 0.8525,
+      "step": 21744
+    },
+    {
+      "epoch": 0.5972096344511274,
+      "grad_norm": 0.0,
+      "learning_rate": 7.369017993843284e-06,
+      "loss": 0.8673,
+      "step": 21745
+    },
+    {
+      "epoch": 0.5972370986789707,
+      "grad_norm": 0.0,
+      "learning_rate": 7.368159826726742e-06,
+      "loss": 0.9029,
+      "step": 21746
+    },
+    {
+      "epoch": 0.5972645629068138,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3673016804343355e-06,
+      "loss": 0.8153,
+      "step": 21747
+    },
+    {
+      "epoch": 0.5972920271346571,
+      "grad_norm": 0.0,
+      "learning_rate": 7.366443554972863e-06,
+      "loss": 0.8615,
+      "step": 21748
+    },
+    {
+      "epoch": 0.5973194913625004,
+      "grad_norm": 0.0,
+      "learning_rate": 7.365585450349104e-06,
+      "loss": 0.7321,
+      "step": 21749
+    },
+    {
+      "epoch": 0.5973469555903436,
+      "grad_norm": 0.0,
+      "learning_rate": 7.364727366569857e-06,
+      "loss": 0.9427,
+      "step": 21750
+    },
+    {
+      "epoch": 0.5973744198181868,
+      "grad_norm": 0.0,
+      "learning_rate": 7.363869303641905e-06,
+      "loss": 0.8548,
+      "step": 21751
+    },
+    {
+      "epoch": 0.59740188404603,
+      "grad_norm": 0.0,
+      "learning_rate": 7.363011261572041e-06,
+      "loss": 0.8329,
+      "step": 21752
+    },
+    {
+      "epoch": 0.5974293482738733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.362153240367055e-06,
+      "loss": 0.9242,
+      "step": 21753
+    },
+    {
+      "epoch": 0.5974568125017166,
+      "grad_norm": 0.0,
+      "learning_rate": 7.361295240033737e-06,
+      "loss": 0.8586,
+      "step": 21754
+    },
+    {
+      "epoch": 0.5974842767295597,
+      "grad_norm": 0.0,
+      "learning_rate": 7.360437260578867e-06,
+      "loss": 0.9122,
+      "step": 21755
+    },
+    {
+      "epoch": 0.597511740957403,
+      "grad_norm": 0.0,
+      "learning_rate": 7.359579302009241e-06,
+      "loss": 0.9046,
+      "step": 21756
+    },
+    {
+      "epoch": 0.5975392051852462,
+      "grad_norm": 0.0,
+      "learning_rate": 7.358721364331649e-06,
+      "loss": 0.8266,
+      "step": 21757
+    },
+    {
+      "epoch": 0.5975666694130894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.357863447552874e-06,
+      "loss": 0.9207,
+      "step": 21758
+    },
+    {
+      "epoch": 0.5975941336409327,
+      "grad_norm": 0.0,
+      "learning_rate": 7.357005551679708e-06,
+      "loss": 0.8136,
+      "step": 21759
+    },
+    {
+      "epoch": 0.5976215978687759,
+      "grad_norm": 0.0,
+      "learning_rate": 7.356147676718941e-06,
+      "loss": 0.9436,
+      "step": 21760
+    },
+    {
+      "epoch": 0.5976490620966192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.355289822677354e-06,
+      "loss": 0.9581,
+      "step": 21761
+    },
+    {
+      "epoch": 0.5976765263244624,
+      "grad_norm": 0.0,
+      "learning_rate": 7.35443198956174e-06,
+      "loss": 0.7716,
+      "step": 21762
+    },
+    {
+      "epoch": 0.5977039905523056,
+      "grad_norm": 0.0,
+      "learning_rate": 7.353574177378883e-06,
+      "loss": 0.7646,
+      "step": 21763
+    },
+    {
+      "epoch": 0.5977314547801489,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3527163861355726e-06,
+      "loss": 0.8561,
+      "step": 21764
+    },
+    {
+      "epoch": 0.597758919007992,
+      "grad_norm": 0.0,
+      "learning_rate": 7.351858615838597e-06,
+      "loss": 0.8259,
+      "step": 21765
+    },
+    {
+      "epoch": 0.5977863832358353,
+      "grad_norm": 0.0,
+      "learning_rate": 7.351000866494744e-06,
+      "loss": 0.7806,
+      "step": 21766
+    },
+    {
+      "epoch": 0.5978138474636786,
+      "grad_norm": 0.0,
+      "learning_rate": 7.350143138110795e-06,
+      "loss": 0.7929,
+      "step": 21767
+    },
+    {
+      "epoch": 0.5978413116915218,
+      "grad_norm": 0.0,
+      "learning_rate": 7.349285430693538e-06,
+      "loss": 0.8672,
+      "step": 21768
+    },
+    {
+      "epoch": 0.597868775919365,
+      "grad_norm": 0.0,
+      "learning_rate": 7.348427744249765e-06,
+      "loss": 0.944,
+      "step": 21769
+    },
+    {
+      "epoch": 0.5978962401472082,
+      "grad_norm": 0.0,
+      "learning_rate": 7.347570078786257e-06,
+      "loss": 0.8549,
+      "step": 21770
+    },
+    {
+      "epoch": 0.5979237043750515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.346712434309806e-06,
+      "loss": 0.8944,
+      "step": 21771
+    },
+    {
+      "epoch": 0.5979511686028948,
+      "grad_norm": 0.0,
+      "learning_rate": 7.345854810827189e-06,
+      "loss": 0.8844,
+      "step": 21772
+    },
+    {
+      "epoch": 0.5979786328307379,
+      "grad_norm": 0.0,
+      "learning_rate": 7.344997208345198e-06,
+      "loss": 0.9098,
+      "step": 21773
+    },
+    {
+      "epoch": 0.5980060970585812,
+      "grad_norm": 0.0,
+      "learning_rate": 7.344139626870619e-06,
+      "loss": 0.9911,
+      "step": 21774
+    },
+    {
+      "epoch": 0.5980335612864245,
+      "grad_norm": 0.0,
+      "learning_rate": 7.343282066410234e-06,
+      "loss": 0.8192,
+      "step": 21775
+    },
+    {
+      "epoch": 0.5980610255142677,
+      "grad_norm": 0.0,
+      "learning_rate": 7.342424526970831e-06,
+      "loss": 0.9282,
+      "step": 21776
+    },
+    {
+      "epoch": 0.5980884897421109,
+      "grad_norm": 0.0,
+      "learning_rate": 7.341567008559198e-06,
+      "loss": 0.8811,
+      "step": 21777
+    },
+    {
+      "epoch": 0.5981159539699541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.340709511182114e-06,
+      "loss": 0.8497,
+      "step": 21778
+    },
+    {
+      "epoch": 0.5981434181977974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3398520348463656e-06,
+      "loss": 0.8997,
+      "step": 21779
+    },
+    {
+      "epoch": 0.5981708824256406,
+      "grad_norm": 0.0,
+      "learning_rate": 7.338994579558738e-06,
+      "loss": 0.7821,
+      "step": 21780
+    },
+    {
+      "epoch": 0.5981983466534838,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3381371453260175e-06,
+      "loss": 0.8666,
+      "step": 21781
+    },
+    {
+      "epoch": 0.5982258108813271,
+      "grad_norm": 0.0,
+      "learning_rate": 7.337279732154985e-06,
+      "loss": 0.8679,
+      "step": 21782
+    },
+    {
+      "epoch": 0.5982532751091703,
+      "grad_norm": 0.0,
+      "learning_rate": 7.336422340052431e-06,
+      "loss": 0.9321,
+      "step": 21783
+    },
+    {
+      "epoch": 0.5982807393370135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3355649690251315e-06,
+      "loss": 0.8272,
+      "step": 21784
+    },
+    {
+      "epoch": 0.5983082035648568,
+      "grad_norm": 0.0,
+      "learning_rate": 7.334707619079875e-06,
+      "loss": 0.8466,
+      "step": 21785
+    },
+    {
+      "epoch": 0.5983356677927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.333850290223444e-06,
+      "loss": 0.8834,
+      "step": 21786
+    },
+    {
+      "epoch": 0.5983631320205433,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3329929824626215e-06,
+      "loss": 0.8543,
+      "step": 21787
+    },
+    {
+      "epoch": 0.5983905962483865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.332135695804193e-06,
+      "loss": 0.7905,
+      "step": 21788
+    },
+    {
+      "epoch": 0.5984180604762297,
+      "grad_norm": 0.0,
+      "learning_rate": 7.331278430254942e-06,
+      "loss": 0.8528,
+      "step": 21789
+    },
+    {
+      "epoch": 0.598445524704073,
+      "grad_norm": 0.0,
+      "learning_rate": 7.330421185821649e-06,
+      "loss": 0.904,
+      "step": 21790
+    },
+    {
+      "epoch": 0.5984729889319161,
+      "grad_norm": 0.0,
+      "learning_rate": 7.329563962511095e-06,
+      "loss": 0.7377,
+      "step": 21791
+    },
+    {
+      "epoch": 0.5985004531597594,
+      "grad_norm": 0.0,
+      "learning_rate": 7.328706760330066e-06,
+      "loss": 0.7709,
+      "step": 21792
+    },
+    {
+      "epoch": 0.5985279173876027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.327849579285346e-06,
+      "loss": 0.835,
+      "step": 21793
+    },
+    {
+      "epoch": 0.5985553816154459,
+      "grad_norm": 0.0,
+      "learning_rate": 7.326992419383712e-06,
+      "loss": 0.9183,
+      "step": 21794
+    },
+    {
+      "epoch": 0.5985828458432891,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3261352806319545e-06,
+      "loss": 0.997,
+      "step": 21795
+    },
+    {
+      "epoch": 0.5986103100711323,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3252781630368464e-06,
+      "loss": 0.7928,
+      "step": 21796
+    },
+    {
+      "epoch": 0.5986377742989756,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3244210666051725e-06,
+      "loss": 0.775,
+      "step": 21797
+    },
+    {
+      "epoch": 0.5986652385268189,
+      "grad_norm": 0.0,
+      "learning_rate": 7.323563991343718e-06,
+      "loss": 0.9007,
+      "step": 21798
+    },
+    {
+      "epoch": 0.598692702754662,
+      "grad_norm": 0.0,
+      "learning_rate": 7.322706937259261e-06,
+      "loss": 0.6923,
+      "step": 21799
+    },
+    {
+      "epoch": 0.5987201669825053,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3218499043585835e-06,
+      "loss": 0.8669,
+      "step": 21800
+    },
+    {
+      "epoch": 0.5987476312103486,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3209928926484706e-06,
+      "loss": 0.7985,
+      "step": 21801
+    },
+    {
+      "epoch": 0.5987750954381917,
+      "grad_norm": 0.0,
+      "learning_rate": 7.320135902135698e-06,
+      "loss": 0.9454,
+      "step": 21802
+    },
+    {
+      "epoch": 0.598802559666035,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319278932827045e-06,
+      "loss": 0.8896,
+      "step": 21803
+    },
+    {
+      "epoch": 0.5988300238938782,
+      "grad_norm": 0.0,
+      "learning_rate": 7.318421984729296e-06,
+      "loss": 0.874,
+      "step": 21804
+    },
+    {
+      "epoch": 0.5988574881217215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.317565057849234e-06,
+      "loss": 0.8998,
+      "step": 21805
+    },
+    {
+      "epoch": 0.5988849523495647,
+      "grad_norm": 0.0,
+      "learning_rate": 7.316708152193633e-06,
+      "loss": 0.8723,
+      "step": 21806
+    },
+    {
+      "epoch": 0.5989124165774079,
+      "grad_norm": 0.0,
+      "learning_rate": 7.315851267769283e-06,
+      "loss": 0.9407,
+      "step": 21807
+    },
+    {
+      "epoch": 0.5989398808052512,
+      "grad_norm": 0.0,
+      "learning_rate": 7.314994404582952e-06,
+      "loss": 0.8475,
+      "step": 21808
+    },
+    {
+      "epoch": 0.5989673450330943,
+      "grad_norm": 0.0,
+      "learning_rate": 7.314137562641424e-06,
+      "loss": 0.8176,
+      "step": 21809
+    },
+    {
+      "epoch": 0.5989948092609376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3132807419514825e-06,
+      "loss": 0.8577,
+      "step": 21810
+    },
+    {
+      "epoch": 0.5990222734887809,
+      "grad_norm": 0.0,
+      "learning_rate": 7.312423942519904e-06,
+      "loss": 0.8202,
+      "step": 21811
+    },
+    {
+      "epoch": 0.5990497377166241,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3115671643534655e-06,
+      "loss": 0.8517,
+      "step": 21812
+    },
+    {
+      "epoch": 0.5990772019444673,
+      "grad_norm": 0.0,
+      "learning_rate": 7.310710407458957e-06,
+      "loss": 0.8929,
+      "step": 21813
+    },
+    {
+      "epoch": 0.5991046661723106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3098536718431445e-06,
+      "loss": 0.886,
+      "step": 21814
+    },
+    {
+      "epoch": 0.5991321304001538,
+      "grad_norm": 0.0,
+      "learning_rate": 7.30899695751281e-06,
+      "loss": 0.8609,
+      "step": 21815
+    },
+    {
+      "epoch": 0.5991595946279971,
+      "grad_norm": 0.0,
+      "learning_rate": 7.308140264474735e-06,
+      "loss": 0.9211,
+      "step": 21816
+    },
+    {
+      "epoch": 0.5991870588558402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.307283592735699e-06,
+      "loss": 0.8067,
+      "step": 21817
+    },
+    {
+      "epoch": 0.5992145230836835,
+      "grad_norm": 0.0,
+      "learning_rate": 7.306426942302476e-06,
+      "loss": 0.8092,
+      "step": 21818
+    },
+    {
+      "epoch": 0.5992419873115268,
+      "grad_norm": 0.0,
+      "learning_rate": 7.305570313181851e-06,
+      "loss": 0.9884,
+      "step": 21819
+    },
+    {
+      "epoch": 0.59926945153937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.304713705380593e-06,
+      "loss": 0.9303,
+      "step": 21820
+    },
+    {
+      "epoch": 0.5992969157672132,
+      "grad_norm": 0.0,
+      "learning_rate": 7.303857118905483e-06,
+      "loss": 0.7657,
+      "step": 21821
+    },
+    {
+      "epoch": 0.5993243799950564,
+      "grad_norm": 0.0,
+      "learning_rate": 7.303000553763303e-06,
+      "loss": 0.7873,
+      "step": 21822
+    },
+    {
+      "epoch": 0.5993518442228997,
+      "grad_norm": 0.0,
+      "learning_rate": 7.302144009960826e-06,
+      "loss": 0.8289,
+      "step": 21823
+    },
+    {
+      "epoch": 0.5993793084507429,
+      "grad_norm": 0.0,
+      "learning_rate": 7.301287487504829e-06,
+      "loss": 0.8157,
+      "step": 21824
+    },
+    {
+      "epoch": 0.5994067726785861,
+      "grad_norm": 0.0,
+      "learning_rate": 7.300430986402096e-06,
+      "loss": 0.888,
+      "step": 21825
+    },
+    {
+      "epoch": 0.5994342369064294,
+      "grad_norm": 0.0,
+      "learning_rate": 7.299574506659396e-06,
+      "loss": 0.9187,
+      "step": 21826
+    },
+    {
+      "epoch": 0.5994617011342727,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2987180482835065e-06,
+      "loss": 0.8829,
+      "step": 21827
+    },
+    {
+      "epoch": 0.5994891653621158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.297861611281206e-06,
+      "loss": 0.8423,
+      "step": 21828
+    },
+    {
+      "epoch": 0.5995166295899591,
+      "grad_norm": 0.0,
+      "learning_rate": 7.297005195659275e-06,
+      "loss": 0.9677,
+      "step": 21829
+    },
+    {
+      "epoch": 0.5995440938178023,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2961488014244844e-06,
+      "loss": 0.9272,
+      "step": 21830
+    },
+    {
+      "epoch": 0.5995715580456455,
+      "grad_norm": 0.0,
+      "learning_rate": 7.295292428583611e-06,
+      "loss": 0.8473,
+      "step": 21831
+    },
+    {
+      "epoch": 0.5995990222734888,
+      "grad_norm": 0.0,
+      "learning_rate": 7.294436077143432e-06,
+      "loss": 0.7968,
+      "step": 21832
+    },
+    {
+      "epoch": 0.599626486501332,
+      "grad_norm": 0.0,
+      "learning_rate": 7.29357974711072e-06,
+      "loss": 0.7524,
+      "step": 21833
+    },
+    {
+      "epoch": 0.5996539507291753,
+      "grad_norm": 0.0,
+      "learning_rate": 7.292723438492258e-06,
+      "loss": 0.888,
+      "step": 21834
+    },
+    {
+      "epoch": 0.5996814149570184,
+      "grad_norm": 0.0,
+      "learning_rate": 7.291867151294814e-06,
+      "loss": 0.8674,
+      "step": 21835
+    },
+    {
+      "epoch": 0.5997088791848617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2910108855251695e-06,
+      "loss": 0.8831,
+      "step": 21836
+    },
+    {
+      "epoch": 0.599736343412705,
+      "grad_norm": 0.0,
+      "learning_rate": 7.290154641190092e-06,
+      "loss": 0.7815,
+      "step": 21837
+    },
+    {
+      "epoch": 0.5997638076405482,
+      "grad_norm": 0.0,
+      "learning_rate": 7.289298418296363e-06,
+      "loss": 0.7893,
+      "step": 21838
+    },
+    {
+      "epoch": 0.5997912718683914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.288442216850753e-06,
+      "loss": 0.8593,
+      "step": 21839
+    },
+    {
+      "epoch": 0.5998187360962347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.287586036860039e-06,
+      "loss": 0.8663,
+      "step": 21840
+    },
+    {
+      "epoch": 0.5998462003240779,
+      "grad_norm": 0.0,
+      "learning_rate": 7.286729878330997e-06,
+      "loss": 0.8898,
+      "step": 21841
+    },
+    {
+      "epoch": 0.5998736645519211,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2858737412703995e-06,
+      "loss": 0.9904,
+      "step": 21842
+    },
+    {
+      "epoch": 0.5999011287797643,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2850176256850205e-06,
+      "loss": 0.9042,
+      "step": 21843
+    },
+    {
+      "epoch": 0.5999285930076076,
+      "grad_norm": 0.0,
+      "learning_rate": 7.28416153158163e-06,
+      "loss": 0.9658,
+      "step": 21844
+    },
+    {
+      "epoch": 0.5999560572354509,
+      "grad_norm": 0.0,
+      "learning_rate": 7.283305458967006e-06,
+      "loss": 0.881,
+      "step": 21845
+    },
+    {
+      "epoch": 0.599983521463294,
+      "grad_norm": 0.0,
+      "learning_rate": 7.282449407847925e-06,
+      "loss": 0.9059,
+      "step": 21846
+    },
+    {
+      "epoch": 0.6000109856911373,
+      "grad_norm": 0.0,
+      "learning_rate": 7.281593378231153e-06,
+      "loss": 0.9205,
+      "step": 21847
+    },
+    {
+      "epoch": 0.6000384499189805,
+      "grad_norm": 0.0,
+      "learning_rate": 7.280737370123471e-06,
+      "loss": 0.8221,
+      "step": 21848
+    },
+    {
+      "epoch": 0.6000659141468238,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2798813835316465e-06,
+      "loss": 0.8174,
+      "step": 21849
+    },
+    {
+      "epoch": 0.600093378374667,
+      "grad_norm": 0.0,
+      "learning_rate": 7.279025418462455e-06,
+      "loss": 0.8411,
+      "step": 21850
+    },
+    {
+      "epoch": 0.6001208426025102,
+      "grad_norm": 0.0,
+      "learning_rate": 7.278169474922667e-06,
+      "loss": 0.8712,
+      "step": 21851
+    },
+    {
+      "epoch": 0.6001483068303535,
+      "grad_norm": 0.0,
+      "learning_rate": 7.277313552919057e-06,
+      "loss": 0.8729,
+      "step": 21852
+    },
+    {
+      "epoch": 0.6001757710581967,
+      "grad_norm": 0.0,
+      "learning_rate": 7.276457652458397e-06,
+      "loss": 0.843,
+      "step": 21853
+    },
+    {
+      "epoch": 0.6002032352860399,
+      "grad_norm": 0.0,
+      "learning_rate": 7.275601773547459e-06,
+      "loss": 0.8123,
+      "step": 21854
+    },
+    {
+      "epoch": 0.6002306995138832,
+      "grad_norm": 0.0,
+      "learning_rate": 7.274745916193017e-06,
+      "loss": 0.7864,
+      "step": 21855
+    },
+    {
+      "epoch": 0.6002581637417264,
+      "grad_norm": 0.0,
+      "learning_rate": 7.273890080401838e-06,
+      "loss": 0.94,
+      "step": 21856
+    },
+    {
+      "epoch": 0.6002856279695696,
+      "grad_norm": 0.0,
+      "learning_rate": 7.273034266180696e-06,
+      "loss": 0.8026,
+      "step": 21857
+    },
+    {
+      "epoch": 0.6003130921974129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272178473536366e-06,
+      "loss": 0.8522,
+      "step": 21858
+    },
+    {
+      "epoch": 0.6003405564252561,
+      "grad_norm": 0.0,
+      "learning_rate": 7.271322702475614e-06,
+      "loss": 0.84,
+      "step": 21859
+    },
+    {
+      "epoch": 0.6003680206530994,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2704669530052185e-06,
+      "loss": 0.9015,
+      "step": 21860
+    },
+    {
+      "epoch": 0.6003954848809425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.269611225131941e-06,
+      "loss": 0.8276,
+      "step": 21861
+    },
+    {
+      "epoch": 0.6004229491087858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.26875551886256e-06,
+      "loss": 0.8821,
+      "step": 21862
+    },
+    {
+      "epoch": 0.6004504133366291,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2678998342038405e-06,
+      "loss": 0.924,
+      "step": 21863
+    },
+    {
+      "epoch": 0.6004778775644722,
+      "grad_norm": 0.0,
+      "learning_rate": 7.267044171162556e-06,
+      "loss": 0.9442,
+      "step": 21864
+    },
+    {
+      "epoch": 0.6005053417923155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.266188529745479e-06,
+      "loss": 0.8351,
+      "step": 21865
+    },
+    {
+      "epoch": 0.6005328060201588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.265332909959377e-06,
+      "loss": 0.8574,
+      "step": 21866
+    },
+    {
+      "epoch": 0.600560270248002,
+      "grad_norm": 0.0,
+      "learning_rate": 7.264477311811021e-06,
+      "loss": 0.8063,
+      "step": 21867
+    },
+    {
+      "epoch": 0.6005877344758452,
+      "grad_norm": 0.0,
+      "learning_rate": 7.263621735307178e-06,
+      "loss": 0.7982,
+      "step": 21868
+    },
+    {
+      "epoch": 0.6006151987036884,
+      "grad_norm": 0.0,
+      "learning_rate": 7.262766180454619e-06,
+      "loss": 0.8846,
+      "step": 21869
+    },
+    {
+      "epoch": 0.6006426629315317,
+      "grad_norm": 0.0,
+      "learning_rate": 7.261910647260117e-06,
+      "loss": 0.9298,
+      "step": 21870
+    },
+    {
+      "epoch": 0.600670127159375,
+      "grad_norm": 0.0,
+      "learning_rate": 7.261055135730437e-06,
+      "loss": 0.8783,
+      "step": 21871
+    },
+    {
+      "epoch": 0.6006975913872181,
+      "grad_norm": 0.0,
+      "learning_rate": 7.260199645872352e-06,
+      "loss": 0.871,
+      "step": 21872
+    },
+    {
+      "epoch": 0.6007250556150614,
+      "grad_norm": 0.0,
+      "learning_rate": 7.259344177692626e-06,
+      "loss": 0.8851,
+      "step": 21873
+    },
+    {
+      "epoch": 0.6007525198429046,
+      "grad_norm": 0.0,
+      "learning_rate": 7.258488731198032e-06,
+      "loss": 0.8721,
+      "step": 21874
+    },
+    {
+      "epoch": 0.6007799840707478,
+      "grad_norm": 0.0,
+      "learning_rate": 7.257633306395335e-06,
+      "loss": 0.8633,
+      "step": 21875
+    },
+    {
+      "epoch": 0.6008074482985911,
+      "grad_norm": 0.0,
+      "learning_rate": 7.256777903291306e-06,
+      "loss": 0.8481,
+      "step": 21876
+    },
+    {
+      "epoch": 0.6008349125264343,
+      "grad_norm": 0.0,
+      "learning_rate": 7.255922521892715e-06,
+      "loss": 0.7857,
+      "step": 21877
+    },
+    {
+      "epoch": 0.6008623767542776,
+      "grad_norm": 0.0,
+      "learning_rate": 7.255067162206331e-06,
+      "loss": 0.9846,
+      "step": 21878
+    },
+    {
+      "epoch": 0.6008898409821208,
+      "grad_norm": 0.0,
+      "learning_rate": 7.254211824238913e-06,
+      "loss": 0.7602,
+      "step": 21879
+    },
+    {
+      "epoch": 0.600917305209964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.253356507997235e-06,
+      "loss": 0.915,
+      "step": 21880
+    },
+    {
+      "epoch": 0.6009447694378073,
+      "grad_norm": 0.0,
+      "learning_rate": 7.252501213488066e-06,
+      "loss": 0.8906,
+      "step": 21881
+    },
+    {
+      "epoch": 0.6009722336656504,
+      "grad_norm": 0.0,
+      "learning_rate": 7.251645940718172e-06,
+      "loss": 0.8424,
+      "step": 21882
+    },
+    {
+      "epoch": 0.6009996978934937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2507906896943185e-06,
+      "loss": 0.932,
+      "step": 21883
+    },
+    {
+      "epoch": 0.601027162121337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.249935460423276e-06,
+      "loss": 0.912,
+      "step": 21884
+    },
+    {
+      "epoch": 0.6010546263491802,
+      "grad_norm": 0.0,
+      "learning_rate": 7.249080252911809e-06,
+      "loss": 0.9204,
+      "step": 21885
+    },
+    {
+      "epoch": 0.6010820905770234,
+      "grad_norm": 0.0,
+      "learning_rate": 7.248225067166683e-06,
+      "loss": 0.8963,
+      "step": 21886
+    },
+    {
+      "epoch": 0.6011095548048666,
+      "grad_norm": 0.0,
+      "learning_rate": 7.247369903194668e-06,
+      "loss": 0.8116,
+      "step": 21887
+    },
+    {
+      "epoch": 0.6011370190327099,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2465147610025255e-06,
+      "loss": 0.8814,
+      "step": 21888
+    },
+    {
+      "epoch": 0.6011644832605532,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2456596405970295e-06,
+      "loss": 0.8398,
+      "step": 21889
+    },
+    {
+      "epoch": 0.6011919474883963,
+      "grad_norm": 0.0,
+      "learning_rate": 7.24480454198494e-06,
+      "loss": 0.876,
+      "step": 21890
+    },
+    {
+      "epoch": 0.6012194117162396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.243949465173021e-06,
+      "loss": 0.916,
+      "step": 21891
+    },
+    {
+      "epoch": 0.6012468759440829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.243094410168044e-06,
+      "loss": 0.8881,
+      "step": 21892
+    },
+    {
+      "epoch": 0.601274340171926,
+      "grad_norm": 0.0,
+      "learning_rate": 7.24223937697677e-06,
+      "loss": 0.837,
+      "step": 21893
+    },
+    {
+      "epoch": 0.6013018043997693,
+      "grad_norm": 0.0,
+      "learning_rate": 7.241384365605968e-06,
+      "loss": 0.8476,
+      "step": 21894
+    },
+    {
+      "epoch": 0.6013292686276125,
+      "grad_norm": 0.0,
+      "learning_rate": 7.240529376062404e-06,
+      "loss": 0.9013,
+      "step": 21895
+    },
+    {
+      "epoch": 0.6013567328554558,
+      "grad_norm": 0.0,
+      "learning_rate": 7.239674408352837e-06,
+      "loss": 0.8809,
+      "step": 21896
+    },
+    {
+      "epoch": 0.601384197083299,
+      "grad_norm": 0.0,
+      "learning_rate": 7.238819462484035e-06,
+      "loss": 0.9016,
+      "step": 21897
+    },
+    {
+      "epoch": 0.6014116613111422,
+      "grad_norm": 0.0,
+      "learning_rate": 7.237964538462765e-06,
+      "loss": 0.7859,
+      "step": 21898
+    },
+    {
+      "epoch": 0.6014391255389855,
+      "grad_norm": 0.0,
+      "learning_rate": 7.237109636295787e-06,
+      "loss": 0.8835,
+      "step": 21899
+    },
+    {
+      "epoch": 0.6014665897668287,
+      "grad_norm": 0.0,
+      "learning_rate": 7.236254755989869e-06,
+      "loss": 0.847,
+      "step": 21900
+    },
+    {
+      "epoch": 0.6014940539946719,
+      "grad_norm": 0.0,
+      "learning_rate": 7.235399897551777e-06,
+      "loss": 0.7919,
+      "step": 21901
+    },
+    {
+      "epoch": 0.6015215182225152,
+      "grad_norm": 0.0,
+      "learning_rate": 7.234545060988269e-06,
+      "loss": 0.8411,
+      "step": 21902
+    },
+    {
+      "epoch": 0.6015489824503584,
+      "grad_norm": 0.0,
+      "learning_rate": 7.233690246306111e-06,
+      "loss": 0.7717,
+      "step": 21903
+    },
+    {
+      "epoch": 0.6015764466782016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.232835453512068e-06,
+      "loss": 0.922,
+      "step": 21904
+    },
+    {
+      "epoch": 0.6016039109060449,
+      "grad_norm": 0.0,
+      "learning_rate": 7.231980682612902e-06,
+      "loss": 0.8675,
+      "step": 21905
+    },
+    {
+      "epoch": 0.6016313751338881,
+      "grad_norm": 0.0,
+      "learning_rate": 7.231125933615378e-06,
+      "loss": 0.8654,
+      "step": 21906
+    },
+    {
+      "epoch": 0.6016588393617314,
+      "grad_norm": 0.0,
+      "learning_rate": 7.230271206526262e-06,
+      "loss": 0.8303,
+      "step": 21907
+    },
+    {
+      "epoch": 0.6016863035895745,
+      "grad_norm": 0.0,
+      "learning_rate": 7.229416501352309e-06,
+      "loss": 0.9016,
+      "step": 21908
+    },
+    {
+      "epoch": 0.6017137678174178,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2285618181002835e-06,
+      "loss": 0.9211,
+      "step": 21909
+    },
+    {
+      "epoch": 0.6017412320452611,
+      "grad_norm": 0.0,
+      "learning_rate": 7.227707156776955e-06,
+      "loss": 0.9057,
+      "step": 21910
+    },
+    {
+      "epoch": 0.6017686962731043,
+      "grad_norm": 0.0,
+      "learning_rate": 7.226852517389078e-06,
+      "loss": 0.8258,
+      "step": 21911
+    },
+    {
+      "epoch": 0.6017961605009475,
+      "grad_norm": 0.0,
+      "learning_rate": 7.225997899943418e-06,
+      "loss": 0.8422,
+      "step": 21912
+    },
+    {
+      "epoch": 0.6018236247287907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.225143304446742e-06,
+      "loss": 0.8993,
+      "step": 21913
+    },
+    {
+      "epoch": 0.601851088956634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.224288730905805e-06,
+      "loss": 0.868,
+      "step": 21914
+    },
+    {
+      "epoch": 0.6018785531844772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.223434179327369e-06,
+      "loss": 0.6976,
+      "step": 21915
+    },
+    {
+      "epoch": 0.6019060174123204,
+      "grad_norm": 0.0,
+      "learning_rate": 7.222579649718197e-06,
+      "loss": 0.9204,
+      "step": 21916
+    },
+    {
+      "epoch": 0.6019334816401637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.221725142085052e-06,
+      "loss": 0.8617,
+      "step": 21917
+    },
+    {
+      "epoch": 0.601960945868007,
+      "grad_norm": 0.0,
+      "learning_rate": 7.220870656434692e-06,
+      "loss": 0.7931,
+      "step": 21918
+    },
+    {
+      "epoch": 0.6019884100958501,
+      "grad_norm": 0.0,
+      "learning_rate": 7.220016192773885e-06,
+      "loss": 0.8342,
+      "step": 21919
+    },
+    {
+      "epoch": 0.6020158743236934,
+      "grad_norm": 0.0,
+      "learning_rate": 7.219161751109382e-06,
+      "loss": 0.8736,
+      "step": 21920
+    },
+    {
+      "epoch": 0.6020433385515366,
+      "grad_norm": 0.0,
+      "learning_rate": 7.218307331447948e-06,
+      "loss": 0.8181,
+      "step": 21921
+    },
+    {
+      "epoch": 0.6020708027793799,
+      "grad_norm": 0.0,
+      "learning_rate": 7.217452933796346e-06,
+      "loss": 0.8878,
+      "step": 21922
+    },
+    {
+      "epoch": 0.6020982670072231,
+      "grad_norm": 0.0,
+      "learning_rate": 7.216598558161333e-06,
+      "loss": 0.8981,
+      "step": 21923
+    },
+    {
+      "epoch": 0.6021257312350663,
+      "grad_norm": 0.0,
+      "learning_rate": 7.215744204549669e-06,
+      "loss": 0.9019,
+      "step": 21924
+    },
+    {
+      "epoch": 0.6021531954629096,
+      "grad_norm": 0.0,
+      "learning_rate": 7.21488987296812e-06,
+      "loss": 0.8613,
+      "step": 21925
+    },
+    {
+      "epoch": 0.6021806596907527,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2140355634234405e-06,
+      "loss": 0.9138,
+      "step": 21926
+    },
+    {
+      "epoch": 0.602208123918596,
+      "grad_norm": 0.0,
+      "learning_rate": 7.213181275922387e-06,
+      "loss": 0.7756,
+      "step": 21927
+    },
+    {
+      "epoch": 0.6022355881464393,
+      "grad_norm": 0.0,
+      "learning_rate": 7.212327010471724e-06,
+      "loss": 0.9107,
+      "step": 21928
+    },
+    {
+      "epoch": 0.6022630523742825,
+      "grad_norm": 0.0,
+      "learning_rate": 7.211472767078211e-06,
+      "loss": 0.8229,
+      "step": 21929
+    },
+    {
+      "epoch": 0.6022905166021257,
+      "grad_norm": 0.0,
+      "learning_rate": 7.210618545748604e-06,
+      "loss": 0.7817,
+      "step": 21930
+    },
+    {
+      "epoch": 0.602317980829969,
+      "grad_norm": 0.0,
+      "learning_rate": 7.209764346489668e-06,
+      "loss": 0.7847,
+      "step": 21931
+    },
+    {
+      "epoch": 0.6023454450578122,
+      "grad_norm": 0.0,
+      "learning_rate": 7.208910169308152e-06,
+      "loss": 0.9391,
+      "step": 21932
+    },
+    {
+      "epoch": 0.6023729092856555,
+      "grad_norm": 0.0,
+      "learning_rate": 7.208056014210821e-06,
+      "loss": 0.8298,
+      "step": 21933
+    },
+    {
+      "epoch": 0.6024003735134986,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2072018812044345e-06,
+      "loss": 0.9156,
+      "step": 21934
+    },
+    {
+      "epoch": 0.6024278377413419,
+      "grad_norm": 0.0,
+      "learning_rate": 7.206347770295745e-06,
+      "loss": 0.8823,
+      "step": 21935
+    },
+    {
+      "epoch": 0.6024553019691852,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2054936814915156e-06,
+      "loss": 0.7818,
+      "step": 21936
+    },
+    {
+      "epoch": 0.6024827661970283,
+      "grad_norm": 0.0,
+      "learning_rate": 7.204639614798505e-06,
+      "loss": 0.8025,
+      "step": 21937
+    },
+    {
+      "epoch": 0.6025102304248716,
+      "grad_norm": 0.0,
+      "learning_rate": 7.20378557022347e-06,
+      "loss": 0.7687,
+      "step": 21938
+    },
+    {
+      "epoch": 0.6025376946527148,
+      "grad_norm": 0.0,
+      "learning_rate": 7.202931547773161e-06,
+      "loss": 0.8127,
+      "step": 21939
+    },
+    {
+      "epoch": 0.6025651588805581,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2020775474543426e-06,
+      "loss": 0.8487,
+      "step": 21940
+    },
+    {
+      "epoch": 0.6025926231084013,
+      "grad_norm": 0.0,
+      "learning_rate": 7.201223569273772e-06,
+      "loss": 0.902,
+      "step": 21941
+    },
+    {
+      "epoch": 0.6026200873362445,
+      "grad_norm": 0.0,
+      "learning_rate": 7.200369613238203e-06,
+      "loss": 0.8105,
+      "step": 21942
+    },
+    {
+      "epoch": 0.6026475515640878,
+      "grad_norm": 0.0,
+      "learning_rate": 7.199515679354399e-06,
+      "loss": 0.8538,
+      "step": 21943
+    },
+    {
+      "epoch": 0.6026750157919311,
+      "grad_norm": 0.0,
+      "learning_rate": 7.198661767629106e-06,
+      "loss": 0.9161,
+      "step": 21944
+    },
+    {
+      "epoch": 0.6027024800197742,
+      "grad_norm": 0.0,
+      "learning_rate": 7.197807878069087e-06,
+      "loss": 0.8156,
+      "step": 21945
+    },
+    {
+      "epoch": 0.6027299442476175,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1969540106811e-06,
+      "loss": 0.8971,
+      "step": 21946
+    },
+    {
+      "epoch": 0.6027574084754607,
+      "grad_norm": 0.0,
+      "learning_rate": 7.196100165471896e-06,
+      "loss": 0.9056,
+      "step": 21947
+    },
+    {
+      "epoch": 0.6027848727033039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.195246342448239e-06,
+      "loss": 0.8909,
+      "step": 21948
+    },
+    {
+      "epoch": 0.6028123369311472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.194392541616872e-06,
+      "loss": 0.8185,
+      "step": 21949
+    },
+    {
+      "epoch": 0.6028398011589904,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1935387629845645e-06,
+      "loss": 0.8727,
+      "step": 21950
+    },
+    {
+      "epoch": 0.6028672653868337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.19268500655806e-06,
+      "loss": 0.903,
+      "step": 21951
+    },
+    {
+      "epoch": 0.6028947296146769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.191831272344121e-06,
+      "loss": 0.7736,
+      "step": 21952
+    },
+    {
+      "epoch": 0.6029221938425201,
+      "grad_norm": 0.0,
+      "learning_rate": 7.190977560349501e-06,
+      "loss": 0.841,
+      "step": 21953
+    },
+    {
+      "epoch": 0.6029496580703634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1901238705809566e-06,
+      "loss": 0.786,
+      "step": 21954
+    },
+    {
+      "epoch": 0.6029771222982065,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1892702030452395e-06,
+      "loss": 0.8193,
+      "step": 21955
+    },
+    {
+      "epoch": 0.6030045865260498,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1884165577491035e-06,
+      "loss": 0.7852,
+      "step": 21956
+    },
+    {
+      "epoch": 0.6030320507538931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.187562934699306e-06,
+      "loss": 0.9456,
+      "step": 21957
+    },
+    {
+      "epoch": 0.6030595149817363,
+      "grad_norm": 0.0,
+      "learning_rate": 7.186709333902599e-06,
+      "loss": 0.8308,
+      "step": 21958
+    },
+    {
+      "epoch": 0.6030869792095795,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1858557553657385e-06,
+      "loss": 0.8473,
+      "step": 21959
+    },
+    {
+      "epoch": 0.6031144434374227,
+      "grad_norm": 0.0,
+      "learning_rate": 7.185002199095481e-06,
+      "loss": 0.9088,
+      "step": 21960
+    },
+    {
+      "epoch": 0.603141907665266,
+      "grad_norm": 0.0,
+      "learning_rate": 7.184148665098571e-06,
+      "loss": 0.87,
+      "step": 21961
+    },
+    {
+      "epoch": 0.6031693718931093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1832951533817705e-06,
+      "loss": 0.9281,
+      "step": 21962
+    },
+    {
+      "epoch": 0.6031968361209524,
+      "grad_norm": 0.0,
+      "learning_rate": 7.182441663951827e-06,
+      "loss": 0.7057,
+      "step": 21963
+    },
+    {
+      "epoch": 0.6032243003487957,
+      "grad_norm": 0.0,
+      "learning_rate": 7.181588196815498e-06,
+      "loss": 0.844,
+      "step": 21964
+    },
+    {
+      "epoch": 0.603251764576639,
+      "grad_norm": 0.0,
+      "learning_rate": 7.180734751979535e-06,
+      "loss": 0.8781,
+      "step": 21965
+    },
+    {
+      "epoch": 0.6032792288044821,
+      "grad_norm": 0.0,
+      "learning_rate": 7.179881329450693e-06,
+      "loss": 0.8464,
+      "step": 21966
+    },
+    {
+      "epoch": 0.6033066930323254,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17902792923572e-06,
+      "loss": 0.8541,
+      "step": 21967
+    },
+    {
+      "epoch": 0.6033341572601686,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17817455134137e-06,
+      "loss": 0.8878,
+      "step": 21968
+    },
+    {
+      "epoch": 0.6033616214880119,
+      "grad_norm": 0.0,
+      "learning_rate": 7.177321195774396e-06,
+      "loss": 0.9208,
+      "step": 21969
+    },
+    {
+      "epoch": 0.6033890857158551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.176467862541552e-06,
+      "loss": 0.8747,
+      "step": 21970
+    },
+    {
+      "epoch": 0.6034165499436983,
+      "grad_norm": 0.0,
+      "learning_rate": 7.175614551649586e-06,
+      "loss": 0.8862,
+      "step": 21971
+    },
+    {
+      "epoch": 0.6034440141715416,
+      "grad_norm": 0.0,
+      "learning_rate": 7.174761263105257e-06,
+      "loss": 0.9468,
+      "step": 21972
+    },
+    {
+      "epoch": 0.6034714783993848,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1739079969153045e-06,
+      "loss": 0.8357,
+      "step": 21973
+    },
+    {
+      "epoch": 0.603498942627228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17305475308649e-06,
+      "loss": 0.7917,
+      "step": 21974
+    },
+    {
+      "epoch": 0.6035264068550713,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17220153162556e-06,
+      "loss": 0.8807,
+      "step": 21975
+    },
+    {
+      "epoch": 0.6035538710829145,
+      "grad_norm": 0.0,
+      "learning_rate": 7.171348332539266e-06,
+      "loss": 0.8289,
+      "step": 21976
+    },
+    {
+      "epoch": 0.6035813353107577,
+      "grad_norm": 0.0,
+      "learning_rate": 7.170495155834363e-06,
+      "loss": 0.8909,
+      "step": 21977
+    },
+    {
+      "epoch": 0.603608799538601,
+      "grad_norm": 0.0,
+      "learning_rate": 7.169642001517597e-06,
+      "loss": 0.8546,
+      "step": 21978
+    },
+    {
+      "epoch": 0.6036362637664442,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16878886959572e-06,
+      "loss": 0.8523,
+      "step": 21979
+    },
+    {
+      "epoch": 0.6036637279942875,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1679357600754805e-06,
+      "loss": 0.8067,
+      "step": 21980
+    },
+    {
+      "epoch": 0.6036911922221306,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16708267296363e-06,
+      "loss": 0.9118,
+      "step": 21981
+    },
+    {
+      "epoch": 0.6037186564499739,
+      "grad_norm": 0.0,
+      "learning_rate": 7.166229608266921e-06,
+      "loss": 0.809,
+      "step": 21982
+    },
+    {
+      "epoch": 0.6037461206778172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1653765659921e-06,
+      "loss": 0.8543,
+      "step": 21983
+    },
+    {
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16452354614592e-06,
+      "loss": 0.7727,
+      "step": 21984
+    },
+    {
+      "epoch": 0.6038010491335036,
+      "grad_norm": 0.0,
+      "learning_rate": 7.163670548735125e-06,
+      "loss": 0.7739,
+      "step": 21985
+    },
+    {
+      "epoch": 0.6038285133613468,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1628175737664695e-06,
+      "loss": 0.9083,
+      "step": 21986
+    },
+    {
+      "epoch": 0.6038559775891901,
+      "grad_norm": 0.0,
+      "learning_rate": 7.161964621246697e-06,
+      "loss": 0.8381,
+      "step": 21987
+    },
+    {
+      "epoch": 0.6038834418170334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16111169118256e-06,
+      "loss": 0.8791,
+      "step": 21988
+    },
+    {
+      "epoch": 0.6039109060448765,
+      "grad_norm": 0.0,
+      "learning_rate": 7.160258783580811e-06,
+      "loss": 0.9397,
+      "step": 21989
+    },
+    {
+      "epoch": 0.6039383702727198,
+      "grad_norm": 0.0,
+      "learning_rate": 7.159405898448193e-06,
+      "loss": 0.8052,
+      "step": 21990
+    },
+    {
+      "epoch": 0.6039658345005631,
+      "grad_norm": 0.0,
+      "learning_rate": 7.158553035791455e-06,
+      "loss": 0.8471,
+      "step": 21991
+    },
+    {
+      "epoch": 0.6039932987284062,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1577001956173465e-06,
+      "loss": 0.7508,
+      "step": 21992
+    },
+    {
+      "epoch": 0.6040207629562495,
+      "grad_norm": 0.0,
+      "learning_rate": 7.156847377932614e-06,
+      "loss": 0.9775,
+      "step": 21993
+    },
+    {
+      "epoch": 0.6040482271840927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.155994582744008e-06,
+      "loss": 0.8198,
+      "step": 21994
+    },
+    {
+      "epoch": 0.604075691411936,
+      "grad_norm": 0.0,
+      "learning_rate": 7.155141810058273e-06,
+      "loss": 0.8819,
+      "step": 21995
+    },
+    {
+      "epoch": 0.6041031556397792,
+      "grad_norm": 0.0,
+      "learning_rate": 7.15428905988216e-06,
+      "loss": 0.9026,
+      "step": 21996
+    },
+    {
+      "epoch": 0.6041306198676224,
+      "grad_norm": 0.0,
+      "learning_rate": 7.153436332222412e-06,
+      "loss": 0.871,
+      "step": 21997
+    },
+    {
+      "epoch": 0.6041580840954657,
+      "grad_norm": 0.0,
+      "learning_rate": 7.152583627085781e-06,
+      "loss": 0.8729,
+      "step": 21998
+    },
+    {
+      "epoch": 0.6041855483233088,
+      "grad_norm": 0.0,
+      "learning_rate": 7.151730944479008e-06,
+      "loss": 0.8502,
+      "step": 21999
+    },
+    {
+      "epoch": 0.6042130125511521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.150878284408844e-06,
+      "loss": 0.9742,
+      "step": 22000
+    },
+    {
+      "epoch": 0.6042404767789954,
+      "grad_norm": 0.0,
+      "learning_rate": 7.150025646882038e-06,
+      "loss": 0.8779,
+      "step": 22001
+    },
+    {
+      "epoch": 0.6042679410068386,
+      "grad_norm": 0.0,
+      "learning_rate": 7.14917303190533e-06,
+      "loss": 0.9062,
+      "step": 22002
+    },
+    {
+      "epoch": 0.6042954052346818,
+      "grad_norm": 0.0,
+      "learning_rate": 7.14832043948547e-06,
+      "loss": 0.9822,
+      "step": 22003
+    },
+    {
+      "epoch": 0.6043228694625251,
+      "grad_norm": 0.0,
+      "learning_rate": 7.147467869629203e-06,
+      "loss": 0.8657,
+      "step": 22004
+    },
+    {
+      "epoch": 0.6043503336903683,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1466153223432734e-06,
+      "loss": 0.8541,
+      "step": 22005
+    },
+    {
+      "epoch": 0.6043777979182116,
+      "grad_norm": 0.0,
+      "learning_rate": 7.145762797634433e-06,
+      "loss": 0.8361,
+      "step": 22006
+    },
+    {
+      "epoch": 0.6044052621460547,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1449102955094214e-06,
+      "loss": 0.9001,
+      "step": 22007
+    },
+    {
+      "epoch": 0.604432726373898,
+      "grad_norm": 0.0,
+      "learning_rate": 7.144057815974984e-06,
+      "loss": 0.8856,
+      "step": 22008
+    },
+    {
+      "epoch": 0.6044601906017413,
+      "grad_norm": 0.0,
+      "learning_rate": 7.143205359037867e-06,
+      "loss": 0.8747,
+      "step": 22009
+    },
+    {
+      "epoch": 0.6044876548295844,
+      "grad_norm": 0.0,
+      "learning_rate": 7.142352924704818e-06,
+      "loss": 0.935,
+      "step": 22010
+    },
+    {
+      "epoch": 0.6045151190574277,
+      "grad_norm": 0.0,
+      "learning_rate": 7.141500512982578e-06,
+      "loss": 0.7832,
+      "step": 22011
+    },
+    {
+      "epoch": 0.6045425832852709,
+      "grad_norm": 0.0,
+      "learning_rate": 7.140648123877891e-06,
+      "loss": 0.8506,
+      "step": 22012
+    },
+    {
+      "epoch": 0.6045700475131142,
+      "grad_norm": 0.0,
+      "learning_rate": 7.13979575739751e-06,
+      "loss": 0.8617,
+      "step": 22013
+    },
+    {
+      "epoch": 0.6045975117409574,
+      "grad_norm": 0.0,
+      "learning_rate": 7.138943413548169e-06,
+      "loss": 0.7822,
+      "step": 22014
+    },
+    {
+      "epoch": 0.6046249759688006,
+      "grad_norm": 0.0,
+      "learning_rate": 7.138091092336617e-06,
+      "loss": 0.8479,
+      "step": 22015
+    },
+    {
+      "epoch": 0.6046524401966439,
+      "grad_norm": 0.0,
+      "learning_rate": 7.137238793769594e-06,
+      "loss": 0.9023,
+      "step": 22016
+    },
+    {
+      "epoch": 0.6046799044244872,
+      "grad_norm": 0.0,
+      "learning_rate": 7.136386517853848e-06,
+      "loss": 0.8278,
+      "step": 22017
+    },
+    {
+      "epoch": 0.6047073686523303,
+      "grad_norm": 0.0,
+      "learning_rate": 7.135534264596123e-06,
+      "loss": 0.7957,
+      "step": 22018
+    },
+    {
+      "epoch": 0.6047348328801736,
+      "grad_norm": 0.0,
+      "learning_rate": 7.134682034003162e-06,
+      "loss": 0.8482,
+      "step": 22019
+    },
+    {
+      "epoch": 0.6047622971080168,
+      "grad_norm": 0.0,
+      "learning_rate": 7.133829826081703e-06,
+      "loss": 0.8838,
+      "step": 22020
+    },
+    {
+      "epoch": 0.60478976133586,
+      "grad_norm": 0.0,
+      "learning_rate": 7.132977640838492e-06,
+      "loss": 0.8408,
+      "step": 22021
+    },
+    {
+      "epoch": 0.6048172255637033,
+      "grad_norm": 0.0,
+      "learning_rate": 7.132125478280275e-06,
+      "loss": 0.9511,
+      "step": 22022
+    },
+    {
+      "epoch": 0.6048446897915465,
+      "grad_norm": 0.0,
+      "learning_rate": 7.13127333841379e-06,
+      "loss": 0.9005,
+      "step": 22023
+    },
+    {
+      "epoch": 0.6048721540193898,
+      "grad_norm": 0.0,
+      "learning_rate": 7.130421221245782e-06,
+      "loss": 0.9139,
+      "step": 22024
+    },
+    {
+      "epoch": 0.6048996182472329,
+      "grad_norm": 0.0,
+      "learning_rate": 7.129569126782996e-06,
+      "loss": 0.8198,
+      "step": 22025
+    },
+    {
+      "epoch": 0.6049270824750762,
+      "grad_norm": 0.0,
+      "learning_rate": 7.128717055032168e-06,
+      "loss": 0.7221,
+      "step": 22026
+    },
+    {
+      "epoch": 0.6049545467029195,
+      "grad_norm": 0.0,
+      "learning_rate": 7.127865006000045e-06,
+      "loss": 0.9076,
+      "step": 22027
+    },
+    {
+      "epoch": 0.6049820109307626,
+      "grad_norm": 0.0,
+      "learning_rate": 7.127012979693364e-06,
+      "loss": 0.7972,
+      "step": 22028
+    },
+    {
+      "epoch": 0.6050094751586059,
+      "grad_norm": 0.0,
+      "learning_rate": 7.126160976118869e-06,
+      "loss": 0.7657,
+      "step": 22029
+    },
+    {
+      "epoch": 0.6050369393864492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1253089952833045e-06,
+      "loss": 0.9493,
+      "step": 22030
+    },
+    {
+      "epoch": 0.6050644036142924,
+      "grad_norm": 0.0,
+      "learning_rate": 7.12445703719341e-06,
+      "loss": 0.8786,
+      "step": 22031
+    },
+    {
+      "epoch": 0.6050918678421356,
+      "grad_norm": 0.0,
+      "learning_rate": 7.123605101855921e-06,
+      "loss": 0.8318,
+      "step": 22032
+    },
+    {
+      "epoch": 0.6051193320699788,
+      "grad_norm": 0.0,
+      "learning_rate": 7.122753189277582e-06,
+      "loss": 0.8868,
+      "step": 22033
+    },
+    {
+      "epoch": 0.6051467962978221,
+      "grad_norm": 0.0,
+      "learning_rate": 7.121901299465138e-06,
+      "loss": 0.9373,
+      "step": 22034
+    },
+    {
+      "epoch": 0.6051742605256654,
+      "grad_norm": 0.0,
+      "learning_rate": 7.121049432425323e-06,
+      "loss": 0.8854,
+      "step": 22035
+    },
+    {
+      "epoch": 0.6052017247535085,
+      "grad_norm": 0.0,
+      "learning_rate": 7.120197588164881e-06,
+      "loss": 0.7962,
+      "step": 22036
+    },
+    {
+      "epoch": 0.6052291889813518,
+      "grad_norm": 0.0,
+      "learning_rate": 7.119345766690554e-06,
+      "loss": 0.8443,
+      "step": 22037
+    },
+    {
+      "epoch": 0.605256653209195,
+      "grad_norm": 0.0,
+      "learning_rate": 7.118493968009077e-06,
+      "loss": 0.9533,
+      "step": 22038
+    },
+    {
+      "epoch": 0.6052841174370382,
+      "grad_norm": 0.0,
+      "learning_rate": 7.117642192127191e-06,
+      "loss": 0.8078,
+      "step": 22039
+    },
+    {
+      "epoch": 0.6053115816648815,
+      "grad_norm": 0.0,
+      "learning_rate": 7.116790439051636e-06,
+      "loss": 0.8051,
+      "step": 22040
+    },
+    {
+      "epoch": 0.6053390458927247,
+      "grad_norm": 0.0,
+      "learning_rate": 7.115938708789152e-06,
+      "loss": 0.8945,
+      "step": 22041
+    },
+    {
+      "epoch": 0.605366510120568,
+      "grad_norm": 0.0,
+      "learning_rate": 7.11508700134648e-06,
+      "loss": 0.7719,
+      "step": 22042
+    },
+    {
+      "epoch": 0.6053939743484112,
+      "grad_norm": 0.0,
+      "learning_rate": 7.11423531673036e-06,
+      "loss": 0.8744,
+      "step": 22043
+    },
+    {
+      "epoch": 0.6054214385762544,
+      "grad_norm": 0.0,
+      "learning_rate": 7.113383654947523e-06,
+      "loss": 0.8819,
+      "step": 22044
+    },
+    {
+      "epoch": 0.6054489028040977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.112532016004711e-06,
+      "loss": 0.9357,
+      "step": 22045
+    },
+    {
+      "epoch": 0.6054763670319409,
+      "grad_norm": 0.0,
+      "learning_rate": 7.11168039990867e-06,
+      "loss": 0.752,
+      "step": 22046
+    },
+    {
+      "epoch": 0.6055038312597841,
+      "grad_norm": 0.0,
+      "learning_rate": 7.110828806666128e-06,
+      "loss": 0.8807,
+      "step": 22047
+    },
+    {
+      "epoch": 0.6055312954876274,
+      "grad_norm": 0.0,
+      "learning_rate": 7.109977236283829e-06,
+      "loss": 0.9572,
+      "step": 22048
+    },
+    {
+      "epoch": 0.6055587597154706,
+      "grad_norm": 0.0,
+      "learning_rate": 7.109125688768512e-06,
+      "loss": 0.8157,
+      "step": 22049
+    },
+    {
+      "epoch": 0.6055862239433139,
+      "grad_norm": 0.0,
+      "learning_rate": 7.108274164126912e-06,
+      "loss": 0.8842,
+      "step": 22050
+    },
+    {
+      "epoch": 0.605613688171157,
+      "grad_norm": 0.0,
+      "learning_rate": 7.107422662365764e-06,
+      "loss": 0.8329,
+      "step": 22051
+    },
+    {
+      "epoch": 0.6056411523990003,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1065711834918085e-06,
+      "loss": 0.8442,
+      "step": 22052
+    },
+    {
+      "epoch": 0.6056686166268436,
+      "grad_norm": 0.0,
+      "learning_rate": 7.105719727511782e-06,
+      "loss": 0.9077,
+      "step": 22053
+    },
+    {
+      "epoch": 0.6056960808546867,
+      "grad_norm": 0.0,
+      "learning_rate": 7.104868294432425e-06,
+      "loss": 0.8556,
+      "step": 22054
+    },
+    {
+      "epoch": 0.60572354508253,
+      "grad_norm": 0.0,
+      "learning_rate": 7.104016884260473e-06,
+      "loss": 0.8836,
+      "step": 22055
+    },
+    {
+      "epoch": 0.6057510093103733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.103165497002658e-06,
+      "loss": 0.8928,
+      "step": 22056
+    },
+    {
+      "epoch": 0.6057784735382165,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1023141326657185e-06,
+      "loss": 0.8435,
+      "step": 22057
+    },
+    {
+      "epoch": 0.6058059377660597,
+      "grad_norm": 0.0,
+      "learning_rate": 7.101462791256395e-06,
+      "loss": 0.8037,
+      "step": 22058
+    },
+    {
+      "epoch": 0.6058334019939029,
+      "grad_norm": 0.0,
+      "learning_rate": 7.100611472781418e-06,
+      "loss": 0.8717,
+      "step": 22059
+    },
+    {
+      "epoch": 0.6058608662217462,
+      "grad_norm": 0.0,
+      "learning_rate": 7.099760177247527e-06,
+      "loss": 0.8301,
+      "step": 22060
+    },
+    {
+      "epoch": 0.6058883304495895,
+      "grad_norm": 0.0,
+      "learning_rate": 7.098908904661462e-06,
+      "loss": 0.7847,
+      "step": 22061
+    },
+    {
+      "epoch": 0.6059157946774326,
+      "grad_norm": 0.0,
+      "learning_rate": 7.098057655029949e-06,
+      "loss": 0.7955,
+      "step": 22062
+    },
+    {
+      "epoch": 0.6059432589052759,
+      "grad_norm": 0.0,
+      "learning_rate": 7.097206428359726e-06,
+      "loss": 0.8492,
+      "step": 22063
+    },
+    {
+      "epoch": 0.6059707231331191,
+      "grad_norm": 0.0,
+      "learning_rate": 7.096355224657532e-06,
+      "loss": 0.8887,
+      "step": 22064
+    },
+    {
+      "epoch": 0.6059981873609623,
+      "grad_norm": 0.0,
+      "learning_rate": 7.095504043930099e-06,
+      "loss": 0.801,
+      "step": 22065
+    },
+    {
+      "epoch": 0.6060256515888056,
+      "grad_norm": 0.0,
+      "learning_rate": 7.094652886184166e-06,
+      "loss": 0.9689,
+      "step": 22066
+    },
+    {
+      "epoch": 0.6060531158166488,
+      "grad_norm": 0.0,
+      "learning_rate": 7.093801751426466e-06,
+      "loss": 0.8116,
+      "step": 22067
+    },
+    {
+      "epoch": 0.6060805800444921,
+      "grad_norm": 0.0,
+      "learning_rate": 7.092950639663728e-06,
+      "loss": 0.8513,
+      "step": 22068
+    },
+    {
+      "epoch": 0.6061080442723353,
+      "grad_norm": 0.0,
+      "learning_rate": 7.092099550902691e-06,
+      "loss": 0.8632,
+      "step": 22069
+    },
+    {
+      "epoch": 0.6061355085001785,
+      "grad_norm": 0.0,
+      "learning_rate": 7.09124848515009e-06,
+      "loss": 0.9172,
+      "step": 22070
+    },
+    {
+      "epoch": 0.6061629727280218,
+      "grad_norm": 0.0,
+      "learning_rate": 7.090397442412657e-06,
+      "loss": 0.8439,
+      "step": 22071
+    },
+    {
+      "epoch": 0.6061904369558649,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0895464226971315e-06,
+      "loss": 0.8542,
+      "step": 22072
+    },
+    {
+      "epoch": 0.6062179011837082,
+      "grad_norm": 0.0,
+      "learning_rate": 7.088695426010237e-06,
+      "loss": 0.9788,
+      "step": 22073
+    },
+    {
+      "epoch": 0.6062453654115515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.087844452358713e-06,
+      "loss": 0.8291,
+      "step": 22074
+    },
+    {
+      "epoch": 0.6062728296393947,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0869935017492905e-06,
+      "loss": 0.9349,
+      "step": 22075
+    },
+    {
+      "epoch": 0.6063002938672379,
+      "grad_norm": 0.0,
+      "learning_rate": 7.086142574188703e-06,
+      "loss": 0.7666,
+      "step": 22076
+    },
+    {
+      "epoch": 0.6063277580950811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.085291669683687e-06,
+      "loss": 0.782,
+      "step": 22077
+    },
+    {
+      "epoch": 0.6063552223229244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.084440788240974e-06,
+      "loss": 0.8575,
+      "step": 22078
+    },
+    {
+      "epoch": 0.6063826865507677,
+      "grad_norm": 0.0,
+      "learning_rate": 7.083589929867293e-06,
+      "loss": 0.8486,
+      "step": 22079
+    },
+    {
+      "epoch": 0.6064101507786108,
+      "grad_norm": 0.0,
+      "learning_rate": 7.082739094569375e-06,
+      "loss": 0.9535,
+      "step": 22080
+    },
+    {
+      "epoch": 0.6064376150064541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.081888282353957e-06,
+      "loss": 0.8246,
+      "step": 22081
+    },
+    {
+      "epoch": 0.6064650792342974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.081037493227771e-06,
+      "loss": 0.949,
+      "step": 22082
+    },
+    {
+      "epoch": 0.6064925434621405,
+      "grad_norm": 0.0,
+      "learning_rate": 7.080186727197545e-06,
+      "loss": 0.7609,
+      "step": 22083
+    },
+    {
+      "epoch": 0.6065200076899838,
+      "grad_norm": 0.0,
+      "learning_rate": 7.079335984270017e-06,
+      "loss": 0.932,
+      "step": 22084
+    },
+    {
+      "epoch": 0.606547471917827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.078485264451911e-06,
+      "loss": 0.8858,
+      "step": 22085
+    },
+    {
+      "epoch": 0.6065749361456703,
+      "grad_norm": 0.0,
+      "learning_rate": 7.077634567749962e-06,
+      "loss": 0.9148,
+      "step": 22086
+    },
+    {
+      "epoch": 0.6066024003735135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.076783894170897e-06,
+      "loss": 0.8891,
+      "step": 22087
+    },
+    {
+      "epoch": 0.6066298646013567,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0759332437214526e-06,
+      "loss": 0.9629,
+      "step": 22088
+    },
+    {
+      "epoch": 0.6066573288292,
+      "grad_norm": 0.0,
+      "learning_rate": 7.075082616408358e-06,
+      "loss": 0.9083,
+      "step": 22089
+    },
+    {
+      "epoch": 0.6066847930570431,
+      "grad_norm": 0.0,
+      "learning_rate": 7.074232012238342e-06,
+      "loss": 0.9308,
+      "step": 22090
+    },
+    {
+      "epoch": 0.6067122572848864,
+      "grad_norm": 0.0,
+      "learning_rate": 7.073381431218137e-06,
+      "loss": 0.8516,
+      "step": 22091
+    },
+    {
+      "epoch": 0.6067397215127297,
+      "grad_norm": 0.0,
+      "learning_rate": 7.072530873354468e-06,
+      "loss": 0.9107,
+      "step": 22092
+    },
+    {
+      "epoch": 0.6067671857405729,
+      "grad_norm": 0.0,
+      "learning_rate": 7.07168033865407e-06,
+      "loss": 0.9018,
+      "step": 22093
+    },
+    {
+      "epoch": 0.6067946499684161,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0708298271236744e-06,
+      "loss": 0.9532,
+      "step": 22094
+    },
+    {
+      "epoch": 0.6068221141962594,
+      "grad_norm": 0.0,
+      "learning_rate": 7.069979338770004e-06,
+      "loss": 0.8424,
+      "step": 22095
+    },
+    {
+      "epoch": 0.6068495784241026,
+      "grad_norm": 0.0,
+      "learning_rate": 7.069128873599797e-06,
+      "loss": 0.8974,
+      "step": 22096
+    },
+    {
+      "epoch": 0.6068770426519459,
+      "grad_norm": 0.0,
+      "learning_rate": 7.068278431619774e-06,
+      "loss": 0.908,
+      "step": 22097
+    },
+    {
+      "epoch": 0.606904506879789,
+      "grad_norm": 0.0,
+      "learning_rate": 7.067428012836669e-06,
+      "loss": 0.8327,
+      "step": 22098
+    },
+    {
+      "epoch": 0.6069319711076323,
+      "grad_norm": 0.0,
+      "learning_rate": 7.066577617257209e-06,
+      "loss": 0.8333,
+      "step": 22099
+    },
+    {
+      "epoch": 0.6069594353354756,
+      "grad_norm": 0.0,
+      "learning_rate": 7.065727244888121e-06,
+      "loss": 0.7463,
+      "step": 22100
+    },
+    {
+      "epoch": 0.6069868995633187,
+      "grad_norm": 0.0,
+      "learning_rate": 7.064876895736139e-06,
+      "loss": 0.8335,
+      "step": 22101
+    },
+    {
+      "epoch": 0.607014363791162,
+      "grad_norm": 0.0,
+      "learning_rate": 7.064026569807987e-06,
+      "loss": 0.9114,
+      "step": 22102
+    },
+    {
+      "epoch": 0.6070418280190052,
+      "grad_norm": 0.0,
+      "learning_rate": 7.063176267110395e-06,
+      "loss": 0.9643,
+      "step": 22103
+    },
+    {
+      "epoch": 0.6070692922468485,
+      "grad_norm": 0.0,
+      "learning_rate": 7.062325987650087e-06,
+      "loss": 0.8604,
+      "step": 22104
+    },
+    {
+      "epoch": 0.6070967564746917,
+      "grad_norm": 0.0,
+      "learning_rate": 7.061475731433795e-06,
+      "loss": 0.833,
+      "step": 22105
+    },
+    {
+      "epoch": 0.6071242207025349,
+      "grad_norm": 0.0,
+      "learning_rate": 7.060625498468246e-06,
+      "loss": 0.9346,
+      "step": 22106
+    },
+    {
+      "epoch": 0.6071516849303782,
+      "grad_norm": 0.0,
+      "learning_rate": 7.059775288760165e-06,
+      "loss": 0.9107,
+      "step": 22107
+    },
+    {
+      "epoch": 0.6071791491582215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.058925102316286e-06,
+      "loss": 0.7804,
+      "step": 22108
+    },
+    {
+      "epoch": 0.6072066133860646,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0580749391433265e-06,
+      "loss": 0.8396,
+      "step": 22109
+    },
+    {
+      "epoch": 0.6072340776139079,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0572247992480195e-06,
+      "loss": 0.7843,
+      "step": 22110
+    },
+    {
+      "epoch": 0.6072615418417511,
+      "grad_norm": 0.0,
+      "learning_rate": 7.056374682637086e-06,
+      "loss": 0.8028,
+      "step": 22111
+    },
+    {
+      "epoch": 0.6072890060695944,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0555245893172585e-06,
+      "loss": 0.8058,
+      "step": 22112
+    },
+    {
+      "epoch": 0.6073164702974376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.054674519295261e-06,
+      "loss": 0.8444,
+      "step": 22113
+    },
+    {
+      "epoch": 0.6073439345252808,
+      "grad_norm": 0.0,
+      "learning_rate": 7.053824472577821e-06,
+      "loss": 0.8876,
+      "step": 22114
+    },
+    {
+      "epoch": 0.6073713987531241,
+      "grad_norm": 0.0,
+      "learning_rate": 7.052974449171663e-06,
+      "loss": 0.8738,
+      "step": 22115
+    },
+    {
+      "epoch": 0.6073988629809672,
+      "grad_norm": 0.0,
+      "learning_rate": 7.052124449083511e-06,
+      "loss": 0.899,
+      "step": 22116
+    },
+    {
+      "epoch": 0.6074263272088105,
+      "grad_norm": 0.0,
+      "learning_rate": 7.051274472320093e-06,
+      "loss": 0.7766,
+      "step": 22117
+    },
+    {
+      "epoch": 0.6074537914366538,
+      "grad_norm": 0.0,
+      "learning_rate": 7.050424518888135e-06,
+      "loss": 0.9327,
+      "step": 22118
+    },
+    {
+      "epoch": 0.607481255664497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.049574588794357e-06,
+      "loss": 0.8658,
+      "step": 22119
+    },
+    {
+      "epoch": 0.6075087198923402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.048724682045494e-06,
+      "loss": 0.8577,
+      "step": 22120
+    },
+    {
+      "epoch": 0.6075361841201835,
+      "grad_norm": 0.0,
+      "learning_rate": 7.04787479864826e-06,
+      "loss": 0.8332,
+      "step": 22121
+    },
+    {
+      "epoch": 0.6075636483480267,
+      "grad_norm": 0.0,
+      "learning_rate": 7.047024938609384e-06,
+      "loss": 0.8023,
+      "step": 22122
+    },
+    {
+      "epoch": 0.60759111257587,
+      "grad_norm": 0.0,
+      "learning_rate": 7.046175101935591e-06,
+      "loss": 0.8543,
+      "step": 22123
+    },
+    {
+      "epoch": 0.6076185768037131,
+      "grad_norm": 0.0,
+      "learning_rate": 7.045325288633605e-06,
+      "loss": 0.907,
+      "step": 22124
+    },
+    {
+      "epoch": 0.6076460410315564,
+      "grad_norm": 0.0,
+      "learning_rate": 7.044475498710151e-06,
+      "loss": 0.786,
+      "step": 22125
+    },
+    {
+      "epoch": 0.6076735052593997,
+      "grad_norm": 0.0,
+      "learning_rate": 7.043625732171951e-06,
+      "loss": 0.9643,
+      "step": 22126
+    },
+    {
+      "epoch": 0.6077009694872428,
+      "grad_norm": 0.0,
+      "learning_rate": 7.04277598902573e-06,
+      "loss": 0.8773,
+      "step": 22127
+    },
+    {
+      "epoch": 0.6077284337150861,
+      "grad_norm": 0.0,
+      "learning_rate": 7.041926269278208e-06,
+      "loss": 0.9219,
+      "step": 22128
+    },
+    {
+      "epoch": 0.6077558979429293,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0410765729361115e-06,
+      "loss": 0.8857,
+      "step": 22129
+    },
+    {
+      "epoch": 0.6077833621707726,
+      "grad_norm": 0.0,
+      "learning_rate": 7.040226900006166e-06,
+      "loss": 0.8462,
+      "step": 22130
+    },
+    {
+      "epoch": 0.6078108263986158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.039377250495094e-06,
+      "loss": 0.853,
+      "step": 22131
+    },
+    {
+      "epoch": 0.607838290626459,
+      "grad_norm": 0.0,
+      "learning_rate": 7.038527624409612e-06,
+      "loss": 0.7581,
+      "step": 22132
+    },
+    {
+      "epoch": 0.6078657548543023,
+      "grad_norm": 0.0,
+      "learning_rate": 7.037678021756446e-06,
+      "loss": 0.8615,
+      "step": 22133
+    },
+    {
+      "epoch": 0.6078932190821456,
+      "grad_norm": 0.0,
+      "learning_rate": 7.036828442542323e-06,
+      "loss": 0.9736,
+      "step": 22134
+    },
+    {
+      "epoch": 0.6079206833099887,
+      "grad_norm": 0.0,
+      "learning_rate": 7.035978886773957e-06,
+      "loss": 0.8471,
+      "step": 22135
+    },
+    {
+      "epoch": 0.607948147537832,
+      "grad_norm": 0.0,
+      "learning_rate": 7.035129354458075e-06,
+      "loss": 0.8763,
+      "step": 22136
+    },
+    {
+      "epoch": 0.6079756117656752,
+      "grad_norm": 0.0,
+      "learning_rate": 7.034279845601402e-06,
+      "loss": 0.8382,
+      "step": 22137
+    },
+    {
+      "epoch": 0.6080030759935184,
+      "grad_norm": 0.0,
+      "learning_rate": 7.03343036021065e-06,
+      "loss": 0.7748,
+      "step": 22138
+    },
+    {
+      "epoch": 0.6080305402213617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.032580898292549e-06,
+      "loss": 0.9285,
+      "step": 22139
+    },
+    {
+      "epoch": 0.6080580044492049,
+      "grad_norm": 0.0,
+      "learning_rate": 7.031731459853817e-06,
+      "loss": 0.9142,
+      "step": 22140
+    },
+    {
+      "epoch": 0.6080854686770482,
+      "grad_norm": 0.0,
+      "learning_rate": 7.030882044901174e-06,
+      "loss": 0.8423,
+      "step": 22141
+    },
+    {
+      "epoch": 0.6081129329048914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.030032653441345e-06,
+      "loss": 0.9005,
+      "step": 22142
+    },
+    {
+      "epoch": 0.6081403971327346,
+      "grad_norm": 0.0,
+      "learning_rate": 7.029183285481049e-06,
+      "loss": 0.8854,
+      "step": 22143
+    },
+    {
+      "epoch": 0.6081678613605779,
+      "grad_norm": 0.0,
+      "learning_rate": 7.028333941027004e-06,
+      "loss": 0.8951,
+      "step": 22144
+    },
+    {
+      "epoch": 0.608195325588421,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0274846200859295e-06,
+      "loss": 0.893,
+      "step": 22145
+    },
+    {
+      "epoch": 0.6082227898162643,
+      "grad_norm": 0.0,
+      "learning_rate": 7.026635322664552e-06,
+      "loss": 0.9022,
+      "step": 22146
+    },
+    {
+      "epoch": 0.6082502540441076,
+      "grad_norm": 0.0,
+      "learning_rate": 7.025786048769585e-06,
+      "loss": 0.8845,
+      "step": 22147
+    },
+    {
+      "epoch": 0.6082777182719508,
+      "grad_norm": 0.0,
+      "learning_rate": 7.02493679840775e-06,
+      "loss": 0.8147,
+      "step": 22148
+    },
+    {
+      "epoch": 0.608305182499794,
+      "grad_norm": 0.0,
+      "learning_rate": 7.024087571585772e-06,
+      "loss": 0.8629,
+      "step": 22149
+    },
+    {
+      "epoch": 0.6083326467276372,
+      "grad_norm": 0.0,
+      "learning_rate": 7.023238368310362e-06,
+      "loss": 0.8375,
+      "step": 22150
+    },
+    {
+      "epoch": 0.6083601109554805,
+      "grad_norm": 0.0,
+      "learning_rate": 7.022389188588246e-06,
+      "loss": 0.9048,
+      "step": 22151
+    },
+    {
+      "epoch": 0.6083875751833238,
+      "grad_norm": 0.0,
+      "learning_rate": 7.021540032426136e-06,
+      "loss": 0.8294,
+      "step": 22152
+    },
+    {
+      "epoch": 0.6084150394111669,
+      "grad_norm": 0.0,
+      "learning_rate": 7.020690899830757e-06,
+      "loss": 0.9216,
+      "step": 22153
+    },
+    {
+      "epoch": 0.6084425036390102,
+      "grad_norm": 0.0,
+      "learning_rate": 7.019841790808827e-06,
+      "loss": 0.8275,
+      "step": 22154
+    },
+    {
+      "epoch": 0.6084699678668535,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0189927053670644e-06,
+      "loss": 0.7291,
+      "step": 22155
+    },
+    {
+      "epoch": 0.6084974320946966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.018143643512183e-06,
+      "loss": 0.7617,
+      "step": 22156
+    },
+    {
+      "epoch": 0.6085248963225399,
+      "grad_norm": 0.0,
+      "learning_rate": 7.017294605250904e-06,
+      "loss": 0.7994,
+      "step": 22157
+    },
+    {
+      "epoch": 0.6085523605503831,
+      "grad_norm": 0.0,
+      "learning_rate": 7.016445590589949e-06,
+      "loss": 0.8569,
+      "step": 22158
+    },
+    {
+      "epoch": 0.6085798247782264,
+      "grad_norm": 0.0,
+      "learning_rate": 7.015596599536027e-06,
+      "loss": 0.8106,
+      "step": 22159
+    },
+    {
+      "epoch": 0.6086072890060696,
+      "grad_norm": 0.0,
+      "learning_rate": 7.014747632095863e-06,
+      "loss": 0.9294,
+      "step": 22160
+    },
+    {
+      "epoch": 0.6086347532339128,
+      "grad_norm": 0.0,
+      "learning_rate": 7.013898688276176e-06,
+      "loss": 0.8358,
+      "step": 22161
+    },
+    {
+      "epoch": 0.6086622174617561,
+      "grad_norm": 0.0,
+      "learning_rate": 7.013049768083675e-06,
+      "loss": 0.8631,
+      "step": 22162
+    },
+    {
+      "epoch": 0.6086896816895992,
+      "grad_norm": 0.0,
+      "learning_rate": 7.012200871525083e-06,
+      "loss": 0.8748,
+      "step": 22163
+    },
+    {
+      "epoch": 0.6087171459174425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0113519986071135e-06,
+      "loss": 0.7673,
+      "step": 22164
+    },
+    {
+      "epoch": 0.6087446101452858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.010503149336484e-06,
+      "loss": 0.8567,
+      "step": 22165
+    },
+    {
+      "epoch": 0.608772074373129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.009654323719914e-06,
+      "loss": 0.9052,
+      "step": 22166
+    },
+    {
+      "epoch": 0.6087995386009722,
+      "grad_norm": 0.0,
+      "learning_rate": 7.008805521764119e-06,
+      "loss": 0.9427,
+      "step": 22167
+    },
+    {
+      "epoch": 0.6088270028288155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.00795674347581e-06,
+      "loss": 0.8655,
+      "step": 22168
+    },
+    {
+      "epoch": 0.6088544670566587,
+      "grad_norm": 0.0,
+      "learning_rate": 7.007107988861707e-06,
+      "loss": 0.8493,
+      "step": 22169
+    },
+    {
+      "epoch": 0.608881931284502,
+      "grad_norm": 0.0,
+      "learning_rate": 7.006259257928525e-06,
+      "loss": 0.8206,
+      "step": 22170
+    },
+    {
+      "epoch": 0.6089093955123451,
+      "grad_norm": 0.0,
+      "learning_rate": 7.005410550682978e-06,
+      "loss": 0.9046,
+      "step": 22171
+    },
+    {
+      "epoch": 0.6089368597401884,
+      "grad_norm": 0.0,
+      "learning_rate": 7.004561867131784e-06,
+      "loss": 0.8194,
+      "step": 22172
+    },
+    {
+      "epoch": 0.6089643239680317,
+      "grad_norm": 0.0,
+      "learning_rate": 7.003713207281659e-06,
+      "loss": 0.7846,
+      "step": 22173
+    },
+    {
+      "epoch": 0.6089917881958749,
+      "grad_norm": 0.0,
+      "learning_rate": 7.002864571139314e-06,
+      "loss": 0.8617,
+      "step": 22174
+    },
+    {
+      "epoch": 0.6090192524237181,
+      "grad_norm": 0.0,
+      "learning_rate": 7.002015958711466e-06,
+      "loss": 0.9338,
+      "step": 22175
+    },
+    {
+      "epoch": 0.6090467166515613,
+      "grad_norm": 0.0,
+      "learning_rate": 7.001167370004826e-06,
+      "loss": 0.9004,
+      "step": 22176
+    },
+    {
+      "epoch": 0.6090741808794046,
+      "grad_norm": 0.0,
+      "learning_rate": 7.000318805026114e-06,
+      "loss": 1.014,
+      "step": 22177
+    },
+    {
+      "epoch": 0.6091016451072478,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9994702637820425e-06,
+      "loss": 0.8384,
+      "step": 22178
+    },
+    {
+      "epoch": 0.609129109335091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.998621746279326e-06,
+      "loss": 0.8527,
+      "step": 22179
+    },
+    {
+      "epoch": 0.6091565735629343,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9977732525246735e-06,
+      "loss": 0.7924,
+      "step": 22180
+    },
+    {
+      "epoch": 0.6091840377907776,
+      "grad_norm": 0.0,
+      "learning_rate": 6.996924782524801e-06,
+      "loss": 0.8134,
+      "step": 22181
+    },
+    {
+      "epoch": 0.6092115020186207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.996076336286426e-06,
+      "loss": 0.9099,
+      "step": 22182
+    },
+    {
+      "epoch": 0.609238966246464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.995227913816256e-06,
+      "loss": 0.8718,
+      "step": 22183
+    },
+    {
+      "epoch": 0.6092664304743072,
+      "grad_norm": 0.0,
+      "learning_rate": 6.994379515121006e-06,
+      "loss": 0.8967,
+      "step": 22184
+    },
+    {
+      "epoch": 0.6092938947021505,
+      "grad_norm": 0.0,
+      "learning_rate": 6.993531140207394e-06,
+      "loss": 0.794,
+      "step": 22185
+    },
+    {
+      "epoch": 0.6093213589299937,
+      "grad_norm": 0.0,
+      "learning_rate": 6.992682789082127e-06,
+      "loss": 0.8468,
+      "step": 22186
+    },
+    {
+      "epoch": 0.6093488231578369,
+      "grad_norm": 0.0,
+      "learning_rate": 6.991834461751916e-06,
+      "loss": 0.7966,
+      "step": 22187
+    },
+    {
+      "epoch": 0.6093762873856802,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9909861582234766e-06,
+      "loss": 0.8968,
+      "step": 22188
+    },
+    {
+      "epoch": 0.6094037516135233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.99013787850352e-06,
+      "loss": 0.7998,
+      "step": 22189
+    },
+    {
+      "epoch": 0.6094312158413666,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9892896225987625e-06,
+      "loss": 0.7895,
+      "step": 22190
+    },
+    {
+      "epoch": 0.6094586800692099,
+      "grad_norm": 0.0,
+      "learning_rate": 6.98844139051591e-06,
+      "loss": 0.849,
+      "step": 22191
+    },
+    {
+      "epoch": 0.6094861442970531,
+      "grad_norm": 0.0,
+      "learning_rate": 6.987593182261673e-06,
+      "loss": 0.8432,
+      "step": 22192
+    },
+    {
+      "epoch": 0.6095136085248963,
+      "grad_norm": 0.0,
+      "learning_rate": 6.986744997842766e-06,
+      "loss": 0.8755,
+      "step": 22193
+    },
+    {
+      "epoch": 0.6095410727527396,
+      "grad_norm": 0.0,
+      "learning_rate": 6.985896837265902e-06,
+      "loss": 0.9406,
+      "step": 22194
+    },
+    {
+      "epoch": 0.6095685369805828,
+      "grad_norm": 0.0,
+      "learning_rate": 6.985048700537786e-06,
+      "loss": 0.8479,
+      "step": 22195
+    },
+    {
+      "epoch": 0.609596001208426,
+      "grad_norm": 0.0,
+      "learning_rate": 6.98420058766514e-06,
+      "loss": 0.8335,
+      "step": 22196
+    },
+    {
+      "epoch": 0.6096234654362692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.983352498654661e-06,
+      "loss": 0.9192,
+      "step": 22197
+    },
+    {
+      "epoch": 0.6096509296641125,
+      "grad_norm": 0.0,
+      "learning_rate": 6.982504433513067e-06,
+      "loss": 0.9208,
+      "step": 22198
+    },
+    {
+      "epoch": 0.6096783938919558,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9816563922470674e-06,
+      "loss": 0.8446,
+      "step": 22199
+    },
+    {
+      "epoch": 0.6097058581197989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.980808374863368e-06,
+      "loss": 0.8927,
+      "step": 22200
+    },
+    {
+      "epoch": 0.6097333223476422,
+      "grad_norm": 0.0,
+      "learning_rate": 6.979960381368685e-06,
+      "loss": 0.8876,
+      "step": 22201
+    },
+    {
+      "epoch": 0.6097607865754854,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9791124117697285e-06,
+      "loss": 1.0196,
+      "step": 22202
+    },
+    {
+      "epoch": 0.6097882508033287,
+      "grad_norm": 0.0,
+      "learning_rate": 6.978264466073203e-06,
+      "loss": 0.9066,
+      "step": 22203
+    },
+    {
+      "epoch": 0.6098157150311719,
+      "grad_norm": 0.0,
+      "learning_rate": 6.977416544285818e-06,
+      "loss": 0.9608,
+      "step": 22204
+    },
+    {
+      "epoch": 0.6098431792590151,
+      "grad_norm": 0.0,
+      "learning_rate": 6.976568646414282e-06,
+      "loss": 0.8904,
+      "step": 22205
+    },
+    {
+      "epoch": 0.6098706434868584,
+      "grad_norm": 0.0,
+      "learning_rate": 6.975720772465311e-06,
+      "loss": 0.9186,
+      "step": 22206
+    },
+    {
+      "epoch": 0.6098981077147017,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9748729224456034e-06,
+      "loss": 0.7424,
+      "step": 22207
+    },
+    {
+      "epoch": 0.6099255719425448,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9740250963618805e-06,
+      "loss": 0.8818,
+      "step": 22208
+    },
+    {
+      "epoch": 0.6099530361703881,
+      "grad_norm": 0.0,
+      "learning_rate": 6.973177294220836e-06,
+      "loss": 0.8253,
+      "step": 22209
+    },
+    {
+      "epoch": 0.6099805003982313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9723295160291905e-06,
+      "loss": 0.8642,
+      "step": 22210
+    },
+    {
+      "epoch": 0.6100079646260745,
+      "grad_norm": 0.0,
+      "learning_rate": 6.971481761793641e-06,
+      "loss": 0.8034,
+      "step": 22211
+    },
+    {
+      "epoch": 0.6100354288539178,
+      "grad_norm": 0.0,
+      "learning_rate": 6.970634031520904e-06,
+      "loss": 0.9091,
+      "step": 22212
+    },
+    {
+      "epoch": 0.610062893081761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.969786325217682e-06,
+      "loss": 0.8661,
+      "step": 22213
+    },
+    {
+      "epoch": 0.6100903573096043,
+      "grad_norm": 0.0,
+      "learning_rate": 6.96893864289069e-06,
+      "loss": 0.8895,
+      "step": 22214
+    },
+    {
+      "epoch": 0.6101178215374474,
+      "grad_norm": 0.0,
+      "learning_rate": 6.968090984546627e-06,
+      "loss": 0.8386,
+      "step": 22215
+    },
+    {
+      "epoch": 0.6101452857652907,
+      "grad_norm": 0.0,
+      "learning_rate": 6.967243350192201e-06,
+      "loss": 0.9031,
+      "step": 22216
+    },
+    {
+      "epoch": 0.610172749993134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.96639573983412e-06,
+      "loss": 0.863,
+      "step": 22217
+    },
+    {
+      "epoch": 0.6102002142209771,
+      "grad_norm": 0.0,
+      "learning_rate": 6.965548153479092e-06,
+      "loss": 0.8424,
+      "step": 22218
+    },
+    {
+      "epoch": 0.6102276784488204,
+      "grad_norm": 0.0,
+      "learning_rate": 6.964700591133822e-06,
+      "loss": 0.801,
+      "step": 22219
+    },
+    {
+      "epoch": 0.6102551426766637,
+      "grad_norm": 0.0,
+      "learning_rate": 6.96385305280502e-06,
+      "loss": 0.882,
+      "step": 22220
+    },
+    {
+      "epoch": 0.6102826069045069,
+      "grad_norm": 0.0,
+      "learning_rate": 6.963005538499385e-06,
+      "loss": 0.8728,
+      "step": 22221
+    },
+    {
+      "epoch": 0.6103100711323501,
+      "grad_norm": 0.0,
+      "learning_rate": 6.962158048223628e-06,
+      "loss": 0.8156,
+      "step": 22222
+    },
+    {
+      "epoch": 0.6103375353601933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.961310581984452e-06,
+      "loss": 0.9015,
+      "step": 22223
+    },
+    {
+      "epoch": 0.6103649995880366,
+      "grad_norm": 0.0,
+      "learning_rate": 6.960463139788562e-06,
+      "loss": 0.7932,
+      "step": 22224
+    },
+    {
+      "epoch": 0.6103924638158799,
+      "grad_norm": 0.0,
+      "learning_rate": 6.959615721642669e-06,
+      "loss": 0.941,
+      "step": 22225
+    },
+    {
+      "epoch": 0.610419928043723,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9587683275534735e-06,
+      "loss": 0.834,
+      "step": 22226
+    },
+    {
+      "epoch": 0.6104473922715663,
+      "grad_norm": 0.0,
+      "learning_rate": 6.957920957527682e-06,
+      "loss": 0.7727,
+      "step": 22227
+    },
+    {
+      "epoch": 0.6104748564994095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.957073611571995e-06,
+      "loss": 0.7213,
+      "step": 22228
+    },
+    {
+      "epoch": 0.6105023207272527,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95622628969312e-06,
+      "loss": 0.8777,
+      "step": 22229
+    },
+    {
+      "epoch": 0.610529784955096,
+      "grad_norm": 0.0,
+      "learning_rate": 6.955378991897765e-06,
+      "loss": 0.8331,
+      "step": 22230
+    },
+    {
+      "epoch": 0.6105572491829392,
+      "grad_norm": 0.0,
+      "learning_rate": 6.954531718192627e-06,
+      "loss": 0.8891,
+      "step": 22231
+    },
+    {
+      "epoch": 0.6105847134107825,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95368446858442e-06,
+      "loss": 0.8341,
+      "step": 22232
+    },
+    {
+      "epoch": 0.6106121776386257,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9528372430798365e-06,
+      "loss": 0.8775,
+      "step": 22233
+    },
+    {
+      "epoch": 0.6106396418664689,
+      "grad_norm": 0.0,
+      "learning_rate": 6.951990041685587e-06,
+      "loss": 0.8289,
+      "step": 22234
+    },
+    {
+      "epoch": 0.6106671060943122,
+      "grad_norm": 0.0,
+      "learning_rate": 6.951142864408371e-06,
+      "loss": 0.8468,
+      "step": 22235
+    },
+    {
+      "epoch": 0.6106945703221554,
+      "grad_norm": 0.0,
+      "learning_rate": 6.950295711254893e-06,
+      "loss": 0.8034,
+      "step": 22236
+    },
+    {
+      "epoch": 0.6107220345499986,
+      "grad_norm": 0.0,
+      "learning_rate": 6.949448582231859e-06,
+      "loss": 0.8054,
+      "step": 22237
+    },
+    {
+      "epoch": 0.6107494987778419,
+      "grad_norm": 0.0,
+      "learning_rate": 6.948601477345971e-06,
+      "loss": 1.0151,
+      "step": 22238
+    },
+    {
+      "epoch": 0.6107769630056851,
+      "grad_norm": 0.0,
+      "learning_rate": 6.947754396603928e-06,
+      "loss": 0.9058,
+      "step": 22239
+    },
+    {
+      "epoch": 0.6108044272335283,
+      "grad_norm": 0.0,
+      "learning_rate": 6.946907340012434e-06,
+      "loss": 0.8315,
+      "step": 22240
+    },
+    {
+      "epoch": 0.6108318914613715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.946060307578191e-06,
+      "loss": 0.7525,
+      "step": 22241
+    },
+    {
+      "epoch": 0.6108593556892148,
+      "grad_norm": 0.0,
+      "learning_rate": 6.945213299307903e-06,
+      "loss": 0.7652,
+      "step": 22242
+    },
+    {
+      "epoch": 0.6108868199170581,
+      "grad_norm": 0.0,
+      "learning_rate": 6.944366315208271e-06,
+      "loss": 0.9465,
+      "step": 22243
+    },
+    {
+      "epoch": 0.6109142841449012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9435193552859995e-06,
+      "loss": 0.8752,
+      "step": 22244
+    },
+    {
+      "epoch": 0.6109417483727445,
+      "grad_norm": 0.0,
+      "learning_rate": 6.942672419547781e-06,
+      "loss": 0.89,
+      "step": 22245
+    },
+    {
+      "epoch": 0.6109692126005878,
+      "grad_norm": 0.0,
+      "learning_rate": 6.941825508000326e-06,
+      "loss": 0.8886,
+      "step": 22246
+    },
+    {
+      "epoch": 0.610996676828431,
+      "grad_norm": 0.0,
+      "learning_rate": 6.940978620650331e-06,
+      "loss": 0.8979,
+      "step": 22247
+    },
+    {
+      "epoch": 0.6110241410562742,
+      "grad_norm": 0.0,
+      "learning_rate": 6.940131757504497e-06,
+      "loss": 0.9412,
+      "step": 22248
+    },
+    {
+      "epoch": 0.6110516052841174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.93928491856953e-06,
+      "loss": 0.8644,
+      "step": 22249
+    },
+    {
+      "epoch": 0.6110790695119607,
+      "grad_norm": 0.0,
+      "learning_rate": 6.938438103852123e-06,
+      "loss": 0.8794,
+      "step": 22250
+    },
+    {
+      "epoch": 0.611106533739804,
+      "grad_norm": 0.0,
+      "learning_rate": 6.93759131335898e-06,
+      "loss": 0.7492,
+      "step": 22251
+    },
+    {
+      "epoch": 0.6111339979676471,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9367445470968005e-06,
+      "loss": 0.8387,
+      "step": 22252
+    },
+    {
+      "epoch": 0.6111614621954904,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9358978050722844e-06,
+      "loss": 0.8737,
+      "step": 22253
+    },
+    {
+      "epoch": 0.6111889264233336,
+      "grad_norm": 0.0,
+      "learning_rate": 6.935051087292133e-06,
+      "loss": 0.9154,
+      "step": 22254
+    },
+    {
+      "epoch": 0.6112163906511768,
+      "grad_norm": 0.0,
+      "learning_rate": 6.934204393763048e-06,
+      "loss": 0.8536,
+      "step": 22255
+    },
+    {
+      "epoch": 0.6112438548790201,
+      "grad_norm": 0.0,
+      "learning_rate": 6.93335772449172e-06,
+      "loss": 0.815,
+      "step": 22256
+    },
+    {
+      "epoch": 0.6112713191068633,
+      "grad_norm": 0.0,
+      "learning_rate": 6.932511079484854e-06,
+      "loss": 0.8438,
+      "step": 22257
+    },
+    {
+      "epoch": 0.6112987833347066,
+      "grad_norm": 0.0,
+      "learning_rate": 6.931664458749151e-06,
+      "loss": 0.9035,
+      "step": 22258
+    },
+    {
+      "epoch": 0.6113262475625498,
+      "grad_norm": 0.0,
+      "learning_rate": 6.930817862291306e-06,
+      "loss": 0.8591,
+      "step": 22259
+    },
+    {
+      "epoch": 0.611353711790393,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9299712901180175e-06,
+      "loss": 0.9186,
+      "step": 22260
+    },
+    {
+      "epoch": 0.6113811760182363,
+      "grad_norm": 0.0,
+      "learning_rate": 6.92912474223599e-06,
+      "loss": 0.9202,
+      "step": 22261
+    },
+    {
+      "epoch": 0.6114086402460794,
+      "grad_norm": 0.0,
+      "learning_rate": 6.928278218651913e-06,
+      "loss": 0.9716,
+      "step": 22262
+    },
+    {
+      "epoch": 0.6114361044739227,
+      "grad_norm": 0.0,
+      "learning_rate": 6.927431719372491e-06,
+      "loss": 0.9704,
+      "step": 22263
+    },
+    {
+      "epoch": 0.611463568701766,
+      "grad_norm": 0.0,
+      "learning_rate": 6.926585244404416e-06,
+      "loss": 0.8322,
+      "step": 22264
+    },
+    {
+      "epoch": 0.6114910329296092,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9257387937543905e-06,
+      "loss": 0.9073,
+      "step": 22265
+    },
+    {
+      "epoch": 0.6115184971574524,
+      "grad_norm": 0.0,
+      "learning_rate": 6.924892367429112e-06,
+      "loss": 0.9313,
+      "step": 22266
+    },
+    {
+      "epoch": 0.6115459613852956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.924045965435279e-06,
+      "loss": 0.961,
+      "step": 22267
+    },
+    {
+      "epoch": 0.6115734256131389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.923199587779581e-06,
+      "loss": 0.8714,
+      "step": 22268
+    },
+    {
+      "epoch": 0.6116008898409822,
+      "grad_norm": 0.0,
+      "learning_rate": 6.92235323446872e-06,
+      "loss": 0.8905,
+      "step": 22269
+    },
+    {
+      "epoch": 0.6116283540688253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.921506905509394e-06,
+      "loss": 0.7908,
+      "step": 22270
+    },
+    {
+      "epoch": 0.6116558182966686,
+      "grad_norm": 0.0,
+      "learning_rate": 6.920660600908297e-06,
+      "loss": 0.9006,
+      "step": 22271
+    },
+    {
+      "epoch": 0.6116832825245119,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9198143206721245e-06,
+      "loss": 0.8131,
+      "step": 22272
+    },
+    {
+      "epoch": 0.611710746752355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.918968064807579e-06,
+      "loss": 0.8958,
+      "step": 22273
+    },
+    {
+      "epoch": 0.6117382109801983,
+      "grad_norm": 0.0,
+      "learning_rate": 6.918121833321347e-06,
+      "loss": 0.8492,
+      "step": 22274
+    },
+    {
+      "epoch": 0.6117656752080415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.917275626220132e-06,
+      "loss": 0.9226,
+      "step": 22275
+    },
+    {
+      "epoch": 0.6117931394358848,
+      "grad_norm": 0.0,
+      "learning_rate": 6.916429443510624e-06,
+      "loss": 0.8899,
+      "step": 22276
+    },
+    {
+      "epoch": 0.611820603663728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.915583285199521e-06,
+      "loss": 0.9277,
+      "step": 22277
+    },
+    {
+      "epoch": 0.6118480678915712,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9147371512935205e-06,
+      "loss": 0.8153,
+      "step": 22278
+    },
+    {
+      "epoch": 0.6118755321194145,
+      "grad_norm": 0.0,
+      "learning_rate": 6.913891041799315e-06,
+      "loss": 0.8294,
+      "step": 22279
+    },
+    {
+      "epoch": 0.6119029963472576,
+      "grad_norm": 0.0,
+      "learning_rate": 6.913044956723595e-06,
+      "loss": 0.8485,
+      "step": 22280
+    },
+    {
+      "epoch": 0.6119304605751009,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9121988960730616e-06,
+      "loss": 0.8287,
+      "step": 22281
+    },
+    {
+      "epoch": 0.6119579248029442,
+      "grad_norm": 0.0,
+      "learning_rate": 6.911352859854405e-06,
+      "loss": 0.8534,
+      "step": 22282
+    },
+    {
+      "epoch": 0.6119853890307874,
+      "grad_norm": 0.0,
+      "learning_rate": 6.910506848074322e-06,
+      "loss": 0.7575,
+      "step": 22283
+    },
+    {
+      "epoch": 0.6120128532586306,
+      "grad_norm": 0.0,
+      "learning_rate": 6.909660860739505e-06,
+      "loss": 0.8047,
+      "step": 22284
+    },
+    {
+      "epoch": 0.6120403174864739,
+      "grad_norm": 0.0,
+      "learning_rate": 6.908814897856652e-06,
+      "loss": 0.8176,
+      "step": 22285
+    },
+    {
+      "epoch": 0.6120677817143171,
+      "grad_norm": 0.0,
+      "learning_rate": 6.907968959432449e-06,
+      "loss": 0.8074,
+      "step": 22286
+    },
+    {
+      "epoch": 0.6120952459421604,
+      "grad_norm": 0.0,
+      "learning_rate": 6.907123045473594e-06,
+      "loss": 0.836,
+      "step": 22287
+    },
+    {
+      "epoch": 0.6121227101700035,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9062771559867795e-06,
+      "loss": 0.8348,
+      "step": 22288
+    },
+    {
+      "epoch": 0.6121501743978468,
+      "grad_norm": 0.0,
+      "learning_rate": 6.905431290978697e-06,
+      "loss": 0.8728,
+      "step": 22289
+    },
+    {
+      "epoch": 0.6121776386256901,
+      "grad_norm": 0.0,
+      "learning_rate": 6.904585450456044e-06,
+      "loss": 0.8661,
+      "step": 22290
+    },
+    {
+      "epoch": 0.6122051028535332,
+      "grad_norm": 0.0,
+      "learning_rate": 6.903739634425511e-06,
+      "loss": 0.9395,
+      "step": 22291
+    },
+    {
+      "epoch": 0.6122325670813765,
+      "grad_norm": 0.0,
+      "learning_rate": 6.902893842893786e-06,
+      "loss": 0.8245,
+      "step": 22292
+    },
+    {
+      "epoch": 0.6122600313092197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.902048075867564e-06,
+      "loss": 0.889,
+      "step": 22293
+    },
+    {
+      "epoch": 0.612287495537063,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9012023333535405e-06,
+      "loss": 0.8623,
+      "step": 22294
+    },
+    {
+      "epoch": 0.6123149597649062,
+      "grad_norm": 0.0,
+      "learning_rate": 6.900356615358402e-06,
+      "loss": 0.8091,
+      "step": 22295
+    },
+    {
+      "epoch": 0.6123424239927494,
+      "grad_norm": 0.0,
+      "learning_rate": 6.899510921888843e-06,
+      "loss": 0.8629,
+      "step": 22296
+    },
+    {
+      "epoch": 0.6123698882205927,
+      "grad_norm": 0.0,
+      "learning_rate": 6.898665252951559e-06,
+      "loss": 0.8785,
+      "step": 22297
+    },
+    {
+      "epoch": 0.612397352448436,
+      "grad_norm": 0.0,
+      "learning_rate": 6.897819608553231e-06,
+      "loss": 0.7819,
+      "step": 22298
+    },
+    {
+      "epoch": 0.6124248166762791,
+      "grad_norm": 0.0,
+      "learning_rate": 6.896973988700559e-06,
+      "loss": 0.7099,
+      "step": 22299
+    },
+    {
+      "epoch": 0.6124522809041224,
+      "grad_norm": 0.0,
+      "learning_rate": 6.89612839340023e-06,
+      "loss": 0.8597,
+      "step": 22300
+    },
+    {
+      "epoch": 0.6124797451319656,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8952828226589345e-06,
+      "loss": 0.9418,
+      "step": 22301
+    },
+    {
+      "epoch": 0.6125072093598088,
+      "grad_norm": 0.0,
+      "learning_rate": 6.894437276483366e-06,
+      "loss": 0.8559,
+      "step": 22302
+    },
+    {
+      "epoch": 0.6125346735876521,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8935917548802145e-06,
+      "loss": 0.7871,
+      "step": 22303
+    },
+    {
+      "epoch": 0.6125621378154953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8927462578561645e-06,
+      "loss": 0.8898,
+      "step": 22304
+    },
+    {
+      "epoch": 0.6125896020433386,
+      "grad_norm": 0.0,
+      "learning_rate": 6.89190078541791e-06,
+      "loss": 0.7498,
+      "step": 22305
+    },
+    {
+      "epoch": 0.6126170662711817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.891055337572142e-06,
+      "loss": 0.9472,
+      "step": 22306
+    },
+    {
+      "epoch": 0.612644530499025,
+      "grad_norm": 0.0,
+      "learning_rate": 6.890209914325548e-06,
+      "loss": 0.8384,
+      "step": 22307
+    },
+    {
+      "epoch": 0.6126719947268683,
+      "grad_norm": 0.0,
+      "learning_rate": 6.88936451568482e-06,
+      "loss": 0.8421,
+      "step": 22308
+    },
+    {
+      "epoch": 0.6126994589547115,
+      "grad_norm": 0.0,
+      "learning_rate": 6.88851914165664e-06,
+      "loss": 0.8499,
+      "step": 22309
+    },
+    {
+      "epoch": 0.6127269231825547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8876737922477046e-06,
+      "loss": 0.7813,
+      "step": 22310
+    },
+    {
+      "epoch": 0.612754387410398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.886828467464699e-06,
+      "loss": 0.8871,
+      "step": 22311
+    },
+    {
+      "epoch": 0.6127818516382412,
+      "grad_norm": 0.0,
+      "learning_rate": 6.885983167314312e-06,
+      "loss": 0.8649,
+      "step": 22312
+    },
+    {
+      "epoch": 0.6128093158660844,
+      "grad_norm": 0.0,
+      "learning_rate": 6.885137891803233e-06,
+      "loss": 0.8622,
+      "step": 22313
+    },
+    {
+      "epoch": 0.6128367800939276,
+      "grad_norm": 0.0,
+      "learning_rate": 6.884292640938152e-06,
+      "loss": 0.8338,
+      "step": 22314
+    },
+    {
+      "epoch": 0.6128642443217709,
+      "grad_norm": 0.0,
+      "learning_rate": 6.883447414725753e-06,
+      "loss": 0.833,
+      "step": 22315
+    },
+    {
+      "epoch": 0.6128917085496142,
+      "grad_norm": 0.0,
+      "learning_rate": 6.882602213172725e-06,
+      "loss": 0.8072,
+      "step": 22316
+    },
+    {
+      "epoch": 0.6129191727774573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8817570362857545e-06,
+      "loss": 0.8242,
+      "step": 22317
+    },
+    {
+      "epoch": 0.6129466370053006,
+      "grad_norm": 0.0,
+      "learning_rate": 6.880911884071532e-06,
+      "loss": 0.839,
+      "step": 22318
+    },
+    {
+      "epoch": 0.6129741012331438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.880066756536741e-06,
+      "loss": 0.7983,
+      "step": 22319
+    },
+    {
+      "epoch": 0.613001565460987,
+      "grad_norm": 0.0,
+      "learning_rate": 6.879221653688074e-06,
+      "loss": 0.8473,
+      "step": 22320
+    },
+    {
+      "epoch": 0.6130290296888303,
+      "grad_norm": 0.0,
+      "learning_rate": 6.878376575532211e-06,
+      "loss": 0.8505,
+      "step": 22321
+    },
+    {
+      "epoch": 0.6130564939166735,
+      "grad_norm": 0.0,
+      "learning_rate": 6.877531522075841e-06,
+      "loss": 0.8395,
+      "step": 22322
+    },
+    {
+      "epoch": 0.6130839581445168,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8766864933256534e-06,
+      "loss": 0.8278,
+      "step": 22323
+    },
+    {
+      "epoch": 0.61311142237236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.87584148928833e-06,
+      "loss": 0.9346,
+      "step": 22324
+    },
+    {
+      "epoch": 0.6131388866002032,
+      "grad_norm": 0.0,
+      "learning_rate": 6.874996509970559e-06,
+      "loss": 0.7787,
+      "step": 22325
+    },
+    {
+      "epoch": 0.6131663508280465,
+      "grad_norm": 0.0,
+      "learning_rate": 6.874151555379032e-06,
+      "loss": 0.7965,
+      "step": 22326
+    },
+    {
+      "epoch": 0.6131938150558897,
+      "grad_norm": 0.0,
+      "learning_rate": 6.873306625520424e-06,
+      "loss": 0.9061,
+      "step": 22327
+    },
+    {
+      "epoch": 0.6132212792837329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.872461720401424e-06,
+      "loss": 0.8153,
+      "step": 22328
+    },
+    {
+      "epoch": 0.6132487435115762,
+      "grad_norm": 0.0,
+      "learning_rate": 6.871616840028718e-06,
+      "loss": 0.8897,
+      "step": 22329
+    },
+    {
+      "epoch": 0.6132762077394194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8707719844089945e-06,
+      "loss": 0.878,
+      "step": 22330
+    },
+    {
+      "epoch": 0.6133036719672627,
+      "grad_norm": 0.0,
+      "learning_rate": 6.869927153548932e-06,
+      "loss": 0.8156,
+      "step": 22331
+    },
+    {
+      "epoch": 0.6133311361951058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.869082347455224e-06,
+      "loss": 0.8962,
+      "step": 22332
+    },
+    {
+      "epoch": 0.6133586004229491,
+      "grad_norm": 0.0,
+      "learning_rate": 6.868237566134543e-06,
+      "loss": 0.7785,
+      "step": 22333
+    },
+    {
+      "epoch": 0.6133860646507924,
+      "grad_norm": 0.0,
+      "learning_rate": 6.867392809593581e-06,
+      "loss": 0.8933,
+      "step": 22334
+    },
+    {
+      "epoch": 0.6134135288786355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.866548077839023e-06,
+      "loss": 0.8743,
+      "step": 22335
+    },
+    {
+      "epoch": 0.6134409931064788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.865703370877548e-06,
+      "loss": 0.8596,
+      "step": 22336
+    },
+    {
+      "epoch": 0.6134684573343221,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8648586887158404e-06,
+      "loss": 0.8724,
+      "step": 22337
+    },
+    {
+      "epoch": 0.6134959215621653,
+      "grad_norm": 0.0,
+      "learning_rate": 6.864014031360591e-06,
+      "loss": 0.909,
+      "step": 22338
+    },
+    {
+      "epoch": 0.6135233857900085,
+      "grad_norm": 0.0,
+      "learning_rate": 6.863169398818474e-06,
+      "loss": 0.8443,
+      "step": 22339
+    },
+    {
+      "epoch": 0.6135508500178517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.862324791096176e-06,
+      "loss": 0.7797,
+      "step": 22340
+    },
+    {
+      "epoch": 0.613578314245695,
+      "grad_norm": 0.0,
+      "learning_rate": 6.861480208200377e-06,
+      "loss": 0.8639,
+      "step": 22341
+    },
+    {
+      "epoch": 0.6136057784735383,
+      "grad_norm": 0.0,
+      "learning_rate": 6.860635650137767e-06,
+      "loss": 0.946,
+      "step": 22342
+    },
+    {
+      "epoch": 0.6136332427013814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.85979111691502e-06,
+      "loss": 0.8585,
+      "step": 22343
+    },
+    {
+      "epoch": 0.6136607069292247,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8589466085388275e-06,
+      "loss": 0.8969,
+      "step": 22344
+    },
+    {
+      "epoch": 0.613688171157068,
+      "grad_norm": 0.0,
+      "learning_rate": 6.858102125015861e-06,
+      "loss": 0.8736,
+      "step": 22345
+    },
+    {
+      "epoch": 0.6137156353849111,
+      "grad_norm": 0.0,
+      "learning_rate": 6.85725766635281e-06,
+      "loss": 0.818,
+      "step": 22346
+    },
+    {
+      "epoch": 0.6137430996127544,
+      "grad_norm": 0.0,
+      "learning_rate": 6.856413232556351e-06,
+      "loss": 0.8329,
+      "step": 22347
+    },
+    {
+      "epoch": 0.6137705638405976,
+      "grad_norm": 0.0,
+      "learning_rate": 6.85556882363317e-06,
+      "loss": 0.9072,
+      "step": 22348
+    },
+    {
+      "epoch": 0.6137980280684409,
+      "grad_norm": 0.0,
+      "learning_rate": 6.854724439589944e-06,
+      "loss": 0.8309,
+      "step": 22349
+    },
+    {
+      "epoch": 0.6138254922962841,
+      "grad_norm": 0.0,
+      "learning_rate": 6.853880080433362e-06,
+      "loss": 0.8413,
+      "step": 22350
+    },
+    {
+      "epoch": 0.6138529565241273,
+      "grad_norm": 0.0,
+      "learning_rate": 6.853035746170098e-06,
+      "loss": 0.8631,
+      "step": 22351
+    },
+    {
+      "epoch": 0.6138804207519706,
+      "grad_norm": 0.0,
+      "learning_rate": 6.852191436806831e-06,
+      "loss": 0.8556,
+      "step": 22352
+    },
+    {
+      "epoch": 0.6139078849798137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8513471523502455e-06,
+      "loss": 0.9662,
+      "step": 22353
+    },
+    {
+      "epoch": 0.613935349207657,
+      "grad_norm": 0.0,
+      "learning_rate": 6.850502892807023e-06,
+      "loss": 0.8402,
+      "step": 22354
+    },
+    {
+      "epoch": 0.6139628134355003,
+      "grad_norm": 0.0,
+      "learning_rate": 6.849658658183839e-06,
+      "loss": 0.8732,
+      "step": 22355
+    },
+    {
+      "epoch": 0.6139902776633435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.84881444848738e-06,
+      "loss": 0.8544,
+      "step": 22356
+    },
+    {
+      "epoch": 0.6140177418911867,
+      "grad_norm": 0.0,
+      "learning_rate": 6.847970263724319e-06,
+      "loss": 0.938,
+      "step": 22357
+    },
+    {
+      "epoch": 0.61404520611903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.847126103901338e-06,
+      "loss": 0.9059,
+      "step": 22358
+    },
+    {
+      "epoch": 0.6140726703468732,
+      "grad_norm": 0.0,
+      "learning_rate": 6.846281969025117e-06,
+      "loss": 0.8535,
+      "step": 22359
+    },
+    {
+      "epoch": 0.6141001345747165,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8454378591023325e-06,
+      "loss": 0.7632,
+      "step": 22360
+    },
+    {
+      "epoch": 0.6141275988025596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.844593774139667e-06,
+      "loss": 0.8842,
+      "step": 22361
+    },
+    {
+      "epoch": 0.6141550630304029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.843749714143802e-06,
+      "loss": 0.7931,
+      "step": 22362
+    },
+    {
+      "epoch": 0.6141825272582462,
+      "grad_norm": 0.0,
+      "learning_rate": 6.84290567912141e-06,
+      "loss": 0.8862,
+      "step": 22363
+    },
+    {
+      "epoch": 0.6142099914860893,
+      "grad_norm": 0.0,
+      "learning_rate": 6.842061669079169e-06,
+      "loss": 0.8124,
+      "step": 22364
+    },
+    {
+      "epoch": 0.6142374557139326,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8412176840237584e-06,
+      "loss": 0.8231,
+      "step": 22365
+    },
+    {
+      "epoch": 0.6142649199417758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8403737239618595e-06,
+      "loss": 0.9072,
+      "step": 22366
+    },
+    {
+      "epoch": 0.6142923841696191,
+      "grad_norm": 0.0,
+      "learning_rate": 6.839529788900146e-06,
+      "loss": 0.7635,
+      "step": 22367
+    },
+    {
+      "epoch": 0.6143198483974623,
+      "grad_norm": 0.0,
+      "learning_rate": 6.838685878845302e-06,
+      "loss": 0.797,
+      "step": 22368
+    },
+    {
+      "epoch": 0.6143473126253055,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8378419938039945e-06,
+      "loss": 0.8269,
+      "step": 22369
+    },
+    {
+      "epoch": 0.6143747768531488,
+      "grad_norm": 0.0,
+      "learning_rate": 6.836998133782908e-06,
+      "loss": 0.8739,
+      "step": 22370
+    },
+    {
+      "epoch": 0.6144022410809921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8361542987887155e-06,
+      "loss": 0.9327,
+      "step": 22371
+    },
+    {
+      "epoch": 0.6144297053088352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.835310488828096e-06,
+      "loss": 0.7885,
+      "step": 22372
+    },
+    {
+      "epoch": 0.6144571695366785,
+      "grad_norm": 0.0,
+      "learning_rate": 6.834466703907729e-06,
+      "loss": 0.8407,
+      "step": 22373
+    },
+    {
+      "epoch": 0.6144846337645217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.833622944034286e-06,
+      "loss": 0.9272,
+      "step": 22374
+    },
+    {
+      "epoch": 0.614512097992365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.832779209214445e-06,
+      "loss": 0.8104,
+      "step": 22375
+    },
+    {
+      "epoch": 0.6145395622202082,
+      "grad_norm": 0.0,
+      "learning_rate": 6.83193549945488e-06,
+      "loss": 0.8181,
+      "step": 22376
+    },
+    {
+      "epoch": 0.6145670264480514,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8310918147622685e-06,
+      "loss": 0.8461,
+      "step": 22377
+    },
+    {
+      "epoch": 0.6145944906758947,
+      "grad_norm": 0.0,
+      "learning_rate": 6.830248155143289e-06,
+      "loss": 0.8178,
+      "step": 22378
+    },
+    {
+      "epoch": 0.6146219549037378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.829404520604614e-06,
+      "loss": 0.9275,
+      "step": 22379
+    },
+    {
+      "epoch": 0.6146494191315811,
+      "grad_norm": 0.0,
+      "learning_rate": 6.828560911152917e-06,
+      "loss": 0.8428,
+      "step": 22380
+    },
+    {
+      "epoch": 0.6146768833594244,
+      "grad_norm": 0.0,
+      "learning_rate": 6.827717326794873e-06,
+      "loss": 0.8678,
+      "step": 22381
+    },
+    {
+      "epoch": 0.6147043475872676,
+      "grad_norm": 0.0,
+      "learning_rate": 6.826873767537161e-06,
+      "loss": 0.9093,
+      "step": 22382
+    },
+    {
+      "epoch": 0.6147318118151108,
+      "grad_norm": 0.0,
+      "learning_rate": 6.826030233386449e-06,
+      "loss": 0.8627,
+      "step": 22383
+    },
+    {
+      "epoch": 0.6147592760429541,
+      "grad_norm": 0.0,
+      "learning_rate": 6.825186724349417e-06,
+      "loss": 0.8946,
+      "step": 22384
+    },
+    {
+      "epoch": 0.6147867402707973,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8243432404327425e-06,
+      "loss": 0.71,
+      "step": 22385
+    },
+    {
+      "epoch": 0.6148142044986405,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8234997816430885e-06,
+      "loss": 0.8476,
+      "step": 22386
+    },
+    {
+      "epoch": 0.6148416687264837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.822656347987137e-06,
+      "loss": 0.8077,
+      "step": 22387
+    },
+    {
+      "epoch": 0.614869132954327,
+      "grad_norm": 0.0,
+      "learning_rate": 6.821812939471559e-06,
+      "loss": 0.7957,
+      "step": 22388
+    },
+    {
+      "epoch": 0.6148965971821703,
+      "grad_norm": 0.0,
+      "learning_rate": 6.820969556103025e-06,
+      "loss": 0.8857,
+      "step": 22389
+    },
+    {
+      "epoch": 0.6149240614100134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.820126197888216e-06,
+      "loss": 0.8339,
+      "step": 22390
+    },
+    {
+      "epoch": 0.6149515256378567,
+      "grad_norm": 0.0,
+      "learning_rate": 6.819282864833802e-06,
+      "loss": 0.9639,
+      "step": 22391
+    },
+    {
+      "epoch": 0.6149789898656999,
+      "grad_norm": 0.0,
+      "learning_rate": 6.818439556946449e-06,
+      "loss": 0.9525,
+      "step": 22392
+    },
+    {
+      "epoch": 0.6150064540935432,
+      "grad_norm": 0.0,
+      "learning_rate": 6.817596274232835e-06,
+      "loss": 0.7408,
+      "step": 22393
+    },
+    {
+      "epoch": 0.6150339183213864,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8167530166996335e-06,
+      "loss": 0.806,
+      "step": 22394
+    },
+    {
+      "epoch": 0.6150613825492296,
+      "grad_norm": 0.0,
+      "learning_rate": 6.815909784353514e-06,
+      "loss": 0.7954,
+      "step": 22395
+    },
+    {
+      "epoch": 0.6150888467770729,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8150665772011495e-06,
+      "loss": 0.9143,
+      "step": 22396
+    },
+    {
+      "epoch": 0.6151163110049161,
+      "grad_norm": 0.0,
+      "learning_rate": 6.814223395249215e-06,
+      "loss": 0.8801,
+      "step": 22397
+    },
+    {
+      "epoch": 0.6151437752327593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.813380238504377e-06,
+      "loss": 0.9325,
+      "step": 22398
+    },
+    {
+      "epoch": 0.6151712394606026,
+      "grad_norm": 0.0,
+      "learning_rate": 6.81253710697331e-06,
+      "loss": 0.916,
+      "step": 22399
+    },
+    {
+      "epoch": 0.6151987036884458,
+      "grad_norm": 0.0,
+      "learning_rate": 6.811694000662682e-06,
+      "loss": 0.8553,
+      "step": 22400
+    },
+    {
+      "epoch": 0.615226167916289,
+      "grad_norm": 0.0,
+      "learning_rate": 6.810850919579166e-06,
+      "loss": 0.832,
+      "step": 22401
+    },
+    {
+      "epoch": 0.6152536321441323,
+      "grad_norm": 0.0,
+      "learning_rate": 6.810007863729435e-06,
+      "loss": 0.8487,
+      "step": 22402
+    },
+    {
+      "epoch": 0.6152810963719755,
+      "grad_norm": 0.0,
+      "learning_rate": 6.80916483312016e-06,
+      "loss": 0.7756,
+      "step": 22403
+    },
+    {
+      "epoch": 0.6153085605998188,
+      "grad_norm": 0.0,
+      "learning_rate": 6.808321827758004e-06,
+      "loss": 0.7485,
+      "step": 22404
+    },
+    {
+      "epoch": 0.6153360248276619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.807478847649643e-06,
+      "loss": 0.8165,
+      "step": 22405
+    },
+    {
+      "epoch": 0.6153634890555052,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8066358928017475e-06,
+      "loss": 0.8375,
+      "step": 22406
+    },
+    {
+      "epoch": 0.6153909532833485,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8057929632209825e-06,
+      "loss": 0.8898,
+      "step": 22407
+    },
+    {
+      "epoch": 0.6154184175111916,
+      "grad_norm": 0.0,
+      "learning_rate": 6.804950058914023e-06,
+      "loss": 0.8248,
+      "step": 22408
+    },
+    {
+      "epoch": 0.6154458817390349,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8041071798875394e-06,
+      "loss": 0.9098,
+      "step": 22409
+    },
+    {
+      "epoch": 0.6154733459668782,
+      "grad_norm": 0.0,
+      "learning_rate": 6.803264326148194e-06,
+      "loss": 0.9785,
+      "step": 22410
+    },
+    {
+      "epoch": 0.6155008101947214,
+      "grad_norm": 0.0,
+      "learning_rate": 6.80242149770266e-06,
+      "loss": 0.8846,
+      "step": 22411
+    },
+    {
+      "epoch": 0.6155282744225646,
+      "grad_norm": 0.0,
+      "learning_rate": 6.801578694557605e-06,
+      "loss": 0.7648,
+      "step": 22412
+    },
+    {
+      "epoch": 0.6155557386504078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.800735916719699e-06,
+      "loss": 0.7483,
+      "step": 22413
+    },
+    {
+      "epoch": 0.6155832028782511,
+      "grad_norm": 0.0,
+      "learning_rate": 6.799893164195611e-06,
+      "loss": 0.8414,
+      "step": 22414
+    },
+    {
+      "epoch": 0.6156106671060944,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7990504369920095e-06,
+      "loss": 0.9058,
+      "step": 22415
+    },
+    {
+      "epoch": 0.6156381313339375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.798207735115558e-06,
+      "loss": 0.9709,
+      "step": 22416
+    },
+    {
+      "epoch": 0.6156655955617808,
+      "grad_norm": 0.0,
+      "learning_rate": 6.797365058572927e-06,
+      "loss": 0.8462,
+      "step": 22417
+    },
+    {
+      "epoch": 0.615693059789624,
+      "grad_norm": 0.0,
+      "learning_rate": 6.796522407370785e-06,
+      "loss": 0.8701,
+      "step": 22418
+    },
+    {
+      "epoch": 0.6157205240174672,
+      "grad_norm": 0.0,
+      "learning_rate": 6.795679781515797e-06,
+      "loss": 0.8329,
+      "step": 22419
+    },
+    {
+      "epoch": 0.6157479882453105,
+      "grad_norm": 0.0,
+      "learning_rate": 6.794837181014633e-06,
+      "loss": 0.8845,
+      "step": 22420
+    },
+    {
+      "epoch": 0.6157754524731537,
+      "grad_norm": 0.0,
+      "learning_rate": 6.793994605873961e-06,
+      "loss": 0.8078,
+      "step": 22421
+    },
+    {
+      "epoch": 0.615802916700997,
+      "grad_norm": 0.0,
+      "learning_rate": 6.793152056100443e-06,
+      "loss": 0.8362,
+      "step": 22422
+    },
+    {
+      "epoch": 0.6158303809288402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.79230953170075e-06,
+      "loss": 0.8905,
+      "step": 22423
+    },
+    {
+      "epoch": 0.6158578451566834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.791467032681544e-06,
+      "loss": 0.8704,
+      "step": 22424
+    },
+    {
+      "epoch": 0.6158853093845267,
+      "grad_norm": 0.0,
+      "learning_rate": 6.790624559049493e-06,
+      "loss": 0.9525,
+      "step": 22425
+    },
+    {
+      "epoch": 0.6159127736123698,
+      "grad_norm": 0.0,
+      "learning_rate": 6.789782110811267e-06,
+      "loss": 0.8061,
+      "step": 22426
+    },
+    {
+      "epoch": 0.6159402378402131,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7889396879735305e-06,
+      "loss": 0.8739,
+      "step": 22427
+    },
+    {
+      "epoch": 0.6159677020680564,
+      "grad_norm": 0.0,
+      "learning_rate": 6.788097290542942e-06,
+      "loss": 0.873,
+      "step": 22428
+    },
+    {
+      "epoch": 0.6159951662958996,
+      "grad_norm": 0.0,
+      "learning_rate": 6.787254918526171e-06,
+      "loss": 0.9089,
+      "step": 22429
+    },
+    {
+      "epoch": 0.6160226305237428,
+      "grad_norm": 0.0,
+      "learning_rate": 6.786412571929888e-06,
+      "loss": 0.8438,
+      "step": 22430
+    },
+    {
+      "epoch": 0.616050094751586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.785570250760751e-06,
+      "loss": 0.8352,
+      "step": 22431
+    },
+    {
+      "epoch": 0.6160775589794293,
+      "grad_norm": 0.0,
+      "learning_rate": 6.784727955025431e-06,
+      "loss": 0.8258,
+      "step": 22432
+    },
+    {
+      "epoch": 0.6161050232072726,
+      "grad_norm": 0.0,
+      "learning_rate": 6.783885684730584e-06,
+      "loss": 0.9249,
+      "step": 22433
+    },
+    {
+      "epoch": 0.6161324874351157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.78304343988288e-06,
+      "loss": 0.8154,
+      "step": 22434
+    },
+    {
+      "epoch": 0.616159951662959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.782201220488985e-06,
+      "loss": 0.838,
+      "step": 22435
+    },
+    {
+      "epoch": 0.6161874158908023,
+      "grad_norm": 0.0,
+      "learning_rate": 6.781359026555558e-06,
+      "loss": 0.8386,
+      "step": 22436
+    },
+    {
+      "epoch": 0.6162148801186454,
+      "grad_norm": 0.0,
+      "learning_rate": 6.780516858089266e-06,
+      "loss": 0.8856,
+      "step": 22437
+    },
+    {
+      "epoch": 0.6162423443464887,
+      "grad_norm": 0.0,
+      "learning_rate": 6.779674715096773e-06,
+      "loss": 0.8141,
+      "step": 22438
+    },
+    {
+      "epoch": 0.6162698085743319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.778832597584741e-06,
+      "loss": 0.8125,
+      "step": 22439
+    },
+    {
+      "epoch": 0.6162972728021752,
+      "grad_norm": 0.0,
+      "learning_rate": 6.777990505559832e-06,
+      "loss": 0.9581,
+      "step": 22440
+    },
+    {
+      "epoch": 0.6163247370300184,
+      "grad_norm": 0.0,
+      "learning_rate": 6.777148439028709e-06,
+      "loss": 0.9598,
+      "step": 22441
+    },
+    {
+      "epoch": 0.6163522012578616,
+      "grad_norm": 0.0,
+      "learning_rate": 6.776306397998038e-06,
+      "loss": 0.7896,
+      "step": 22442
+    },
+    {
+      "epoch": 0.6163796654857049,
+      "grad_norm": 0.0,
+      "learning_rate": 6.775464382474478e-06,
+      "loss": 0.7771,
+      "step": 22443
+    },
+    {
+      "epoch": 0.616407129713548,
+      "grad_norm": 0.0,
+      "learning_rate": 6.774622392464697e-06,
+      "loss": 0.7946,
+      "step": 22444
+    },
+    {
+      "epoch": 0.6164345939413913,
+      "grad_norm": 0.0,
+      "learning_rate": 6.77378042797535e-06,
+      "loss": 0.8839,
+      "step": 22445
+    },
+    {
+      "epoch": 0.6164620581692346,
+      "grad_norm": 0.0,
+      "learning_rate": 6.772938489013099e-06,
+      "loss": 0.841,
+      "step": 22446
+    },
+    {
+      "epoch": 0.6164895223970778,
+      "grad_norm": 0.0,
+      "learning_rate": 6.772096575584613e-06,
+      "loss": 0.7364,
+      "step": 22447
+    },
+    {
+      "epoch": 0.616516986624921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.771254687696547e-06,
+      "loss": 0.8371,
+      "step": 22448
+    },
+    {
+      "epoch": 0.6165444508527643,
+      "grad_norm": 0.0,
+      "learning_rate": 6.770412825355563e-06,
+      "loss": 0.9053,
+      "step": 22449
+    },
+    {
+      "epoch": 0.6165719150806075,
+      "grad_norm": 0.0,
+      "learning_rate": 6.76957098856833e-06,
+      "loss": 0.8559,
+      "step": 22450
+    },
+    {
+      "epoch": 0.6165993793084508,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7687291773414995e-06,
+      "loss": 0.85,
+      "step": 22451
+    },
+    {
+      "epoch": 0.6166268435362939,
+      "grad_norm": 0.0,
+      "learning_rate": 6.767887391681733e-06,
+      "loss": 0.7262,
+      "step": 22452
+    },
+    {
+      "epoch": 0.6166543077641372,
+      "grad_norm": 0.0,
+      "learning_rate": 6.767045631595694e-06,
+      "loss": 0.8795,
+      "step": 22453
+    },
+    {
+      "epoch": 0.6166817719919805,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7662038970900444e-06,
+      "loss": 0.9442,
+      "step": 22454
+    },
+    {
+      "epoch": 0.6167092362198237,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7653621881714404e-06,
+      "loss": 0.8575,
+      "step": 22455
+    },
+    {
+      "epoch": 0.6167367004476669,
+      "grad_norm": 0.0,
+      "learning_rate": 6.764520504846548e-06,
+      "loss": 0.813,
+      "step": 22456
+    },
+    {
+      "epoch": 0.6167641646755101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.763678847122019e-06,
+      "loss": 0.8331,
+      "step": 22457
+    },
+    {
+      "epoch": 0.6167916289033534,
+      "grad_norm": 0.0,
+      "learning_rate": 6.762837215004514e-06,
+      "loss": 0.8167,
+      "step": 22458
+    },
+    {
+      "epoch": 0.6168190931311966,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7619956085007e-06,
+      "loss": 0.9055,
+      "step": 22459
+    },
+    {
+      "epoch": 0.6168465573590398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.761154027617227e-06,
+      "loss": 0.7706,
+      "step": 22460
+    },
+    {
+      "epoch": 0.6168740215868831,
+      "grad_norm": 0.0,
+      "learning_rate": 6.76031247236076e-06,
+      "loss": 0.8357,
+      "step": 22461
+    },
+    {
+      "epoch": 0.6169014858147264,
+      "grad_norm": 0.0,
+      "learning_rate": 6.759470942737958e-06,
+      "loss": 0.9476,
+      "step": 22462
+    },
+    {
+      "epoch": 0.6169289500425695,
+      "grad_norm": 0.0,
+      "learning_rate": 6.758629438755475e-06,
+      "loss": 0.8106,
+      "step": 22463
+    },
+    {
+      "epoch": 0.6169564142704128,
+      "grad_norm": 0.0,
+      "learning_rate": 6.75778796041997e-06,
+      "loss": 0.7609,
+      "step": 22464
+    },
+    {
+      "epoch": 0.616983878498256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.756946507738103e-06,
+      "loss": 0.8778,
+      "step": 22465
+    },
+    {
+      "epoch": 0.6170113427260993,
+      "grad_norm": 0.0,
+      "learning_rate": 6.756105080716533e-06,
+      "loss": 0.9303,
+      "step": 22466
+    },
+    {
+      "epoch": 0.6170388069539425,
+      "grad_norm": 0.0,
+      "learning_rate": 6.755263679361914e-06,
+      "loss": 0.7808,
+      "step": 22467
+    },
+    {
+      "epoch": 0.6170662711817857,
+      "grad_norm": 0.0,
+      "learning_rate": 6.75442230368091e-06,
+      "loss": 0.8786,
+      "step": 22468
+    },
+    {
+      "epoch": 0.617093735409629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.75358095368017e-06,
+      "loss": 0.844,
+      "step": 22469
+    },
+    {
+      "epoch": 0.6171211996374721,
+      "grad_norm": 0.0,
+      "learning_rate": 6.752739629366354e-06,
+      "loss": 0.7795,
+      "step": 22470
+    },
+    {
+      "epoch": 0.6171486638653154,
+      "grad_norm": 0.0,
+      "learning_rate": 6.751898330746123e-06,
+      "loss": 0.8891,
+      "step": 22471
+    },
+    {
+      "epoch": 0.6171761280931587,
+      "grad_norm": 0.0,
+      "learning_rate": 6.751057057826128e-06,
+      "loss": 0.8461,
+      "step": 22472
+    },
+    {
+      "epoch": 0.6172035923210019,
+      "grad_norm": 0.0,
+      "learning_rate": 6.750215810613028e-06,
+      "loss": 0.8373,
+      "step": 22473
+    },
+    {
+      "epoch": 0.6172310565488451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.749374589113484e-06,
+      "loss": 0.8493,
+      "step": 22474
+    },
+    {
+      "epoch": 0.6172585207766884,
+      "grad_norm": 0.0,
+      "learning_rate": 6.748533393334145e-06,
+      "loss": 0.856,
+      "step": 22475
+    },
+    {
+      "epoch": 0.6172859850045316,
+      "grad_norm": 0.0,
+      "learning_rate": 6.747692223281667e-06,
+      "loss": 0.9169,
+      "step": 22476
+    },
+    {
+      "epoch": 0.6173134492323749,
+      "grad_norm": 0.0,
+      "learning_rate": 6.746851078962709e-06,
+      "loss": 0.7681,
+      "step": 22477
+    },
+    {
+      "epoch": 0.617340913460218,
+      "grad_norm": 0.0,
+      "learning_rate": 6.746009960383928e-06,
+      "loss": 0.8637,
+      "step": 22478
+    },
+    {
+      "epoch": 0.6173683776880613,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7451688675519725e-06,
+      "loss": 0.923,
+      "step": 22479
+    },
+    {
+      "epoch": 0.6173958419159046,
+      "grad_norm": 0.0,
+      "learning_rate": 6.744327800473507e-06,
+      "loss": 0.8086,
+      "step": 22480
+    },
+    {
+      "epoch": 0.6174233061437477,
+      "grad_norm": 0.0,
+      "learning_rate": 6.743486759155177e-06,
+      "loss": 0.8441,
+      "step": 22481
+    },
+    {
+      "epoch": 0.617450770371591,
+      "grad_norm": 0.0,
+      "learning_rate": 6.742645743603641e-06,
+      "loss": 0.8276,
+      "step": 22482
+    },
+    {
+      "epoch": 0.6174782345994342,
+      "grad_norm": 0.0,
+      "learning_rate": 6.741804753825556e-06,
+      "loss": 0.9091,
+      "step": 22483
+    },
+    {
+      "epoch": 0.6175056988272775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.74096378982757e-06,
+      "loss": 0.9026,
+      "step": 22484
+    },
+    {
+      "epoch": 0.6175331630551207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.740122851616344e-06,
+      "loss": 0.9533,
+      "step": 22485
+    },
+    {
+      "epoch": 0.6175606272829639,
+      "grad_norm": 0.0,
+      "learning_rate": 6.739281939198531e-06,
+      "loss": 0.9165,
+      "step": 22486
+    },
+    {
+      "epoch": 0.6175880915108072,
+      "grad_norm": 0.0,
+      "learning_rate": 6.738441052580781e-06,
+      "loss": 0.8305,
+      "step": 22487
+    },
+    {
+      "epoch": 0.6176155557386505,
+      "grad_norm": 0.0,
+      "learning_rate": 6.737600191769746e-06,
+      "loss": 0.9115,
+      "step": 22488
+    },
+    {
+      "epoch": 0.6176430199664936,
+      "grad_norm": 0.0,
+      "learning_rate": 6.736759356772082e-06,
+      "loss": 0.8296,
+      "step": 22489
+    },
+    {
+      "epoch": 0.6176704841943369,
+      "grad_norm": 0.0,
+      "learning_rate": 6.735918547594444e-06,
+      "loss": 0.9272,
+      "step": 22490
+    },
+    {
+      "epoch": 0.6176979484221801,
+      "grad_norm": 0.0,
+      "learning_rate": 6.735077764243486e-06,
+      "loss": 0.7829,
+      "step": 22491
+    },
+    {
+      "epoch": 0.6177254126500233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.734237006725853e-06,
+      "loss": 0.7857,
+      "step": 22492
+    },
+    {
+      "epoch": 0.6177528768778666,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7333962750482e-06,
+      "loss": 0.9046,
+      "step": 22493
+    },
+    {
+      "epoch": 0.6177803411057098,
+      "grad_norm": 0.0,
+      "learning_rate": 6.732555569217184e-06,
+      "loss": 0.8744,
+      "step": 22494
+    },
+    {
+      "epoch": 0.6178078053335531,
+      "grad_norm": 0.0,
+      "learning_rate": 6.731714889239455e-06,
+      "loss": 0.8393,
+      "step": 22495
+    },
+    {
+      "epoch": 0.6178352695613962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.73087423512166e-06,
+      "loss": 0.8266,
+      "step": 22496
+    },
+    {
+      "epoch": 0.6178627337892395,
+      "grad_norm": 0.0,
+      "learning_rate": 6.730033606870461e-06,
+      "loss": 0.8818,
+      "step": 22497
+    },
+    {
+      "epoch": 0.6178901980170828,
+      "grad_norm": 0.0,
+      "learning_rate": 6.729193004492498e-06,
+      "loss": 0.8838,
+      "step": 22498
+    },
+    {
+      "epoch": 0.617917662244926,
+      "grad_norm": 0.0,
+      "learning_rate": 6.728352427994429e-06,
+      "loss": 0.8356,
+      "step": 22499
+    },
+    {
+      "epoch": 0.6179451264727692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7275118773829e-06,
+      "loss": 0.699,
+      "step": 22500
+    },
+    {
+      "epoch": 0.6179725907006125,
+      "grad_norm": 0.0,
+      "learning_rate": 6.726671352664564e-06,
+      "loss": 0.8363,
+      "step": 22501
+    },
+    {
+      "epoch": 0.6180000549284557,
+      "grad_norm": 0.0,
+      "learning_rate": 6.725830853846076e-06,
+      "loss": 0.9431,
+      "step": 22502
+    },
+    {
+      "epoch": 0.6180275191562989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7249903809340845e-06,
+      "loss": 0.9106,
+      "step": 22503
+    },
+    {
+      "epoch": 0.6180549833841421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.724149933935233e-06,
+      "loss": 0.8206,
+      "step": 22504
+    },
+    {
+      "epoch": 0.6180824476119854,
+      "grad_norm": 0.0,
+      "learning_rate": 6.723309512856175e-06,
+      "loss": 0.8487,
+      "step": 22505
+    },
+    {
+      "epoch": 0.6181099118398287,
+      "grad_norm": 0.0,
+      "learning_rate": 6.722469117703565e-06,
+      "loss": 0.8422,
+      "step": 22506
+    },
+    {
+      "epoch": 0.6181373760676718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.721628748484048e-06,
+      "loss": 0.884,
+      "step": 22507
+    },
+    {
+      "epoch": 0.6181648402955151,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7207884052042724e-06,
+      "loss": 0.8752,
+      "step": 22508
+    },
+    {
+      "epoch": 0.6181923045233583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.719948087870893e-06,
+      "loss": 0.9061,
+      "step": 22509
+    },
+    {
+      "epoch": 0.6182197687512015,
+      "grad_norm": 0.0,
+      "learning_rate": 6.719107796490551e-06,
+      "loss": 0.7812,
+      "step": 22510
+    },
+    {
+      "epoch": 0.6182472329790448,
+      "grad_norm": 0.0,
+      "learning_rate": 6.718267531069903e-06,
+      "loss": 0.8467,
+      "step": 22511
+    },
+    {
+      "epoch": 0.618274697206888,
+      "grad_norm": 0.0,
+      "learning_rate": 6.71742729161559e-06,
+      "loss": 0.8785,
+      "step": 22512
+    },
+    {
+      "epoch": 0.6183021614347313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.716587078134263e-06,
+      "loss": 0.9199,
+      "step": 22513
+    },
+    {
+      "epoch": 0.6183296256625745,
+      "grad_norm": 0.0,
+      "learning_rate": 6.715746890632575e-06,
+      "loss": 0.8609,
+      "step": 22514
+    },
+    {
+      "epoch": 0.6183570898904177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.71490672911717e-06,
+      "loss": 0.9191,
+      "step": 22515
+    },
+    {
+      "epoch": 0.618384554118261,
+      "grad_norm": 0.0,
+      "learning_rate": 6.714066593594693e-06,
+      "loss": 0.8923,
+      "step": 22516
+    },
+    {
+      "epoch": 0.6184120183461042,
+      "grad_norm": 0.0,
+      "learning_rate": 6.713226484071793e-06,
+      "loss": 0.8662,
+      "step": 22517
+    },
+    {
+      "epoch": 0.6184394825739474,
+      "grad_norm": 0.0,
+      "learning_rate": 6.71238640055512e-06,
+      "loss": 0.8889,
+      "step": 22518
+    },
+    {
+      "epoch": 0.6184669468017907,
+      "grad_norm": 0.0,
+      "learning_rate": 6.711546343051318e-06,
+      "loss": 0.8321,
+      "step": 22519
+    },
+    {
+      "epoch": 0.6184944110296339,
+      "grad_norm": 0.0,
+      "learning_rate": 6.710706311567035e-06,
+      "loss": 0.9342,
+      "step": 22520
+    },
+    {
+      "epoch": 0.6185218752574771,
+      "grad_norm": 0.0,
+      "learning_rate": 6.709866306108921e-06,
+      "loss": 0.8492,
+      "step": 22521
+    },
+    {
+      "epoch": 0.6185493394853203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.709026326683615e-06,
+      "loss": 0.8736,
+      "step": 22522
+    },
+    {
+      "epoch": 0.6185768037131636,
+      "grad_norm": 0.0,
+      "learning_rate": 6.708186373297771e-06,
+      "loss": 0.9413,
+      "step": 22523
+    },
+    {
+      "epoch": 0.6186042679410069,
+      "grad_norm": 0.0,
+      "learning_rate": 6.70734644595803e-06,
+      "loss": 0.8917,
+      "step": 22524
+    },
+    {
+      "epoch": 0.61863173216885,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7065065446710365e-06,
+      "loss": 0.8612,
+      "step": 22525
+    },
+    {
+      "epoch": 0.6186591963966933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.705666669443444e-06,
+      "loss": 0.8773,
+      "step": 22526
+    },
+    {
+      "epoch": 0.6186866606245366,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7048268202818935e-06,
+      "loss": 0.8417,
+      "step": 22527
+    },
+    {
+      "epoch": 0.6187141248523798,
+      "grad_norm": 0.0,
+      "learning_rate": 6.703986997193026e-06,
+      "loss": 0.8597,
+      "step": 22528
+    },
+    {
+      "epoch": 0.618741589080223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.703147200183489e-06,
+      "loss": 0.8406,
+      "step": 22529
+    },
+    {
+      "epoch": 0.6187690533080662,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7023074292599315e-06,
+      "loss": 0.8373,
+      "step": 22530
+    },
+    {
+      "epoch": 0.6187965175359095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.701467684428993e-06,
+      "loss": 0.9199,
+      "step": 22531
+    },
+    {
+      "epoch": 0.6188239817637528,
+      "grad_norm": 0.0,
+      "learning_rate": 6.700627965697319e-06,
+      "loss": 0.8648,
+      "step": 22532
+    },
+    {
+      "epoch": 0.6188514459915959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.69978827307156e-06,
+      "loss": 0.7589,
+      "step": 22533
+    },
+    {
+      "epoch": 0.6188789102194392,
+      "grad_norm": 0.0,
+      "learning_rate": 6.698948606558351e-06,
+      "loss": 0.7931,
+      "step": 22534
+    },
+    {
+      "epoch": 0.6189063744472824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.69810896616434e-06,
+      "loss": 0.8589,
+      "step": 22535
+    },
+    {
+      "epoch": 0.6189338386751256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.697269351896167e-06,
+      "loss": 0.9524,
+      "step": 22536
+    },
+    {
+      "epoch": 0.6189613029029689,
+      "grad_norm": 0.0,
+      "learning_rate": 6.696429763760479e-06,
+      "loss": 0.8215,
+      "step": 22537
+    },
+    {
+      "epoch": 0.6189887671308121,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6955902017639215e-06,
+      "loss": 0.8377,
+      "step": 22538
+    },
+    {
+      "epoch": 0.6190162313586554,
+      "grad_norm": 0.0,
+      "learning_rate": 6.694750665913137e-06,
+      "loss": 0.9064,
+      "step": 22539
+    },
+    {
+      "epoch": 0.6190436955864986,
+      "grad_norm": 0.0,
+      "learning_rate": 6.693911156214758e-06,
+      "loss": 0.8486,
+      "step": 22540
+    },
+    {
+      "epoch": 0.6190711598143418,
+      "grad_norm": 0.0,
+      "learning_rate": 6.693071672675438e-06,
+      "loss": 0.7421,
+      "step": 22541
+    },
+    {
+      "epoch": 0.6190986240421851,
+      "grad_norm": 0.0,
+      "learning_rate": 6.692232215301817e-06,
+      "loss": 0.8131,
+      "step": 22542
+    },
+    {
+      "epoch": 0.6191260882700282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.691392784100534e-06,
+      "loss": 0.8589,
+      "step": 22543
+    },
+    {
+      "epoch": 0.6191535524978715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6905533790782315e-06,
+      "loss": 0.9251,
+      "step": 22544
+    },
+    {
+      "epoch": 0.6191810167257148,
+      "grad_norm": 0.0,
+      "learning_rate": 6.689714000241558e-06,
+      "loss": 0.8543,
+      "step": 22545
+    },
+    {
+      "epoch": 0.619208480953558,
+      "grad_norm": 0.0,
+      "learning_rate": 6.688874647597145e-06,
+      "loss": 0.8654,
+      "step": 22546
+    },
+    {
+      "epoch": 0.6192359451814012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.688035321151642e-06,
+      "loss": 0.9508,
+      "step": 22547
+    },
+    {
+      "epoch": 0.6192634094092445,
+      "grad_norm": 0.0,
+      "learning_rate": 6.687196020911682e-06,
+      "loss": 0.7793,
+      "step": 22548
+    },
+    {
+      "epoch": 0.6192908736370877,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6863567468839116e-06,
+      "loss": 0.8357,
+      "step": 22549
+    },
+    {
+      "epoch": 0.619318337864931,
+      "grad_norm": 0.0,
+      "learning_rate": 6.685517499074975e-06,
+      "loss": 0.8092,
+      "step": 22550
+    },
+    {
+      "epoch": 0.6193458020927741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.684678277491506e-06,
+      "loss": 0.8607,
+      "step": 22551
+    },
+    {
+      "epoch": 0.6193732663206174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.683839082140142e-06,
+      "loss": 0.8172,
+      "step": 22552
+    },
+    {
+      "epoch": 0.6194007305484607,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6829999130275305e-06,
+      "loss": 0.8031,
+      "step": 22553
+    },
+    {
+      "epoch": 0.6194281947763038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6821607701603086e-06,
+      "loss": 1.0292,
+      "step": 22554
+    },
+    {
+      "epoch": 0.6194556590041471,
+      "grad_norm": 0.0,
+      "learning_rate": 6.681321653545115e-06,
+      "loss": 0.8898,
+      "step": 22555
+    },
+    {
+      "epoch": 0.6194831232319903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.680482563188595e-06,
+      "loss": 0.8625,
+      "step": 22556
+    },
+    {
+      "epoch": 0.6195105874598336,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6796434990973765e-06,
+      "loss": 0.8168,
+      "step": 22557
+    },
+    {
+      "epoch": 0.6195380516876768,
+      "grad_norm": 0.0,
+      "learning_rate": 6.678804461278106e-06,
+      "loss": 0.7277,
+      "step": 22558
+    },
+    {
+      "epoch": 0.61956551591552,
+      "grad_norm": 0.0,
+      "learning_rate": 6.677965449737423e-06,
+      "loss": 0.7577,
+      "step": 22559
+    },
+    {
+      "epoch": 0.6195929801433633,
+      "grad_norm": 0.0,
+      "learning_rate": 6.677126464481962e-06,
+      "loss": 0.7277,
+      "step": 22560
+    },
+    {
+      "epoch": 0.6196204443712066,
+      "grad_norm": 0.0,
+      "learning_rate": 6.676287505518364e-06,
+      "loss": 0.7872,
+      "step": 22561
+    },
+    {
+      "epoch": 0.6196479085990497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6754485728532714e-06,
+      "loss": 0.7752,
+      "step": 22562
+    },
+    {
+      "epoch": 0.619675372826893,
+      "grad_norm": 0.0,
+      "learning_rate": 6.674609666493314e-06,
+      "loss": 0.9068,
+      "step": 22563
+    },
+    {
+      "epoch": 0.6197028370547362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.673770786445133e-06,
+      "loss": 0.799,
+      "step": 22564
+    },
+    {
+      "epoch": 0.6197303012825794,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6729319327153645e-06,
+      "loss": 0.8638,
+      "step": 22565
+    },
+    {
+      "epoch": 0.6197577655104227,
+      "grad_norm": 0.0,
+      "learning_rate": 6.672093105310648e-06,
+      "loss": 0.8393,
+      "step": 22566
+    },
+    {
+      "epoch": 0.6197852297382659,
+      "grad_norm": 0.0,
+      "learning_rate": 6.67125430423762e-06,
+      "loss": 0.7497,
+      "step": 22567
+    },
+    {
+      "epoch": 0.6198126939661092,
+      "grad_norm": 0.0,
+      "learning_rate": 6.670415529502921e-06,
+      "loss": 0.8481,
+      "step": 22568
+    },
+    {
+      "epoch": 0.6198401581939523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6695767811131805e-06,
+      "loss": 0.8999,
+      "step": 22569
+    },
+    {
+      "epoch": 0.6198676224217956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.668738059075037e-06,
+      "loss": 0.7766,
+      "step": 22570
+    },
+    {
+      "epoch": 0.6198950866496389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.667899363395133e-06,
+      "loss": 0.8507,
+      "step": 22571
+    },
+    {
+      "epoch": 0.619922550877482,
+      "grad_norm": 0.0,
+      "learning_rate": 6.667060694080096e-06,
+      "loss": 0.8702,
+      "step": 22572
+    },
+    {
+      "epoch": 0.6199500151053253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.666222051136566e-06,
+      "loss": 0.9196,
+      "step": 22573
+    },
+    {
+      "epoch": 0.6199774793331686,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6653834345711835e-06,
+      "loss": 0.8744,
+      "step": 22574
+    },
+    {
+      "epoch": 0.6200049435610118,
+      "grad_norm": 0.0,
+      "learning_rate": 6.664544844390576e-06,
+      "loss": 0.6687,
+      "step": 22575
+    },
+    {
+      "epoch": 0.620032407788855,
+      "grad_norm": 0.0,
+      "learning_rate": 6.66370628060138e-06,
+      "loss": 0.9589,
+      "step": 22576
+    },
+    {
+      "epoch": 0.6200598720166982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6628677432102325e-06,
+      "loss": 0.8591,
+      "step": 22577
+    },
+    {
+      "epoch": 0.6200873362445415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.66202923222377e-06,
+      "loss": 0.82,
+      "step": 22578
+    },
+    {
+      "epoch": 0.6201148004723848,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6611907476486225e-06,
+      "loss": 0.7693,
+      "step": 22579
+    },
+    {
+      "epoch": 0.6201422647002279,
+      "grad_norm": 0.0,
+      "learning_rate": 6.660352289491432e-06,
+      "loss": 0.8578,
+      "step": 22580
+    },
+    {
+      "epoch": 0.6201697289280712,
+      "grad_norm": 0.0,
+      "learning_rate": 6.659513857758824e-06,
+      "loss": 0.9004,
+      "step": 22581
+    },
+    {
+      "epoch": 0.6201971931559144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.658675452457435e-06,
+      "loss": 0.8085,
+      "step": 22582
+    },
+    {
+      "epoch": 0.6202246573837576,
+      "grad_norm": 0.0,
+      "learning_rate": 6.657837073593903e-06,
+      "loss": 0.8968,
+      "step": 22583
+    },
+    {
+      "epoch": 0.6202521216116009,
+      "grad_norm": 0.0,
+      "learning_rate": 6.656998721174856e-06,
+      "loss": 0.7726,
+      "step": 22584
+    },
+    {
+      "epoch": 0.6202795858394441,
+      "grad_norm": 0.0,
+      "learning_rate": 6.656160395206932e-06,
+      "loss": 0.9247,
+      "step": 22585
+    },
+    {
+      "epoch": 0.6203070500672874,
+      "grad_norm": 0.0,
+      "learning_rate": 6.655322095696765e-06,
+      "loss": 0.7818,
+      "step": 22586
+    },
+    {
+      "epoch": 0.6203345142951306,
+      "grad_norm": 0.0,
+      "learning_rate": 6.654483822650983e-06,
+      "loss": 0.8597,
+      "step": 22587
+    },
+    {
+      "epoch": 0.6203619785229738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.65364557607622e-06,
+      "loss": 0.8475,
+      "step": 22588
+    },
+    {
+      "epoch": 0.6203894427508171,
+      "grad_norm": 0.0,
+      "learning_rate": 6.652807355979107e-06,
+      "loss": 0.8264,
+      "step": 22589
+    },
+    {
+      "epoch": 0.6204169069786603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.651969162366283e-06,
+      "loss": 0.9234,
+      "step": 22590
+    },
+    {
+      "epoch": 0.6204443712065035,
+      "grad_norm": 0.0,
+      "learning_rate": 6.651130995244373e-06,
+      "loss": 0.8391,
+      "step": 22591
+    },
+    {
+      "epoch": 0.6204718354343468,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6502928546200165e-06,
+      "loss": 0.8474,
+      "step": 22592
+    },
+    {
+      "epoch": 0.62049929966219,
+      "grad_norm": 0.0,
+      "learning_rate": 6.649454740499837e-06,
+      "loss": 0.8561,
+      "step": 22593
+    },
+    {
+      "epoch": 0.6205267638900333,
+      "grad_norm": 0.0,
+      "learning_rate": 6.648616652890467e-06,
+      "loss": 0.9344,
+      "step": 22594
+    },
+    {
+      "epoch": 0.6205542281178764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.647778591798544e-06,
+      "loss": 0.7869,
+      "step": 22595
+    },
+    {
+      "epoch": 0.6205816923457197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.646940557230693e-06,
+      "loss": 0.8869,
+      "step": 22596
+    },
+    {
+      "epoch": 0.620609156573563,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6461025491935475e-06,
+      "loss": 0.8468,
+      "step": 22597
+    },
+    {
+      "epoch": 0.6206366208014061,
+      "grad_norm": 0.0,
+      "learning_rate": 6.645264567693741e-06,
+      "loss": 0.9524,
+      "step": 22598
+    },
+    {
+      "epoch": 0.6206640850292494,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6444266127379e-06,
+      "loss": 0.856,
+      "step": 22599
+    },
+    {
+      "epoch": 0.6206915492570927,
+      "grad_norm": 0.0,
+      "learning_rate": 6.643588684332652e-06,
+      "loss": 0.8453,
+      "step": 22600
+    },
+    {
+      "epoch": 0.6207190134849359,
+      "grad_norm": 0.0,
+      "learning_rate": 6.642750782484631e-06,
+      "loss": 0.8688,
+      "step": 22601
+    },
+    {
+      "epoch": 0.6207464777127791,
+      "grad_norm": 0.0,
+      "learning_rate": 6.641912907200469e-06,
+      "loss": 0.7558,
+      "step": 22602
+    },
+    {
+      "epoch": 0.6207739419406223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.64107505848679e-06,
+      "loss": 0.8694,
+      "step": 22603
+    },
+    {
+      "epoch": 0.6208014061684656,
+      "grad_norm": 0.0,
+      "learning_rate": 6.640237236350231e-06,
+      "loss": 0.9146,
+      "step": 22604
+    },
+    {
+      "epoch": 0.6208288703963089,
+      "grad_norm": 0.0,
+      "learning_rate": 6.639399440797413e-06,
+      "loss": 0.8954,
+      "step": 22605
+    },
+    {
+      "epoch": 0.620856334624152,
+      "grad_norm": 0.0,
+      "learning_rate": 6.638561671834967e-06,
+      "loss": 0.8764,
+      "step": 22606
+    },
+    {
+      "epoch": 0.6208837988519953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.637723929469526e-06,
+      "loss": 0.9214,
+      "step": 22607
+    },
+    {
+      "epoch": 0.6209112630798385,
+      "grad_norm": 0.0,
+      "learning_rate": 6.636886213707713e-06,
+      "loss": 0.8973,
+      "step": 22608
+    },
+    {
+      "epoch": 0.6209387273076817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.636048524556159e-06,
+      "loss": 0.9268,
+      "step": 22609
+    },
+    {
+      "epoch": 0.620966191535525,
+      "grad_norm": 0.0,
+      "learning_rate": 6.635210862021497e-06,
+      "loss": 0.8467,
+      "step": 22610
+    },
+    {
+      "epoch": 0.6209936557633682,
+      "grad_norm": 0.0,
+      "learning_rate": 6.634373226110347e-06,
+      "loss": 0.8406,
+      "step": 22611
+    },
+    {
+      "epoch": 0.6210211199912115,
+      "grad_norm": 0.0,
+      "learning_rate": 6.633535616829338e-06,
+      "loss": 0.807,
+      "step": 22612
+    },
+    {
+      "epoch": 0.6210485842190547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6326980341851e-06,
+      "loss": 0.7169,
+      "step": 22613
+    },
+    {
+      "epoch": 0.6210760484468979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.631860478184261e-06,
+      "loss": 0.815,
+      "step": 22614
+    },
+    {
+      "epoch": 0.6211035126747412,
+      "grad_norm": 0.0,
+      "learning_rate": 6.631022948833449e-06,
+      "loss": 0.8322,
+      "step": 22615
+    },
+    {
+      "epoch": 0.6211309769025843,
+      "grad_norm": 0.0,
+      "learning_rate": 6.630185446139283e-06,
+      "loss": 0.9119,
+      "step": 22616
+    },
+    {
+      "epoch": 0.6211584411304276,
+      "grad_norm": 0.0,
+      "learning_rate": 6.629347970108398e-06,
+      "loss": 0.8455,
+      "step": 22617
+    },
+    {
+      "epoch": 0.6211859053582709,
+      "grad_norm": 0.0,
+      "learning_rate": 6.628510520747415e-06,
+      "loss": 0.8265,
+      "step": 22618
+    },
+    {
+      "epoch": 0.6212133695861141,
+      "grad_norm": 0.0,
+      "learning_rate": 6.627673098062966e-06,
+      "loss": 0.8012,
+      "step": 22619
+    },
+    {
+      "epoch": 0.6212408338139573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62683570206167e-06,
+      "loss": 0.8339,
+      "step": 22620
+    },
+    {
+      "epoch": 0.6212682980418005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.625998332750162e-06,
+      "loss": 0.8607,
+      "step": 22621
+    },
+    {
+      "epoch": 0.6212957622696438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6251609901350576e-06,
+      "loss": 0.8553,
+      "step": 22622
+    },
+    {
+      "epoch": 0.6213232264974871,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62432367422299e-06,
+      "loss": 0.7822,
+      "step": 22623
+    },
+    {
+      "epoch": 0.6213506907253302,
+      "grad_norm": 0.0,
+      "learning_rate": 6.623486385020578e-06,
+      "loss": 0.7745,
+      "step": 22624
+    },
+    {
+      "epoch": 0.6213781549531735,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62264912253445e-06,
+      "loss": 0.7751,
+      "step": 22625
+    },
+    {
+      "epoch": 0.6214056191810168,
+      "grad_norm": 0.0,
+      "learning_rate": 6.621811886771232e-06,
+      "loss": 0.9855,
+      "step": 22626
+    },
+    {
+      "epoch": 0.6214330834088599,
+      "grad_norm": 0.0,
+      "learning_rate": 6.620974677737548e-06,
+      "loss": 0.9029,
+      "step": 22627
+    },
+    {
+      "epoch": 0.6214605476367032,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62013749544002e-06,
+      "loss": 0.8303,
+      "step": 22628
+    },
+    {
+      "epoch": 0.6214880118645464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6193003398852705e-06,
+      "loss": 0.875,
+      "step": 22629
+    },
+    {
+      "epoch": 0.6215154760923897,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6184632110799285e-06,
+      "loss": 0.8753,
+      "step": 22630
+    },
+    {
+      "epoch": 0.6215429403202329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.617626109030617e-06,
+      "loss": 0.8596,
+      "step": 22631
+    },
+    {
+      "epoch": 0.6215704045480761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.616789033743954e-06,
+      "loss": 0.8574,
+      "step": 22632
+    },
+    {
+      "epoch": 0.6215978687759194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.615951985226573e-06,
+      "loss": 0.8716,
+      "step": 22633
+    },
+    {
+      "epoch": 0.6216253330037625,
+      "grad_norm": 0.0,
+      "learning_rate": 6.615114963485085e-06,
+      "loss": 0.8799,
+      "step": 22634
+    },
+    {
+      "epoch": 0.6216527972316058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.614277968526122e-06,
+      "loss": 0.8747,
+      "step": 22635
+    },
+    {
+      "epoch": 0.6216802614594491,
+      "grad_norm": 0.0,
+      "learning_rate": 6.613441000356301e-06,
+      "loss": 0.9268,
+      "step": 22636
+    },
+    {
+      "epoch": 0.6217077256872923,
+      "grad_norm": 0.0,
+      "learning_rate": 6.612604058982246e-06,
+      "loss": 0.8248,
+      "step": 22637
+    },
+    {
+      "epoch": 0.6217351899151355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6117671444105825e-06,
+      "loss": 0.8756,
+      "step": 22638
+    },
+    {
+      "epoch": 0.6217626541429788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.610930256647931e-06,
+      "loss": 0.7395,
+      "step": 22639
+    },
+    {
+      "epoch": 0.621790118370822,
+      "grad_norm": 0.0,
+      "learning_rate": 6.610093395700909e-06,
+      "loss": 0.8806,
+      "step": 22640
+    },
+    {
+      "epoch": 0.6218175825986653,
+      "grad_norm": 0.0,
+      "learning_rate": 6.609256561576141e-06,
+      "loss": 0.9273,
+      "step": 22641
+    },
+    {
+      "epoch": 0.6218450468265084,
+      "grad_norm": 0.0,
+      "learning_rate": 6.608419754280249e-06,
+      "loss": 0.8745,
+      "step": 22642
+    },
+    {
+      "epoch": 0.6218725110543517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.607582973819855e-06,
+      "loss": 0.758,
+      "step": 22643
+    },
+    {
+      "epoch": 0.621899975282195,
+      "grad_norm": 0.0,
+      "learning_rate": 6.606746220201576e-06,
+      "loss": 0.7937,
+      "step": 22644
+    },
+    {
+      "epoch": 0.6219274395100381,
+      "grad_norm": 0.0,
+      "learning_rate": 6.60590949343204e-06,
+      "loss": 0.824,
+      "step": 22645
+    },
+    {
+      "epoch": 0.6219549037378814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.605072793517859e-06,
+      "loss": 0.8385,
+      "step": 22646
+    },
+    {
+      "epoch": 0.6219823679657246,
+      "grad_norm": 0.0,
+      "learning_rate": 6.604236120465659e-06,
+      "loss": 0.9085,
+      "step": 22647
+    },
+    {
+      "epoch": 0.6220098321935679,
+      "grad_norm": 0.0,
+      "learning_rate": 6.603399474282055e-06,
+      "loss": 0.9376,
+      "step": 22648
+    },
+    {
+      "epoch": 0.6220372964214111,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6025628549736706e-06,
+      "loss": 0.8575,
+      "step": 22649
+    },
+    {
+      "epoch": 0.6220647606492543,
+      "grad_norm": 0.0,
+      "learning_rate": 6.601726262547128e-06,
+      "loss": 0.9308,
+      "step": 22650
+    },
+    {
+      "epoch": 0.6220922248770976,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6008896970090454e-06,
+      "loss": 0.8938,
+      "step": 22651
+    },
+    {
+      "epoch": 0.6221196891049409,
+      "grad_norm": 0.0,
+      "learning_rate": 6.600053158366035e-06,
+      "loss": 0.873,
+      "step": 22652
+    },
+    {
+      "epoch": 0.622147153332784,
+      "grad_norm": 0.0,
+      "learning_rate": 6.599216646624723e-06,
+      "loss": 0.9091,
+      "step": 22653
+    },
+    {
+      "epoch": 0.6221746175606273,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5983801617917265e-06,
+      "loss": 0.8458,
+      "step": 22654
+    },
+    {
+      "epoch": 0.6222020817884705,
+      "grad_norm": 0.0,
+      "learning_rate": 6.597543703873663e-06,
+      "loss": 0.8111,
+      "step": 22655
+    },
+    {
+      "epoch": 0.6222295460163138,
+      "grad_norm": 0.0,
+      "learning_rate": 6.596707272877152e-06,
+      "loss": 0.8821,
+      "step": 22656
+    },
+    {
+      "epoch": 0.622257010244157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5958708688088136e-06,
+      "loss": 0.7853,
+      "step": 22657
+    },
+    {
+      "epoch": 0.6222844744720002,
+      "grad_norm": 0.0,
+      "learning_rate": 6.595034491675261e-06,
+      "loss": 0.8317,
+      "step": 22658
+    },
+    {
+      "epoch": 0.6223119386998435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.594198141483116e-06,
+      "loss": 0.816,
+      "step": 22659
+    },
+    {
+      "epoch": 0.6223394029276866,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5933618182389925e-06,
+      "loss": 0.851,
+      "step": 22660
+    },
+    {
+      "epoch": 0.6223668671555299,
+      "grad_norm": 0.0,
+      "learning_rate": 6.592525521949508e-06,
+      "loss": 0.9529,
+      "step": 22661
+    },
+    {
+      "epoch": 0.6223943313833732,
+      "grad_norm": 0.0,
+      "learning_rate": 6.591689252621287e-06,
+      "loss": 0.8768,
+      "step": 22662
+    },
+    {
+      "epoch": 0.6224217956112164,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5908530102609415e-06,
+      "loss": 0.8299,
+      "step": 22663
+    },
+    {
+      "epoch": 0.6224492598390596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.590016794875083e-06,
+      "loss": 0.8029,
+      "step": 22664
+    },
+    {
+      "epoch": 0.6224767240669029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.589180606470333e-06,
+      "loss": 0.8394,
+      "step": 22665
+    },
+    {
+      "epoch": 0.6225041882947461,
+      "grad_norm": 0.0,
+      "learning_rate": 6.58834444505331e-06,
+      "loss": 1.0068,
+      "step": 22666
+    },
+    {
+      "epoch": 0.6225316525225894,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5875083106306235e-06,
+      "loss": 0.8113,
+      "step": 22667
+    },
+    {
+      "epoch": 0.6225591167504325,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5866722032088945e-06,
+      "loss": 0.855,
+      "step": 22668
+    },
+    {
+      "epoch": 0.6225865809782758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.58583612279474e-06,
+      "loss": 0.8228,
+      "step": 22669
+    },
+    {
+      "epoch": 0.6226140452061191,
+      "grad_norm": 0.0,
+      "learning_rate": 6.585000069394769e-06,
+      "loss": 0.9767,
+      "step": 22670
+    },
+    {
+      "epoch": 0.6226415094339622,
+      "grad_norm": 0.0,
+      "learning_rate": 6.584164043015604e-06,
+      "loss": 0.8741,
+      "step": 22671
+    },
+    {
+      "epoch": 0.6226689736618055,
+      "grad_norm": 0.0,
+      "learning_rate": 6.583328043663852e-06,
+      "loss": 0.766,
+      "step": 22672
+    },
+    {
+      "epoch": 0.6226964378896487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.582492071346134e-06,
+      "loss": 0.8228,
+      "step": 22673
+    },
+    {
+      "epoch": 0.622723902117492,
+      "grad_norm": 0.0,
+      "learning_rate": 6.581656126069065e-06,
+      "loss": 0.8427,
+      "step": 22674
+    },
+    {
+      "epoch": 0.6227513663453352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.580820207839255e-06,
+      "loss": 0.7146,
+      "step": 22675
+    },
+    {
+      "epoch": 0.6227788305731784,
+      "grad_norm": 0.0,
+      "learning_rate": 6.579984316663319e-06,
+      "loss": 0.8749,
+      "step": 22676
+    },
+    {
+      "epoch": 0.6228062948010217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.579148452547871e-06,
+      "loss": 0.7245,
+      "step": 22677
+    },
+    {
+      "epoch": 0.622833759028865,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5783126154995266e-06,
+      "loss": 0.84,
+      "step": 22678
+    },
+    {
+      "epoch": 0.6228612232567081,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5774768055248984e-06,
+      "loss": 0.8556,
+      "step": 22679
+    },
+    {
+      "epoch": 0.6228886874845514,
+      "grad_norm": 0.0,
+      "learning_rate": 6.576641022630603e-06,
+      "loss": 0.8736,
+      "step": 22680
+    },
+    {
+      "epoch": 0.6229161517123946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.575805266823245e-06,
+      "loss": 0.858,
+      "step": 22681
+    },
+    {
+      "epoch": 0.6229436159402378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5749695381094426e-06,
+      "loss": 0.8032,
+      "step": 22682
+    },
+    {
+      "epoch": 0.6229710801680811,
+      "grad_norm": 0.0,
+      "learning_rate": 6.57413383649581e-06,
+      "loss": 0.7777,
+      "step": 22683
+    },
+    {
+      "epoch": 0.6229985443959243,
+      "grad_norm": 0.0,
+      "learning_rate": 6.573298161988955e-06,
+      "loss": 0.9758,
+      "step": 22684
+    },
+    {
+      "epoch": 0.6230260086237676,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5724625145954925e-06,
+      "loss": 0.8316,
+      "step": 22685
+    },
+    {
+      "epoch": 0.6230534728516107,
+      "grad_norm": 0.0,
+      "learning_rate": 6.571626894322039e-06,
+      "loss": 0.9407,
+      "step": 22686
+    },
+    {
+      "epoch": 0.623080937079454,
+      "grad_norm": 0.0,
+      "learning_rate": 6.570791301175199e-06,
+      "loss": 0.9044,
+      "step": 22687
+    },
+    {
+      "epoch": 0.6231084013072973,
+      "grad_norm": 0.0,
+      "learning_rate": 6.569955735161584e-06,
+      "loss": 0.806,
+      "step": 22688
+    },
+    {
+      "epoch": 0.6231358655351404,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5691201962878094e-06,
+      "loss": 0.8801,
+      "step": 22689
+    },
+    {
+      "epoch": 0.6231633297629837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.568284684560485e-06,
+      "loss": 0.8188,
+      "step": 22690
+    },
+    {
+      "epoch": 0.623190793990827,
+      "grad_norm": 0.0,
+      "learning_rate": 6.567449199986221e-06,
+      "loss": 0.7806,
+      "step": 22691
+    },
+    {
+      "epoch": 0.6232182582186702,
+      "grad_norm": 0.0,
+      "learning_rate": 6.566613742571633e-06,
+      "loss": 0.9242,
+      "step": 22692
+    },
+    {
+      "epoch": 0.6232457224465134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.565778312323321e-06,
+      "loss": 0.9269,
+      "step": 22693
+    },
+    {
+      "epoch": 0.6232731866743566,
+      "grad_norm": 0.0,
+      "learning_rate": 6.564942909247902e-06,
+      "loss": 0.8592,
+      "step": 22694
+    },
+    {
+      "epoch": 0.6233006509021999,
+      "grad_norm": 0.0,
+      "learning_rate": 6.564107533351988e-06,
+      "loss": 0.8745,
+      "step": 22695
+    },
+    {
+      "epoch": 0.6233281151300432,
+      "grad_norm": 0.0,
+      "learning_rate": 6.563272184642184e-06,
+      "loss": 0.9215,
+      "step": 22696
+    },
+    {
+      "epoch": 0.6233555793578863,
+      "grad_norm": 0.0,
+      "learning_rate": 6.562436863125101e-06,
+      "loss": 0.8804,
+      "step": 22697
+    },
+    {
+      "epoch": 0.6233830435857296,
+      "grad_norm": 0.0,
+      "learning_rate": 6.561601568807354e-06,
+      "loss": 0.9157,
+      "step": 22698
+    },
+    {
+      "epoch": 0.6234105078135728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.560766301695545e-06,
+      "loss": 0.9076,
+      "step": 22699
+    },
+    {
+      "epoch": 0.623437972041416,
+      "grad_norm": 0.0,
+      "learning_rate": 6.559931061796283e-06,
+      "loss": 0.912,
+      "step": 22700
+    },
+    {
+      "epoch": 0.6234654362692593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.559095849116179e-06,
+      "loss": 0.7878,
+      "step": 22701
+    },
+    {
+      "epoch": 0.6234929004971025,
+      "grad_norm": 0.0,
+      "learning_rate": 6.558260663661844e-06,
+      "loss": 0.9101,
+      "step": 22702
+    },
+    {
+      "epoch": 0.6235203647249458,
+      "grad_norm": 0.0,
+      "learning_rate": 6.557425505439881e-06,
+      "loss": 0.8141,
+      "step": 22703
+    },
+    {
+      "epoch": 0.623547828952789,
+      "grad_norm": 0.0,
+      "learning_rate": 6.556590374456906e-06,
+      "loss": 0.9152,
+      "step": 22704
+    },
+    {
+      "epoch": 0.6235752931806322,
+      "grad_norm": 0.0,
+      "learning_rate": 6.555755270719517e-06,
+      "loss": 0.8411,
+      "step": 22705
+    },
+    {
+      "epoch": 0.6236027574084755,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5549201942343265e-06,
+      "loss": 0.911,
+      "step": 22706
+    },
+    {
+      "epoch": 0.6236302216363186,
+      "grad_norm": 0.0,
+      "learning_rate": 6.554085145007945e-06,
+      "loss": 0.8025,
+      "step": 22707
+    },
+    {
+      "epoch": 0.6236576858641619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.553250123046972e-06,
+      "loss": 0.9628,
+      "step": 22708
+    },
+    {
+      "epoch": 0.6236851500920052,
+      "grad_norm": 0.0,
+      "learning_rate": 6.552415128358022e-06,
+      "loss": 0.871,
+      "step": 22709
+    },
+    {
+      "epoch": 0.6237126143198484,
+      "grad_norm": 0.0,
+      "learning_rate": 6.551580160947701e-06,
+      "loss": 0.8654,
+      "step": 22710
+    },
+    {
+      "epoch": 0.6237400785476916,
+      "grad_norm": 0.0,
+      "learning_rate": 6.550745220822614e-06,
+      "loss": 0.8732,
+      "step": 22711
+    },
+    {
+      "epoch": 0.6237675427755348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.549910307989362e-06,
+      "loss": 0.8647,
+      "step": 22712
+    },
+    {
+      "epoch": 0.6237950070033781,
+      "grad_norm": 0.0,
+      "learning_rate": 6.549075422454558e-06,
+      "loss": 0.8257,
+      "step": 22713
+    },
+    {
+      "epoch": 0.6238224712312214,
+      "grad_norm": 0.0,
+      "learning_rate": 6.548240564224808e-06,
+      "loss": 0.802,
+      "step": 22714
+    },
+    {
+      "epoch": 0.6238499354590645,
+      "grad_norm": 0.0,
+      "learning_rate": 6.547405733306713e-06,
+      "loss": 0.929,
+      "step": 22715
+    },
+    {
+      "epoch": 0.6238773996869078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5465709297068854e-06,
+      "loss": 0.8528,
+      "step": 22716
+    },
+    {
+      "epoch": 0.6239048639147511,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545736153431923e-06,
+      "loss": 0.8036,
+      "step": 22717
+    },
+    {
+      "epoch": 0.6239323281425943,
+      "grad_norm": 0.0,
+      "learning_rate": 6.544901404488433e-06,
+      "loss": 0.8849,
+      "step": 22718
+    },
+    {
+      "epoch": 0.6239597923704375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.544066682883025e-06,
+      "loss": 0.7838,
+      "step": 22719
+    },
+    {
+      "epoch": 0.6239872565982807,
+      "grad_norm": 0.0,
+      "learning_rate": 6.543231988622296e-06,
+      "loss": 0.8717,
+      "step": 22720
+    },
+    {
+      "epoch": 0.624014720826124,
+      "grad_norm": 0.0,
+      "learning_rate": 6.542397321712856e-06,
+      "loss": 0.8446,
+      "step": 22721
+    },
+    {
+      "epoch": 0.6240421850539672,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5415626821613125e-06,
+      "loss": 0.7886,
+      "step": 22722
+    },
+    {
+      "epoch": 0.6240696492818104,
+      "grad_norm": 0.0,
+      "learning_rate": 6.540728069974261e-06,
+      "loss": 0.8713,
+      "step": 22723
+    },
+    {
+      "epoch": 0.6240971135096537,
+      "grad_norm": 0.0,
+      "learning_rate": 6.539893485158309e-06,
+      "loss": 0.8553,
+      "step": 22724
+    },
+    {
+      "epoch": 0.6241245777374969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.539058927720056e-06,
+      "loss": 0.8849,
+      "step": 22725
+    },
+    {
+      "epoch": 0.6241520419653401,
+      "grad_norm": 0.0,
+      "learning_rate": 6.538224397666115e-06,
+      "loss": 0.9082,
+      "step": 22726
+    },
+    {
+      "epoch": 0.6241795061931834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.537389895003081e-06,
+      "loss": 0.9322,
+      "step": 22727
+    },
+    {
+      "epoch": 0.6242069704210266,
+      "grad_norm": 0.0,
+      "learning_rate": 6.536555419737564e-06,
+      "loss": 0.7848,
+      "step": 22728
+    },
+    {
+      "epoch": 0.6242344346488699,
+      "grad_norm": 0.0,
+      "learning_rate": 6.535720971876156e-06,
+      "loss": 0.8511,
+      "step": 22729
+    },
+    {
+      "epoch": 0.6242618988767131,
+      "grad_norm": 0.0,
+      "learning_rate": 6.534886551425468e-06,
+      "loss": 0.7711,
+      "step": 22730
+    },
+    {
+      "epoch": 0.6242893631045563,
+      "grad_norm": 0.0,
+      "learning_rate": 6.534052158392101e-06,
+      "loss": 0.9094,
+      "step": 22731
+    },
+    {
+      "epoch": 0.6243168273323996,
+      "grad_norm": 0.0,
+      "learning_rate": 6.533217792782653e-06,
+      "loss": 0.7788,
+      "step": 22732
+    },
+    {
+      "epoch": 0.6243442915602427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5323834546037336e-06,
+      "loss": 0.7654,
+      "step": 22733
+    },
+    {
+      "epoch": 0.624371755788086,
+      "grad_norm": 0.0,
+      "learning_rate": 6.531549143861935e-06,
+      "loss": 0.9486,
+      "step": 22734
+    },
+    {
+      "epoch": 0.6243992200159293,
+      "grad_norm": 0.0,
+      "learning_rate": 6.530714860563865e-06,
+      "loss": 0.8672,
+      "step": 22735
+    },
+    {
+      "epoch": 0.6244266842437725,
+      "grad_norm": 0.0,
+      "learning_rate": 6.529880604716121e-06,
+      "loss": 0.9359,
+      "step": 22736
+    },
+    {
+      "epoch": 0.6244541484716157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.529046376325307e-06,
+      "loss": 0.8159,
+      "step": 22737
+    },
+    {
+      "epoch": 0.624481612699459,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5282121753980236e-06,
+      "loss": 0.8024,
+      "step": 22738
+    },
+    {
+      "epoch": 0.6245090769273022,
+      "grad_norm": 0.0,
+      "learning_rate": 6.52737800194087e-06,
+      "loss": 0.8507,
+      "step": 22739
+    },
+    {
+      "epoch": 0.6245365411551455,
+      "grad_norm": 0.0,
+      "learning_rate": 6.526543855960448e-06,
+      "loss": 0.8859,
+      "step": 22740
+    },
+    {
+      "epoch": 0.6245640053829886,
+      "grad_norm": 0.0,
+      "learning_rate": 6.525709737463353e-06,
+      "loss": 0.8161,
+      "step": 22741
+    },
+    {
+      "epoch": 0.6245914696108319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.52487564645619e-06,
+      "loss": 0.8715,
+      "step": 22742
+    },
+    {
+      "epoch": 0.6246189338386752,
+      "grad_norm": 0.0,
+      "learning_rate": 6.524041582945558e-06,
+      "loss": 0.8583,
+      "step": 22743
+    },
+    {
+      "epoch": 0.6246463980665183,
+      "grad_norm": 0.0,
+      "learning_rate": 6.523207546938053e-06,
+      "loss": 0.9276,
+      "step": 22744
+    },
+    {
+      "epoch": 0.6246738622943616,
+      "grad_norm": 0.0,
+      "learning_rate": 6.522373538440282e-06,
+      "loss": 0.8431,
+      "step": 22745
+    },
+    {
+      "epoch": 0.6247013265222048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.521539557458834e-06,
+      "loss": 0.8146,
+      "step": 22746
+    },
+    {
+      "epoch": 0.6247287907500481,
+      "grad_norm": 0.0,
+      "learning_rate": 6.520705604000315e-06,
+      "loss": 0.7352,
+      "step": 22747
+    },
+    {
+      "epoch": 0.6247562549778913,
+      "grad_norm": 0.0,
+      "learning_rate": 6.519871678071318e-06,
+      "loss": 0.8671,
+      "step": 22748
+    },
+    {
+      "epoch": 0.6247837192057345,
+      "grad_norm": 0.0,
+      "learning_rate": 6.519037779678444e-06,
+      "loss": 0.7703,
+      "step": 22749
+    },
+    {
+      "epoch": 0.6248111834335778,
+      "grad_norm": 0.0,
+      "learning_rate": 6.518203908828296e-06,
+      "loss": 0.7995,
+      "step": 22750
+    },
+    {
+      "epoch": 0.624838647661421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.517370065527464e-06,
+      "loss": 0.8195,
+      "step": 22751
+    },
+    {
+      "epoch": 0.6248661118892642,
+      "grad_norm": 0.0,
+      "learning_rate": 6.516536249782552e-06,
+      "loss": 0.7625,
+      "step": 22752
+    },
+    {
+      "epoch": 0.6248935761171075,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5157024616001504e-06,
+      "loss": 0.8313,
+      "step": 22753
+    },
+    {
+      "epoch": 0.6249210403449507,
+      "grad_norm": 0.0,
+      "learning_rate": 6.514868700986862e-06,
+      "loss": 0.889,
+      "step": 22754
+    },
+    {
+      "epoch": 0.6249485045727939,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5140349679492835e-06,
+      "loss": 0.8945,
+      "step": 22755
+    },
+    {
+      "epoch": 0.6249759688006372,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5132012624940085e-06,
+      "loss": 0.7871,
+      "step": 22756
+    },
+    {
+      "epoch": 0.6250034330284804,
+      "grad_norm": 0.0,
+      "learning_rate": 6.512367584627641e-06,
+      "loss": 0.8247,
+      "step": 22757
+    },
+    {
+      "epoch": 0.6250308972563237,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5115339343567685e-06,
+      "loss": 0.8515,
+      "step": 22758
+    },
+    {
+      "epoch": 0.6250583614841668,
+      "grad_norm": 0.0,
+      "learning_rate": 6.510700311687992e-06,
+      "loss": 0.855,
+      "step": 22759
+    },
+    {
+      "epoch": 0.6250858257120101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.509866716627904e-06,
+      "loss": 0.7569,
+      "step": 22760
+    },
+    {
+      "epoch": 0.6251132899398534,
+      "grad_norm": 0.0,
+      "learning_rate": 6.509033149183102e-06,
+      "loss": 0.9295,
+      "step": 22761
+    },
+    {
+      "epoch": 0.6251407541676965,
+      "grad_norm": 0.0,
+      "learning_rate": 6.508199609360184e-06,
+      "loss": 0.9115,
+      "step": 22762
+    },
+    {
+      "epoch": 0.6251682183955398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.507366097165744e-06,
+      "loss": 0.8607,
+      "step": 22763
+    },
+    {
+      "epoch": 0.6251956826233831,
+      "grad_norm": 0.0,
+      "learning_rate": 6.506532612606377e-06,
+      "loss": 0.9244,
+      "step": 22764
+    },
+    {
+      "epoch": 0.6252231468512263,
+      "grad_norm": 0.0,
+      "learning_rate": 6.505699155688674e-06,
+      "loss": 0.8775,
+      "step": 22765
+    },
+    {
+      "epoch": 0.6252506110790695,
+      "grad_norm": 0.0,
+      "learning_rate": 6.504865726419234e-06,
+      "loss": 0.9019,
+      "step": 22766
+    },
+    {
+      "epoch": 0.6252780753069127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.50403232480465e-06,
+      "loss": 0.8405,
+      "step": 22767
+    },
+    {
+      "epoch": 0.625305539534756,
+      "grad_norm": 0.0,
+      "learning_rate": 6.503198950851517e-06,
+      "loss": 0.8513,
+      "step": 22768
+    },
+    {
+      "epoch": 0.6253330037625993,
+      "grad_norm": 0.0,
+      "learning_rate": 6.502365604566432e-06,
+      "loss": 0.8552,
+      "step": 22769
+    },
+    {
+      "epoch": 0.6253604679904424,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5015322859559805e-06,
+      "loss": 0.8655,
+      "step": 22770
+    },
+    {
+      "epoch": 0.6253879322182857,
+      "grad_norm": 0.0,
+      "learning_rate": 6.500698995026762e-06,
+      "loss": 0.7527,
+      "step": 22771
+    },
+    {
+      "epoch": 0.6254153964461289,
+      "grad_norm": 0.0,
+      "learning_rate": 6.499865731785368e-06,
+      "loss": 0.9726,
+      "step": 22772
+    },
+    {
+      "epoch": 0.6254428606739721,
+      "grad_norm": 0.0,
+      "learning_rate": 6.499032496238391e-06,
+      "loss": 0.8577,
+      "step": 22773
+    },
+    {
+      "epoch": 0.6254703249018154,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4981992883924285e-06,
+      "loss": 0.8234,
+      "step": 22774
+    },
+    {
+      "epoch": 0.6254977891296586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.497366108254071e-06,
+      "loss": 0.8549,
+      "step": 22775
+    },
+    {
+      "epoch": 0.6255252533575019,
+      "grad_norm": 0.0,
+      "learning_rate": 6.496532955829904e-06,
+      "loss": 0.8282,
+      "step": 22776
+    },
+    {
+      "epoch": 0.6255527175853451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4956998311265275e-06,
+      "loss": 0.8419,
+      "step": 22777
+    },
+    {
+      "epoch": 0.6255801818131883,
+      "grad_norm": 0.0,
+      "learning_rate": 6.49486673415053e-06,
+      "loss": 0.9224,
+      "step": 22778
+    },
+    {
+      "epoch": 0.6256076460410316,
+      "grad_norm": 0.0,
+      "learning_rate": 6.494033664908508e-06,
+      "loss": 0.8429,
+      "step": 22779
+    },
+    {
+      "epoch": 0.6256351102688748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.493200623407047e-06,
+      "loss": 0.8189,
+      "step": 22780
+    },
+    {
+      "epoch": 0.625662574496718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.492367609652745e-06,
+      "loss": 0.8457,
+      "step": 22781
+    },
+    {
+      "epoch": 0.6256900387245613,
+      "grad_norm": 0.0,
+      "learning_rate": 6.491534623652184e-06,
+      "loss": 0.8894,
+      "step": 22782
+    },
+    {
+      "epoch": 0.6257175029524045,
+      "grad_norm": 0.0,
+      "learning_rate": 6.490701665411965e-06,
+      "loss": 0.8512,
+      "step": 22783
+    },
+    {
+      "epoch": 0.6257449671802477,
+      "grad_norm": 0.0,
+      "learning_rate": 6.489868734938669e-06,
+      "loss": 0.8588,
+      "step": 22784
+    },
+    {
+      "epoch": 0.6257724314080909,
+      "grad_norm": 0.0,
+      "learning_rate": 6.489035832238893e-06,
+      "loss": 0.7904,
+      "step": 22785
+    },
+    {
+      "epoch": 0.6257998956359342,
+      "grad_norm": 0.0,
+      "learning_rate": 6.488202957319228e-06,
+      "loss": 0.7762,
+      "step": 22786
+    },
+    {
+      "epoch": 0.6258273598637775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.487370110186262e-06,
+      "loss": 0.7995,
+      "step": 22787
+    },
+    {
+      "epoch": 0.6258548240916206,
+      "grad_norm": 0.0,
+      "learning_rate": 6.486537290846581e-06,
+      "loss": 0.8733,
+      "step": 22788
+    },
+    {
+      "epoch": 0.6258822883194639,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4857044993067775e-06,
+      "loss": 0.8298,
+      "step": 22789
+    },
+    {
+      "epoch": 0.6259097525473072,
+      "grad_norm": 0.0,
+      "learning_rate": 6.484871735573441e-06,
+      "loss": 0.8263,
+      "step": 22790
+    },
+    {
+      "epoch": 0.6259372167751504,
+      "grad_norm": 0.0,
+      "learning_rate": 6.484038999653164e-06,
+      "loss": 0.8184,
+      "step": 22791
+    },
+    {
+      "epoch": 0.6259646810029936,
+      "grad_norm": 0.0,
+      "learning_rate": 6.483206291552535e-06,
+      "loss": 0.8003,
+      "step": 22792
+    },
+    {
+      "epoch": 0.6259921452308368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.482373611278134e-06,
+      "loss": 0.8728,
+      "step": 22793
+    },
+    {
+      "epoch": 0.6260196094586801,
+      "grad_norm": 0.0,
+      "learning_rate": 6.481540958836556e-06,
+      "loss": 0.9754,
+      "step": 22794
+    },
+    {
+      "epoch": 0.6260470736865233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.480708334234393e-06,
+      "loss": 0.829,
+      "step": 22795
+    },
+    {
+      "epoch": 0.6260745379143665,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4798757374782255e-06,
+      "loss": 0.9153,
+      "step": 22796
+    },
+    {
+      "epoch": 0.6261020021422098,
+      "grad_norm": 0.0,
+      "learning_rate": 6.479043168574643e-06,
+      "loss": 0.7893,
+      "step": 22797
+    },
+    {
+      "epoch": 0.626129466370053,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4782106275302415e-06,
+      "loss": 0.8461,
+      "step": 22798
+    },
+    {
+      "epoch": 0.6261569305978962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.477378114351599e-06,
+      "loss": 0.8611,
+      "step": 22799
+    },
+    {
+      "epoch": 0.6261843948257395,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4765456290453035e-06,
+      "loss": 0.8195,
+      "step": 22800
+    },
+    {
+      "epoch": 0.6262118590535827,
+      "grad_norm": 0.0,
+      "learning_rate": 6.475713171617944e-06,
+      "loss": 0.7475,
+      "step": 22801
+    },
+    {
+      "epoch": 0.626239323281426,
+      "grad_norm": 0.0,
+      "learning_rate": 6.474880742076106e-06,
+      "loss": 0.7551,
+      "step": 22802
+    },
+    {
+      "epoch": 0.6262667875092692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.47404834042638e-06,
+      "loss": 0.8704,
+      "step": 22803
+    },
+    {
+      "epoch": 0.6262942517371124,
+      "grad_norm": 0.0,
+      "learning_rate": 6.47321596667535e-06,
+      "loss": 0.7106,
+      "step": 22804
+    },
+    {
+      "epoch": 0.6263217159649557,
+      "grad_norm": 0.0,
+      "learning_rate": 6.472383620829599e-06,
+      "loss": 0.9125,
+      "step": 22805
+    },
+    {
+      "epoch": 0.6263491801927988,
+      "grad_norm": 0.0,
+      "learning_rate": 6.471551302895713e-06,
+      "loss": 0.8924,
+      "step": 22806
+    },
+    {
+      "epoch": 0.6263766444206421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.470719012880285e-06,
+      "loss": 0.8569,
+      "step": 22807
+    },
+    {
+      "epoch": 0.6264041086484854,
+      "grad_norm": 0.0,
+      "learning_rate": 6.469886750789891e-06,
+      "loss": 0.7393,
+      "step": 22808
+    },
+    {
+      "epoch": 0.6264315728763286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.469054516631121e-06,
+      "loss": 0.7541,
+      "step": 22809
+    },
+    {
+      "epoch": 0.6264590371041718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4682223104105645e-06,
+      "loss": 0.8537,
+      "step": 22810
+    },
+    {
+      "epoch": 0.626486501332015,
+      "grad_norm": 0.0,
+      "learning_rate": 6.467390132134799e-06,
+      "loss": 0.7497,
+      "step": 22811
+    },
+    {
+      "epoch": 0.6265139655598583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.466557981810408e-06,
+      "loss": 0.747,
+      "step": 22812
+    },
+    {
+      "epoch": 0.6265414297877016,
+      "grad_norm": 0.0,
+      "learning_rate": 6.465725859443979e-06,
+      "loss": 0.8181,
+      "step": 22813
+    },
+    {
+      "epoch": 0.6265688940155447,
+      "grad_norm": 0.0,
+      "learning_rate": 6.464893765042098e-06,
+      "loss": 0.7636,
+      "step": 22814
+    },
+    {
+      "epoch": 0.626596358243388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4640616986113454e-06,
+      "loss": 0.7786,
+      "step": 22815
+    },
+    {
+      "epoch": 0.6266238224712313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.463229660158311e-06,
+      "loss": 0.8478,
+      "step": 22816
+    },
+    {
+      "epoch": 0.6266512866990744,
+      "grad_norm": 0.0,
+      "learning_rate": 6.462397649689569e-06,
+      "loss": 0.8297,
+      "step": 22817
+    },
+    {
+      "epoch": 0.6266787509269177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.461565667211707e-06,
+      "loss": 0.7872,
+      "step": 22818
+    },
+    {
+      "epoch": 0.6267062151547609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.460733712731311e-06,
+      "loss": 0.8908,
+      "step": 22819
+    },
+    {
+      "epoch": 0.6267336793826042,
+      "grad_norm": 0.0,
+      "learning_rate": 6.459901786254957e-06,
+      "loss": 0.8222,
+      "step": 22820
+    },
+    {
+      "epoch": 0.6267611436104474,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4590698877892325e-06,
+      "loss": 0.8453,
+      "step": 22821
+    },
+    {
+      "epoch": 0.6267886078382906,
+      "grad_norm": 0.0,
+      "learning_rate": 6.458238017340724e-06,
+      "loss": 0.7961,
+      "step": 22822
+    },
+    {
+      "epoch": 0.6268160720661339,
+      "grad_norm": 0.0,
+      "learning_rate": 6.457406174916003e-06,
+      "loss": 0.8625,
+      "step": 22823
+    },
+    {
+      "epoch": 0.626843536293977,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4565743605216566e-06,
+      "loss": 0.8829,
+      "step": 22824
+    },
+    {
+      "epoch": 0.6268710005218203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4557425741642655e-06,
+      "loss": 0.7563,
+      "step": 22825
+    },
+    {
+      "epoch": 0.6268984647496636,
+      "grad_norm": 0.0,
+      "learning_rate": 6.454910815850415e-06,
+      "loss": 0.7575,
+      "step": 22826
+    },
+    {
+      "epoch": 0.6269259289775068,
+      "grad_norm": 0.0,
+      "learning_rate": 6.454079085586682e-06,
+      "loss": 0.8617,
+      "step": 22827
+    },
+    {
+      "epoch": 0.62695339320535,
+      "grad_norm": 0.0,
+      "learning_rate": 6.453247383379652e-06,
+      "loss": 0.8862,
+      "step": 22828
+    },
+    {
+      "epoch": 0.6269808574331933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.452415709235898e-06,
+      "loss": 0.8192,
+      "step": 22829
+    },
+    {
+      "epoch": 0.6270083216610365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.451584063162007e-06,
+      "loss": 0.7973,
+      "step": 22830
+    },
+    {
+      "epoch": 0.6270357858888798,
+      "grad_norm": 0.0,
+      "learning_rate": 6.450752445164557e-06,
+      "loss": 0.9539,
+      "step": 22831
+    },
+    {
+      "epoch": 0.6270632501167229,
+      "grad_norm": 0.0,
+      "learning_rate": 6.449920855250128e-06,
+      "loss": 0.7895,
+      "step": 22832
+    },
+    {
+      "epoch": 0.6270907143445662,
+      "grad_norm": 0.0,
+      "learning_rate": 6.449089293425299e-06,
+      "loss": 0.8195,
+      "step": 22833
+    },
+    {
+      "epoch": 0.6271181785724095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.448257759696657e-06,
+      "loss": 0.868,
+      "step": 22834
+    },
+    {
+      "epoch": 0.6271456428002526,
+      "grad_norm": 0.0,
+      "learning_rate": 6.447426254070772e-06,
+      "loss": 0.8347,
+      "step": 22835
+    },
+    {
+      "epoch": 0.6271731070280959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.446594776554225e-06,
+      "loss": 0.884,
+      "step": 22836
+    },
+    {
+      "epoch": 0.6272005712559391,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4457633271535965e-06,
+      "loss": 0.9045,
+      "step": 22837
+    },
+    {
+      "epoch": 0.6272280354837824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.444931905875468e-06,
+      "loss": 0.8781,
+      "step": 22838
+    },
+    {
+      "epoch": 0.6272554997116256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.444100512726412e-06,
+      "loss": 0.8404,
+      "step": 22839
+    },
+    {
+      "epoch": 0.6272829639394688,
+      "grad_norm": 0.0,
+      "learning_rate": 6.443269147713016e-06,
+      "loss": 0.7884,
+      "step": 22840
+    },
+    {
+      "epoch": 0.6273104281673121,
+      "grad_norm": 0.0,
+      "learning_rate": 6.442437810841846e-06,
+      "loss": 0.9007,
+      "step": 22841
+    },
+    {
+      "epoch": 0.6273378923951554,
+      "grad_norm": 0.0,
+      "learning_rate": 6.441606502119489e-06,
+      "loss": 0.8977,
+      "step": 22842
+    },
+    {
+      "epoch": 0.6273653566229985,
+      "grad_norm": 0.0,
+      "learning_rate": 6.440775221552519e-06,
+      "loss": 0.8057,
+      "step": 22843
+    },
+    {
+      "epoch": 0.6273928208508418,
+      "grad_norm": 0.0,
+      "learning_rate": 6.439943969147513e-06,
+      "loss": 0.8547,
+      "step": 22844
+    },
+    {
+      "epoch": 0.627420285078685,
+      "grad_norm": 0.0,
+      "learning_rate": 6.439112744911049e-06,
+      "loss": 0.8589,
+      "step": 22845
+    },
+    {
+      "epoch": 0.6274477493065282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.43828154884971e-06,
+      "loss": 0.8794,
+      "step": 22846
+    },
+    {
+      "epoch": 0.6274752135343715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.437450380970064e-06,
+      "loss": 0.9037,
+      "step": 22847
+    },
+    {
+      "epoch": 0.6275026777622147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4366192412786875e-06,
+      "loss": 0.8279,
+      "step": 22848
+    },
+    {
+      "epoch": 0.627530141990058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.435788129782161e-06,
+      "loss": 0.7471,
+      "step": 22849
+    },
+    {
+      "epoch": 0.6275576062179011,
+      "grad_norm": 0.0,
+      "learning_rate": 6.434957046487062e-06,
+      "loss": 0.8139,
+      "step": 22850
+    },
+    {
+      "epoch": 0.6275850704457444,
+      "grad_norm": 0.0,
+      "learning_rate": 6.434125991399963e-06,
+      "loss": 0.8205,
+      "step": 22851
+    },
+    {
+      "epoch": 0.6276125346735877,
+      "grad_norm": 0.0,
+      "learning_rate": 6.433294964527441e-06,
+      "loss": 0.7945,
+      "step": 22852
+    },
+    {
+      "epoch": 0.6276399989014309,
+      "grad_norm": 0.0,
+      "learning_rate": 6.432463965876068e-06,
+      "loss": 0.8686,
+      "step": 22853
+    },
+    {
+      "epoch": 0.6276674631292741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.431632995452422e-06,
+      "loss": 0.8467,
+      "step": 22854
+    },
+    {
+      "epoch": 0.6276949273571174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.430802053263081e-06,
+      "loss": 0.8444,
+      "step": 22855
+    },
+    {
+      "epoch": 0.6277223915849606,
+      "grad_norm": 0.0,
+      "learning_rate": 6.429971139314615e-06,
+      "loss": 0.8988,
+      "step": 22856
+    },
+    {
+      "epoch": 0.6277498558128038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.429140253613604e-06,
+      "loss": 0.7473,
+      "step": 22857
+    },
+    {
+      "epoch": 0.627777320040647,
+      "grad_norm": 0.0,
+      "learning_rate": 6.428309396166613e-06,
+      "loss": 0.9481,
+      "step": 22858
+    },
+    {
+      "epoch": 0.6278047842684903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.427478566980225e-06,
+      "loss": 0.8538,
+      "step": 22859
+    },
+    {
+      "epoch": 0.6278322484963336,
+      "grad_norm": 0.0,
+      "learning_rate": 6.426647766061007e-06,
+      "loss": 0.9019,
+      "step": 22860
+    },
+    {
+      "epoch": 0.6278597127241767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.425816993415538e-06,
+      "loss": 0.8723,
+      "step": 22861
+    },
+    {
+      "epoch": 0.62788717695202,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4249862490503916e-06,
+      "loss": 0.9003,
+      "step": 22862
+    },
+    {
+      "epoch": 0.6279146411798632,
+      "grad_norm": 0.0,
+      "learning_rate": 6.424155532972136e-06,
+      "loss": 0.8715,
+      "step": 22863
+    },
+    {
+      "epoch": 0.6279421054077065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4233248451873495e-06,
+      "loss": 0.8476,
+      "step": 22864
+    },
+    {
+      "epoch": 0.6279695696355497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.422494185702599e-06,
+      "loss": 0.8263,
+      "step": 22865
+    },
+    {
+      "epoch": 0.6279970338633929,
+      "grad_norm": 0.0,
+      "learning_rate": 6.421663554524462e-06,
+      "loss": 0.8835,
+      "step": 22866
+    },
+    {
+      "epoch": 0.6280244980912362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.420832951659509e-06,
+      "loss": 0.8,
+      "step": 22867
+    },
+    {
+      "epoch": 0.6280519623190794,
+      "grad_norm": 0.0,
+      "learning_rate": 6.420002377114311e-06,
+      "loss": 0.9254,
+      "step": 22868
+    },
+    {
+      "epoch": 0.6280794265469226,
+      "grad_norm": 0.0,
+      "learning_rate": 6.419171830895445e-06,
+      "loss": 0.8831,
+      "step": 22869
+    },
+    {
+      "epoch": 0.6281068907747659,
+      "grad_norm": 0.0,
+      "learning_rate": 6.418341313009475e-06,
+      "loss": 0.8508,
+      "step": 22870
+    },
+    {
+      "epoch": 0.6281343550026091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.417510823462978e-06,
+      "loss": 0.7852,
+      "step": 22871
+    },
+    {
+      "epoch": 0.6281618192304523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.41668036226252e-06,
+      "loss": 0.9375,
+      "step": 22872
+    },
+    {
+      "epoch": 0.6281892834582956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.415849929414676e-06,
+      "loss": 0.873,
+      "step": 22873
+    },
+    {
+      "epoch": 0.6282167476861388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.415019524926018e-06,
+      "loss": 0.8157,
+      "step": 22874
+    },
+    {
+      "epoch": 0.628244211913982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.414189148803113e-06,
+      "loss": 0.8716,
+      "step": 22875
+    },
+    {
+      "epoch": 0.6282716761418252,
+      "grad_norm": 0.0,
+      "learning_rate": 6.413358801052532e-06,
+      "loss": 0.9025,
+      "step": 22876
+    },
+    {
+      "epoch": 0.6282991403696685,
+      "grad_norm": 0.0,
+      "learning_rate": 6.412528481680845e-06,
+      "loss": 0.7906,
+      "step": 22877
+    },
+    {
+      "epoch": 0.6283266045975118,
+      "grad_norm": 0.0,
+      "learning_rate": 6.411698190694622e-06,
+      "loss": 0.8984,
+      "step": 22878
+    },
+    {
+      "epoch": 0.6283540688253549,
+      "grad_norm": 0.0,
+      "learning_rate": 6.410867928100435e-06,
+      "loss": 0.9683,
+      "step": 22879
+    },
+    {
+      "epoch": 0.6283815330531982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.41003769390485e-06,
+      "loss": 0.9661,
+      "step": 22880
+    },
+    {
+      "epoch": 0.6284089972810415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40920748811444e-06,
+      "loss": 0.7692,
+      "step": 22881
+    },
+    {
+      "epoch": 0.6284364615088847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.408377310735767e-06,
+      "loss": 0.8455,
+      "step": 22882
+    },
+    {
+      "epoch": 0.6284639257367279,
+      "grad_norm": 0.0,
+      "learning_rate": 6.407547161775406e-06,
+      "loss": 0.8467,
+      "step": 22883
+    },
+    {
+      "epoch": 0.6284913899645711,
+      "grad_norm": 0.0,
+      "learning_rate": 6.406717041239922e-06,
+      "loss": 0.776,
+      "step": 22884
+    },
+    {
+      "epoch": 0.6285188541924144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.405886949135884e-06,
+      "loss": 0.8582,
+      "step": 22885
+    },
+    {
+      "epoch": 0.6285463184202577,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4050568854698645e-06,
+      "loss": 0.8471,
+      "step": 22886
+    },
+    {
+      "epoch": 0.6285737826481008,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4042268502484265e-06,
+      "loss": 0.7845,
+      "step": 22887
+    },
+    {
+      "epoch": 0.6286012468759441,
+      "grad_norm": 0.0,
+      "learning_rate": 6.403396843478138e-06,
+      "loss": 0.8162,
+      "step": 22888
+    },
+    {
+      "epoch": 0.6286287111037873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.402566865165565e-06,
+      "loss": 0.8736,
+      "step": 22889
+    },
+    {
+      "epoch": 0.6286561753316305,
+      "grad_norm": 0.0,
+      "learning_rate": 6.401736915317277e-06,
+      "loss": 0.7806,
+      "step": 22890
+    },
+    {
+      "epoch": 0.6286836395594738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.400906993939842e-06,
+      "loss": 0.8157,
+      "step": 22891
+    },
+    {
+      "epoch": 0.628711103787317,
+      "grad_norm": 0.0,
+      "learning_rate": 6.400077101039823e-06,
+      "loss": 0.8511,
+      "step": 22892
+    },
+    {
+      "epoch": 0.6287385680151603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3992472366237935e-06,
+      "loss": 0.7733,
+      "step": 22893
+    },
+    {
+      "epoch": 0.6287660322430035,
+      "grad_norm": 0.0,
+      "learning_rate": 6.398417400698309e-06,
+      "loss": 0.8294,
+      "step": 22894
+    },
+    {
+      "epoch": 0.6287934964708467,
+      "grad_norm": 0.0,
+      "learning_rate": 6.397587593269944e-06,
+      "loss": 0.849,
+      "step": 22895
+    },
+    {
+      "epoch": 0.62882096069869,
+      "grad_norm": 0.0,
+      "learning_rate": 6.396757814345259e-06,
+      "loss": 0.8377,
+      "step": 22896
+    },
+    {
+      "epoch": 0.6288484249265331,
+      "grad_norm": 0.0,
+      "learning_rate": 6.395928063930823e-06,
+      "loss": 0.9019,
+      "step": 22897
+    },
+    {
+      "epoch": 0.6288758891543764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.395098342033201e-06,
+      "loss": 0.9084,
+      "step": 22898
+    },
+    {
+      "epoch": 0.6289033533822197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3942686486589574e-06,
+      "loss": 0.8943,
+      "step": 22899
+    },
+    {
+      "epoch": 0.6289308176100629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.393438983814657e-06,
+      "loss": 0.872,
+      "step": 22900
+    },
+    {
+      "epoch": 0.6289582818379061,
+      "grad_norm": 0.0,
+      "learning_rate": 6.392609347506864e-06,
+      "loss": 0.9503,
+      "step": 22901
+    },
+    {
+      "epoch": 0.6289857460657493,
+      "grad_norm": 0.0,
+      "learning_rate": 6.391779739742141e-06,
+      "loss": 0.7547,
+      "step": 22902
+    },
+    {
+      "epoch": 0.6290132102935926,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3909501605270586e-06,
+      "loss": 0.8498,
+      "step": 22903
+    },
+    {
+      "epoch": 0.6290406745214359,
+      "grad_norm": 0.0,
+      "learning_rate": 6.390120609868174e-06,
+      "loss": 0.9358,
+      "step": 22904
+    },
+    {
+      "epoch": 0.629068138749279,
+      "grad_norm": 0.0,
+      "learning_rate": 6.389291087772056e-06,
+      "loss": 0.8535,
+      "step": 22905
+    },
+    {
+      "epoch": 0.6290956029771223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.388461594245261e-06,
+      "loss": 0.8243,
+      "step": 22906
+    },
+    {
+      "epoch": 0.6291230672049656,
+      "grad_norm": 0.0,
+      "learning_rate": 6.387632129294361e-06,
+      "loss": 0.9317,
+      "step": 22907
+    },
+    {
+      "epoch": 0.6291505314328087,
+      "grad_norm": 0.0,
+      "learning_rate": 6.386802692925912e-06,
+      "loss": 0.876,
+      "step": 22908
+    },
+    {
+      "epoch": 0.629177995660652,
+      "grad_norm": 0.0,
+      "learning_rate": 6.385973285146481e-06,
+      "loss": 0.8276,
+      "step": 22909
+    },
+    {
+      "epoch": 0.6292054598884952,
+      "grad_norm": 0.0,
+      "learning_rate": 6.38514390596263e-06,
+      "loss": 0.8969,
+      "step": 22910
+    },
+    {
+      "epoch": 0.6292329241163385,
+      "grad_norm": 0.0,
+      "learning_rate": 6.38431455538092e-06,
+      "loss": 0.883,
+      "step": 22911
+    },
+    {
+      "epoch": 0.6292603883441817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.383485233407914e-06,
+      "loss": 0.923,
+      "step": 22912
+    },
+    {
+      "epoch": 0.6292878525720249,
+      "grad_norm": 0.0,
+      "learning_rate": 6.382655940050173e-06,
+      "loss": 0.8555,
+      "step": 22913
+    },
+    {
+      "epoch": 0.6293153167998682,
+      "grad_norm": 0.0,
+      "learning_rate": 6.381826675314258e-06,
+      "loss": 0.8875,
+      "step": 22914
+    },
+    {
+      "epoch": 0.6293427810277114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3809974392067354e-06,
+      "loss": 0.9189,
+      "step": 22915
+    },
+    {
+      "epoch": 0.6293702452555546,
+      "grad_norm": 0.0,
+      "learning_rate": 6.380168231734164e-06,
+      "loss": 0.8797,
+      "step": 22916
+    },
+    {
+      "epoch": 0.6293977094833979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3793390529031e-06,
+      "loss": 0.8511,
+      "step": 22917
+    },
+    {
+      "epoch": 0.6294251737112411,
+      "grad_norm": 0.0,
+      "learning_rate": 6.378509902720107e-06,
+      "loss": 0.8182,
+      "step": 22918
+    },
+    {
+      "epoch": 0.6294526379390843,
+      "grad_norm": 0.0,
+      "learning_rate": 6.377680781191751e-06,
+      "loss": 0.9979,
+      "step": 22919
+    },
+    {
+      "epoch": 0.6294801021669276,
+      "grad_norm": 0.0,
+      "learning_rate": 6.376851688324584e-06,
+      "loss": 0.9551,
+      "step": 22920
+    },
+    {
+      "epoch": 0.6295075663947708,
+      "grad_norm": 0.0,
+      "learning_rate": 6.376022624125169e-06,
+      "loss": 0.7668,
+      "step": 22921
+    },
+    {
+      "epoch": 0.6295350306226141,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3751935886000724e-06,
+      "loss": 0.9371,
+      "step": 22922
+    },
+    {
+      "epoch": 0.6295624948504572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.374364581755843e-06,
+      "loss": 0.8243,
+      "step": 22923
+    },
+    {
+      "epoch": 0.6295899590783005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3735356035990485e-06,
+      "loss": 0.7883,
+      "step": 22924
+    },
+    {
+      "epoch": 0.6296174233061438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3727066541362425e-06,
+      "loss": 0.8208,
+      "step": 22925
+    },
+    {
+      "epoch": 0.629644887533987,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3718777333739874e-06,
+      "loss": 0.9019,
+      "step": 22926
+    },
+    {
+      "epoch": 0.6296723517618302,
+      "grad_norm": 0.0,
+      "learning_rate": 6.371048841318842e-06,
+      "loss": 0.9032,
+      "step": 22927
+    },
+    {
+      "epoch": 0.6296998159896734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.370219977977366e-06,
+      "loss": 0.9207,
+      "step": 22928
+    },
+    {
+      "epoch": 0.6297272802175167,
+      "grad_norm": 0.0,
+      "learning_rate": 6.369391143356111e-06,
+      "loss": 0.8346,
+      "step": 22929
+    },
+    {
+      "epoch": 0.62975474444536,
+      "grad_norm": 0.0,
+      "learning_rate": 6.368562337461641e-06,
+      "loss": 0.8246,
+      "step": 22930
+    },
+    {
+      "epoch": 0.6297822086732031,
+      "grad_norm": 0.0,
+      "learning_rate": 6.367733560300515e-06,
+      "loss": 0.9939,
+      "step": 22931
+    },
+    {
+      "epoch": 0.6298096729010464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.366904811879284e-06,
+      "loss": 0.8462,
+      "step": 22932
+    },
+    {
+      "epoch": 0.6298371371288897,
+      "grad_norm": 0.0,
+      "learning_rate": 6.366076092204512e-06,
+      "loss": 0.787,
+      "step": 22933
+    },
+    {
+      "epoch": 0.6298646013567328,
+      "grad_norm": 0.0,
+      "learning_rate": 6.365247401282756e-06,
+      "loss": 0.8467,
+      "step": 22934
+    },
+    {
+      "epoch": 0.6298920655845761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.364418739120568e-06,
+      "loss": 0.8515,
+      "step": 22935
+    },
+    {
+      "epoch": 0.6299195298124193,
+      "grad_norm": 0.0,
+      "learning_rate": 6.363590105724507e-06,
+      "loss": 0.8421,
+      "step": 22936
+    },
+    {
+      "epoch": 0.6299469940402626,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3627615011011276e-06,
+      "loss": 0.9706,
+      "step": 22937
+    },
+    {
+      "epoch": 0.6299744582681058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.36193292525699e-06,
+      "loss": 0.9785,
+      "step": 22938
+    },
+    {
+      "epoch": 0.630001922495949,
+      "grad_norm": 0.0,
+      "learning_rate": 6.361104378198648e-06,
+      "loss": 0.9245,
+      "step": 22939
+    },
+    {
+      "epoch": 0.6300293867237923,
+      "grad_norm": 0.0,
+      "learning_rate": 6.360275859932662e-06,
+      "loss": 0.859,
+      "step": 22940
+    },
+    {
+      "epoch": 0.6300568509516355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3594473704655794e-06,
+      "loss": 0.8619,
+      "step": 22941
+    },
+    {
+      "epoch": 0.6300843151794787,
+      "grad_norm": 0.0,
+      "learning_rate": 6.358618909803957e-06,
+      "loss": 0.7615,
+      "step": 22942
+    },
+    {
+      "epoch": 0.630111779407322,
+      "grad_norm": 0.0,
+      "learning_rate": 6.357790477954357e-06,
+      "loss": 0.7978,
+      "step": 22943
+    },
+    {
+      "epoch": 0.6301392436351652,
+      "grad_norm": 0.0,
+      "learning_rate": 6.356962074923325e-06,
+      "loss": 0.8268,
+      "step": 22944
+    },
+    {
+      "epoch": 0.6301667078630084,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3561337007174215e-06,
+      "loss": 0.8235,
+      "step": 22945
+    },
+    {
+      "epoch": 0.6301941720908517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.355305355343204e-06,
+      "loss": 0.9632,
+      "step": 22946
+    },
+    {
+      "epoch": 0.6302216363186949,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3544770388072205e-06,
+      "loss": 0.7634,
+      "step": 22947
+    },
+    {
+      "epoch": 0.6302491005465382,
+      "grad_norm": 0.0,
+      "learning_rate": 6.353648751116025e-06,
+      "loss": 0.8972,
+      "step": 22948
+    },
+    {
+      "epoch": 0.6302765647743813,
+      "grad_norm": 0.0,
+      "learning_rate": 6.352820492276172e-06,
+      "loss": 0.8632,
+      "step": 22949
+    },
+    {
+      "epoch": 0.6303040290022246,
+      "grad_norm": 0.0,
+      "learning_rate": 6.351992262294217e-06,
+      "loss": 0.7406,
+      "step": 22950
+    },
+    {
+      "epoch": 0.6303314932300679,
+      "grad_norm": 0.0,
+      "learning_rate": 6.351164061176713e-06,
+      "loss": 0.8396,
+      "step": 22951
+    },
+    {
+      "epoch": 0.630358957457911,
+      "grad_norm": 0.0,
+      "learning_rate": 6.350335888930217e-06,
+      "loss": 0.8295,
+      "step": 22952
+    },
+    {
+      "epoch": 0.6303864216857543,
+      "grad_norm": 0.0,
+      "learning_rate": 6.349507745561271e-06,
+      "loss": 0.7756,
+      "step": 22953
+    },
+    {
+      "epoch": 0.6304138859135976,
+      "grad_norm": 0.0,
+      "learning_rate": 6.348679631076434e-06,
+      "loss": 0.9166,
+      "step": 22954
+    },
+    {
+      "epoch": 0.6304413501414408,
+      "grad_norm": 0.0,
+      "learning_rate": 6.34785154548226e-06,
+      "loss": 0.8058,
+      "step": 22955
+    },
+    {
+      "epoch": 0.630468814369284,
+      "grad_norm": 0.0,
+      "learning_rate": 6.347023488785298e-06,
+      "loss": 0.8156,
+      "step": 22956
+    },
+    {
+      "epoch": 0.6304962785971272,
+      "grad_norm": 0.0,
+      "learning_rate": 6.346195460992102e-06,
+      "loss": 0.8525,
+      "step": 22957
+    },
+    {
+      "epoch": 0.6305237428249705,
+      "grad_norm": 0.0,
+      "learning_rate": 6.345367462109224e-06,
+      "loss": 0.7457,
+      "step": 22958
+    },
+    {
+      "epoch": 0.6305512070528138,
+      "grad_norm": 0.0,
+      "learning_rate": 6.344539492143214e-06,
+      "loss": 0.7661,
+      "step": 22959
+    },
+    {
+      "epoch": 0.6305786712806569,
+      "grad_norm": 0.0,
+      "learning_rate": 6.343711551100622e-06,
+      "loss": 0.8394,
+      "step": 22960
+    },
+    {
+      "epoch": 0.6306061355085002,
+      "grad_norm": 0.0,
+      "learning_rate": 6.342883638988e-06,
+      "loss": 0.87,
+      "step": 22961
+    },
+    {
+      "epoch": 0.6306335997363434,
+      "grad_norm": 0.0,
+      "learning_rate": 6.342055755811903e-06,
+      "loss": 0.7983,
+      "step": 22962
+    },
+    {
+      "epoch": 0.6306610639641866,
+      "grad_norm": 0.0,
+      "learning_rate": 6.341227901578872e-06,
+      "loss": 0.8417,
+      "step": 22963
+    },
+    {
+      "epoch": 0.6306885281920299,
+      "grad_norm": 0.0,
+      "learning_rate": 6.34040007629547e-06,
+      "loss": 0.8489,
+      "step": 22964
+    },
+    {
+      "epoch": 0.6307159924198731,
+      "grad_norm": 0.0,
+      "learning_rate": 6.339572279968236e-06,
+      "loss": 0.905,
+      "step": 22965
+    },
+    {
+      "epoch": 0.6307434566477164,
+      "grad_norm": 0.0,
+      "learning_rate": 6.338744512603722e-06,
+      "loss": 0.818,
+      "step": 22966
+    },
+    {
+      "epoch": 0.6307709208755596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3379167742084844e-06,
+      "loss": 0.9493,
+      "step": 22967
+    },
+    {
+      "epoch": 0.6307983851034028,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3370890647890645e-06,
+      "loss": 0.8391,
+      "step": 22968
+    },
+    {
+      "epoch": 0.6308258493312461,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3362613843520145e-06,
+      "loss": 0.7227,
+      "step": 22969
+    },
+    {
+      "epoch": 0.6308533135590892,
+      "grad_norm": 0.0,
+      "learning_rate": 6.335433732903889e-06,
+      "loss": 0.9121,
+      "step": 22970
+    },
+    {
+      "epoch": 0.6308807777869325,
+      "grad_norm": 0.0,
+      "learning_rate": 6.334606110451228e-06,
+      "loss": 0.9099,
+      "step": 22971
+    },
+    {
+      "epoch": 0.6309082420147758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.333778517000582e-06,
+      "loss": 0.9016,
+      "step": 22972
+    },
+    {
+      "epoch": 0.630935706242619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3329509525585e-06,
+      "loss": 0.7974,
+      "step": 22973
+    },
+    {
+      "epoch": 0.6309631704704622,
+      "grad_norm": 0.0,
+      "learning_rate": 6.332123417131533e-06,
+      "loss": 0.8445,
+      "step": 22974
+    },
+    {
+      "epoch": 0.6309906346983054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.331295910726225e-06,
+      "loss": 0.8254,
+      "step": 22975
+    },
+    {
+      "epoch": 0.6310180989261487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.330468433349127e-06,
+      "loss": 0.8682,
+      "step": 22976
+    },
+    {
+      "epoch": 0.631045563153992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.32964098500678e-06,
+      "loss": 0.8391,
+      "step": 22977
+    },
+    {
+      "epoch": 0.6310730273818351,
+      "grad_norm": 0.0,
+      "learning_rate": 6.328813565705737e-06,
+      "loss": 0.8266,
+      "step": 22978
+    },
+    {
+      "epoch": 0.6311004916096784,
+      "grad_norm": 0.0,
+      "learning_rate": 6.327986175452545e-06,
+      "loss": 0.8271,
+      "step": 22979
+    },
+    {
+      "epoch": 0.6311279558375217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.327158814253746e-06,
+      "loss": 0.8291,
+      "step": 22980
+    },
+    {
+      "epoch": 0.6311554200653648,
+      "grad_norm": 0.0,
+      "learning_rate": 6.326331482115893e-06,
+      "loss": 0.8937,
+      "step": 22981
+    },
+    {
+      "epoch": 0.6311828842932081,
+      "grad_norm": 0.0,
+      "learning_rate": 6.325504179045525e-06,
+      "loss": 0.8436,
+      "step": 22982
+    },
+    {
+      "epoch": 0.6312103485210513,
+      "grad_norm": 0.0,
+      "learning_rate": 6.324676905049193e-06,
+      "loss": 0.9083,
+      "step": 22983
+    },
+    {
+      "epoch": 0.6312378127488946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.323849660133439e-06,
+      "loss": 0.7757,
+      "step": 22984
+    },
+    {
+      "epoch": 0.6312652769767378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3230224443048115e-06,
+      "loss": 0.8911,
+      "step": 22985
+    },
+    {
+      "epoch": 0.631292741204581,
+      "grad_norm": 0.0,
+      "learning_rate": 6.322195257569855e-06,
+      "loss": 0.9751,
+      "step": 22986
+    },
+    {
+      "epoch": 0.6313202054324243,
+      "grad_norm": 0.0,
+      "learning_rate": 6.321368099935116e-06,
+      "loss": 0.923,
+      "step": 22987
+    },
+    {
+      "epoch": 0.6313476696602675,
+      "grad_norm": 0.0,
+      "learning_rate": 6.320540971407136e-06,
+      "loss": 0.8623,
+      "step": 22988
+    },
+    {
+      "epoch": 0.6313751338881107,
+      "grad_norm": 0.0,
+      "learning_rate": 6.31971387199246e-06,
+      "loss": 0.947,
+      "step": 22989
+    },
+    {
+      "epoch": 0.631402598115954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.318886801697634e-06,
+      "loss": 0.8103,
+      "step": 22990
+    },
+    {
+      "epoch": 0.6314300623437972,
+      "grad_norm": 0.0,
+      "learning_rate": 6.318059760529203e-06,
+      "loss": 0.8102,
+      "step": 22991
+    },
+    {
+      "epoch": 0.6314575265716404,
+      "grad_norm": 0.0,
+      "learning_rate": 6.317232748493707e-06,
+      "loss": 0.8566,
+      "step": 22992
+    },
+    {
+      "epoch": 0.6314849907994837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.316405765597696e-06,
+      "loss": 0.8607,
+      "step": 22993
+    },
+    {
+      "epoch": 0.6315124550273269,
+      "grad_norm": 0.0,
+      "learning_rate": 6.315578811847706e-06,
+      "loss": 0.8882,
+      "step": 22994
+    },
+    {
+      "epoch": 0.6315399192551702,
+      "grad_norm": 0.0,
+      "learning_rate": 6.314751887250285e-06,
+      "loss": 0.7235,
+      "step": 22995
+    },
+    {
+      "epoch": 0.6315673834830133,
+      "grad_norm": 0.0,
+      "learning_rate": 6.313924991811973e-06,
+      "loss": 0.8965,
+      "step": 22996
+    },
+    {
+      "epoch": 0.6315948477108566,
+      "grad_norm": 0.0,
+      "learning_rate": 6.313098125539314e-06,
+      "loss": 0.7873,
+      "step": 22997
+    },
+    {
+      "epoch": 0.6316223119386999,
+      "grad_norm": 0.0,
+      "learning_rate": 6.312271288438853e-06,
+      "loss": 0.912,
+      "step": 22998
+    },
+    {
+      "epoch": 0.631649776166543,
+      "grad_norm": 0.0,
+      "learning_rate": 6.31144448051713e-06,
+      "loss": 0.919,
+      "step": 22999
+    },
+    {
+      "epoch": 0.6316772403943863,
+      "grad_norm": 0.0,
+      "learning_rate": 6.310617701780688e-06,
+      "loss": 0.8816,
+      "step": 23000
+    },
+    {
+      "epoch": 0.6317047046222295,
+      "grad_norm": 0.0,
+      "learning_rate": 6.309790952236066e-06,
+      "loss": 0.8456,
+      "step": 23001
+    },
+    {
+      "epoch": 0.6317321688500728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.308964231889806e-06,
+      "loss": 0.873,
+      "step": 23002
+    },
+    {
+      "epoch": 0.631759633077916,
+      "grad_norm": 0.0,
+      "learning_rate": 6.308137540748454e-06,
+      "loss": 0.9753,
+      "step": 23003
+    },
+    {
+      "epoch": 0.6317870973057592,
+      "grad_norm": 0.0,
+      "learning_rate": 6.307310878818544e-06,
+      "loss": 0.9432,
+      "step": 23004
+    },
+    {
+      "epoch": 0.6318145615336025,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3064842461066265e-06,
+      "loss": 0.7695,
+      "step": 23005
+    },
+    {
+      "epoch": 0.6318420257614458,
+      "grad_norm": 0.0,
+      "learning_rate": 6.305657642619232e-06,
+      "loss": 0.8744,
+      "step": 23006
+    },
+    {
+      "epoch": 0.6318694899892889,
+      "grad_norm": 0.0,
+      "learning_rate": 6.304831068362907e-06,
+      "loss": 0.9533,
+      "step": 23007
+    },
+    {
+      "epoch": 0.6318969542171322,
+      "grad_norm": 0.0,
+      "learning_rate": 6.304004523344188e-06,
+      "loss": 0.8185,
+      "step": 23008
+    },
+    {
+      "epoch": 0.6319244184449754,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3031780075696165e-06,
+      "loss": 0.8262,
+      "step": 23009
+    },
+    {
+      "epoch": 0.6319518826728187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.302351521045735e-06,
+      "loss": 0.88,
+      "step": 23010
+    },
+    {
+      "epoch": 0.6319793469006619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.301525063779081e-06,
+      "loss": 0.825,
+      "step": 23011
+    },
+    {
+      "epoch": 0.6320068111285051,
+      "grad_norm": 0.0,
+      "learning_rate": 6.300698635776195e-06,
+      "loss": 0.7896,
+      "step": 23012
+    },
+    {
+      "epoch": 0.6320342753563484,
+      "grad_norm": 0.0,
+      "learning_rate": 6.299872237043609e-06,
+      "loss": 0.856,
+      "step": 23013
+    },
+    {
+      "epoch": 0.6320617395841915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2990458675878695e-06,
+      "loss": 0.8666,
+      "step": 23014
+    },
+    {
+      "epoch": 0.6320892038120348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.298219527415513e-06,
+      "loss": 0.9009,
+      "step": 23015
+    },
+    {
+      "epoch": 0.6321166680398781,
+      "grad_norm": 0.0,
+      "learning_rate": 6.297393216533077e-06,
+      "loss": 0.8632,
+      "step": 23016
+    },
+    {
+      "epoch": 0.6321441322677213,
+      "grad_norm": 0.0,
+      "learning_rate": 6.296566934947104e-06,
+      "loss": 0.9893,
+      "step": 23017
+    },
+    {
+      "epoch": 0.6321715964955645,
+      "grad_norm": 0.0,
+      "learning_rate": 6.295740682664122e-06,
+      "loss": 0.9399,
+      "step": 23018
+    },
+    {
+      "epoch": 0.6321990607234078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.294914459690679e-06,
+      "loss": 0.8711,
+      "step": 23019
+    },
+    {
+      "epoch": 0.632226524951251,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2940882660333045e-06,
+      "loss": 0.8117,
+      "step": 23020
+    },
+    {
+      "epoch": 0.6322539891790943,
+      "grad_norm": 0.0,
+      "learning_rate": 6.29326210169854e-06,
+      "loss": 0.7902,
+      "step": 23021
+    },
+    {
+      "epoch": 0.6322814534069374,
+      "grad_norm": 0.0,
+      "learning_rate": 6.292435966692922e-06,
+      "loss": 0.8371,
+      "step": 23022
+    },
+    {
+      "epoch": 0.6323089176347807,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2916098610229894e-06,
+      "loss": 0.7741,
+      "step": 23023
+    },
+    {
+      "epoch": 0.632336381862624,
+      "grad_norm": 0.0,
+      "learning_rate": 6.290783784695275e-06,
+      "loss": 0.8639,
+      "step": 23024
+    },
+    {
+      "epoch": 0.6323638460904671,
+      "grad_norm": 0.0,
+      "learning_rate": 6.289957737716313e-06,
+      "loss": 0.8964,
+      "step": 23025
+    },
+    {
+      "epoch": 0.6323913103183104,
+      "grad_norm": 0.0,
+      "learning_rate": 6.289131720092644e-06,
+      "loss": 0.8571,
+      "step": 23026
+    },
+    {
+      "epoch": 0.6324187745461536,
+      "grad_norm": 0.0,
+      "learning_rate": 6.288305731830804e-06,
+      "loss": 0.8632,
+      "step": 23027
+    },
+    {
+      "epoch": 0.6324462387739969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2874797729373235e-06,
+      "loss": 0.9415,
+      "step": 23028
+    },
+    {
+      "epoch": 0.6324737030018401,
+      "grad_norm": 0.0,
+      "learning_rate": 6.286653843418747e-06,
+      "loss": 0.8006,
+      "step": 23029
+    },
+    {
+      "epoch": 0.6325011672296833,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2858279432816e-06,
+      "loss": 0.7972,
+      "step": 23030
+    },
+    {
+      "epoch": 0.6325286314575266,
+      "grad_norm": 0.0,
+      "learning_rate": 6.28500207253242e-06,
+      "loss": 0.7591,
+      "step": 23031
+    },
+    {
+      "epoch": 0.6325560956853699,
+      "grad_norm": 0.0,
+      "learning_rate": 6.284176231177743e-06,
+      "loss": 0.7628,
+      "step": 23032
+    },
+    {
+      "epoch": 0.632583559913213,
+      "grad_norm": 0.0,
+      "learning_rate": 6.283350419224103e-06,
+      "loss": 0.7914,
+      "step": 23033
+    },
+    {
+      "epoch": 0.6326110241410563,
+      "grad_norm": 0.0,
+      "learning_rate": 6.282524636678038e-06,
+      "loss": 0.847,
+      "step": 23034
+    },
+    {
+      "epoch": 0.6326384883688995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2816988835460735e-06,
+      "loss": 0.8291,
+      "step": 23035
+    },
+    {
+      "epoch": 0.6326659525967427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.28087315983475e-06,
+      "loss": 0.7824,
+      "step": 23036
+    },
+    {
+      "epoch": 0.632693416824586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.280047465550597e-06,
+      "loss": 0.8791,
+      "step": 23037
+    },
+    {
+      "epoch": 0.6327208810524292,
+      "grad_norm": 0.0,
+      "learning_rate": 6.279221800700149e-06,
+      "loss": 0.8102,
+      "step": 23038
+    },
+    {
+      "epoch": 0.6327483452802725,
+      "grad_norm": 0.0,
+      "learning_rate": 6.278396165289941e-06,
+      "loss": 0.9076,
+      "step": 23039
+    },
+    {
+      "epoch": 0.6327758095081156,
+      "grad_norm": 0.0,
+      "learning_rate": 6.277570559326508e-06,
+      "loss": 0.8765,
+      "step": 23040
+    },
+    {
+      "epoch": 0.6328032737359589,
+      "grad_norm": 0.0,
+      "learning_rate": 6.276744982816374e-06,
+      "loss": 0.8456,
+      "step": 23041
+    },
+    {
+      "epoch": 0.6328307379638022,
+      "grad_norm": 0.0,
+      "learning_rate": 6.275919435766075e-06,
+      "loss": 0.8261,
+      "step": 23042
+    },
+    {
+      "epoch": 0.6328582021916453,
+      "grad_norm": 0.0,
+      "learning_rate": 6.275093918182147e-06,
+      "loss": 0.9399,
+      "step": 23043
+    },
+    {
+      "epoch": 0.6328856664194886,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2742684300711155e-06,
+      "loss": 0.8964,
+      "step": 23044
+    },
+    {
+      "epoch": 0.6329131306473319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.273442971439516e-06,
+      "loss": 0.9623,
+      "step": 23045
+    },
+    {
+      "epoch": 0.6329405948751751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.272617542293884e-06,
+      "loss": 0.8665,
+      "step": 23046
+    },
+    {
+      "epoch": 0.6329680591030183,
+      "grad_norm": 0.0,
+      "learning_rate": 6.27179214264074e-06,
+      "loss": 0.8324,
+      "step": 23047
+    },
+    {
+      "epoch": 0.6329955233308615,
+      "grad_norm": 0.0,
+      "learning_rate": 6.270966772486624e-06,
+      "loss": 0.7612,
+      "step": 23048
+    },
+    {
+      "epoch": 0.6330229875587048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2701414318380615e-06,
+      "loss": 0.8135,
+      "step": 23049
+    },
+    {
+      "epoch": 0.6330504517865481,
+      "grad_norm": 0.0,
+      "learning_rate": 6.269316120701585e-06,
+      "loss": 0.8611,
+      "step": 23050
+    },
+    {
+      "epoch": 0.6330779160143912,
+      "grad_norm": 0.0,
+      "learning_rate": 6.268490839083726e-06,
+      "loss": 0.917,
+      "step": 23051
+    },
+    {
+      "epoch": 0.6331053802422345,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2676655869910155e-06,
+      "loss": 0.8274,
+      "step": 23052
+    },
+    {
+      "epoch": 0.6331328444700777,
+      "grad_norm": 0.0,
+      "learning_rate": 6.266840364429976e-06,
+      "loss": 0.7944,
+      "step": 23053
+    },
+    {
+      "epoch": 0.633160308697921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.266015171407142e-06,
+      "loss": 0.9206,
+      "step": 23054
+    },
+    {
+      "epoch": 0.6331877729257642,
+      "grad_norm": 0.0,
+      "learning_rate": 6.265190007929044e-06,
+      "loss": 0.8552,
+      "step": 23055
+    },
+    {
+      "epoch": 0.6332152371536074,
+      "grad_norm": 0.0,
+      "learning_rate": 6.264364874002208e-06,
+      "loss": 0.7933,
+      "step": 23056
+    },
+    {
+      "epoch": 0.6332427013814507,
+      "grad_norm": 0.0,
+      "learning_rate": 6.263539769633163e-06,
+      "loss": 0.7448,
+      "step": 23057
+    },
+    {
+      "epoch": 0.633270165609294,
+      "grad_norm": 0.0,
+      "learning_rate": 6.262714694828444e-06,
+      "loss": 0.8917,
+      "step": 23058
+    },
+    {
+      "epoch": 0.6332976298371371,
+      "grad_norm": 0.0,
+      "learning_rate": 6.261889649594569e-06,
+      "loss": 0.8218,
+      "step": 23059
+    },
+    {
+      "epoch": 0.6333250940649804,
+      "grad_norm": 0.0,
+      "learning_rate": 6.261064633938075e-06,
+      "loss": 0.9091,
+      "step": 23060
+    },
+    {
+      "epoch": 0.6333525582928236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2602396478654815e-06,
+      "loss": 0.8733,
+      "step": 23061
+    },
+    {
+      "epoch": 0.6333800225206668,
+      "grad_norm": 0.0,
+      "learning_rate": 6.259414691383321e-06,
+      "loss": 0.7749,
+      "step": 23062
+    },
+    {
+      "epoch": 0.6334074867485101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.258589764498124e-06,
+      "loss": 0.8714,
+      "step": 23063
+    },
+    {
+      "epoch": 0.6334349509763533,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2577648672164146e-06,
+      "loss": 0.8236,
+      "step": 23064
+    },
+    {
+      "epoch": 0.6334624152041965,
+      "grad_norm": 0.0,
+      "learning_rate": 6.256939999544715e-06,
+      "loss": 0.8675,
+      "step": 23065
+    },
+    {
+      "epoch": 0.6334898794320397,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2561151614895575e-06,
+      "loss": 0.871,
+      "step": 23066
+    },
+    {
+      "epoch": 0.633517343659883,
+      "grad_norm": 0.0,
+      "learning_rate": 6.255290353057468e-06,
+      "loss": 0.9366,
+      "step": 23067
+    },
+    {
+      "epoch": 0.6335448078877263,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2544655742549695e-06,
+      "loss": 0.8142,
+      "step": 23068
+    },
+    {
+      "epoch": 0.6335722721155694,
+      "grad_norm": 0.0,
+      "learning_rate": 6.253640825088592e-06,
+      "loss": 0.8075,
+      "step": 23069
+    },
+    {
+      "epoch": 0.6335997363434127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.252816105564862e-06,
+      "loss": 0.8126,
+      "step": 23070
+    },
+    {
+      "epoch": 0.633627200571256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.251991415690298e-06,
+      "loss": 0.9213,
+      "step": 23071
+    },
+    {
+      "epoch": 0.6336546647990992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2511667554714325e-06,
+      "loss": 0.8384,
+      "step": 23072
+    },
+    {
+      "epoch": 0.6336821290269424,
+      "grad_norm": 0.0,
+      "learning_rate": 6.250342124914786e-06,
+      "loss": 0.8163,
+      "step": 23073
+    },
+    {
+      "epoch": 0.6337095932547856,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2495175240268845e-06,
+      "loss": 0.8303,
+      "step": 23074
+    },
+    {
+      "epoch": 0.6337370574826289,
+      "grad_norm": 0.0,
+      "learning_rate": 6.248692952814257e-06,
+      "loss": 0.7664,
+      "step": 23075
+    },
+    {
+      "epoch": 0.6337645217104722,
+      "grad_norm": 0.0,
+      "learning_rate": 6.247868411283425e-06,
+      "loss": 0.9125,
+      "step": 23076
+    },
+    {
+      "epoch": 0.6337919859383153,
+      "grad_norm": 0.0,
+      "learning_rate": 6.247043899440906e-06,
+      "loss": 0.8672,
+      "step": 23077
+    },
+    {
+      "epoch": 0.6338194501661586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.246219417293231e-06,
+      "loss": 0.7972,
+      "step": 23078
+    },
+    {
+      "epoch": 0.6338469143940018,
+      "grad_norm": 0.0,
+      "learning_rate": 6.245394964846922e-06,
+      "loss": 0.8633,
+      "step": 23079
+    },
+    {
+      "epoch": 0.633874378621845,
+      "grad_norm": 0.0,
+      "learning_rate": 6.244570542108504e-06,
+      "loss": 0.9417,
+      "step": 23080
+    },
+    {
+      "epoch": 0.6339018428496883,
+      "grad_norm": 0.0,
+      "learning_rate": 6.243746149084495e-06,
+      "loss": 0.8313,
+      "step": 23081
+    },
+    {
+      "epoch": 0.6339293070775315,
+      "grad_norm": 0.0,
+      "learning_rate": 6.242921785781428e-06,
+      "loss": 0.8038,
+      "step": 23082
+    },
+    {
+      "epoch": 0.6339567713053748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.242097452205817e-06,
+      "loss": 0.888,
+      "step": 23083
+    },
+    {
+      "epoch": 0.633984235533218,
+      "grad_norm": 0.0,
+      "learning_rate": 6.241273148364183e-06,
+      "loss": 0.8433,
+      "step": 23084
+    },
+    {
+      "epoch": 0.6340116997610612,
+      "grad_norm": 0.0,
+      "learning_rate": 6.240448874263053e-06,
+      "loss": 0.9313,
+      "step": 23085
+    },
+    {
+      "epoch": 0.6340391639889045,
+      "grad_norm": 0.0,
+      "learning_rate": 6.239624629908947e-06,
+      "loss": 0.8116,
+      "step": 23086
+    },
+    {
+      "epoch": 0.6340666282167476,
+      "grad_norm": 0.0,
+      "learning_rate": 6.23880041530839e-06,
+      "loss": 0.8568,
+      "step": 23087
+    },
+    {
+      "epoch": 0.6340940924445909,
+      "grad_norm": 0.0,
+      "learning_rate": 6.237976230467903e-06,
+      "loss": 0.8301,
+      "step": 23088
+    },
+    {
+      "epoch": 0.6341215566724342,
+      "grad_norm": 0.0,
+      "learning_rate": 6.237152075394002e-06,
+      "loss": 0.9017,
+      "step": 23089
+    },
+    {
+      "epoch": 0.6341490209002774,
+      "grad_norm": 0.0,
+      "learning_rate": 6.236327950093211e-06,
+      "loss": 0.8196,
+      "step": 23090
+    },
+    {
+      "epoch": 0.6341764851281206,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2355038545720535e-06,
+      "loss": 0.8222,
+      "step": 23091
+    },
+    {
+      "epoch": 0.6342039493559638,
+      "grad_norm": 0.0,
+      "learning_rate": 6.234679788837046e-06,
+      "loss": 0.8013,
+      "step": 23092
+    },
+    {
+      "epoch": 0.6342314135838071,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2338557528947134e-06,
+      "loss": 0.8955,
+      "step": 23093
+    },
+    {
+      "epoch": 0.6342588778116504,
+      "grad_norm": 0.0,
+      "learning_rate": 6.23303174675157e-06,
+      "loss": 0.9592,
+      "step": 23094
+    },
+    {
+      "epoch": 0.6342863420394935,
+      "grad_norm": 0.0,
+      "learning_rate": 6.232207770414139e-06,
+      "loss": 0.7943,
+      "step": 23095
+    },
+    {
+      "epoch": 0.6343138062673368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.231383823888939e-06,
+      "loss": 0.8852,
+      "step": 23096
+    },
+    {
+      "epoch": 0.6343412704951801,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2305599071824895e-06,
+      "loss": 0.8655,
+      "step": 23097
+    },
+    {
+      "epoch": 0.6343687347230232,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2297360203013104e-06,
+      "loss": 0.8025,
+      "step": 23098
+    },
+    {
+      "epoch": 0.6343961989508665,
+      "grad_norm": 0.0,
+      "learning_rate": 6.228912163251924e-06,
+      "loss": 0.8236,
+      "step": 23099
+    },
+    {
+      "epoch": 0.6344236631787097,
+      "grad_norm": 0.0,
+      "learning_rate": 6.228088336040844e-06,
+      "loss": 0.8417,
+      "step": 23100
+    },
+    {
+      "epoch": 0.634451127406553,
+      "grad_norm": 0.0,
+      "learning_rate": 6.227264538674587e-06,
+      "loss": 0.7028,
+      "step": 23101
+    },
+    {
+      "epoch": 0.6344785916343962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.226440771159677e-06,
+      "loss": 0.8033,
+      "step": 23102
+    },
+    {
+      "epoch": 0.6345060558622394,
+      "grad_norm": 0.0,
+      "learning_rate": 6.225617033502629e-06,
+      "loss": 0.8257,
+      "step": 23103
+    },
+    {
+      "epoch": 0.6345335200900827,
+      "grad_norm": 0.0,
+      "learning_rate": 6.224793325709959e-06,
+      "loss": 0.8954,
+      "step": 23104
+    },
+    {
+      "epoch": 0.6345609843179258,
+      "grad_norm": 0.0,
+      "learning_rate": 6.223969647788191e-06,
+      "loss": 0.8475,
+      "step": 23105
+    },
+    {
+      "epoch": 0.6345884485457691,
+      "grad_norm": 0.0,
+      "learning_rate": 6.223145999743834e-06,
+      "loss": 0.83,
+      "step": 23106
+    },
+    {
+      "epoch": 0.6346159127736124,
+      "grad_norm": 0.0,
+      "learning_rate": 6.222322381583411e-06,
+      "loss": 0.9325,
+      "step": 23107
+    },
+    {
+      "epoch": 0.6346433770014556,
+      "grad_norm": 0.0,
+      "learning_rate": 6.221498793313434e-06,
+      "loss": 0.8918,
+      "step": 23108
+    },
+    {
+      "epoch": 0.6346708412292988,
+      "grad_norm": 0.0,
+      "learning_rate": 6.220675234940423e-06,
+      "loss": 0.9079,
+      "step": 23109
+    },
+    {
+      "epoch": 0.6346983054571421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.219851706470894e-06,
+      "loss": 0.863,
+      "step": 23110
+    },
+    {
+      "epoch": 0.6347257696849853,
+      "grad_norm": 0.0,
+      "learning_rate": 6.219028207911366e-06,
+      "loss": 0.8842,
+      "step": 23111
+    },
+    {
+      "epoch": 0.6347532339128286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.218204739268349e-06,
+      "loss": 0.7714,
+      "step": 23112
+    },
+    {
+      "epoch": 0.6347806981406717,
+      "grad_norm": 0.0,
+      "learning_rate": 6.217381300548359e-06,
+      "loss": 0.9036,
+      "step": 23113
+    },
+    {
+      "epoch": 0.634808162368515,
+      "grad_norm": 0.0,
+      "learning_rate": 6.216557891757914e-06,
+      "loss": 0.9039,
+      "step": 23114
+    },
+    {
+      "epoch": 0.6348356265963583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.21573451290353e-06,
+      "loss": 0.8522,
+      "step": 23115
+    },
+    {
+      "epoch": 0.6348630908242014,
+      "grad_norm": 0.0,
+      "learning_rate": 6.214911163991719e-06,
+      "loss": 0.7821,
+      "step": 23116
+    },
+    {
+      "epoch": 0.6348905550520447,
+      "grad_norm": 0.0,
+      "learning_rate": 6.214087845029e-06,
+      "loss": 0.8547,
+      "step": 23117
+    },
+    {
+      "epoch": 0.6349180192798879,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2132645560218814e-06,
+      "loss": 0.8743,
+      "step": 23118
+    },
+    {
+      "epoch": 0.6349454835077312,
+      "grad_norm": 0.0,
+      "learning_rate": 6.212441296976883e-06,
+      "loss": 0.807,
+      "step": 23119
+    },
+    {
+      "epoch": 0.6349729477355744,
+      "grad_norm": 0.0,
+      "learning_rate": 6.211618067900513e-06,
+      "loss": 0.9101,
+      "step": 23120
+    },
+    {
+      "epoch": 0.6350004119634176,
+      "grad_norm": 0.0,
+      "learning_rate": 6.210794868799288e-06,
+      "loss": 0.9363,
+      "step": 23121
+    },
+    {
+      "epoch": 0.6350278761912609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.209971699679725e-06,
+      "loss": 0.8785,
+      "step": 23122
+    },
+    {
+      "epoch": 0.6350553404191042,
+      "grad_norm": 0.0,
+      "learning_rate": 6.209148560548335e-06,
+      "loss": 0.8276,
+      "step": 23123
+    },
+    {
+      "epoch": 0.6350828046469473,
+      "grad_norm": 0.0,
+      "learning_rate": 6.208325451411628e-06,
+      "loss": 0.8244,
+      "step": 23124
+    },
+    {
+      "epoch": 0.6351102688747906,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2075023722761175e-06,
+      "loss": 0.7773,
+      "step": 23125
+    },
+    {
+      "epoch": 0.6351377331026338,
+      "grad_norm": 0.0,
+      "learning_rate": 6.206679323148317e-06,
+      "loss": 0.9285,
+      "step": 23126
+    },
+    {
+      "epoch": 0.635165197330477,
+      "grad_norm": 0.0,
+      "learning_rate": 6.205856304034741e-06,
+      "loss": 0.9151,
+      "step": 23127
+    },
+    {
+      "epoch": 0.6351926615583203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.205033314941898e-06,
+      "loss": 0.9123,
+      "step": 23128
+    },
+    {
+      "epoch": 0.6352201257861635,
+      "grad_norm": 0.0,
+      "learning_rate": 6.204210355876306e-06,
+      "loss": 0.8413,
+      "step": 23129
+    },
+    {
+      "epoch": 0.6352475900140068,
+      "grad_norm": 0.0,
+      "learning_rate": 6.203387426844467e-06,
+      "loss": 0.7841,
+      "step": 23130
+    },
+    {
+      "epoch": 0.63527505424185,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2025645278529e-06,
+      "loss": 0.9087,
+      "step": 23131
+    },
+    {
+      "epoch": 0.6353025184696932,
+      "grad_norm": 0.0,
+      "learning_rate": 6.201741658908111e-06,
+      "loss": 0.8419,
+      "step": 23132
+    },
+    {
+      "epoch": 0.6353299826975365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.200918820016614e-06,
+      "loss": 0.749,
+      "step": 23133
+    },
+    {
+      "epoch": 0.6353574469253797,
+      "grad_norm": 0.0,
+      "learning_rate": 6.200096011184921e-06,
+      "loss": 0.8579,
+      "step": 23134
+    },
+    {
+      "epoch": 0.6353849111532229,
+      "grad_norm": 0.0,
+      "learning_rate": 6.199273232419539e-06,
+      "loss": 0.9475,
+      "step": 23135
+    },
+    {
+      "epoch": 0.6354123753810662,
+      "grad_norm": 0.0,
+      "learning_rate": 6.198450483726981e-06,
+      "loss": 0.9651,
+      "step": 23136
+    },
+    {
+      "epoch": 0.6354398396089094,
+      "grad_norm": 0.0,
+      "learning_rate": 6.197627765113752e-06,
+      "loss": 0.8929,
+      "step": 23137
+    },
+    {
+      "epoch": 0.6354673038367527,
+      "grad_norm": 0.0,
+      "learning_rate": 6.196805076586365e-06,
+      "loss": 0.8723,
+      "step": 23138
+    },
+    {
+      "epoch": 0.6354947680645958,
+      "grad_norm": 0.0,
+      "learning_rate": 6.195982418151332e-06,
+      "loss": 0.8488,
+      "step": 23139
+    },
+    {
+      "epoch": 0.6355222322924391,
+      "grad_norm": 0.0,
+      "learning_rate": 6.195159789815158e-06,
+      "loss": 0.8845,
+      "step": 23140
+    },
+    {
+      "epoch": 0.6355496965202824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.194337191584358e-06,
+      "loss": 0.8979,
+      "step": 23141
+    },
+    {
+      "epoch": 0.6355771607481255,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1935146234654305e-06,
+      "loss": 0.8331,
+      "step": 23142
+    },
+    {
+      "epoch": 0.6356046249759688,
+      "grad_norm": 0.0,
+      "learning_rate": 6.192692085464893e-06,
+      "loss": 0.886,
+      "step": 23143
+    },
+    {
+      "epoch": 0.6356320892038121,
+      "grad_norm": 0.0,
+      "learning_rate": 6.191869577589249e-06,
+      "loss": 0.8682,
+      "step": 23144
+    },
+    {
+      "epoch": 0.6356595534316553,
+      "grad_norm": 0.0,
+      "learning_rate": 6.191047099845006e-06,
+      "loss": 0.7748,
+      "step": 23145
+    },
+    {
+      "epoch": 0.6356870176594985,
+      "grad_norm": 0.0,
+      "learning_rate": 6.190224652238679e-06,
+      "loss": 0.8451,
+      "step": 23146
+    },
+    {
+      "epoch": 0.6357144818873417,
+      "grad_norm": 0.0,
+      "learning_rate": 6.189402234776768e-06,
+      "loss": 0.8094,
+      "step": 23147
+    },
+    {
+      "epoch": 0.635741946115185,
+      "grad_norm": 0.0,
+      "learning_rate": 6.188579847465783e-06,
+      "loss": 0.9606,
+      "step": 23148
+    },
+    {
+      "epoch": 0.6357694103430283,
+      "grad_norm": 0.0,
+      "learning_rate": 6.187757490312229e-06,
+      "loss": 0.9888,
+      "step": 23149
+    },
+    {
+      "epoch": 0.6357968745708714,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1869351633226134e-06,
+      "loss": 0.8471,
+      "step": 23150
+    },
+    {
+      "epoch": 0.6358243387987147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.186112866503447e-06,
+      "loss": 0.819,
+      "step": 23151
+    },
+    {
+      "epoch": 0.6358518030265579,
+      "grad_norm": 0.0,
+      "learning_rate": 6.185290599861235e-06,
+      "loss": 0.894,
+      "step": 23152
+    },
+    {
+      "epoch": 0.6358792672544011,
+      "grad_norm": 0.0,
+      "learning_rate": 6.184468363402477e-06,
+      "loss": 0.9192,
+      "step": 23153
+    },
+    {
+      "epoch": 0.6359067314822444,
+      "grad_norm": 0.0,
+      "learning_rate": 6.183646157133684e-06,
+      "loss": 0.8048,
+      "step": 23154
+    },
+    {
+      "epoch": 0.6359341957100876,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1828239810613624e-06,
+      "loss": 0.7825,
+      "step": 23155
+    },
+    {
+      "epoch": 0.6359616599379309,
+      "grad_norm": 0.0,
+      "learning_rate": 6.182001835192013e-06,
+      "loss": 0.8562,
+      "step": 23156
+    },
+    {
+      "epoch": 0.6359891241657741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.181179719532145e-06,
+      "loss": 0.9531,
+      "step": 23157
+    },
+    {
+      "epoch": 0.6360165883936173,
+      "grad_norm": 0.0,
+      "learning_rate": 6.180357634088268e-06,
+      "loss": 0.796,
+      "step": 23158
+    },
+    {
+      "epoch": 0.6360440526214606,
+      "grad_norm": 0.0,
+      "learning_rate": 6.179535578866875e-06,
+      "loss": 0.9255,
+      "step": 23159
+    },
+    {
+      "epoch": 0.6360715168493037,
+      "grad_norm": 0.0,
+      "learning_rate": 6.178713553874478e-06,
+      "loss": 0.7923,
+      "step": 23160
+    },
+    {
+      "epoch": 0.636098981077147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.177891559117579e-06,
+      "loss": 0.8343,
+      "step": 23161
+    },
+    {
+      "epoch": 0.6361264453049903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.177069594602682e-06,
+      "loss": 0.8721,
+      "step": 23162
+    },
+    {
+      "epoch": 0.6361539095328335,
+      "grad_norm": 0.0,
+      "learning_rate": 6.176247660336292e-06,
+      "loss": 0.8335,
+      "step": 23163
+    },
+    {
+      "epoch": 0.6361813737606767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.175425756324915e-06,
+      "loss": 0.9064,
+      "step": 23164
+    },
+    {
+      "epoch": 0.6362088379885199,
+      "grad_norm": 0.0,
+      "learning_rate": 6.174603882575048e-06,
+      "loss": 0.8394,
+      "step": 23165
+    },
+    {
+      "epoch": 0.6362363022163632,
+      "grad_norm": 0.0,
+      "learning_rate": 6.173782039093195e-06,
+      "loss": 0.881,
+      "step": 23166
+    },
+    {
+      "epoch": 0.6362637664442065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.172960225885863e-06,
+      "loss": 0.8,
+      "step": 23167
+    },
+    {
+      "epoch": 0.6362912306720496,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1721384429595495e-06,
+      "loss": 0.8415,
+      "step": 23168
+    },
+    {
+      "epoch": 0.6363186948998929,
+      "grad_norm": 0.0,
+      "learning_rate": 6.171316690320761e-06,
+      "loss": 0.8447,
+      "step": 23169
+    },
+    {
+      "epoch": 0.6363461591277362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.170494967976002e-06,
+      "loss": 0.799,
+      "step": 23170
+    },
+    {
+      "epoch": 0.6363736233555793,
+      "grad_norm": 0.0,
+      "learning_rate": 6.169673275931764e-06,
+      "loss": 0.9278,
+      "step": 23171
+    },
+    {
+      "epoch": 0.6364010875834226,
+      "grad_norm": 0.0,
+      "learning_rate": 6.168851614194559e-06,
+      "loss": 0.9089,
+      "step": 23172
+    },
+    {
+      "epoch": 0.6364285518112658,
+      "grad_norm": 0.0,
+      "learning_rate": 6.168029982770881e-06,
+      "loss": 0.9135,
+      "step": 23173
+    },
+    {
+      "epoch": 0.6364560160391091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.167208381667235e-06,
+      "loss": 0.8302,
+      "step": 23174
+    },
+    {
+      "epoch": 0.6364834802669523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.166386810890122e-06,
+      "loss": 0.8122,
+      "step": 23175
+    },
+    {
+      "epoch": 0.6365109444947955,
+      "grad_norm": 0.0,
+      "learning_rate": 6.165565270446044e-06,
+      "loss": 0.9411,
+      "step": 23176
+    },
+    {
+      "epoch": 0.6365384087226388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.164743760341495e-06,
+      "loss": 0.9404,
+      "step": 23177
+    },
+    {
+      "epoch": 0.636565872950482,
+      "grad_norm": 0.0,
+      "learning_rate": 6.163922280582979e-06,
+      "loss": 0.8559,
+      "step": 23178
+    },
+    {
+      "epoch": 0.6365933371783252,
+      "grad_norm": 0.0,
+      "learning_rate": 6.163100831176998e-06,
+      "loss": 0.8741,
+      "step": 23179
+    },
+    {
+      "epoch": 0.6366208014061685,
+      "grad_norm": 0.0,
+      "learning_rate": 6.162279412130046e-06,
+      "loss": 0.927,
+      "step": 23180
+    },
+    {
+      "epoch": 0.6366482656340117,
+      "grad_norm": 0.0,
+      "learning_rate": 6.161458023448628e-06,
+      "loss": 0.7684,
+      "step": 23181
+    },
+    {
+      "epoch": 0.6366757298618549,
+      "grad_norm": 0.0,
+      "learning_rate": 6.160636665139243e-06,
+      "loss": 0.8831,
+      "step": 23182
+    },
+    {
+      "epoch": 0.6367031940896982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.159815337208386e-06,
+      "loss": 0.9098,
+      "step": 23183
+    },
+    {
+      "epoch": 0.6367306583175414,
+      "grad_norm": 0.0,
+      "learning_rate": 6.158994039662558e-06,
+      "loss": 0.795,
+      "step": 23184
+    },
+    {
+      "epoch": 0.6367581225453847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.158172772508257e-06,
+      "loss": 0.8754,
+      "step": 23185
+    },
+    {
+      "epoch": 0.6367855867732278,
+      "grad_norm": 0.0,
+      "learning_rate": 6.157351535751979e-06,
+      "loss": 0.8196,
+      "step": 23186
+    },
+    {
+      "epoch": 0.6368130510010711,
+      "grad_norm": 0.0,
+      "learning_rate": 6.156530329400228e-06,
+      "loss": 0.9562,
+      "step": 23187
+    },
+    {
+      "epoch": 0.6368405152289144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.155709153459498e-06,
+      "loss": 0.9712,
+      "step": 23188
+    },
+    {
+      "epoch": 0.6368679794567575,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1548880079362835e-06,
+      "loss": 0.893,
+      "step": 23189
+    },
+    {
+      "epoch": 0.6368954436846008,
+      "grad_norm": 0.0,
+      "learning_rate": 6.154066892837085e-06,
+      "loss": 0.9311,
+      "step": 23190
+    },
+    {
+      "epoch": 0.636922907912444,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1532458081684e-06,
+      "loss": 0.8538,
+      "step": 23191
+    },
+    {
+      "epoch": 0.6369503721402873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.152424753936722e-06,
+      "loss": 0.8047,
+      "step": 23192
+    },
+    {
+      "epoch": 0.6369778363681305,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1516037301485495e-06,
+      "loss": 0.7824,
+      "step": 23193
+    },
+    {
+      "epoch": 0.6370053005959737,
+      "grad_norm": 0.0,
+      "learning_rate": 6.150782736810385e-06,
+      "loss": 0.9795,
+      "step": 23194
+    },
+    {
+      "epoch": 0.637032764823817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149961773928713e-06,
+      "loss": 0.871,
+      "step": 23195
+    },
+    {
+      "epoch": 0.6370602290516603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149140841510037e-06,
+      "loss": 0.8133,
+      "step": 23196
+    },
+    {
+      "epoch": 0.6370876932795034,
+      "grad_norm": 0.0,
+      "learning_rate": 6.148319939560848e-06,
+      "loss": 0.8608,
+      "step": 23197
+    },
+    {
+      "epoch": 0.6371151575073467,
+      "grad_norm": 0.0,
+      "learning_rate": 6.147499068087645e-06,
+      "loss": 0.8322,
+      "step": 23198
+    },
+    {
+      "epoch": 0.6371426217351899,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1466782270969224e-06,
+      "loss": 0.8339,
+      "step": 23199
+    },
+    {
+      "epoch": 0.6371700859630332,
+      "grad_norm": 0.0,
+      "learning_rate": 6.145857416595177e-06,
+      "loss": 0.8002,
+      "step": 23200
+    },
+    {
+      "epoch": 0.6371975501908764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.145036636588897e-06,
+      "loss": 0.849,
+      "step": 23201
+    },
+    {
+      "epoch": 0.6372250144187196,
+      "grad_norm": 0.0,
+      "learning_rate": 6.144215887084579e-06,
+      "loss": 0.9025,
+      "step": 23202
+    },
+    {
+      "epoch": 0.6372524786465629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.143395168088721e-06,
+      "loss": 0.8493,
+      "step": 23203
+    },
+    {
+      "epoch": 0.637279942874406,
+      "grad_norm": 0.0,
+      "learning_rate": 6.142574479607813e-06,
+      "loss": 0.8851,
+      "step": 23204
+    },
+    {
+      "epoch": 0.6373074071022493,
+      "grad_norm": 0.0,
+      "learning_rate": 6.141753821648348e-06,
+      "loss": 0.8652,
+      "step": 23205
+    },
+    {
+      "epoch": 0.6373348713300926,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1409331942168284e-06,
+      "loss": 0.8047,
+      "step": 23206
+    },
+    {
+      "epoch": 0.6373623355579358,
+      "grad_norm": 0.0,
+      "learning_rate": 6.140112597319734e-06,
+      "loss": 0.928,
+      "step": 23207
+    },
+    {
+      "epoch": 0.637389799785779,
+      "grad_norm": 0.0,
+      "learning_rate": 6.139292030963566e-06,
+      "loss": 0.821,
+      "step": 23208
+    },
+    {
+      "epoch": 0.6374172640136223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.138471495154814e-06,
+      "loss": 0.9409,
+      "step": 23209
+    },
+    {
+      "epoch": 0.6374447282414655,
+      "grad_norm": 0.0,
+      "learning_rate": 6.137650989899971e-06,
+      "loss": 0.8716,
+      "step": 23210
+    },
+    {
+      "epoch": 0.6374721924693088,
+      "grad_norm": 0.0,
+      "learning_rate": 6.136830515205532e-06,
+      "loss": 0.8037,
+      "step": 23211
+    },
+    {
+      "epoch": 0.6374996566971519,
+      "grad_norm": 0.0,
+      "learning_rate": 6.136010071077987e-06,
+      "loss": 0.8236,
+      "step": 23212
+    },
+    {
+      "epoch": 0.6375271209249952,
+      "grad_norm": 0.0,
+      "learning_rate": 6.135189657523825e-06,
+      "loss": 0.7764,
+      "step": 23213
+    },
+    {
+      "epoch": 0.6375545851528385,
+      "grad_norm": 0.0,
+      "learning_rate": 6.134369274549537e-06,
+      "loss": 0.8251,
+      "step": 23214
+    },
+    {
+      "epoch": 0.6375820493806816,
+      "grad_norm": 0.0,
+      "learning_rate": 6.133548922161621e-06,
+      "loss": 0.8908,
+      "step": 23215
+    },
+    {
+      "epoch": 0.6376095136085249,
+      "grad_norm": 0.0,
+      "learning_rate": 6.13272860036656e-06,
+      "loss": 0.8379,
+      "step": 23216
+    },
+    {
+      "epoch": 0.6376369778363681,
+      "grad_norm": 0.0,
+      "learning_rate": 6.131908309170853e-06,
+      "loss": 0.8405,
+      "step": 23217
+    },
+    {
+      "epoch": 0.6376644420642114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.131088048580982e-06,
+      "loss": 0.7714,
+      "step": 23218
+    },
+    {
+      "epoch": 0.6376919062920546,
+      "grad_norm": 0.0,
+      "learning_rate": 6.130267818603441e-06,
+      "loss": 0.6918,
+      "step": 23219
+    },
+    {
+      "epoch": 0.6377193705198978,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1294476192447216e-06,
+      "loss": 0.8944,
+      "step": 23220
+    },
+    {
+      "epoch": 0.6377468347477411,
+      "grad_norm": 0.0,
+      "learning_rate": 6.128627450511311e-06,
+      "loss": 0.8902,
+      "step": 23221
+    },
+    {
+      "epoch": 0.6377742989755844,
+      "grad_norm": 0.0,
+      "learning_rate": 6.127807312409699e-06,
+      "loss": 0.9599,
+      "step": 23222
+    },
+    {
+      "epoch": 0.6378017632034275,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1269872049463794e-06,
+      "loss": 0.8024,
+      "step": 23223
+    },
+    {
+      "epoch": 0.6378292274312708,
+      "grad_norm": 0.0,
+      "learning_rate": 6.126167128127836e-06,
+      "loss": 0.7929,
+      "step": 23224
+    },
+    {
+      "epoch": 0.637856691659114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.125347081960555e-06,
+      "loss": 0.7358,
+      "step": 23225
+    },
+    {
+      "epoch": 0.6378841558869572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.124527066451031e-06,
+      "loss": 0.821,
+      "step": 23226
+    },
+    {
+      "epoch": 0.6379116201148005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.12370708160575e-06,
+      "loss": 0.928,
+      "step": 23227
+    },
+    {
+      "epoch": 0.6379390843426437,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1228871274312e-06,
+      "loss": 0.8006,
+      "step": 23228
+    },
+    {
+      "epoch": 0.637966548570487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1220672039338716e-06,
+      "loss": 0.816,
+      "step": 23229
+    },
+    {
+      "epoch": 0.6379940127983301,
+      "grad_norm": 0.0,
+      "learning_rate": 6.121247311120247e-06,
+      "loss": 0.8609,
+      "step": 23230
+    },
+    {
+      "epoch": 0.6380214770261734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.120427448996814e-06,
+      "loss": 0.8603,
+      "step": 23231
+    },
+    {
+      "epoch": 0.6380489412540167,
+      "grad_norm": 0.0,
+      "learning_rate": 6.119607617570065e-06,
+      "loss": 0.8088,
+      "step": 23232
+    },
+    {
+      "epoch": 0.6380764054818598,
+      "grad_norm": 0.0,
+      "learning_rate": 6.118787816846482e-06,
+      "loss": 0.8494,
+      "step": 23233
+    },
+    {
+      "epoch": 0.6381038697097031,
+      "grad_norm": 0.0,
+      "learning_rate": 6.117968046832552e-06,
+      "loss": 0.8563,
+      "step": 23234
+    },
+    {
+      "epoch": 0.6381313339375464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.117148307534769e-06,
+      "loss": 0.8906,
+      "step": 23235
+    },
+    {
+      "epoch": 0.6381587981653896,
+      "grad_norm": 0.0,
+      "learning_rate": 6.116328598959609e-06,
+      "loss": 0.7803,
+      "step": 23236
+    },
+    {
+      "epoch": 0.6381862623932328,
+      "grad_norm": 0.0,
+      "learning_rate": 6.11550892111356e-06,
+      "loss": 0.8661,
+      "step": 23237
+    },
+    {
+      "epoch": 0.638213726621076,
+      "grad_norm": 0.0,
+      "learning_rate": 6.114689274003107e-06,
+      "loss": 0.89,
+      "step": 23238
+    },
+    {
+      "epoch": 0.6382411908489193,
+      "grad_norm": 0.0,
+      "learning_rate": 6.113869657634742e-06,
+      "loss": 0.8103,
+      "step": 23239
+    },
+    {
+      "epoch": 0.6382686550767626,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1130500720149435e-06,
+      "loss": 0.8008,
+      "step": 23240
+    },
+    {
+      "epoch": 0.6382961193046057,
+      "grad_norm": 0.0,
+      "learning_rate": 6.112230517150202e-06,
+      "loss": 0.7933,
+      "step": 23241
+    },
+    {
+      "epoch": 0.638323583532449,
+      "grad_norm": 0.0,
+      "learning_rate": 6.111410993046993e-06,
+      "loss": 0.7663,
+      "step": 23242
+    },
+    {
+      "epoch": 0.6383510477602922,
+      "grad_norm": 0.0,
+      "learning_rate": 6.110591499711811e-06,
+      "loss": 0.8705,
+      "step": 23243
+    },
+    {
+      "epoch": 0.6383785119881354,
+      "grad_norm": 0.0,
+      "learning_rate": 6.109772037151132e-06,
+      "loss": 0.8913,
+      "step": 23244
+    },
+    {
+      "epoch": 0.6384059762159787,
+      "grad_norm": 0.0,
+      "learning_rate": 6.108952605371445e-06,
+      "loss": 0.8387,
+      "step": 23245
+    },
+    {
+      "epoch": 0.6384334404438219,
+      "grad_norm": 0.0,
+      "learning_rate": 6.108133204379232e-06,
+      "loss": 0.7684,
+      "step": 23246
+    },
+    {
+      "epoch": 0.6384609046716652,
+      "grad_norm": 0.0,
+      "learning_rate": 6.107313834180979e-06,
+      "loss": 0.9158,
+      "step": 23247
+    },
+    {
+      "epoch": 0.6384883688995084,
+      "grad_norm": 0.0,
+      "learning_rate": 6.106494494783166e-06,
+      "loss": 0.9304,
+      "step": 23248
+    },
+    {
+      "epoch": 0.6385158331273516,
+      "grad_norm": 0.0,
+      "learning_rate": 6.105675186192275e-06,
+      "loss": 0.8204,
+      "step": 23249
+    },
+    {
+      "epoch": 0.6385432973551949,
+      "grad_norm": 0.0,
+      "learning_rate": 6.104855908414791e-06,
+      "loss": 0.8529,
+      "step": 23250
+    },
+    {
+      "epoch": 0.638570761583038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1040366614571964e-06,
+      "loss": 0.794,
+      "step": 23251
+    },
+    {
+      "epoch": 0.6385982258108813,
+      "grad_norm": 0.0,
+      "learning_rate": 6.103217445325972e-06,
+      "loss": 0.8093,
+      "step": 23252
+    },
+    {
+      "epoch": 0.6386256900387246,
+      "grad_norm": 0.0,
+      "learning_rate": 6.102398260027604e-06,
+      "loss": 0.8382,
+      "step": 23253
+    },
+    {
+      "epoch": 0.6386531542665678,
+      "grad_norm": 0.0,
+      "learning_rate": 6.101579105568567e-06,
+      "loss": 0.8513,
+      "step": 23254
+    },
+    {
+      "epoch": 0.638680618494411,
+      "grad_norm": 0.0,
+      "learning_rate": 6.100759981955349e-06,
+      "loss": 0.871,
+      "step": 23255
+    },
+    {
+      "epoch": 0.6387080827222542,
+      "grad_norm": 0.0,
+      "learning_rate": 6.099940889194425e-06,
+      "loss": 0.9125,
+      "step": 23256
+    },
+    {
+      "epoch": 0.6387355469500975,
+      "grad_norm": 0.0,
+      "learning_rate": 6.09912182729228e-06,
+      "loss": 0.8648,
+      "step": 23257
+    },
+    {
+      "epoch": 0.6387630111779408,
+      "grad_norm": 0.0,
+      "learning_rate": 6.098302796255394e-06,
+      "loss": 0.9642,
+      "step": 23258
+    },
+    {
+      "epoch": 0.6387904754057839,
+      "grad_norm": 0.0,
+      "learning_rate": 6.097483796090252e-06,
+      "loss": 0.8236,
+      "step": 23259
+    },
+    {
+      "epoch": 0.6388179396336272,
+      "grad_norm": 0.0,
+      "learning_rate": 6.096664826803328e-06,
+      "loss": 0.8121,
+      "step": 23260
+    },
+    {
+      "epoch": 0.6388454038614705,
+      "grad_norm": 0.0,
+      "learning_rate": 6.095845888401102e-06,
+      "loss": 0.8602,
+      "step": 23261
+    },
+    {
+      "epoch": 0.6388728680893137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0950269808900556e-06,
+      "loss": 0.8109,
+      "step": 23262
+    },
+    {
+      "epoch": 0.6389003323171569,
+      "grad_norm": 0.0,
+      "learning_rate": 6.094208104276669e-06,
+      "loss": 0.8484,
+      "step": 23263
+    },
+    {
+      "epoch": 0.6389277965450001,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0933892585674195e-06,
+      "loss": 0.9294,
+      "step": 23264
+    },
+    {
+      "epoch": 0.6389552607728434,
+      "grad_norm": 0.0,
+      "learning_rate": 6.092570443768793e-06,
+      "loss": 0.8218,
+      "step": 23265
+    },
+    {
+      "epoch": 0.6389827250006866,
+      "grad_norm": 0.0,
+      "learning_rate": 6.091751659887258e-06,
+      "loss": 0.8621,
+      "step": 23266
+    },
+    {
+      "epoch": 0.6390101892285298,
+      "grad_norm": 0.0,
+      "learning_rate": 6.090932906929299e-06,
+      "loss": 0.8707,
+      "step": 23267
+    },
+    {
+      "epoch": 0.6390376534563731,
+      "grad_norm": 0.0,
+      "learning_rate": 6.09011418490139e-06,
+      "loss": 0.8133,
+      "step": 23268
+    },
+    {
+      "epoch": 0.6390651176842163,
+      "grad_norm": 0.0,
+      "learning_rate": 6.089295493810015e-06,
+      "loss": 0.8852,
+      "step": 23269
+    },
+    {
+      "epoch": 0.6390925819120595,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0884768336616475e-06,
+      "loss": 0.8998,
+      "step": 23270
+    },
+    {
+      "epoch": 0.6391200461399028,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0876582044627694e-06,
+      "loss": 0.8893,
+      "step": 23271
+    },
+    {
+      "epoch": 0.639147510367746,
+      "grad_norm": 0.0,
+      "learning_rate": 6.086839606219854e-06,
+      "loss": 0.9502,
+      "step": 23272
+    },
+    {
+      "epoch": 0.6391749745955893,
+      "grad_norm": 0.0,
+      "learning_rate": 6.086021038939378e-06,
+      "loss": 0.8206,
+      "step": 23273
+    },
+    {
+      "epoch": 0.6392024388234325,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08520250262782e-06,
+      "loss": 0.888,
+      "step": 23274
+    },
+    {
+      "epoch": 0.6392299030512757,
+      "grad_norm": 0.0,
+      "learning_rate": 6.084383997291656e-06,
+      "loss": 0.8044,
+      "step": 23275
+    },
+    {
+      "epoch": 0.639257367279119,
+      "grad_norm": 0.0,
+      "learning_rate": 6.083565522937367e-06,
+      "loss": 0.8598,
+      "step": 23276
+    },
+    {
+      "epoch": 0.6392848315069621,
+      "grad_norm": 0.0,
+      "learning_rate": 6.082747079571421e-06,
+      "loss": 0.9075,
+      "step": 23277
+    },
+    {
+      "epoch": 0.6393122957348054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.081928667200295e-06,
+      "loss": 0.8518,
+      "step": 23278
+    },
+    {
+      "epoch": 0.6393397599626487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08111028583047e-06,
+      "loss": 0.8386,
+      "step": 23279
+    },
+    {
+      "epoch": 0.6393672241904919,
+      "grad_norm": 0.0,
+      "learning_rate": 6.080291935468416e-06,
+      "loss": 0.8648,
+      "step": 23280
+    },
+    {
+      "epoch": 0.6393946884183351,
+      "grad_norm": 0.0,
+      "learning_rate": 6.079473616120612e-06,
+      "loss": 0.8626,
+      "step": 23281
+    },
+    {
+      "epoch": 0.6394221526461783,
+      "grad_norm": 0.0,
+      "learning_rate": 6.078655327793535e-06,
+      "loss": 0.8918,
+      "step": 23282
+    },
+    {
+      "epoch": 0.6394496168740216,
+      "grad_norm": 0.0,
+      "learning_rate": 6.077837070493653e-06,
+      "loss": 0.8377,
+      "step": 23283
+    },
+    {
+      "epoch": 0.6394770811018649,
+      "grad_norm": 0.0,
+      "learning_rate": 6.077018844227443e-06,
+      "loss": 0.8439,
+      "step": 23284
+    },
+    {
+      "epoch": 0.639504545329708,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0762006490013795e-06,
+      "loss": 0.7388,
+      "step": 23285
+    },
+    {
+      "epoch": 0.6395320095575513,
+      "grad_norm": 0.0,
+      "learning_rate": 6.075382484821934e-06,
+      "loss": 0.78,
+      "step": 23286
+    },
+    {
+      "epoch": 0.6395594737853946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.074564351695588e-06,
+      "loss": 0.6946,
+      "step": 23287
+    },
+    {
+      "epoch": 0.6395869380132377,
+      "grad_norm": 0.0,
+      "learning_rate": 6.073746249628809e-06,
+      "loss": 0.9054,
+      "step": 23288
+    },
+    {
+      "epoch": 0.639614402241081,
+      "grad_norm": 0.0,
+      "learning_rate": 6.072928178628067e-06,
+      "loss": 0.8156,
+      "step": 23289
+    },
+    {
+      "epoch": 0.6396418664689242,
+      "grad_norm": 0.0,
+      "learning_rate": 6.072110138699838e-06,
+      "loss": 0.9084,
+      "step": 23290
+    },
+    {
+      "epoch": 0.6396693306967675,
+      "grad_norm": 0.0,
+      "learning_rate": 6.071292129850599e-06,
+      "loss": 0.8157,
+      "step": 23291
+    },
+    {
+      "epoch": 0.6396967949246107,
+      "grad_norm": 0.0,
+      "learning_rate": 6.070474152086814e-06,
+      "loss": 0.8414,
+      "step": 23292
+    },
+    {
+      "epoch": 0.6397242591524539,
+      "grad_norm": 0.0,
+      "learning_rate": 6.069656205414962e-06,
+      "loss": 0.916,
+      "step": 23293
+    },
+    {
+      "epoch": 0.6397517233802972,
+      "grad_norm": 0.0,
+      "learning_rate": 6.068838289841514e-06,
+      "loss": 0.815,
+      "step": 23294
+    },
+    {
+      "epoch": 0.6397791876081403,
+      "grad_norm": 0.0,
+      "learning_rate": 6.068020405372939e-06,
+      "loss": 0.7794,
+      "step": 23295
+    },
+    {
+      "epoch": 0.6398066518359836,
+      "grad_norm": 0.0,
+      "learning_rate": 6.06720255201571e-06,
+      "loss": 0.8156,
+      "step": 23296
+    },
+    {
+      "epoch": 0.6398341160638269,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0663847297762956e-06,
+      "loss": 0.8441,
+      "step": 23297
+    },
+    {
+      "epoch": 0.6398615802916701,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0655669386611695e-06,
+      "loss": 0.8309,
+      "step": 23298
+    },
+    {
+      "epoch": 0.6398890445195133,
+      "grad_norm": 0.0,
+      "learning_rate": 6.064749178676804e-06,
+      "loss": 0.842,
+      "step": 23299
+    },
+    {
+      "epoch": 0.6399165087473566,
+      "grad_norm": 0.0,
+      "learning_rate": 6.063931449829669e-06,
+      "loss": 0.9071,
+      "step": 23300
+    },
+    {
+      "epoch": 0.6399439729751998,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0631137521262294e-06,
+      "loss": 0.8713,
+      "step": 23301
+    },
+    {
+      "epoch": 0.6399714372030431,
+      "grad_norm": 0.0,
+      "learning_rate": 6.062296085572958e-06,
+      "loss": 0.7427,
+      "step": 23302
+    },
+    {
+      "epoch": 0.6399989014308862,
+      "grad_norm": 0.0,
+      "learning_rate": 6.061478450176329e-06,
+      "loss": 0.809,
+      "step": 23303
+    },
+    {
+      "epoch": 0.6400263656587295,
+      "grad_norm": 0.0,
+      "learning_rate": 6.060660845942805e-06,
+      "loss": 0.8253,
+      "step": 23304
+    },
+    {
+      "epoch": 0.6400538298865728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.059843272878858e-06,
+      "loss": 0.7599,
+      "step": 23305
+    },
+    {
+      "epoch": 0.6400812941144159,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0590257309909615e-06,
+      "loss": 0.7822,
+      "step": 23306
+    },
+    {
+      "epoch": 0.6401087583422592,
+      "grad_norm": 0.0,
+      "learning_rate": 6.058208220285578e-06,
+      "loss": 0.8418,
+      "step": 23307
+    },
+    {
+      "epoch": 0.6401362225701024,
+      "grad_norm": 0.0,
+      "learning_rate": 6.057390740769178e-06,
+      "loss": 0.8898,
+      "step": 23308
+    },
+    {
+      "epoch": 0.6401636867979457,
+      "grad_norm": 0.0,
+      "learning_rate": 6.056573292448227e-06,
+      "loss": 0.7534,
+      "step": 23309
+    },
+    {
+      "epoch": 0.6401911510257889,
+      "grad_norm": 0.0,
+      "learning_rate": 6.055755875329197e-06,
+      "loss": 0.7635,
+      "step": 23310
+    },
+    {
+      "epoch": 0.6402186152536321,
+      "grad_norm": 0.0,
+      "learning_rate": 6.054938489418556e-06,
+      "loss": 0.8046,
+      "step": 23311
+    },
+    {
+      "epoch": 0.6402460794814754,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0541211347227715e-06,
+      "loss": 0.7731,
+      "step": 23312
+    },
+    {
+      "epoch": 0.6402735437093187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.053303811248306e-06,
+      "loss": 0.8318,
+      "step": 23313
+    },
+    {
+      "epoch": 0.6403010079371618,
+      "grad_norm": 0.0,
+      "learning_rate": 6.052486519001628e-06,
+      "loss": 0.8199,
+      "step": 23314
+    },
+    {
+      "epoch": 0.6403284721650051,
+      "grad_norm": 0.0,
+      "learning_rate": 6.051669257989208e-06,
+      "loss": 0.8322,
+      "step": 23315
+    },
+    {
+      "epoch": 0.6403559363928483,
+      "grad_norm": 0.0,
+      "learning_rate": 6.050852028217509e-06,
+      "loss": 0.8437,
+      "step": 23316
+    },
+    {
+      "epoch": 0.6403834006206915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.050034829692999e-06,
+      "loss": 0.7573,
+      "step": 23317
+    },
+    {
+      "epoch": 0.6404108648485348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.049217662422146e-06,
+      "loss": 0.8315,
+      "step": 23318
+    },
+    {
+      "epoch": 0.640438329076378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.048400526411411e-06,
+      "loss": 0.7233,
+      "step": 23319
+    },
+    {
+      "epoch": 0.6404657933042213,
+      "grad_norm": 0.0,
+      "learning_rate": 6.047583421667262e-06,
+      "loss": 0.9624,
+      "step": 23320
+    },
+    {
+      "epoch": 0.6404932575320644,
+      "grad_norm": 0.0,
+      "learning_rate": 6.046766348196163e-06,
+      "loss": 0.8607,
+      "step": 23321
+    },
+    {
+      "epoch": 0.6405207217599077,
+      "grad_norm": 0.0,
+      "learning_rate": 6.045949306004581e-06,
+      "loss": 0.8419,
+      "step": 23322
+    },
+    {
+      "epoch": 0.640548185987751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04513229509898e-06,
+      "loss": 0.8335,
+      "step": 23323
+    },
+    {
+      "epoch": 0.6405756502155942,
+      "grad_norm": 0.0,
+      "learning_rate": 6.044315315485827e-06,
+      "loss": 0.7433,
+      "step": 23324
+    },
+    {
+      "epoch": 0.6406031144434374,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04349836717158e-06,
+      "loss": 0.7906,
+      "step": 23325
+    },
+    {
+      "epoch": 0.6406305786712807,
+      "grad_norm": 0.0,
+      "learning_rate": 6.042681450162705e-06,
+      "loss": 0.8507,
+      "step": 23326
+    },
+    {
+      "epoch": 0.6406580428991239,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04186456446567e-06,
+      "loss": 0.8851,
+      "step": 23327
+    },
+    {
+      "epoch": 0.6406855071269671,
+      "grad_norm": 0.0,
+      "learning_rate": 6.041047710086934e-06,
+      "loss": 0.8722,
+      "step": 23328
+    },
+    {
+      "epoch": 0.6407129713548103,
+      "grad_norm": 0.0,
+      "learning_rate": 6.040230887032963e-06,
+      "loss": 0.9392,
+      "step": 23329
+    },
+    {
+      "epoch": 0.6407404355826536,
+      "grad_norm": 0.0,
+      "learning_rate": 6.039414095310221e-06,
+      "loss": 0.8717,
+      "step": 23330
+    },
+    {
+      "epoch": 0.6407678998104969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.038597334925167e-06,
+      "loss": 0.8638,
+      "step": 23331
+    },
+    {
+      "epoch": 0.64079536403834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.037780605884266e-06,
+      "loss": 0.9129,
+      "step": 23332
+    },
+    {
+      "epoch": 0.6408228282661833,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036963908193978e-06,
+      "loss": 0.8559,
+      "step": 23333
+    },
+    {
+      "epoch": 0.6408502924940266,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036147241860767e-06,
+      "loss": 0.7569,
+      "step": 23334
+    },
+    {
+      "epoch": 0.6408777567218698,
+      "grad_norm": 0.0,
+      "learning_rate": 6.035330606891099e-06,
+      "loss": 0.8603,
+      "step": 23335
+    },
+    {
+      "epoch": 0.640905220949713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.034514003291427e-06,
+      "loss": 0.8446,
+      "step": 23336
+    },
+    {
+      "epoch": 0.6409326851775562,
+      "grad_norm": 0.0,
+      "learning_rate": 6.033697431068215e-06,
+      "loss": 0.7776,
+      "step": 23337
+    },
+    {
+      "epoch": 0.6409601494053995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.032880890227927e-06,
+      "loss": 0.9792,
+      "step": 23338
+    },
+    {
+      "epoch": 0.6409876136332427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.032064380777023e-06,
+      "loss": 0.7921,
+      "step": 23339
+    },
+    {
+      "epoch": 0.6410150778610859,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0312479027219595e-06,
+      "loss": 0.8254,
+      "step": 23340
+    },
+    {
+      "epoch": 0.6410425420889292,
+      "grad_norm": 0.0,
+      "learning_rate": 6.030431456069206e-06,
+      "loss": 0.8616,
+      "step": 23341
+    },
+    {
+      "epoch": 0.6410700063167724,
+      "grad_norm": 0.0,
+      "learning_rate": 6.029615040825212e-06,
+      "loss": 0.8225,
+      "step": 23342
+    },
+    {
+      "epoch": 0.6410974705446156,
+      "grad_norm": 0.0,
+      "learning_rate": 6.028798656996441e-06,
+      "loss": 0.7984,
+      "step": 23343
+    },
+    {
+      "epoch": 0.6411249347724589,
+      "grad_norm": 0.0,
+      "learning_rate": 6.027982304589355e-06,
+      "loss": 0.8853,
+      "step": 23344
+    },
+    {
+      "epoch": 0.6411523990003021,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0271659836104105e-06,
+      "loss": 0.773,
+      "step": 23345
+    },
+    {
+      "epoch": 0.6411798632281454,
+      "grad_norm": 0.0,
+      "learning_rate": 6.026349694066067e-06,
+      "loss": 0.7998,
+      "step": 23346
+    },
+    {
+      "epoch": 0.6412073274559886,
+      "grad_norm": 0.0,
+      "learning_rate": 6.025533435962789e-06,
+      "loss": 0.9328,
+      "step": 23347
+    },
+    {
+      "epoch": 0.6412347916838318,
+      "grad_norm": 0.0,
+      "learning_rate": 6.024717209307028e-06,
+      "loss": 0.9277,
+      "step": 23348
+    },
+    {
+      "epoch": 0.6412622559116751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.02390101410524e-06,
+      "loss": 0.8433,
+      "step": 23349
+    },
+    {
+      "epoch": 0.6412897201395182,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0230848503638895e-06,
+      "loss": 0.8808,
+      "step": 23350
+    },
+    {
+      "epoch": 0.6413171843673615,
+      "grad_norm": 0.0,
+      "learning_rate": 6.022268718089435e-06,
+      "loss": 0.7874,
+      "step": 23351
+    },
+    {
+      "epoch": 0.6413446485952048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.021452617288327e-06,
+      "loss": 0.8755,
+      "step": 23352
+    },
+    {
+      "epoch": 0.641372112823048,
+      "grad_norm": 0.0,
+      "learning_rate": 6.020636547967031e-06,
+      "loss": 0.9235,
+      "step": 23353
+    },
+    {
+      "epoch": 0.6413995770508912,
+      "grad_norm": 0.0,
+      "learning_rate": 6.019820510131997e-06,
+      "loss": 0.847,
+      "step": 23354
+    },
+    {
+      "epoch": 0.6414270412787344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.019004503789686e-06,
+      "loss": 0.8838,
+      "step": 23355
+    },
+    {
+      "epoch": 0.6414545055065777,
+      "grad_norm": 0.0,
+      "learning_rate": 6.018188528946554e-06,
+      "loss": 0.8074,
+      "step": 23356
+    },
+    {
+      "epoch": 0.641481969734421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.017372585609056e-06,
+      "loss": 0.7777,
+      "step": 23357
+    },
+    {
+      "epoch": 0.6415094339622641,
+      "grad_norm": 0.0,
+      "learning_rate": 6.016556673783648e-06,
+      "loss": 0.8974,
+      "step": 23358
+    },
+    {
+      "epoch": 0.6415368981901074,
+      "grad_norm": 0.0,
+      "learning_rate": 6.01574079347679e-06,
+      "loss": 0.8071,
+      "step": 23359
+    },
+    {
+      "epoch": 0.6415643624179507,
+      "grad_norm": 0.0,
+      "learning_rate": 6.014924944694933e-06,
+      "loss": 0.8103,
+      "step": 23360
+    },
+    {
+      "epoch": 0.6415918266457938,
+      "grad_norm": 0.0,
+      "learning_rate": 6.014109127444532e-06,
+      "loss": 0.8465,
+      "step": 23361
+    },
+    {
+      "epoch": 0.6416192908736371,
+      "grad_norm": 0.0,
+      "learning_rate": 6.013293341732043e-06,
+      "loss": 0.9071,
+      "step": 23362
+    },
+    {
+      "epoch": 0.6416467551014803,
+      "grad_norm": 0.0,
+      "learning_rate": 6.012477587563923e-06,
+      "loss": 0.9139,
+      "step": 23363
+    },
+    {
+      "epoch": 0.6416742193293236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.011661864946623e-06,
+      "loss": 0.8846,
+      "step": 23364
+    },
+    {
+      "epoch": 0.6417016835571668,
+      "grad_norm": 0.0,
+      "learning_rate": 6.010846173886604e-06,
+      "loss": 0.9153,
+      "step": 23365
+    },
+    {
+      "epoch": 0.64172914778501,
+      "grad_norm": 0.0,
+      "learning_rate": 6.010030514390311e-06,
+      "loss": 0.82,
+      "step": 23366
+    },
+    {
+      "epoch": 0.6417566120128533,
+      "grad_norm": 0.0,
+      "learning_rate": 6.009214886464202e-06,
+      "loss": 0.9415,
+      "step": 23367
+    },
+    {
+      "epoch": 0.6417840762406964,
+      "grad_norm": 0.0,
+      "learning_rate": 6.008399290114733e-06,
+      "loss": 0.9384,
+      "step": 23368
+    },
+    {
+      "epoch": 0.6418115404685397,
+      "grad_norm": 0.0,
+      "learning_rate": 6.007583725348352e-06,
+      "loss": 0.9346,
+      "step": 23369
+    },
+    {
+      "epoch": 0.641839004696383,
+      "grad_norm": 0.0,
+      "learning_rate": 6.006768192171516e-06,
+      "loss": 0.844,
+      "step": 23370
+    },
+    {
+      "epoch": 0.6418664689242262,
+      "grad_norm": 0.0,
+      "learning_rate": 6.005952690590681e-06,
+      "loss": 0.7816,
+      "step": 23371
+    },
+    {
+      "epoch": 0.6418939331520694,
+      "grad_norm": 0.0,
+      "learning_rate": 6.005137220612292e-06,
+      "loss": 0.7901,
+      "step": 23372
+    },
+    {
+      "epoch": 0.6419213973799127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.004321782242804e-06,
+      "loss": 0.8663,
+      "step": 23373
+    },
+    {
+      "epoch": 0.6419488616077559,
+      "grad_norm": 0.0,
+      "learning_rate": 6.00350637548867e-06,
+      "loss": 0.8488,
+      "step": 23374
+    },
+    {
+      "epoch": 0.6419763258355992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.002691000356345e-06,
+      "loss": 0.822,
+      "step": 23375
+    },
+    {
+      "epoch": 0.6420037900634423,
+      "grad_norm": 0.0,
+      "learning_rate": 6.001875656852272e-06,
+      "loss": 0.8452,
+      "step": 23376
+    },
+    {
+      "epoch": 0.6420312542912856,
+      "grad_norm": 0.0,
+      "learning_rate": 6.001060344982914e-06,
+      "loss": 0.8016,
+      "step": 23377
+    },
+    {
+      "epoch": 0.6420587185191289,
+      "grad_norm": 0.0,
+      "learning_rate": 6.00024506475471e-06,
+      "loss": 0.8484,
+      "step": 23378
+    },
+    {
+      "epoch": 0.642086182746972,
+      "grad_norm": 0.0,
+      "learning_rate": 5.999429816174117e-06,
+      "loss": 0.8331,
+      "step": 23379
+    },
+    {
+      "epoch": 0.6421136469748153,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9986145992475875e-06,
+      "loss": 0.9247,
+      "step": 23380
+    },
+    {
+      "epoch": 0.6421411112026585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.997799413981566e-06,
+      "loss": 0.8348,
+      "step": 23381
+    },
+    {
+      "epoch": 0.6421685754305018,
+      "grad_norm": 0.0,
+      "learning_rate": 5.996984260382508e-06,
+      "loss": 0.793,
+      "step": 23382
+    },
+    {
+      "epoch": 0.642196039658345,
+      "grad_norm": 0.0,
+      "learning_rate": 5.996169138456864e-06,
+      "loss": 0.8998,
+      "step": 23383
+    },
+    {
+      "epoch": 0.6422235038861882,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9953540482110785e-06,
+      "loss": 0.7698,
+      "step": 23384
+    },
+    {
+      "epoch": 0.6422509681140315,
+      "grad_norm": 0.0,
+      "learning_rate": 5.994538989651602e-06,
+      "loss": 0.8078,
+      "step": 23385
+    },
+    {
+      "epoch": 0.6422784323418748,
+      "grad_norm": 0.0,
+      "learning_rate": 5.993723962784883e-06,
+      "loss": 0.8247,
+      "step": 23386
+    },
+    {
+      "epoch": 0.6423058965697179,
+      "grad_norm": 0.0,
+      "learning_rate": 5.992908967617374e-06,
+      "loss": 0.8725,
+      "step": 23387
+    },
+    {
+      "epoch": 0.6423333607975612,
+      "grad_norm": 0.0,
+      "learning_rate": 5.99209400415552e-06,
+      "loss": 0.9111,
+      "step": 23388
+    },
+    {
+      "epoch": 0.6423608250254044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.991279072405775e-06,
+      "loss": 0.7969,
+      "step": 23389
+    },
+    {
+      "epoch": 0.6423882892532476,
+      "grad_norm": 0.0,
+      "learning_rate": 5.990464172374578e-06,
+      "loss": 0.9941,
+      "step": 23390
+    },
+    {
+      "epoch": 0.6424157534810909,
+      "grad_norm": 0.0,
+      "learning_rate": 5.989649304068385e-06,
+      "loss": 0.7912,
+      "step": 23391
+    },
+    {
+      "epoch": 0.6424432177089341,
+      "grad_norm": 0.0,
+      "learning_rate": 5.988834467493637e-06,
+      "loss": 0.832,
+      "step": 23392
+    },
+    {
+      "epoch": 0.6424706819367774,
+      "grad_norm": 0.0,
+      "learning_rate": 5.988019662656786e-06,
+      "loss": 0.9095,
+      "step": 23393
+    },
+    {
+      "epoch": 0.6424981461646205,
+      "grad_norm": 0.0,
+      "learning_rate": 5.98720488956428e-06,
+      "loss": 0.8109,
+      "step": 23394
+    },
+    {
+      "epoch": 0.6425256103924638,
+      "grad_norm": 0.0,
+      "learning_rate": 5.986390148222558e-06,
+      "loss": 0.7795,
+      "step": 23395
+    },
+    {
+      "epoch": 0.6425530746203071,
+      "grad_norm": 0.0,
+      "learning_rate": 5.985575438638076e-06,
+      "loss": 0.8747,
+      "step": 23396
+    },
+    {
+      "epoch": 0.6425805388481503,
+      "grad_norm": 0.0,
+      "learning_rate": 5.984760760817274e-06,
+      "loss": 0.8424,
+      "step": 23397
+    },
+    {
+      "epoch": 0.6426080030759935,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9839461147666e-06,
+      "loss": 0.8314,
+      "step": 23398
+    },
+    {
+      "epoch": 0.6426354673038368,
+      "grad_norm": 0.0,
+      "learning_rate": 5.983131500492503e-06,
+      "loss": 0.85,
+      "step": 23399
+    },
+    {
+      "epoch": 0.64266293153168,
+      "grad_norm": 0.0,
+      "learning_rate": 5.982316918001425e-06,
+      "loss": 0.9598,
+      "step": 23400
+    },
+    {
+      "epoch": 0.6426903957595232,
+      "grad_norm": 0.0,
+      "learning_rate": 5.98150236729981e-06,
+      "loss": 0.8722,
+      "step": 23401
+    },
+    {
+      "epoch": 0.6427178599873664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.980687848394104e-06,
+      "loss": 0.826,
+      "step": 23402
+    },
+    {
+      "epoch": 0.6427453242152097,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9798733612907554e-06,
+      "loss": 0.8311,
+      "step": 23403
+    },
+    {
+      "epoch": 0.642772788443053,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9790589059962035e-06,
+      "loss": 0.9034,
+      "step": 23404
+    },
+    {
+      "epoch": 0.6428002526708961,
+      "grad_norm": 0.0,
+      "learning_rate": 5.978244482516895e-06,
+      "loss": 0.8871,
+      "step": 23405
+    },
+    {
+      "epoch": 0.6428277168987394,
+      "grad_norm": 0.0,
+      "learning_rate": 5.977430090859278e-06,
+      "loss": 0.8803,
+      "step": 23406
+    },
+    {
+      "epoch": 0.6428551811265826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.976615731029789e-06,
+      "loss": 0.8802,
+      "step": 23407
+    },
+    {
+      "epoch": 0.6428826453544259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9758014030348755e-06,
+      "loss": 0.7999,
+      "step": 23408
+    },
+    {
+      "epoch": 0.6429101095822691,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9749871068809795e-06,
+      "loss": 0.7568,
+      "step": 23409
+    },
+    {
+      "epoch": 0.6429375738101123,
+      "grad_norm": 0.0,
+      "learning_rate": 5.974172842574544e-06,
+      "loss": 0.6804,
+      "step": 23410
+    },
+    {
+      "epoch": 0.6429650380379556,
+      "grad_norm": 0.0,
+      "learning_rate": 5.973358610122014e-06,
+      "loss": 0.9027,
+      "step": 23411
+    },
+    {
+      "epoch": 0.6429925022657988,
+      "grad_norm": 0.0,
+      "learning_rate": 5.972544409529833e-06,
+      "loss": 0.7856,
+      "step": 23412
+    },
+    {
+      "epoch": 0.643019966493642,
+      "grad_norm": 0.0,
+      "learning_rate": 5.971730240804436e-06,
+      "loss": 0.8857,
+      "step": 23413
+    },
+    {
+      "epoch": 0.6430474307214853,
+      "grad_norm": 0.0,
+      "learning_rate": 5.970916103952271e-06,
+      "loss": 0.7468,
+      "step": 23414
+    },
+    {
+      "epoch": 0.6430748949493285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9701019989797804e-06,
+      "loss": 0.9416,
+      "step": 23415
+    },
+    {
+      "epoch": 0.6431023591771717,
+      "grad_norm": 0.0,
+      "learning_rate": 5.969287925893401e-06,
+      "loss": 0.8809,
+      "step": 23416
+    },
+    {
+      "epoch": 0.643129823405015,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9684738846995775e-06,
+      "loss": 0.9323,
+      "step": 23417
+    },
+    {
+      "epoch": 0.6431572876328582,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9676598754047546e-06,
+      "loss": 0.9099,
+      "step": 23418
+    },
+    {
+      "epoch": 0.6431847518607015,
+      "grad_norm": 0.0,
+      "learning_rate": 5.966845898015363e-06,
+      "loss": 0.7324,
+      "step": 23419
+    },
+    {
+      "epoch": 0.6432122160885446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.966031952537853e-06,
+      "loss": 0.8083,
+      "step": 23420
+    },
+    {
+      "epoch": 0.6432396803163879,
+      "grad_norm": 0.0,
+      "learning_rate": 5.965218038978657e-06,
+      "loss": 0.8592,
+      "step": 23421
+    },
+    {
+      "epoch": 0.6432671445442312,
+      "grad_norm": 0.0,
+      "learning_rate": 5.964404157344221e-06,
+      "loss": 0.9001,
+      "step": 23422
+    },
+    {
+      "epoch": 0.6432946087720743,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9635903076409825e-06,
+      "loss": 0.917,
+      "step": 23423
+    },
+    {
+      "epoch": 0.6433220729999176,
+      "grad_norm": 0.0,
+      "learning_rate": 5.962776489875384e-06,
+      "loss": 0.9069,
+      "step": 23424
+    },
+    {
+      "epoch": 0.6433495372277609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.961962704053858e-06,
+      "loss": 0.877,
+      "step": 23425
+    },
+    {
+      "epoch": 0.6433770014556041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.961148950182847e-06,
+      "loss": 0.8123,
+      "step": 23426
+    },
+    {
+      "epoch": 0.6434044656834473,
+      "grad_norm": 0.0,
+      "learning_rate": 5.960335228268792e-06,
+      "loss": 0.927,
+      "step": 23427
+    },
+    {
+      "epoch": 0.6434319299112905,
+      "grad_norm": 0.0,
+      "learning_rate": 5.959521538318127e-06,
+      "loss": 0.8135,
+      "step": 23428
+    },
+    {
+      "epoch": 0.6434593941391338,
+      "grad_norm": 0.0,
+      "learning_rate": 5.958707880337294e-06,
+      "loss": 0.8134,
+      "step": 23429
+    },
+    {
+      "epoch": 0.643486858366977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.957894254332732e-06,
+      "loss": 0.9001,
+      "step": 23430
+    },
+    {
+      "epoch": 0.6435143225948202,
+      "grad_norm": 0.0,
+      "learning_rate": 5.957080660310872e-06,
+      "loss": 0.8674,
+      "step": 23431
+    },
+    {
+      "epoch": 0.6435417868226635,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9562670982781605e-06,
+      "loss": 0.8714,
+      "step": 23432
+    },
+    {
+      "epoch": 0.6435692510505067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.955453568241027e-06,
+      "loss": 0.9387,
+      "step": 23433
+    },
+    {
+      "epoch": 0.6435967152783499,
+      "grad_norm": 0.0,
+      "learning_rate": 5.954640070205911e-06,
+      "loss": 0.8531,
+      "step": 23434
+    },
+    {
+      "epoch": 0.6436241795061932,
+      "grad_norm": 0.0,
+      "learning_rate": 5.953826604179252e-06,
+      "loss": 0.8001,
+      "step": 23435
+    },
+    {
+      "epoch": 0.6436516437340364,
+      "grad_norm": 0.0,
+      "learning_rate": 5.953013170167485e-06,
+      "loss": 0.8684,
+      "step": 23436
+    },
+    {
+      "epoch": 0.6436791079618797,
+      "grad_norm": 0.0,
+      "learning_rate": 5.952199768177044e-06,
+      "loss": 0.8481,
+      "step": 23437
+    },
+    {
+      "epoch": 0.6437065721897229,
+      "grad_norm": 0.0,
+      "learning_rate": 5.951386398214365e-06,
+      "loss": 0.8666,
+      "step": 23438
+    },
+    {
+      "epoch": 0.6437340364175661,
+      "grad_norm": 0.0,
+      "learning_rate": 5.950573060285887e-06,
+      "loss": 0.8478,
+      "step": 23439
+    },
+    {
+      "epoch": 0.6437615006454094,
+      "grad_norm": 0.0,
+      "learning_rate": 5.949759754398041e-06,
+      "loss": 0.8527,
+      "step": 23440
+    },
+    {
+      "epoch": 0.6437889648732525,
+      "grad_norm": 0.0,
+      "learning_rate": 5.948946480557266e-06,
+      "loss": 0.8033,
+      "step": 23441
+    },
+    {
+      "epoch": 0.6438164291010958,
+      "grad_norm": 0.0,
+      "learning_rate": 5.94813323877e-06,
+      "loss": 0.8578,
+      "step": 23442
+    },
+    {
+      "epoch": 0.6438438933289391,
+      "grad_norm": 0.0,
+      "learning_rate": 5.947320029042668e-06,
+      "loss": 0.9175,
+      "step": 23443
+    },
+    {
+      "epoch": 0.6438713575567823,
+      "grad_norm": 0.0,
+      "learning_rate": 5.946506851381713e-06,
+      "loss": 0.8361,
+      "step": 23444
+    },
+    {
+      "epoch": 0.6438988217846255,
+      "grad_norm": 0.0,
+      "learning_rate": 5.945693705793562e-06,
+      "loss": 0.8338,
+      "step": 23445
+    },
+    {
+      "epoch": 0.6439262860124687,
+      "grad_norm": 0.0,
+      "learning_rate": 5.944880592284655e-06,
+      "loss": 0.8393,
+      "step": 23446
+    },
+    {
+      "epoch": 0.643953750240312,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9440675108614245e-06,
+      "loss": 0.7396,
+      "step": 23447
+    },
+    {
+      "epoch": 0.6439812144681553,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9432544615303055e-06,
+      "loss": 0.9319,
+      "step": 23448
+    },
+    {
+      "epoch": 0.6440086786959984,
+      "grad_norm": 0.0,
+      "learning_rate": 5.942441444297724e-06,
+      "loss": 0.9144,
+      "step": 23449
+    },
+    {
+      "epoch": 0.6440361429238417,
+      "grad_norm": 0.0,
+      "learning_rate": 5.941628459170119e-06,
+      "loss": 0.8318,
+      "step": 23450
+    },
+    {
+      "epoch": 0.644063607151685,
+      "grad_norm": 0.0,
+      "learning_rate": 5.940815506153922e-06,
+      "loss": 0.7924,
+      "step": 23451
+    },
+    {
+      "epoch": 0.6440910713795281,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9400025852555645e-06,
+      "loss": 0.8761,
+      "step": 23452
+    },
+    {
+      "epoch": 0.6441185356073714,
+      "grad_norm": 0.0,
+      "learning_rate": 5.939189696481478e-06,
+      "loss": 0.9294,
+      "step": 23453
+    },
+    {
+      "epoch": 0.6441459998352146,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9383768398381e-06,
+      "loss": 0.9099,
+      "step": 23454
+    },
+    {
+      "epoch": 0.6441734640630579,
+      "grad_norm": 0.0,
+      "learning_rate": 5.937564015331854e-06,
+      "loss": 0.9717,
+      "step": 23455
+    },
+    {
+      "epoch": 0.6442009282909011,
+      "grad_norm": 0.0,
+      "learning_rate": 5.936751222969177e-06,
+      "loss": 0.8444,
+      "step": 23456
+    },
+    {
+      "epoch": 0.6442283925187443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.935938462756496e-06,
+      "loss": 0.8575,
+      "step": 23457
+    },
+    {
+      "epoch": 0.6442558567465876,
+      "grad_norm": 0.0,
+      "learning_rate": 5.935125734700244e-06,
+      "loss": 0.883,
+      "step": 23458
+    },
+    {
+      "epoch": 0.6442833209744308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.934313038806856e-06,
+      "loss": 0.8255,
+      "step": 23459
+    },
+    {
+      "epoch": 0.644310785202274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.933500375082757e-06,
+      "loss": 0.7912,
+      "step": 23460
+    },
+    {
+      "epoch": 0.6443382494301173,
+      "grad_norm": 0.0,
+      "learning_rate": 5.932687743534375e-06,
+      "loss": 0.8007,
+      "step": 23461
+    },
+    {
+      "epoch": 0.6443657136579605,
+      "grad_norm": 0.0,
+      "learning_rate": 5.931875144168144e-06,
+      "loss": 0.9153,
+      "step": 23462
+    },
+    {
+      "epoch": 0.6443931778858037,
+      "grad_norm": 0.0,
+      "learning_rate": 5.931062576990494e-06,
+      "loss": 0.7583,
+      "step": 23463
+    },
+    {
+      "epoch": 0.644420642113647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9302500420078525e-06,
+      "loss": 0.825,
+      "step": 23464
+    },
+    {
+      "epoch": 0.6444481063414902,
+      "grad_norm": 0.0,
+      "learning_rate": 5.929437539226651e-06,
+      "loss": 0.9307,
+      "step": 23465
+    },
+    {
+      "epoch": 0.6444755705693335,
+      "grad_norm": 0.0,
+      "learning_rate": 5.928625068653313e-06,
+      "loss": 0.9352,
+      "step": 23466
+    },
+    {
+      "epoch": 0.6445030347971766,
+      "grad_norm": 0.0,
+      "learning_rate": 5.927812630294272e-06,
+      "loss": 0.8075,
+      "step": 23467
+    },
+    {
+      "epoch": 0.6445304990250199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.927000224155957e-06,
+      "loss": 0.8323,
+      "step": 23468
+    },
+    {
+      "epoch": 0.6445579632528632,
+      "grad_norm": 0.0,
+      "learning_rate": 5.926187850244789e-06,
+      "loss": 0.8758,
+      "step": 23469
+    },
+    {
+      "epoch": 0.6445854274807064,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9253755085672035e-06,
+      "loss": 0.8237,
+      "step": 23470
+    },
+    {
+      "epoch": 0.6446128917085496,
+      "grad_norm": 0.0,
+      "learning_rate": 5.924563199129629e-06,
+      "loss": 0.8491,
+      "step": 23471
+    },
+    {
+      "epoch": 0.6446403559363928,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9237509219384855e-06,
+      "loss": 0.8015,
+      "step": 23472
+    },
+    {
+      "epoch": 0.6446678201642361,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9229386770002025e-06,
+      "loss": 0.868,
+      "step": 23473
+    },
+    {
+      "epoch": 0.6446952843920793,
+      "grad_norm": 0.0,
+      "learning_rate": 5.922126464321209e-06,
+      "loss": 0.8126,
+      "step": 23474
+    },
+    {
+      "epoch": 0.6447227486199225,
+      "grad_norm": 0.0,
+      "learning_rate": 5.921314283907933e-06,
+      "loss": 0.8947,
+      "step": 23475
+    },
+    {
+      "epoch": 0.6447502128477658,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9205021357667945e-06,
+      "loss": 0.8613,
+      "step": 23476
+    },
+    {
+      "epoch": 0.6447776770756091,
+      "grad_norm": 0.0,
+      "learning_rate": 5.919690019904228e-06,
+      "loss": 0.7237,
+      "step": 23477
+    },
+    {
+      "epoch": 0.6448051413034522,
+      "grad_norm": 0.0,
+      "learning_rate": 5.91887793632665e-06,
+      "loss": 0.805,
+      "step": 23478
+    },
+    {
+      "epoch": 0.6448326055312955,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9180658850404915e-06,
+      "loss": 0.9022,
+      "step": 23479
+    },
+    {
+      "epoch": 0.6448600697591387,
+      "grad_norm": 0.0,
+      "learning_rate": 5.917253866052179e-06,
+      "loss": 0.8488,
+      "step": 23480
+    },
+    {
+      "epoch": 0.644887533986982,
+      "grad_norm": 0.0,
+      "learning_rate": 5.916441879368133e-06,
+      "loss": 0.7904,
+      "step": 23481
+    },
+    {
+      "epoch": 0.6449149982148252,
+      "grad_norm": 0.0,
+      "learning_rate": 5.915629924994781e-06,
+      "loss": 0.8813,
+      "step": 23482
+    },
+    {
+      "epoch": 0.6449424624426684,
+      "grad_norm": 0.0,
+      "learning_rate": 5.914818002938552e-06,
+      "loss": 0.7841,
+      "step": 23483
+    },
+    {
+      "epoch": 0.6449699266705117,
+      "grad_norm": 0.0,
+      "learning_rate": 5.914006113205861e-06,
+      "loss": 0.8079,
+      "step": 23484
+    },
+    {
+      "epoch": 0.6449973908983548,
+      "grad_norm": 0.0,
+      "learning_rate": 5.913194255803137e-06,
+      "loss": 0.8971,
+      "step": 23485
+    },
+    {
+      "epoch": 0.6450248551261981,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9123824307368004e-06,
+      "loss": 0.9027,
+      "step": 23486
+    },
+    {
+      "epoch": 0.6450523193540414,
+      "grad_norm": 0.0,
+      "learning_rate": 5.911570638013282e-06,
+      "loss": 1.0243,
+      "step": 23487
+    },
+    {
+      "epoch": 0.6450797835818846,
+      "grad_norm": 0.0,
+      "learning_rate": 5.910758877638997e-06,
+      "loss": 0.8573,
+      "step": 23488
+    },
+    {
+      "epoch": 0.6451072478097278,
+      "grad_norm": 0.0,
+      "learning_rate": 5.909947149620376e-06,
+      "loss": 0.8669,
+      "step": 23489
+    },
+    {
+      "epoch": 0.6451347120375711,
+      "grad_norm": 0.0,
+      "learning_rate": 5.909135453963833e-06,
+      "loss": 0.8383,
+      "step": 23490
+    },
+    {
+      "epoch": 0.6451621762654143,
+      "grad_norm": 0.0,
+      "learning_rate": 5.908323790675796e-06,
+      "loss": 0.7867,
+      "step": 23491
+    },
+    {
+      "epoch": 0.6451896404932576,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9075121597626875e-06,
+      "loss": 0.9122,
+      "step": 23492
+    },
+    {
+      "epoch": 0.6452171047211007,
+      "grad_norm": 0.0,
+      "learning_rate": 5.906700561230927e-06,
+      "loss": 0.933,
+      "step": 23493
+    },
+    {
+      "epoch": 0.645244568948944,
+      "grad_norm": 0.0,
+      "learning_rate": 5.905888995086936e-06,
+      "loss": 0.7968,
+      "step": 23494
+    },
+    {
+      "epoch": 0.6452720331767873,
+      "grad_norm": 0.0,
+      "learning_rate": 5.905077461337141e-06,
+      "loss": 0.9022,
+      "step": 23495
+    },
+    {
+      "epoch": 0.6452994974046304,
+      "grad_norm": 0.0,
+      "learning_rate": 5.904265959987958e-06,
+      "loss": 0.8556,
+      "step": 23496
+    },
+    {
+      "epoch": 0.6453269616324737,
+      "grad_norm": 0.0,
+      "learning_rate": 5.903454491045808e-06,
+      "loss": 0.7637,
+      "step": 23497
+    },
+    {
+      "epoch": 0.6453544258603169,
+      "grad_norm": 0.0,
+      "learning_rate": 5.902643054517112e-06,
+      "loss": 0.8046,
+      "step": 23498
+    },
+    {
+      "epoch": 0.6453818900881602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.901831650408294e-06,
+      "loss": 0.8453,
+      "step": 23499
+    },
+    {
+      "epoch": 0.6454093543160034,
+      "grad_norm": 0.0,
+      "learning_rate": 5.901020278725768e-06,
+      "loss": 0.7756,
+      "step": 23500
+    },
+    {
+      "epoch": 0.6454368185438466,
+      "grad_norm": 0.0,
+      "learning_rate": 5.900208939475963e-06,
+      "loss": 0.8694,
+      "step": 23501
+    },
+    {
+      "epoch": 0.6454642827716899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.899397632665288e-06,
+      "loss": 0.818,
+      "step": 23502
+    },
+    {
+      "epoch": 0.6454917469995332,
+      "grad_norm": 0.0,
+      "learning_rate": 5.898586358300168e-06,
+      "loss": 0.9157,
+      "step": 23503
+    },
+    {
+      "epoch": 0.6455192112273763,
+      "grad_norm": 0.0,
+      "learning_rate": 5.897775116387023e-06,
+      "loss": 0.8582,
+      "step": 23504
+    },
+    {
+      "epoch": 0.6455466754552196,
+      "grad_norm": 0.0,
+      "learning_rate": 5.896963906932268e-06,
+      "loss": 0.7684,
+      "step": 23505
+    },
+    {
+      "epoch": 0.6455741396830628,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8961527299423245e-06,
+      "loss": 0.7693,
+      "step": 23506
+    },
+    {
+      "epoch": 0.645601603910906,
+      "grad_norm": 0.0,
+      "learning_rate": 5.895341585423615e-06,
+      "loss": 0.9294,
+      "step": 23507
+    },
+    {
+      "epoch": 0.6456290681387493,
+      "grad_norm": 0.0,
+      "learning_rate": 5.894530473382549e-06,
+      "loss": 0.9418,
+      "step": 23508
+    },
+    {
+      "epoch": 0.6456565323665925,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8937193938255475e-06,
+      "loss": 0.8611,
+      "step": 23509
+    },
+    {
+      "epoch": 0.6456839965944358,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892908346759028e-06,
+      "loss": 0.757,
+      "step": 23510
+    },
+    {
+      "epoch": 0.6457114608222789,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892097332189412e-06,
+      "loss": 0.8355,
+      "step": 23511
+    },
+    {
+      "epoch": 0.6457389250501222,
+      "grad_norm": 0.0,
+      "learning_rate": 5.89128635012311e-06,
+      "loss": 0.809,
+      "step": 23512
+    },
+    {
+      "epoch": 0.6457663892779655,
+      "grad_norm": 0.0,
+      "learning_rate": 5.890475400566547e-06,
+      "loss": 0.8319,
+      "step": 23513
+    },
+    {
+      "epoch": 0.6457938535058086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.88966448352613e-06,
+      "loss": 0.7708,
+      "step": 23514
+    },
+    {
+      "epoch": 0.6458213177336519,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8888535990082785e-06,
+      "loss": 0.8538,
+      "step": 23515
+    },
+    {
+      "epoch": 0.6458487819614952,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8880427470194135e-06,
+      "loss": 0.8897,
+      "step": 23516
+    },
+    {
+      "epoch": 0.6458762461893384,
+      "grad_norm": 0.0,
+      "learning_rate": 5.887231927565944e-06,
+      "loss": 0.8487,
+      "step": 23517
+    },
+    {
+      "epoch": 0.6459037104171816,
+      "grad_norm": 0.0,
+      "learning_rate": 5.886421140654292e-06,
+      "loss": 0.8846,
+      "step": 23518
+    },
+    {
+      "epoch": 0.6459311746450248,
+      "grad_norm": 0.0,
+      "learning_rate": 5.885610386290868e-06,
+      "loss": 0.907,
+      "step": 23519
+    },
+    {
+      "epoch": 0.6459586388728681,
+      "grad_norm": 0.0,
+      "learning_rate": 5.884799664482088e-06,
+      "loss": 0.8892,
+      "step": 23520
+    },
+    {
+      "epoch": 0.6459861031007114,
+      "grad_norm": 0.0,
+      "learning_rate": 5.883988975234366e-06,
+      "loss": 0.8565,
+      "step": 23521
+    },
+    {
+      "epoch": 0.6460135673285545,
+      "grad_norm": 0.0,
+      "learning_rate": 5.883178318554117e-06,
+      "loss": 0.8093,
+      "step": 23522
+    },
+    {
+      "epoch": 0.6460410315563978,
+      "grad_norm": 0.0,
+      "learning_rate": 5.882367694447759e-06,
+      "loss": 0.8232,
+      "step": 23523
+    },
+    {
+      "epoch": 0.6460684957842411,
+      "grad_norm": 0.0,
+      "learning_rate": 5.881557102921704e-06,
+      "loss": 0.8675,
+      "step": 23524
+    },
+    {
+      "epoch": 0.6460959600120842,
+      "grad_norm": 0.0,
+      "learning_rate": 5.880746543982361e-06,
+      "loss": 0.832,
+      "step": 23525
+    },
+    {
+      "epoch": 0.6461234242399275,
+      "grad_norm": 0.0,
+      "learning_rate": 5.879936017636148e-06,
+      "loss": 0.783,
+      "step": 23526
+    },
+    {
+      "epoch": 0.6461508884677707,
+      "grad_norm": 0.0,
+      "learning_rate": 5.879125523889475e-06,
+      "loss": 0.8378,
+      "step": 23527
+    },
+    {
+      "epoch": 0.646178352695614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.878315062748761e-06,
+      "loss": 0.812,
+      "step": 23528
+    },
+    {
+      "epoch": 0.6462058169234572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.87750463422041e-06,
+      "loss": 0.8432,
+      "step": 23529
+    },
+    {
+      "epoch": 0.6462332811513004,
+      "grad_norm": 0.0,
+      "learning_rate": 5.876694238310846e-06,
+      "loss": 0.819,
+      "step": 23530
+    },
+    {
+      "epoch": 0.6462607453791437,
+      "grad_norm": 0.0,
+      "learning_rate": 5.875883875026469e-06,
+      "loss": 0.8846,
+      "step": 23531
+    },
+    {
+      "epoch": 0.6462882096069869,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8750735443736995e-06,
+      "loss": 0.8318,
+      "step": 23532
+    },
+    {
+      "epoch": 0.6463156738348301,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8742632463589434e-06,
+      "loss": 0.7697,
+      "step": 23533
+    },
+    {
+      "epoch": 0.6463431380626734,
+      "grad_norm": 0.0,
+      "learning_rate": 5.873452980988614e-06,
+      "loss": 0.934,
+      "step": 23534
+    },
+    {
+      "epoch": 0.6463706022905166,
+      "grad_norm": 0.0,
+      "learning_rate": 5.872642748269126e-06,
+      "loss": 0.9033,
+      "step": 23535
+    },
+    {
+      "epoch": 0.6463980665183598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.871832548206888e-06,
+      "loss": 0.9113,
+      "step": 23536
+    },
+    {
+      "epoch": 0.6464255307462031,
+      "grad_norm": 0.0,
+      "learning_rate": 5.871022380808309e-06,
+      "loss": 0.9264,
+      "step": 23537
+    },
+    {
+      "epoch": 0.6464529949740463,
+      "grad_norm": 0.0,
+      "learning_rate": 5.870212246079797e-06,
+      "loss": 0.7737,
+      "step": 23538
+    },
+    {
+      "epoch": 0.6464804592018896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.869402144027767e-06,
+      "loss": 0.9104,
+      "step": 23539
+    },
+    {
+      "epoch": 0.6465079234297327,
+      "grad_norm": 0.0,
+      "learning_rate": 5.86859207465863e-06,
+      "loss": 0.8749,
+      "step": 23540
+    },
+    {
+      "epoch": 0.646535387657576,
+      "grad_norm": 0.0,
+      "learning_rate": 5.867782037978789e-06,
+      "loss": 0.7692,
+      "step": 23541
+    },
+    {
+      "epoch": 0.6465628518854193,
+      "grad_norm": 0.0,
+      "learning_rate": 5.866972033994662e-06,
+      "loss": 0.8335,
+      "step": 23542
+    },
+    {
+      "epoch": 0.6465903161132625,
+      "grad_norm": 0.0,
+      "learning_rate": 5.86616206271265e-06,
+      "loss": 0.8123,
+      "step": 23543
+    },
+    {
+      "epoch": 0.6466177803411057,
+      "grad_norm": 0.0,
+      "learning_rate": 5.865352124139165e-06,
+      "loss": 0.899,
+      "step": 23544
+    },
+    {
+      "epoch": 0.6466452445689489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.864542218280615e-06,
+      "loss": 0.8972,
+      "step": 23545
+    },
+    {
+      "epoch": 0.6466727087967922,
+      "grad_norm": 0.0,
+      "learning_rate": 5.863732345143409e-06,
+      "loss": 0.881,
+      "step": 23546
+    },
+    {
+      "epoch": 0.6467001730246354,
+      "grad_norm": 0.0,
+      "learning_rate": 5.862922504733958e-06,
+      "loss": 0.8654,
+      "step": 23547
+    },
+    {
+      "epoch": 0.6467276372524786,
+      "grad_norm": 0.0,
+      "learning_rate": 5.862112697058667e-06,
+      "loss": 0.8787,
+      "step": 23548
+    },
+    {
+      "epoch": 0.6467551014803219,
+      "grad_norm": 0.0,
+      "learning_rate": 5.861302922123939e-06,
+      "loss": 0.7734,
+      "step": 23549
+    },
+    {
+      "epoch": 0.6467825657081652,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8604931799361865e-06,
+      "loss": 0.7667,
+      "step": 23550
+    },
+    {
+      "epoch": 0.6468100299360083,
+      "grad_norm": 0.0,
+      "learning_rate": 5.859683470501818e-06,
+      "loss": 0.8087,
+      "step": 23551
+    },
+    {
+      "epoch": 0.6468374941638516,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8588737938272344e-06,
+      "loss": 0.8013,
+      "step": 23552
+    },
+    {
+      "epoch": 0.6468649583916948,
+      "grad_norm": 0.0,
+      "learning_rate": 5.858064149918845e-06,
+      "loss": 0.8398,
+      "step": 23553
+    },
+    {
+      "epoch": 0.646892422619538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8572545387830615e-06,
+      "loss": 0.8438,
+      "step": 23554
+    },
+    {
+      "epoch": 0.6469198868473813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.856444960426282e-06,
+      "loss": 0.705,
+      "step": 23555
+    },
+    {
+      "epoch": 0.6469473510752245,
+      "grad_norm": 0.0,
+      "learning_rate": 5.855635414854914e-06,
+      "loss": 0.8871,
+      "step": 23556
+    },
+    {
+      "epoch": 0.6469748153030678,
+      "grad_norm": 0.0,
+      "learning_rate": 5.854825902075365e-06,
+      "loss": 0.8793,
+      "step": 23557
+    },
+    {
+      "epoch": 0.6470022795309109,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8540164220940374e-06,
+      "loss": 0.897,
+      "step": 23558
+    },
+    {
+      "epoch": 0.6470297437587542,
+      "grad_norm": 0.0,
+      "learning_rate": 5.85320697491734e-06,
+      "loss": 0.8077,
+      "step": 23559
+    },
+    {
+      "epoch": 0.6470572079865975,
+      "grad_norm": 0.0,
+      "learning_rate": 5.852397560551678e-06,
+      "loss": 0.9038,
+      "step": 23560
+    },
+    {
+      "epoch": 0.6470846722144407,
+      "grad_norm": 0.0,
+      "learning_rate": 5.851588179003449e-06,
+      "loss": 0.8554,
+      "step": 23561
+    },
+    {
+      "epoch": 0.6471121364422839,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8507788302790605e-06,
+      "loss": 0.8473,
+      "step": 23562
+    },
+    {
+      "epoch": 0.6471396006701272,
+      "grad_norm": 0.0,
+      "learning_rate": 5.84996951438492e-06,
+      "loss": 0.8072,
+      "step": 23563
+    },
+    {
+      "epoch": 0.6471670648979704,
+      "grad_norm": 0.0,
+      "learning_rate": 5.849160231327427e-06,
+      "loss": 0.8324,
+      "step": 23564
+    },
+    {
+      "epoch": 0.6471945291258137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.848350981112985e-06,
+      "loss": 0.8897,
+      "step": 23565
+    },
+    {
+      "epoch": 0.6472219933536568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8475417637480016e-06,
+      "loss": 0.8565,
+      "step": 23566
+    },
+    {
+      "epoch": 0.6472494575815001,
+      "grad_norm": 0.0,
+      "learning_rate": 5.846732579238873e-06,
+      "loss": 0.7926,
+      "step": 23567
+    },
+    {
+      "epoch": 0.6472769218093434,
+      "grad_norm": 0.0,
+      "learning_rate": 5.845923427592006e-06,
+      "loss": 0.8748,
+      "step": 23568
+    },
+    {
+      "epoch": 0.6473043860371865,
+      "grad_norm": 0.0,
+      "learning_rate": 5.845114308813802e-06,
+      "loss": 0.8222,
+      "step": 23569
+    },
+    {
+      "epoch": 0.6473318502650298,
+      "grad_norm": 0.0,
+      "learning_rate": 5.844305222910663e-06,
+      "loss": 0.8184,
+      "step": 23570
+    },
+    {
+      "epoch": 0.647359314492873,
+      "grad_norm": 0.0,
+      "learning_rate": 5.843496169888992e-06,
+      "loss": 0.7666,
+      "step": 23571
+    },
+    {
+      "epoch": 0.6473867787207163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.842687149755191e-06,
+      "loss": 0.9343,
+      "step": 23572
+    },
+    {
+      "epoch": 0.6474142429485595,
+      "grad_norm": 0.0,
+      "learning_rate": 5.841878162515655e-06,
+      "loss": 0.7832,
+      "step": 23573
+    },
+    {
+      "epoch": 0.6474417071764027,
+      "grad_norm": 0.0,
+      "learning_rate": 5.841069208176791e-06,
+      "loss": 0.8468,
+      "step": 23574
+    },
+    {
+      "epoch": 0.647469171404246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.840260286745e-06,
+      "loss": 0.9189,
+      "step": 23575
+    },
+    {
+      "epoch": 0.6474966356320893,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8394513982266786e-06,
+      "loss": 0.8352,
+      "step": 23576
+    },
+    {
+      "epoch": 0.6475240998599324,
+      "grad_norm": 0.0,
+      "learning_rate": 5.838642542628235e-06,
+      "loss": 0.8016,
+      "step": 23577
+    },
+    {
+      "epoch": 0.6475515640877757,
+      "grad_norm": 0.0,
+      "learning_rate": 5.837833719956058e-06,
+      "loss": 0.9012,
+      "step": 23578
+    },
+    {
+      "epoch": 0.6475790283156189,
+      "grad_norm": 0.0,
+      "learning_rate": 5.837024930216555e-06,
+      "loss": 0.7736,
+      "step": 23579
+    },
+    {
+      "epoch": 0.6476064925434621,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8362161734161204e-06,
+      "loss": 0.9739,
+      "step": 23580
+    },
+    {
+      "epoch": 0.6476339567713054,
+      "grad_norm": 0.0,
+      "learning_rate": 5.835407449561158e-06,
+      "loss": 0.9454,
+      "step": 23581
+    },
+    {
+      "epoch": 0.6476614209991486,
+      "grad_norm": 0.0,
+      "learning_rate": 5.834598758658066e-06,
+      "loss": 0.8654,
+      "step": 23582
+    },
+    {
+      "epoch": 0.6476888852269919,
+      "grad_norm": 0.0,
+      "learning_rate": 5.833790100713243e-06,
+      "loss": 0.7334,
+      "step": 23583
+    },
+    {
+      "epoch": 0.647716349454835,
+      "grad_norm": 0.0,
+      "learning_rate": 5.832981475733085e-06,
+      "loss": 0.9142,
+      "step": 23584
+    },
+    {
+      "epoch": 0.6477438136826783,
+      "grad_norm": 0.0,
+      "learning_rate": 5.83217288372399e-06,
+      "loss": 0.9122,
+      "step": 23585
+    },
+    {
+      "epoch": 0.6477712779105216,
+      "grad_norm": 0.0,
+      "learning_rate": 5.831364324692363e-06,
+      "loss": 0.8912,
+      "step": 23586
+    },
+    {
+      "epoch": 0.6477987421383647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.830555798644593e-06,
+      "loss": 0.8438,
+      "step": 23587
+    },
+    {
+      "epoch": 0.647826206366208,
+      "grad_norm": 0.0,
+      "learning_rate": 5.829747305587079e-06,
+      "loss": 0.7606,
+      "step": 23588
+    },
+    {
+      "epoch": 0.6478536705940513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.828938845526224e-06,
+      "loss": 0.7704,
+      "step": 23589
+    },
+    {
+      "epoch": 0.6478811348218945,
+      "grad_norm": 0.0,
+      "learning_rate": 5.828130418468416e-06,
+      "loss": 0.8988,
+      "step": 23590
+    },
+    {
+      "epoch": 0.6479085990497377,
+      "grad_norm": 0.0,
+      "learning_rate": 5.827322024420057e-06,
+      "loss": 0.8621,
+      "step": 23591
+    },
+    {
+      "epoch": 0.6479360632775809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.826513663387543e-06,
+      "loss": 0.8554,
+      "step": 23592
+    },
+    {
+      "epoch": 0.6479635275054242,
+      "grad_norm": 0.0,
+      "learning_rate": 5.825705335377272e-06,
+      "loss": 0.8661,
+      "step": 23593
+    },
+    {
+      "epoch": 0.6479909917332675,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824897040395634e-06,
+      "loss": 0.7342,
+      "step": 23594
+    },
+    {
+      "epoch": 0.6480184559611106,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824088778449033e-06,
+      "loss": 0.784,
+      "step": 23595
+    },
+    {
+      "epoch": 0.6480459201889539,
+      "grad_norm": 0.0,
+      "learning_rate": 5.823280549543854e-06,
+      "loss": 0.8,
+      "step": 23596
+    },
+    {
+      "epoch": 0.6480733844167971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.822472353686497e-06,
+      "loss": 0.8236,
+      "step": 23597
+    },
+    {
+      "epoch": 0.6481008486446403,
+      "grad_norm": 0.0,
+      "learning_rate": 5.821664190883358e-06,
+      "loss": 0.8972,
+      "step": 23598
+    },
+    {
+      "epoch": 0.6481283128724836,
+      "grad_norm": 0.0,
+      "learning_rate": 5.82085606114083e-06,
+      "loss": 0.8357,
+      "step": 23599
+    },
+    {
+      "epoch": 0.6481557771003268,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8200479644653075e-06,
+      "loss": 0.7825,
+      "step": 23600
+    },
+    {
+      "epoch": 0.6481832413281701,
+      "grad_norm": 0.0,
+      "learning_rate": 5.81923990086319e-06,
+      "loss": 0.8107,
+      "step": 23601
+    },
+    {
+      "epoch": 0.6482107055560133,
+      "grad_norm": 0.0,
+      "learning_rate": 5.818431870340864e-06,
+      "loss": 0.9489,
+      "step": 23602
+    },
+    {
+      "epoch": 0.6482381697838565,
+      "grad_norm": 0.0,
+      "learning_rate": 5.81762387290472e-06,
+      "loss": 0.7976,
+      "step": 23603
+    },
+    {
+      "epoch": 0.6482656340116998,
+      "grad_norm": 0.0,
+      "learning_rate": 5.816815908561159e-06,
+      "loss": 0.8774,
+      "step": 23604
+    },
+    {
+      "epoch": 0.648293098239543,
+      "grad_norm": 0.0,
+      "learning_rate": 5.816007977316569e-06,
+      "loss": 0.8884,
+      "step": 23605
+    },
+    {
+      "epoch": 0.6483205624673862,
+      "grad_norm": 0.0,
+      "learning_rate": 5.815200079177346e-06,
+      "loss": 0.8441,
+      "step": 23606
+    },
+    {
+      "epoch": 0.6483480266952295,
+      "grad_norm": 0.0,
+      "learning_rate": 5.814392214149883e-06,
+      "loss": 0.8762,
+      "step": 23607
+    },
+    {
+      "epoch": 0.6483754909230727,
+      "grad_norm": 0.0,
+      "learning_rate": 5.813584382240567e-06,
+      "loss": 0.8675,
+      "step": 23608
+    },
+    {
+      "epoch": 0.648402955150916,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8127765834557934e-06,
+      "loss": 0.8322,
+      "step": 23609
+    },
+    {
+      "epoch": 0.6484304193787591,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8119688178019575e-06,
+      "loss": 0.7731,
+      "step": 23610
+    },
+    {
+      "epoch": 0.6484578836066024,
+      "grad_norm": 0.0,
+      "learning_rate": 5.811161085285441e-06,
+      "loss": 0.8047,
+      "step": 23611
+    },
+    {
+      "epoch": 0.6484853478344457,
+      "grad_norm": 0.0,
+      "learning_rate": 5.810353385912643e-06,
+      "loss": 0.8793,
+      "step": 23612
+    },
+    {
+      "epoch": 0.6485128120622888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.809545719689955e-06,
+      "loss": 0.9055,
+      "step": 23613
+    },
+    {
+      "epoch": 0.6485402762901321,
+      "grad_norm": 0.0,
+      "learning_rate": 5.80873808662376e-06,
+      "loss": 0.7745,
+      "step": 23614
+    },
+    {
+      "epoch": 0.6485677405179754,
+      "grad_norm": 0.0,
+      "learning_rate": 5.807930486720455e-06,
+      "loss": 0.8666,
+      "step": 23615
+    },
+    {
+      "epoch": 0.6485952047458186,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8071229199864254e-06,
+      "loss": 0.7603,
+      "step": 23616
+    },
+    {
+      "epoch": 0.6486226689736618,
+      "grad_norm": 0.0,
+      "learning_rate": 5.806315386428067e-06,
+      "loss": 0.9146,
+      "step": 23617
+    },
+    {
+      "epoch": 0.648650133201505,
+      "grad_norm": 0.0,
+      "learning_rate": 5.805507886051764e-06,
+      "loss": 0.8395,
+      "step": 23618
+    },
+    {
+      "epoch": 0.6486775974293483,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8047004188639085e-06,
+      "loss": 0.9335,
+      "step": 23619
+    },
+    {
+      "epoch": 0.6487050616571916,
+      "grad_norm": 0.0,
+      "learning_rate": 5.803892984870887e-06,
+      "loss": 0.891,
+      "step": 23620
+    },
+    {
+      "epoch": 0.6487325258850347,
+      "grad_norm": 0.0,
+      "learning_rate": 5.803085584079089e-06,
+      "loss": 0.7184,
+      "step": 23621
+    },
+    {
+      "epoch": 0.648759990112878,
+      "grad_norm": 0.0,
+      "learning_rate": 5.802278216494903e-06,
+      "loss": 0.7976,
+      "step": 23622
+    },
+    {
+      "epoch": 0.6487874543407212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.801470882124718e-06,
+      "loss": 0.7535,
+      "step": 23623
+    },
+    {
+      "epoch": 0.6488149185685644,
+      "grad_norm": 0.0,
+      "learning_rate": 5.800663580974921e-06,
+      "loss": 0.9549,
+      "step": 23624
+    },
+    {
+      "epoch": 0.6488423827964077,
+      "grad_norm": 0.0,
+      "learning_rate": 5.799856313051906e-06,
+      "loss": 0.7313,
+      "step": 23625
+    },
+    {
+      "epoch": 0.6488698470242509,
+      "grad_norm": 0.0,
+      "learning_rate": 5.799049078362054e-06,
+      "loss": 0.8612,
+      "step": 23626
+    },
+    {
+      "epoch": 0.6488973112520942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.79824187691175e-06,
+      "loss": 0.8461,
+      "step": 23627
+    },
+    {
+      "epoch": 0.6489247754799374,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7974347087073825e-06,
+      "loss": 0.8313,
+      "step": 23628
+    },
+    {
+      "epoch": 0.6489522397077806,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7966275737553404e-06,
+      "loss": 0.8047,
+      "step": 23629
+    },
+    {
+      "epoch": 0.6489797039356239,
+      "grad_norm": 0.0,
+      "learning_rate": 5.795820472062008e-06,
+      "loss": 0.9265,
+      "step": 23630
+    },
+    {
+      "epoch": 0.649007168163467,
+      "grad_norm": 0.0,
+      "learning_rate": 5.795013403633778e-06,
+      "loss": 0.8783,
+      "step": 23631
+    },
+    {
+      "epoch": 0.6490346323913103,
+      "grad_norm": 0.0,
+      "learning_rate": 5.794206368477026e-06,
+      "loss": 0.8374,
+      "step": 23632
+    },
+    {
+      "epoch": 0.6490620966191536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.793399366598142e-06,
+      "loss": 0.8609,
+      "step": 23633
+    },
+    {
+      "epoch": 0.6490895608469968,
+      "grad_norm": 0.0,
+      "learning_rate": 5.792592398003516e-06,
+      "loss": 0.8172,
+      "step": 23634
+    },
+    {
+      "epoch": 0.64911702507484,
+      "grad_norm": 0.0,
+      "learning_rate": 5.791785462699525e-06,
+      "loss": 0.8265,
+      "step": 23635
+    },
+    {
+      "epoch": 0.6491444893026832,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7909785606925615e-06,
+      "loss": 0.8424,
+      "step": 23636
+    },
+    {
+      "epoch": 0.6491719535305265,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7901716919890016e-06,
+      "loss": 0.8336,
+      "step": 23637
+    },
+    {
+      "epoch": 0.6491994177583698,
+      "grad_norm": 0.0,
+      "learning_rate": 5.789364856595234e-06,
+      "loss": 0.8174,
+      "step": 23638
+    },
+    {
+      "epoch": 0.6492268819862129,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7885580545176435e-06,
+      "loss": 0.8616,
+      "step": 23639
+    },
+    {
+      "epoch": 0.6492543462140562,
+      "grad_norm": 0.0,
+      "learning_rate": 5.787751285762613e-06,
+      "loss": 0.828,
+      "step": 23640
+    },
+    {
+      "epoch": 0.6492818104418995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.786944550336529e-06,
+      "loss": 0.8454,
+      "step": 23641
+    },
+    {
+      "epoch": 0.6493092746697426,
+      "grad_norm": 0.0,
+      "learning_rate": 5.786137848245769e-06,
+      "loss": 0.7825,
+      "step": 23642
+    },
+    {
+      "epoch": 0.6493367388975859,
+      "grad_norm": 0.0,
+      "learning_rate": 5.785331179496721e-06,
+      "loss": 0.8892,
+      "step": 23643
+    },
+    {
+      "epoch": 0.6493642031254291,
+      "grad_norm": 0.0,
+      "learning_rate": 5.784524544095762e-06,
+      "loss": 0.8757,
+      "step": 23644
+    },
+    {
+      "epoch": 0.6493916673532724,
+      "grad_norm": 0.0,
+      "learning_rate": 5.783717942049278e-06,
+      "loss": 0.8566,
+      "step": 23645
+    },
+    {
+      "epoch": 0.6494191315811156,
+      "grad_norm": 0.0,
+      "learning_rate": 5.782911373363651e-06,
+      "loss": 0.7924,
+      "step": 23646
+    },
+    {
+      "epoch": 0.6494465958089588,
+      "grad_norm": 0.0,
+      "learning_rate": 5.782104838045264e-06,
+      "loss": 0.7935,
+      "step": 23647
+    },
+    {
+      "epoch": 0.6494740600368021,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7812983361004995e-06,
+      "loss": 0.8307,
+      "step": 23648
+    },
+    {
+      "epoch": 0.6495015242646452,
+      "grad_norm": 0.0,
+      "learning_rate": 5.780491867535733e-06,
+      "loss": 0.794,
+      "step": 23649
+    },
+    {
+      "epoch": 0.6495289884924885,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7796854323573535e-06,
+      "loss": 0.8166,
+      "step": 23650
+    },
+    {
+      "epoch": 0.6495564527203318,
+      "grad_norm": 0.0,
+      "learning_rate": 5.778879030571734e-06,
+      "loss": 0.9596,
+      "step": 23651
+    },
+    {
+      "epoch": 0.649583916948175,
+      "grad_norm": 0.0,
+      "learning_rate": 5.778072662185258e-06,
+      "loss": 0.928,
+      "step": 23652
+    },
+    {
+      "epoch": 0.6496113811760182,
+      "grad_norm": 0.0,
+      "learning_rate": 5.777266327204307e-06,
+      "loss": 0.833,
+      "step": 23653
+    },
+    {
+      "epoch": 0.6496388454038615,
+      "grad_norm": 0.0,
+      "learning_rate": 5.776460025635264e-06,
+      "loss": 0.9199,
+      "step": 23654
+    },
+    {
+      "epoch": 0.6496663096317047,
+      "grad_norm": 0.0,
+      "learning_rate": 5.775653757484503e-06,
+      "loss": 0.8211,
+      "step": 23655
+    },
+    {
+      "epoch": 0.649693773859548,
+      "grad_norm": 0.0,
+      "learning_rate": 5.774847522758404e-06,
+      "loss": 0.8081,
+      "step": 23656
+    },
+    {
+      "epoch": 0.6497212380873911,
+      "grad_norm": 0.0,
+      "learning_rate": 5.774041321463349e-06,
+      "loss": 0.8849,
+      "step": 23657
+    },
+    {
+      "epoch": 0.6497487023152344,
+      "grad_norm": 0.0,
+      "learning_rate": 5.773235153605718e-06,
+      "loss": 0.775,
+      "step": 23658
+    },
+    {
+      "epoch": 0.6497761665430777,
+      "grad_norm": 0.0,
+      "learning_rate": 5.772429019191887e-06,
+      "loss": 0.9043,
+      "step": 23659
+    },
+    {
+      "epoch": 0.6498036307709208,
+      "grad_norm": 0.0,
+      "learning_rate": 5.771622918228237e-06,
+      "loss": 0.8148,
+      "step": 23660
+    },
+    {
+      "epoch": 0.6498310949987641,
+      "grad_norm": 0.0,
+      "learning_rate": 5.770816850721142e-06,
+      "loss": 0.7797,
+      "step": 23661
+    },
+    {
+      "epoch": 0.6498585592266073,
+      "grad_norm": 0.0,
+      "learning_rate": 5.770010816676981e-06,
+      "loss": 0.8632,
+      "step": 23662
+    },
+    {
+      "epoch": 0.6498860234544506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.769204816102133e-06,
+      "loss": 0.9154,
+      "step": 23663
+    },
+    {
+      "epoch": 0.6499134876822938,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7683988490029765e-06,
+      "loss": 0.8494,
+      "step": 23664
+    },
+    {
+      "epoch": 0.649940951910137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.76759291538589e-06,
+      "loss": 0.8404,
+      "step": 23665
+    },
+    {
+      "epoch": 0.6499684161379803,
+      "grad_norm": 0.0,
+      "learning_rate": 5.766787015257249e-06,
+      "loss": 0.8744,
+      "step": 23666
+    },
+    {
+      "epoch": 0.6499958803658236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.765981148623424e-06,
+      "loss": 0.7999,
+      "step": 23667
+    },
+    {
+      "epoch": 0.6500233445936667,
+      "grad_norm": 0.0,
+      "learning_rate": 5.765175315490796e-06,
+      "loss": 0.8888,
+      "step": 23668
+    },
+    {
+      "epoch": 0.65005080882151,
+      "grad_norm": 0.0,
+      "learning_rate": 5.764369515865742e-06,
+      "loss": 0.8813,
+      "step": 23669
+    },
+    {
+      "epoch": 0.6500782730493532,
+      "grad_norm": 0.0,
+      "learning_rate": 5.763563749754637e-06,
+      "loss": 0.9143,
+      "step": 23670
+    },
+    {
+      "epoch": 0.6501057372771964,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7627580171638574e-06,
+      "loss": 0.8462,
+      "step": 23671
+    },
+    {
+      "epoch": 0.6501332015050397,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7619523180997814e-06,
+      "loss": 0.8808,
+      "step": 23672
+    },
+    {
+      "epoch": 0.6501606657328829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.761146652568777e-06,
+      "loss": 0.9519,
+      "step": 23673
+    },
+    {
+      "epoch": 0.6501881299607262,
+      "grad_norm": 0.0,
+      "learning_rate": 5.760341020577225e-06,
+      "loss": 0.8444,
+      "step": 23674
+    },
+    {
+      "epoch": 0.6502155941885693,
+      "grad_norm": 0.0,
+      "learning_rate": 5.759535422131492e-06,
+      "loss": 0.8546,
+      "step": 23675
+    },
+    {
+      "epoch": 0.6502430584164126,
+      "grad_norm": 0.0,
+      "learning_rate": 5.758729857237959e-06,
+      "loss": 0.8334,
+      "step": 23676
+    },
+    {
+      "epoch": 0.6502705226442559,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7579243259029985e-06,
+      "loss": 0.8034,
+      "step": 23677
+    },
+    {
+      "epoch": 0.650297986872099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.757118828132987e-06,
+      "loss": 0.845,
+      "step": 23678
+    },
+    {
+      "epoch": 0.6503254510999423,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7563133639342915e-06,
+      "loss": 0.7964,
+      "step": 23679
+    },
+    {
+      "epoch": 0.6503529153277856,
+      "grad_norm": 0.0,
+      "learning_rate": 5.755507933313289e-06,
+      "loss": 0.8973,
+      "step": 23680
+    },
+    {
+      "epoch": 0.6503803795556288,
+      "grad_norm": 0.0,
+      "learning_rate": 5.754702536276351e-06,
+      "loss": 0.8937,
+      "step": 23681
+    },
+    {
+      "epoch": 0.650407843783472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.753897172829855e-06,
+      "loss": 0.8116,
+      "step": 23682
+    },
+    {
+      "epoch": 0.6504353080113152,
+      "grad_norm": 0.0,
+      "learning_rate": 5.753091842980165e-06,
+      "loss": 0.7765,
+      "step": 23683
+    },
+    {
+      "epoch": 0.6504627722391585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.752286546733663e-06,
+      "loss": 0.8931,
+      "step": 23684
+    },
+    {
+      "epoch": 0.6504902364670018,
+      "grad_norm": 0.0,
+      "learning_rate": 5.751481284096709e-06,
+      "loss": 0.7939,
+      "step": 23685
+    },
+    {
+      "epoch": 0.6505177006948449,
+      "grad_norm": 0.0,
+      "learning_rate": 5.750676055075683e-06,
+      "loss": 0.8014,
+      "step": 23686
+    },
+    {
+      "epoch": 0.6505451649226882,
+      "grad_norm": 0.0,
+      "learning_rate": 5.749870859676954e-06,
+      "loss": 0.8748,
+      "step": 23687
+    },
+    {
+      "epoch": 0.6505726291505314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.749065697906891e-06,
+      "loss": 0.8133,
+      "step": 23688
+    },
+    {
+      "epoch": 0.6506000933783747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.748260569771873e-06,
+      "loss": 0.9148,
+      "step": 23689
+    },
+    {
+      "epoch": 0.6506275576062179,
+      "grad_norm": 0.0,
+      "learning_rate": 5.747455475278263e-06,
+      "loss": 0.8511,
+      "step": 23690
+    },
+    {
+      "epoch": 0.6506550218340611,
+      "grad_norm": 0.0,
+      "learning_rate": 5.746650414432428e-06,
+      "loss": 0.8013,
+      "step": 23691
+    },
+    {
+      "epoch": 0.6506824860619044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.745845387240745e-06,
+      "loss": 0.9065,
+      "step": 23692
+    },
+    {
+      "epoch": 0.6507099502897477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7450403937095785e-06,
+      "loss": 0.7727,
+      "step": 23693
+    },
+    {
+      "epoch": 0.6507374145175908,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7442354338453026e-06,
+      "loss": 0.8795,
+      "step": 23694
+    },
+    {
+      "epoch": 0.6507648787454341,
+      "grad_norm": 0.0,
+      "learning_rate": 5.743430507654287e-06,
+      "loss": 0.8339,
+      "step": 23695
+    },
+    {
+      "epoch": 0.6507923429732773,
+      "grad_norm": 0.0,
+      "learning_rate": 5.742625615142896e-06,
+      "loss": 0.797,
+      "step": 23696
+    },
+    {
+      "epoch": 0.6508198072011205,
+      "grad_norm": 0.0,
+      "learning_rate": 5.741820756317498e-06,
+      "loss": 0.822,
+      "step": 23697
+    },
+    {
+      "epoch": 0.6508472714289638,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7410159311844694e-06,
+      "loss": 0.8537,
+      "step": 23698
+    },
+    {
+      "epoch": 0.650874735656807,
+      "grad_norm": 0.0,
+      "learning_rate": 5.740211139750167e-06,
+      "loss": 0.829,
+      "step": 23699
+    },
+    {
+      "epoch": 0.6509021998846503,
+      "grad_norm": 0.0,
+      "learning_rate": 5.739406382020966e-06,
+      "loss": 0.8478,
+      "step": 23700
+    },
+    {
+      "epoch": 0.6509296641124934,
+      "grad_norm": 0.0,
+      "learning_rate": 5.738601658003234e-06,
+      "loss": 0.8217,
+      "step": 23701
+    },
+    {
+      "epoch": 0.6509571283403367,
+      "grad_norm": 0.0,
+      "learning_rate": 5.737796967703334e-06,
+      "loss": 0.8285,
+      "step": 23702
+    },
+    {
+      "epoch": 0.65098459256818,
+      "grad_norm": 0.0,
+      "learning_rate": 5.736992311127635e-06,
+      "loss": 0.8536,
+      "step": 23703
+    },
+    {
+      "epoch": 0.6510120567960231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.736187688282504e-06,
+      "loss": 0.861,
+      "step": 23704
+    },
+    {
+      "epoch": 0.6510395210238664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.735383099174312e-06,
+      "loss": 0.8891,
+      "step": 23705
+    },
+    {
+      "epoch": 0.6510669852517097,
+      "grad_norm": 0.0,
+      "learning_rate": 5.734578543809416e-06,
+      "loss": 0.8767,
+      "step": 23706
+    },
+    {
+      "epoch": 0.6510944494795529,
+      "grad_norm": 0.0,
+      "learning_rate": 5.733774022194192e-06,
+      "loss": 0.8217,
+      "step": 23707
+    },
+    {
+      "epoch": 0.6511219137073961,
+      "grad_norm": 0.0,
+      "learning_rate": 5.732969534334995e-06,
+      "loss": 0.8024,
+      "step": 23708
+    },
+    {
+      "epoch": 0.6511493779352393,
+      "grad_norm": 0.0,
+      "learning_rate": 5.732165080238197e-06,
+      "loss": 0.8844,
+      "step": 23709
+    },
+    {
+      "epoch": 0.6511768421630826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.731360659910163e-06,
+      "loss": 0.9038,
+      "step": 23710
+    },
+    {
+      "epoch": 0.6512043063909259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.730556273357255e-06,
+      "loss": 0.7681,
+      "step": 23711
+    },
+    {
+      "epoch": 0.651231770618769,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72975192058584e-06,
+      "loss": 0.963,
+      "step": 23712
+    },
+    {
+      "epoch": 0.6512592348466123,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7289476016022875e-06,
+      "loss": 0.8666,
+      "step": 23713
+    },
+    {
+      "epoch": 0.6512866990744555,
+      "grad_norm": 0.0,
+      "learning_rate": 5.728143316412955e-06,
+      "loss": 0.9095,
+      "step": 23714
+    },
+    {
+      "epoch": 0.6513141633022987,
+      "grad_norm": 0.0,
+      "learning_rate": 5.727339065024202e-06,
+      "loss": 0.8865,
+      "step": 23715
+    },
+    {
+      "epoch": 0.651341627530142,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7265348474424e-06,
+      "loss": 0.8087,
+      "step": 23716
+    },
+    {
+      "epoch": 0.6513690917579852,
+      "grad_norm": 0.0,
+      "learning_rate": 5.725730663673907e-06,
+      "loss": 0.9254,
+      "step": 23717
+    },
+    {
+      "epoch": 0.6513965559858285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.724926513725091e-06,
+      "loss": 0.8596,
+      "step": 23718
+    },
+    {
+      "epoch": 0.6514240202136717,
+      "grad_norm": 0.0,
+      "learning_rate": 5.724122397602316e-06,
+      "loss": 0.8404,
+      "step": 23719
+    },
+    {
+      "epoch": 0.6514514844415149,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7233183153119355e-06,
+      "loss": 0.8719,
+      "step": 23720
+    },
+    {
+      "epoch": 0.6514789486693582,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7225142668603175e-06,
+      "loss": 0.8488,
+      "step": 23721
+    },
+    {
+      "epoch": 0.6515064128972013,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72171025225383e-06,
+      "loss": 0.7517,
+      "step": 23722
+    },
+    {
+      "epoch": 0.6515338771250446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72090627149882e-06,
+      "loss": 0.9261,
+      "step": 23723
+    },
+    {
+      "epoch": 0.6515613413528879,
+      "grad_norm": 0.0,
+      "learning_rate": 5.720102324601659e-06,
+      "loss": 0.8616,
+      "step": 23724
+    },
+    {
+      "epoch": 0.6515888055807311,
+      "grad_norm": 0.0,
+      "learning_rate": 5.719298411568712e-06,
+      "loss": 0.8087,
+      "step": 23725
+    },
+    {
+      "epoch": 0.6516162698085743,
+      "grad_norm": 0.0,
+      "learning_rate": 5.718494532406329e-06,
+      "loss": 0.9094,
+      "step": 23726
+    },
+    {
+      "epoch": 0.6516437340364176,
+      "grad_norm": 0.0,
+      "learning_rate": 5.717690687120876e-06,
+      "loss": 0.9171,
+      "step": 23727
+    },
+    {
+      "epoch": 0.6516711982642608,
+      "grad_norm": 0.0,
+      "learning_rate": 5.716886875718713e-06,
+      "loss": 0.9107,
+      "step": 23728
+    },
+    {
+      "epoch": 0.6516986624921041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.716083098206203e-06,
+      "loss": 0.8868,
+      "step": 23729
+    },
+    {
+      "epoch": 0.6517261267199472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7152793545897e-06,
+      "loss": 0.853,
+      "step": 23730
+    },
+    {
+      "epoch": 0.6517535909477905,
+      "grad_norm": 0.0,
+      "learning_rate": 5.71447564487557e-06,
+      "loss": 0.8911,
+      "step": 23731
+    },
+    {
+      "epoch": 0.6517810551756338,
+      "grad_norm": 0.0,
+      "learning_rate": 5.713671969070165e-06,
+      "loss": 0.9242,
+      "step": 23732
+    },
+    {
+      "epoch": 0.651808519403477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.712868327179848e-06,
+      "loss": 0.8753,
+      "step": 23733
+    },
+    {
+      "epoch": 0.6518359836313202,
+      "grad_norm": 0.0,
+      "learning_rate": 5.712064719210977e-06,
+      "loss": 0.9175,
+      "step": 23734
+    },
+    {
+      "epoch": 0.6518634478591634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.71126114516991e-06,
+      "loss": 0.9475,
+      "step": 23735
+    },
+    {
+      "epoch": 0.6518909120870067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.710457605063009e-06,
+      "loss": 0.7866,
+      "step": 23736
+    },
+    {
+      "epoch": 0.65191837631485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.709654098896631e-06,
+      "loss": 0.9477,
+      "step": 23737
+    },
+    {
+      "epoch": 0.6519458405426931,
+      "grad_norm": 0.0,
+      "learning_rate": 5.708850626677132e-06,
+      "loss": 0.8842,
+      "step": 23738
+    },
+    {
+      "epoch": 0.6519733047705364,
+      "grad_norm": 0.0,
+      "learning_rate": 5.708047188410866e-06,
+      "loss": 0.8618,
+      "step": 23739
+    },
+    {
+      "epoch": 0.6520007689983797,
+      "grad_norm": 0.0,
+      "learning_rate": 5.707243784104192e-06,
+      "loss": 0.8515,
+      "step": 23740
+    },
+    {
+      "epoch": 0.6520282332262228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.706440413763467e-06,
+      "loss": 0.8701,
+      "step": 23741
+    },
+    {
+      "epoch": 0.6520556974540661,
+      "grad_norm": 0.0,
+      "learning_rate": 5.705637077395051e-06,
+      "loss": 0.9501,
+      "step": 23742
+    },
+    {
+      "epoch": 0.6520831616819093,
+      "grad_norm": 0.0,
+      "learning_rate": 5.704833775005301e-06,
+      "loss": 0.9045,
+      "step": 23743
+    },
+    {
+      "epoch": 0.6521106259097526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.704030506600566e-06,
+      "loss": 0.8866,
+      "step": 23744
+    },
+    {
+      "epoch": 0.6521380901375958,
+      "grad_norm": 0.0,
+      "learning_rate": 5.703227272187205e-06,
+      "loss": 0.91,
+      "step": 23745
+    },
+    {
+      "epoch": 0.652165554365439,
+      "grad_norm": 0.0,
+      "learning_rate": 5.70242407177158e-06,
+      "loss": 0.9098,
+      "step": 23746
+    },
+    {
+      "epoch": 0.6521930185932823,
+      "grad_norm": 0.0,
+      "learning_rate": 5.701620905360034e-06,
+      "loss": 0.84,
+      "step": 23747
+    },
+    {
+      "epoch": 0.6522204828211254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.70081777295893e-06,
+      "loss": 0.7364,
+      "step": 23748
+    },
+    {
+      "epoch": 0.6522479470489687,
+      "grad_norm": 0.0,
+      "learning_rate": 5.700014674574626e-06,
+      "loss": 0.8276,
+      "step": 23749
+    },
+    {
+      "epoch": 0.652275411276812,
+      "grad_norm": 0.0,
+      "learning_rate": 5.699211610213464e-06,
+      "loss": 0.8052,
+      "step": 23750
+    },
+    {
+      "epoch": 0.6523028755046552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.698408579881809e-06,
+      "loss": 0.8743,
+      "step": 23751
+    },
+    {
+      "epoch": 0.6523303397324984,
+      "grad_norm": 0.0,
+      "learning_rate": 5.69760558358601e-06,
+      "loss": 0.8425,
+      "step": 23752
+    },
+    {
+      "epoch": 0.6523578039603417,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696802621332428e-06,
+      "loss": 0.8281,
+      "step": 23753
+    },
+    {
+      "epoch": 0.6523852681881849,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6959996931274054e-06,
+      "loss": 0.7856,
+      "step": 23754
+    },
+    {
+      "epoch": 0.6524127324160282,
+      "grad_norm": 0.0,
+      "learning_rate": 5.695196798977305e-06,
+      "loss": 0.8107,
+      "step": 23755
+    },
+    {
+      "epoch": 0.6524401966438713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.69439393888847e-06,
+      "loss": 0.8576,
+      "step": 23756
+    },
+    {
+      "epoch": 0.6524676608717146,
+      "grad_norm": 0.0,
+      "learning_rate": 5.693591112867258e-06,
+      "loss": 0.8092,
+      "step": 23757
+    },
+    {
+      "epoch": 0.6524951250995579,
+      "grad_norm": 0.0,
+      "learning_rate": 5.692788320920022e-06,
+      "loss": 0.8258,
+      "step": 23758
+    },
+    {
+      "epoch": 0.652522589327401,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6919855630531125e-06,
+      "loss": 0.7822,
+      "step": 23759
+    },
+    {
+      "epoch": 0.6525500535552443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.691182839272886e-06,
+      "loss": 0.8563,
+      "step": 23760
+    },
+    {
+      "epoch": 0.6525775177830875,
+      "grad_norm": 0.0,
+      "learning_rate": 5.690380149585688e-06,
+      "loss": 0.8444,
+      "step": 23761
+    },
+    {
+      "epoch": 0.6526049820109308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6895774939978735e-06,
+      "loss": 0.7997,
+      "step": 23762
+    },
+    {
+      "epoch": 0.652632446238774,
+      "grad_norm": 0.0,
+      "learning_rate": 5.688774872515788e-06,
+      "loss": 1.0216,
+      "step": 23763
+    },
+    {
+      "epoch": 0.6526599104666172,
+      "grad_norm": 0.0,
+      "learning_rate": 5.687972285145786e-06,
+      "loss": 0.978,
+      "step": 23764
+    },
+    {
+      "epoch": 0.6526873746944605,
+      "grad_norm": 0.0,
+      "learning_rate": 5.687169731894218e-06,
+      "loss": 0.8075,
+      "step": 23765
+    },
+    {
+      "epoch": 0.6527148389223038,
+      "grad_norm": 0.0,
+      "learning_rate": 5.686367212767436e-06,
+      "loss": 0.8081,
+      "step": 23766
+    },
+    {
+      "epoch": 0.6527423031501469,
+      "grad_norm": 0.0,
+      "learning_rate": 5.685564727771787e-06,
+      "loss": 0.9587,
+      "step": 23767
+    },
+    {
+      "epoch": 0.6527697673779902,
+      "grad_norm": 0.0,
+      "learning_rate": 5.684762276913618e-06,
+      "loss": 0.8758,
+      "step": 23768
+    },
+    {
+      "epoch": 0.6527972316058334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.683959860199282e-06,
+      "loss": 0.8841,
+      "step": 23769
+    },
+    {
+      "epoch": 0.6528246958336766,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6831574776351314e-06,
+      "loss": 0.9263,
+      "step": 23770
+    },
+    {
+      "epoch": 0.6528521600615199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.682355129227508e-06,
+      "loss": 0.749,
+      "step": 23771
+    },
+    {
+      "epoch": 0.6528796242893631,
+      "grad_norm": 0.0,
+      "learning_rate": 5.681552814982767e-06,
+      "loss": 0.9114,
+      "step": 23772
+    },
+    {
+      "epoch": 0.6529070885172064,
+      "grad_norm": 0.0,
+      "learning_rate": 5.680750534907248e-06,
+      "loss": 0.8282,
+      "step": 23773
+    },
+    {
+      "epoch": 0.6529345527450495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6799482890073046e-06,
+      "loss": 0.7481,
+      "step": 23774
+    },
+    {
+      "epoch": 0.6529620169728928,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6791460772892835e-06,
+      "loss": 0.891,
+      "step": 23775
+    },
+    {
+      "epoch": 0.6529894812007361,
+      "grad_norm": 0.0,
+      "learning_rate": 5.678343899759533e-06,
+      "loss": 0.9167,
+      "step": 23776
+    },
+    {
+      "epoch": 0.6530169454285792,
+      "grad_norm": 0.0,
+      "learning_rate": 5.677541756424403e-06,
+      "loss": 0.8688,
+      "step": 23777
+    },
+    {
+      "epoch": 0.6530444096564225,
+      "grad_norm": 0.0,
+      "learning_rate": 5.676739647290232e-06,
+      "loss": 0.8245,
+      "step": 23778
+    },
+    {
+      "epoch": 0.6530718738842658,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6759375723633785e-06,
+      "loss": 0.8619,
+      "step": 23779
+    },
+    {
+      "epoch": 0.653099338112109,
+      "grad_norm": 0.0,
+      "learning_rate": 5.675135531650175e-06,
+      "loss": 0.8204,
+      "step": 23780
+    },
+    {
+      "epoch": 0.6531268023399522,
+      "grad_norm": 0.0,
+      "learning_rate": 5.674333525156977e-06,
+      "loss": 0.87,
+      "step": 23781
+    },
+    {
+      "epoch": 0.6531542665677954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.673531552890126e-06,
+      "loss": 0.9131,
+      "step": 23782
+    },
+    {
+      "epoch": 0.6531817307956387,
+      "grad_norm": 0.0,
+      "learning_rate": 5.67272961485597e-06,
+      "loss": 0.7851,
+      "step": 23783
+    },
+    {
+      "epoch": 0.653209195023482,
+      "grad_norm": 0.0,
+      "learning_rate": 5.671927711060856e-06,
+      "loss": 0.7735,
+      "step": 23784
+    },
+    {
+      "epoch": 0.6532366592513251,
+      "grad_norm": 0.0,
+      "learning_rate": 5.671125841511125e-06,
+      "loss": 0.9153,
+      "step": 23785
+    },
+    {
+      "epoch": 0.6532641234791684,
+      "grad_norm": 0.0,
+      "learning_rate": 5.670324006213125e-06,
+      "loss": 0.8025,
+      "step": 23786
+    },
+    {
+      "epoch": 0.6532915877070116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.669522205173195e-06,
+      "loss": 0.8647,
+      "step": 23787
+    },
+    {
+      "epoch": 0.6533190519348548,
+      "grad_norm": 0.0,
+      "learning_rate": 5.668720438397683e-06,
+      "loss": 0.7928,
+      "step": 23788
+    },
+    {
+      "epoch": 0.6533465161626981,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6679187058929345e-06,
+      "loss": 0.876,
+      "step": 23789
+    },
+    {
+      "epoch": 0.6533739803905413,
+      "grad_norm": 0.0,
+      "learning_rate": 5.667117007665293e-06,
+      "loss": 0.8097,
+      "step": 23790
+    },
+    {
+      "epoch": 0.6534014446183846,
+      "grad_norm": 0.0,
+      "learning_rate": 5.666315343721096e-06,
+      "loss": 0.9762,
+      "step": 23791
+    },
+    {
+      "epoch": 0.6534289088462278,
+      "grad_norm": 0.0,
+      "learning_rate": 5.665513714066693e-06,
+      "loss": 0.8122,
+      "step": 23792
+    },
+    {
+      "epoch": 0.653456373074071,
+      "grad_norm": 0.0,
+      "learning_rate": 5.664712118708422e-06,
+      "loss": 0.8735,
+      "step": 23793
+    },
+    {
+      "epoch": 0.6534838373019143,
+      "grad_norm": 0.0,
+      "learning_rate": 5.663910557652633e-06,
+      "loss": 0.8463,
+      "step": 23794
+    },
+    {
+      "epoch": 0.6535113015297574,
+      "grad_norm": 0.0,
+      "learning_rate": 5.66310903090566e-06,
+      "loss": 0.7897,
+      "step": 23795
+    },
+    {
+      "epoch": 0.6535387657576007,
+      "grad_norm": 0.0,
+      "learning_rate": 5.662307538473851e-06,
+      "loss": 0.8187,
+      "step": 23796
+    },
+    {
+      "epoch": 0.653566229985444,
+      "grad_norm": 0.0,
+      "learning_rate": 5.661506080363541e-06,
+      "loss": 0.8164,
+      "step": 23797
+    },
+    {
+      "epoch": 0.6535936942132872,
+      "grad_norm": 0.0,
+      "learning_rate": 5.660704656581077e-06,
+      "loss": 0.844,
+      "step": 23798
+    },
+    {
+      "epoch": 0.6536211584411304,
+      "grad_norm": 0.0,
+      "learning_rate": 5.659903267132798e-06,
+      "loss": 0.8563,
+      "step": 23799
+    },
+    {
+      "epoch": 0.6536486226689736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.659101912025045e-06,
+      "loss": 0.8258,
+      "step": 23800
+    },
+    {
+      "epoch": 0.6536760868968169,
+      "grad_norm": 0.0,
+      "learning_rate": 5.658300591264163e-06,
+      "loss": 0.796,
+      "step": 23801
+    },
+    {
+      "epoch": 0.6537035511246602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.657499304856486e-06,
+      "loss": 0.8407,
+      "step": 23802
+    },
+    {
+      "epoch": 0.6537310153525033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.65669805280836e-06,
+      "loss": 0.8785,
+      "step": 23803
+    },
+    {
+      "epoch": 0.6537584795803466,
+      "grad_norm": 0.0,
+      "learning_rate": 5.655896835126116e-06,
+      "loss": 0.814,
+      "step": 23804
+    },
+    {
+      "epoch": 0.6537859438081899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6550956518161e-06,
+      "loss": 0.809,
+      "step": 23805
+    },
+    {
+      "epoch": 0.653813408036033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.654294502884652e-06,
+      "loss": 0.8474,
+      "step": 23806
+    },
+    {
+      "epoch": 0.6538408722638763,
+      "grad_norm": 0.0,
+      "learning_rate": 5.653493388338107e-06,
+      "loss": 0.9024,
+      "step": 23807
+    },
+    {
+      "epoch": 0.6538683364917195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.65269230818281e-06,
+      "loss": 0.7581,
+      "step": 23808
+    },
+    {
+      "epoch": 0.6538958007195628,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6518912624250935e-06,
+      "loss": 0.8974,
+      "step": 23809
+    },
+    {
+      "epoch": 0.653923264947406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6510902510712985e-06,
+      "loss": 0.7999,
+      "step": 23810
+    },
+    {
+      "epoch": 0.6539507291752492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.65028927412776e-06,
+      "loss": 0.8469,
+      "step": 23811
+    },
+    {
+      "epoch": 0.6539781934030925,
+      "grad_norm": 0.0,
+      "learning_rate": 5.649488331600818e-06,
+      "loss": 0.8944,
+      "step": 23812
+    },
+    {
+      "epoch": 0.6540056576309357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.648687423496808e-06,
+      "loss": 0.8796,
+      "step": 23813
+    },
+    {
+      "epoch": 0.6540331218587789,
+      "grad_norm": 0.0,
+      "learning_rate": 5.647886549822075e-06,
+      "loss": 0.6798,
+      "step": 23814
+    },
+    {
+      "epoch": 0.6540605860866222,
+      "grad_norm": 0.0,
+      "learning_rate": 5.647085710582944e-06,
+      "loss": 0.8213,
+      "step": 23815
+    },
+    {
+      "epoch": 0.6540880503144654,
+      "grad_norm": 0.0,
+      "learning_rate": 5.646284905785757e-06,
+      "loss": 0.8561,
+      "step": 23816
+    },
+    {
+      "epoch": 0.6541155145423087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.64548413543685e-06,
+      "loss": 0.8591,
+      "step": 23817
+    },
+    {
+      "epoch": 0.6541429787701519,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6446833995425635e-06,
+      "loss": 0.8718,
+      "step": 23818
+    },
+    {
+      "epoch": 0.6541704429979951,
+      "grad_norm": 0.0,
+      "learning_rate": 5.643882698109229e-06,
+      "loss": 0.9095,
+      "step": 23819
+    },
+    {
+      "epoch": 0.6541979072258384,
+      "grad_norm": 0.0,
+      "learning_rate": 5.643082031143179e-06,
+      "loss": 0.9361,
+      "step": 23820
+    },
+    {
+      "epoch": 0.6542253714536815,
+      "grad_norm": 0.0,
+      "learning_rate": 5.642281398650751e-06,
+      "loss": 0.9064,
+      "step": 23821
+    },
+    {
+      "epoch": 0.6542528356815248,
+      "grad_norm": 0.0,
+      "learning_rate": 5.641480800638279e-06,
+      "loss": 0.8359,
+      "step": 23822
+    },
+    {
+      "epoch": 0.6542802999093681,
+      "grad_norm": 0.0,
+      "learning_rate": 5.640680237112102e-06,
+      "loss": 0.8717,
+      "step": 23823
+    },
+    {
+      "epoch": 0.6543077641372113,
+      "grad_norm": 0.0,
+      "learning_rate": 5.639879708078551e-06,
+      "loss": 0.8529,
+      "step": 23824
+    },
+    {
+      "epoch": 0.6543352283650545,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6390792135439635e-06,
+      "loss": 0.9485,
+      "step": 23825
+    },
+    {
+      "epoch": 0.6543626925928977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.63827875351467e-06,
+      "loss": 0.9426,
+      "step": 23826
+    },
+    {
+      "epoch": 0.654390156820741,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6374783279970015e-06,
+      "loss": 0.7828,
+      "step": 23827
+    },
+    {
+      "epoch": 0.6544176210485843,
+      "grad_norm": 0.0,
+      "learning_rate": 5.636677936997294e-06,
+      "loss": 0.8086,
+      "step": 23828
+    },
+    {
+      "epoch": 0.6544450852764274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6358775805218825e-06,
+      "loss": 0.7821,
+      "step": 23829
+    },
+    {
+      "epoch": 0.6544725495042707,
+      "grad_norm": 0.0,
+      "learning_rate": 5.635077258577097e-06,
+      "loss": 0.8976,
+      "step": 23830
+    },
+    {
+      "epoch": 0.654500013732114,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6342769711692744e-06,
+      "loss": 0.7588,
+      "step": 23831
+    },
+    {
+      "epoch": 0.6545274779599571,
+      "grad_norm": 0.0,
+      "learning_rate": 5.63347671830474e-06,
+      "loss": 0.8301,
+      "step": 23832
+    },
+    {
+      "epoch": 0.6545549421878004,
+      "grad_norm": 0.0,
+      "learning_rate": 5.63267649998983e-06,
+      "loss": 0.8241,
+      "step": 23833
+    },
+    {
+      "epoch": 0.6545824064156436,
+      "grad_norm": 0.0,
+      "learning_rate": 5.631876316230878e-06,
+      "loss": 0.8404,
+      "step": 23834
+    },
+    {
+      "epoch": 0.6546098706434869,
+      "grad_norm": 0.0,
+      "learning_rate": 5.631076167034208e-06,
+      "loss": 0.8489,
+      "step": 23835
+    },
+    {
+      "epoch": 0.6546373348713301,
+      "grad_norm": 0.0,
+      "learning_rate": 5.630276052406158e-06,
+      "loss": 0.7628,
+      "step": 23836
+    },
+    {
+      "epoch": 0.6546647990991733,
+      "grad_norm": 0.0,
+      "learning_rate": 5.62947597235306e-06,
+      "loss": 0.7894,
+      "step": 23837
+    },
+    {
+      "epoch": 0.6546922633270166,
+      "grad_norm": 0.0,
+      "learning_rate": 5.628675926881236e-06,
+      "loss": 0.8303,
+      "step": 23838
+    },
+    {
+      "epoch": 0.6547197275548597,
+      "grad_norm": 0.0,
+      "learning_rate": 5.627875915997022e-06,
+      "loss": 0.8037,
+      "step": 23839
+    },
+    {
+      "epoch": 0.654747191782703,
+      "grad_norm": 0.0,
+      "learning_rate": 5.627075939706747e-06,
+      "loss": 0.8083,
+      "step": 23840
+    },
+    {
+      "epoch": 0.6547746560105463,
+      "grad_norm": 0.0,
+      "learning_rate": 5.626275998016742e-06,
+      "loss": 0.8555,
+      "step": 23841
+    },
+    {
+      "epoch": 0.6548021202383895,
+      "grad_norm": 0.0,
+      "learning_rate": 5.625476090933338e-06,
+      "loss": 0.8574,
+      "step": 23842
+    },
+    {
+      "epoch": 0.6548295844662327,
+      "grad_norm": 0.0,
+      "learning_rate": 5.624676218462863e-06,
+      "loss": 0.8736,
+      "step": 23843
+    },
+    {
+      "epoch": 0.654857048694076,
+      "grad_norm": 0.0,
+      "learning_rate": 5.623876380611641e-06,
+      "loss": 0.8983,
+      "step": 23844
+    },
+    {
+      "epoch": 0.6548845129219192,
+      "grad_norm": 0.0,
+      "learning_rate": 5.623076577386002e-06,
+      "loss": 0.8314,
+      "step": 23845
+    },
+    {
+      "epoch": 0.6549119771497625,
+      "grad_norm": 0.0,
+      "learning_rate": 5.622276808792278e-06,
+      "loss": 0.7937,
+      "step": 23846
+    },
+    {
+      "epoch": 0.6549394413776056,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6214770748367965e-06,
+      "loss": 0.8302,
+      "step": 23847
+    },
+    {
+      "epoch": 0.6549669056054489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.620677375525883e-06,
+      "loss": 0.8657,
+      "step": 23848
+    },
+    {
+      "epoch": 0.6549943698332922,
+      "grad_norm": 0.0,
+      "learning_rate": 5.619877710865871e-06,
+      "loss": 0.8459,
+      "step": 23849
+    },
+    {
+      "epoch": 0.6550218340611353,
+      "grad_norm": 0.0,
+      "learning_rate": 5.619078080863083e-06,
+      "loss": 0.7754,
+      "step": 23850
+    },
+    {
+      "epoch": 0.6550492982889786,
+      "grad_norm": 0.0,
+      "learning_rate": 5.618278485523842e-06,
+      "loss": 0.758,
+      "step": 23851
+    },
+    {
+      "epoch": 0.6550767625168218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.61747892485448e-06,
+      "loss": 0.8335,
+      "step": 23852
+    },
+    {
+      "epoch": 0.6551042267446651,
+      "grad_norm": 0.0,
+      "learning_rate": 5.616679398861322e-06,
+      "loss": 0.9013,
+      "step": 23853
+    },
+    {
+      "epoch": 0.6551316909725083,
+      "grad_norm": 0.0,
+      "learning_rate": 5.615879907550694e-06,
+      "loss": 0.826,
+      "step": 23854
+    },
+    {
+      "epoch": 0.6551591552003515,
+      "grad_norm": 0.0,
+      "learning_rate": 5.615080450928927e-06,
+      "loss": 0.8073,
+      "step": 23855
+    },
+    {
+      "epoch": 0.6551866194281948,
+      "grad_norm": 0.0,
+      "learning_rate": 5.614281029002337e-06,
+      "loss": 0.8623,
+      "step": 23856
+    },
+    {
+      "epoch": 0.6552140836560381,
+      "grad_norm": 0.0,
+      "learning_rate": 5.613481641777256e-06,
+      "loss": 0.9022,
+      "step": 23857
+    },
+    {
+      "epoch": 0.6552415478838812,
+      "grad_norm": 0.0,
+      "learning_rate": 5.612682289260012e-06,
+      "loss": 0.8824,
+      "step": 23858
+    },
+    {
+      "epoch": 0.6552690121117245,
+      "grad_norm": 0.0,
+      "learning_rate": 5.611882971456919e-06,
+      "loss": 0.8408,
+      "step": 23859
+    },
+    {
+      "epoch": 0.6552964763395677,
+      "grad_norm": 0.0,
+      "learning_rate": 5.611083688374308e-06,
+      "loss": 0.8551,
+      "step": 23860
+    },
+    {
+      "epoch": 0.655323940567411,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6102844400185074e-06,
+      "loss": 0.8349,
+      "step": 23861
+    },
+    {
+      "epoch": 0.6553514047952542,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6094852263958336e-06,
+      "loss": 0.8366,
+      "step": 23862
+    },
+    {
+      "epoch": 0.6553788690230974,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6086860475126126e-06,
+      "loss": 0.8659,
+      "step": 23863
+    },
+    {
+      "epoch": 0.6554063332509407,
+      "grad_norm": 0.0,
+      "learning_rate": 5.607886903375168e-06,
+      "loss": 0.8027,
+      "step": 23864
+    },
+    {
+      "epoch": 0.6554337974787838,
+      "grad_norm": 0.0,
+      "learning_rate": 5.607087793989829e-06,
+      "loss": 0.8114,
+      "step": 23865
+    },
+    {
+      "epoch": 0.6554612617066271,
+      "grad_norm": 0.0,
+      "learning_rate": 5.606288719362908e-06,
+      "loss": 0.9886,
+      "step": 23866
+    },
+    {
+      "epoch": 0.6554887259344704,
+      "grad_norm": 0.0,
+      "learning_rate": 5.605489679500737e-06,
+      "loss": 0.8549,
+      "step": 23867
+    },
+    {
+      "epoch": 0.6555161901623136,
+      "grad_norm": 0.0,
+      "learning_rate": 5.60469067440963e-06,
+      "loss": 0.799,
+      "step": 23868
+    },
+    {
+      "epoch": 0.6555436543901568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6038917040959115e-06,
+      "loss": 0.9097,
+      "step": 23869
+    },
+    {
+      "epoch": 0.6555711186180001,
+      "grad_norm": 0.0,
+      "learning_rate": 5.603092768565908e-06,
+      "loss": 0.9285,
+      "step": 23870
+    },
+    {
+      "epoch": 0.6555985828458433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.602293867825934e-06,
+      "loss": 0.8384,
+      "step": 23871
+    },
+    {
+      "epoch": 0.6556260470736865,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6014950018823166e-06,
+      "loss": 0.8676,
+      "step": 23872
+    },
+    {
+      "epoch": 0.6556535113015297,
+      "grad_norm": 0.0,
+      "learning_rate": 5.600696170741377e-06,
+      "loss": 0.8595,
+      "step": 23873
+    },
+    {
+      "epoch": 0.655680975529373,
+      "grad_norm": 0.0,
+      "learning_rate": 5.599897374409436e-06,
+      "loss": 0.7751,
+      "step": 23874
+    },
+    {
+      "epoch": 0.6557084397572163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.599098612892806e-06,
+      "loss": 0.8287,
+      "step": 23875
+    },
+    {
+      "epoch": 0.6557359039850594,
+      "grad_norm": 0.0,
+      "learning_rate": 5.598299886197812e-06,
+      "loss": 0.9271,
+      "step": 23876
+    },
+    {
+      "epoch": 0.6557633682129027,
+      "grad_norm": 0.0,
+      "learning_rate": 5.597501194330774e-06,
+      "loss": 0.7932,
+      "step": 23877
+    },
+    {
+      "epoch": 0.6557908324407459,
+      "grad_norm": 0.0,
+      "learning_rate": 5.596702537298018e-06,
+      "loss": 0.8277,
+      "step": 23878
+    },
+    {
+      "epoch": 0.6558182966685892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.595903915105851e-06,
+      "loss": 0.8168,
+      "step": 23879
+    },
+    {
+      "epoch": 0.6558457608964324,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5951053277605995e-06,
+      "loss": 0.8986,
+      "step": 23880
+    },
+    {
+      "epoch": 0.6558732251242756,
+      "grad_norm": 0.0,
+      "learning_rate": 5.594306775268581e-06,
+      "loss": 0.8307,
+      "step": 23881
+    },
+    {
+      "epoch": 0.6559006893521189,
+      "grad_norm": 0.0,
+      "learning_rate": 5.593508257636118e-06,
+      "loss": 0.8245,
+      "step": 23882
+    },
+    {
+      "epoch": 0.6559281535799621,
+      "grad_norm": 0.0,
+      "learning_rate": 5.592709774869519e-06,
+      "loss": 0.8011,
+      "step": 23883
+    },
+    {
+      "epoch": 0.6559556178078053,
+      "grad_norm": 0.0,
+      "learning_rate": 5.591911326975112e-06,
+      "loss": 0.7488,
+      "step": 23884
+    },
+    {
+      "epoch": 0.6559830820356486,
+      "grad_norm": 0.0,
+      "learning_rate": 5.591112913959207e-06,
+      "loss": 0.9382,
+      "step": 23885
+    },
+    {
+      "epoch": 0.6560105462634918,
+      "grad_norm": 0.0,
+      "learning_rate": 5.590314535828125e-06,
+      "loss": 0.803,
+      "step": 23886
+    },
+    {
+      "epoch": 0.656038010491335,
+      "grad_norm": 0.0,
+      "learning_rate": 5.58951619258818e-06,
+      "loss": 0.8277,
+      "step": 23887
+    },
+    {
+      "epoch": 0.6560654747191783,
+      "grad_norm": 0.0,
+      "learning_rate": 5.588717884245693e-06,
+      "loss": 0.7912,
+      "step": 23888
+    },
+    {
+      "epoch": 0.6560929389470215,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5879196108069845e-06,
+      "loss": 0.9601,
+      "step": 23889
+    },
+    {
+      "epoch": 0.6561204031748648,
+      "grad_norm": 0.0,
+      "learning_rate": 5.587121372278359e-06,
+      "loss": 0.8145,
+      "step": 23890
+    },
+    {
+      "epoch": 0.6561478674027079,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5863231686661435e-06,
+      "loss": 0.8508,
+      "step": 23891
+    },
+    {
+      "epoch": 0.6561753316305512,
+      "grad_norm": 0.0,
+      "learning_rate": 5.585524999976643e-06,
+      "loss": 0.8371,
+      "step": 23892
+    },
+    {
+      "epoch": 0.6562027958583945,
+      "grad_norm": 0.0,
+      "learning_rate": 5.584726866216181e-06,
+      "loss": 0.829,
+      "step": 23893
+    },
+    {
+      "epoch": 0.6562302600862376,
+      "grad_norm": 0.0,
+      "learning_rate": 5.583928767391069e-06,
+      "loss": 0.8468,
+      "step": 23894
+    },
+    {
+      "epoch": 0.6562577243140809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.583130703507624e-06,
+      "loss": 0.8282,
+      "step": 23895
+    },
+    {
+      "epoch": 0.6562851885419242,
+      "grad_norm": 0.0,
+      "learning_rate": 5.582332674572163e-06,
+      "loss": 0.8373,
+      "step": 23896
+    },
+    {
+      "epoch": 0.6563126527697674,
+      "grad_norm": 0.0,
+      "learning_rate": 5.581534680590995e-06,
+      "loss": 0.8837,
+      "step": 23897
+    },
+    {
+      "epoch": 0.6563401169976106,
+      "grad_norm": 0.0,
+      "learning_rate": 5.580736721570436e-06,
+      "loss": 0.8629,
+      "step": 23898
+    },
+    {
+      "epoch": 0.6563675812254538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5799387975168e-06,
+      "loss": 0.8269,
+      "step": 23899
+    },
+    {
+      "epoch": 0.6563950454532971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.579140908436399e-06,
+      "loss": 0.8594,
+      "step": 23900
+    },
+    {
+      "epoch": 0.6564225096811404,
+      "grad_norm": 0.0,
+      "learning_rate": 5.578343054335548e-06,
+      "loss": 0.7822,
+      "step": 23901
+    },
+    {
+      "epoch": 0.6564499739089835,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5775452352205626e-06,
+      "loss": 0.8468,
+      "step": 23902
+    },
+    {
+      "epoch": 0.6564774381368268,
+      "grad_norm": 0.0,
+      "learning_rate": 5.576747451097748e-06,
+      "loss": 0.8578,
+      "step": 23903
+    },
+    {
+      "epoch": 0.65650490236467,
+      "grad_norm": 0.0,
+      "learning_rate": 5.575949701973422e-06,
+      "loss": 0.853,
+      "step": 23904
+    },
+    {
+      "epoch": 0.6565323665925132,
+      "grad_norm": 0.0,
+      "learning_rate": 5.575151987853897e-06,
+      "loss": 0.7583,
+      "step": 23905
+    },
+    {
+      "epoch": 0.6565598308203565,
+      "grad_norm": 0.0,
+      "learning_rate": 5.574354308745486e-06,
+      "loss": 0.8764,
+      "step": 23906
+    },
+    {
+      "epoch": 0.6565872950481997,
+      "grad_norm": 0.0,
+      "learning_rate": 5.573556664654494e-06,
+      "loss": 0.7635,
+      "step": 23907
+    },
+    {
+      "epoch": 0.656614759276043,
+      "grad_norm": 0.0,
+      "learning_rate": 5.57275905558724e-06,
+      "loss": 0.882,
+      "step": 23908
+    },
+    {
+      "epoch": 0.6566422235038862,
+      "grad_norm": 0.0,
+      "learning_rate": 5.571961481550028e-06,
+      "loss": 0.7984,
+      "step": 23909
+    },
+    {
+      "epoch": 0.6566696877317294,
+      "grad_norm": 0.0,
+      "learning_rate": 5.571163942549171e-06,
+      "loss": 0.8166,
+      "step": 23910
+    },
+    {
+      "epoch": 0.6566971519595727,
+      "grad_norm": 0.0,
+      "learning_rate": 5.570366438590981e-06,
+      "loss": 0.8912,
+      "step": 23911
+    },
+    {
+      "epoch": 0.6567246161874158,
+      "grad_norm": 0.0,
+      "learning_rate": 5.569568969681769e-06,
+      "loss": 0.8521,
+      "step": 23912
+    },
+    {
+      "epoch": 0.6567520804152591,
+      "grad_norm": 0.0,
+      "learning_rate": 5.568771535827846e-06,
+      "loss": 0.8252,
+      "step": 23913
+    },
+    {
+      "epoch": 0.6567795446431024,
+      "grad_norm": 0.0,
+      "learning_rate": 5.567974137035516e-06,
+      "loss": 0.8765,
+      "step": 23914
+    },
+    {
+      "epoch": 0.6568070088709456,
+      "grad_norm": 0.0,
+      "learning_rate": 5.567176773311093e-06,
+      "loss": 0.888,
+      "step": 23915
+    },
+    {
+      "epoch": 0.6568344730987888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.566379444660883e-06,
+      "loss": 0.7876,
+      "step": 23916
+    },
+    {
+      "epoch": 0.6568619373266321,
+      "grad_norm": 0.0,
+      "learning_rate": 5.565582151091196e-06,
+      "loss": 0.8302,
+      "step": 23917
+    },
+    {
+      "epoch": 0.6568894015544753,
+      "grad_norm": 0.0,
+      "learning_rate": 5.56478489260834e-06,
+      "loss": 0.9038,
+      "step": 23918
+    },
+    {
+      "epoch": 0.6569168657823186,
+      "grad_norm": 0.0,
+      "learning_rate": 5.563987669218625e-06,
+      "loss": 0.9002,
+      "step": 23919
+    },
+    {
+      "epoch": 0.6569443300101617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.563190480928361e-06,
+      "loss": 0.8207,
+      "step": 23920
+    },
+    {
+      "epoch": 0.656971794238005,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5623933277438494e-06,
+      "loss": 0.7536,
+      "step": 23921
+    },
+    {
+      "epoch": 0.6569992584658483,
+      "grad_norm": 0.0,
+      "learning_rate": 5.561596209671403e-06,
+      "loss": 0.8625,
+      "step": 23922
+    },
+    {
+      "epoch": 0.6570267226936914,
+      "grad_norm": 0.0,
+      "learning_rate": 5.560799126717325e-06,
+      "loss": 0.7966,
+      "step": 23923
+    },
+    {
+      "epoch": 0.6570541869215347,
+      "grad_norm": 0.0,
+      "learning_rate": 5.560002078887922e-06,
+      "loss": 0.6946,
+      "step": 23924
+    },
+    {
+      "epoch": 0.6570816511493779,
+      "grad_norm": 0.0,
+      "learning_rate": 5.559205066189504e-06,
+      "loss": 0.8506,
+      "step": 23925
+    },
+    {
+      "epoch": 0.6571091153772212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.558408088628378e-06,
+      "loss": 0.8834,
+      "step": 23926
+    },
+    {
+      "epoch": 0.6571365796050644,
+      "grad_norm": 0.0,
+      "learning_rate": 5.557611146210844e-06,
+      "loss": 0.9362,
+      "step": 23927
+    },
+    {
+      "epoch": 0.6571640438329076,
+      "grad_norm": 0.0,
+      "learning_rate": 5.556814238943212e-06,
+      "loss": 0.8177,
+      "step": 23928
+    },
+    {
+      "epoch": 0.6571915080607509,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5560173668317876e-06,
+      "loss": 0.8328,
+      "step": 23929
+    },
+    {
+      "epoch": 0.6572189722885942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5552205298828786e-06,
+      "loss": 0.754,
+      "step": 23930
+    },
+    {
+      "epoch": 0.6572464365164373,
+      "grad_norm": 0.0,
+      "learning_rate": 5.554423728102783e-06,
+      "loss": 0.8867,
+      "step": 23931
+    },
+    {
+      "epoch": 0.6572739007442806,
+      "grad_norm": 0.0,
+      "learning_rate": 5.553626961497811e-06,
+      "loss": 0.8972,
+      "step": 23932
+    },
+    {
+      "epoch": 0.6573013649721238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.552830230074262e-06,
+      "loss": 0.8693,
+      "step": 23933
+    },
+    {
+      "epoch": 0.657328829199967,
+      "grad_norm": 0.0,
+      "learning_rate": 5.552033533838443e-06,
+      "loss": 0.9183,
+      "step": 23934
+    },
+    {
+      "epoch": 0.6573562934278103,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5512368727966584e-06,
+      "loss": 0.8118,
+      "step": 23935
+    },
+    {
+      "epoch": 0.6573837576556535,
+      "grad_norm": 0.0,
+      "learning_rate": 5.55044024695521e-06,
+      "loss": 0.8404,
+      "step": 23936
+    },
+    {
+      "epoch": 0.6574112218834968,
+      "grad_norm": 0.0,
+      "learning_rate": 5.549643656320406e-06,
+      "loss": 0.8761,
+      "step": 23937
+    },
+    {
+      "epoch": 0.6574386861113399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.548847100898542e-06,
+      "loss": 0.9109,
+      "step": 23938
+    },
+    {
+      "epoch": 0.6574661503391832,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5480505806959275e-06,
+      "loss": 0.8081,
+      "step": 23939
+    },
+    {
+      "epoch": 0.6574936145670265,
+      "grad_norm": 0.0,
+      "learning_rate": 5.547254095718857e-06,
+      "loss": 0.9033,
+      "step": 23940
+    },
+    {
+      "epoch": 0.6575210787948697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.546457645973638e-06,
+      "loss": 0.8547,
+      "step": 23941
+    },
+    {
+      "epoch": 0.6575485430227129,
+      "grad_norm": 0.0,
+      "learning_rate": 5.54566123146657e-06,
+      "loss": 0.8788,
+      "step": 23942
+    },
+    {
+      "epoch": 0.6575760072505562,
+      "grad_norm": 0.0,
+      "learning_rate": 5.54486485220396e-06,
+      "loss": 0.9037,
+      "step": 23943
+    },
+    {
+      "epoch": 0.6576034714783994,
+      "grad_norm": 0.0,
+      "learning_rate": 5.544068508192101e-06,
+      "loss": 0.8857,
+      "step": 23944
+    },
+    {
+      "epoch": 0.6576309357062426,
+      "grad_norm": 0.0,
+      "learning_rate": 5.543272199437298e-06,
+      "loss": 0.8846,
+      "step": 23945
+    },
+    {
+      "epoch": 0.6576583999340858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.542475925945856e-06,
+      "loss": 0.7797,
+      "step": 23946
+    },
+    {
+      "epoch": 0.6576858641619291,
+      "grad_norm": 0.0,
+      "learning_rate": 5.541679687724068e-06,
+      "loss": 0.8174,
+      "step": 23947
+    },
+    {
+      "epoch": 0.6577133283897724,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5408834847782365e-06,
+      "loss": 0.8433,
+      "step": 23948
+    },
+    {
+      "epoch": 0.6577407926176155,
+      "grad_norm": 0.0,
+      "learning_rate": 5.540087317114666e-06,
+      "loss": 0.8677,
+      "step": 23949
+    },
+    {
+      "epoch": 0.6577682568454588,
+      "grad_norm": 0.0,
+      "learning_rate": 5.539291184739648e-06,
+      "loss": 0.8216,
+      "step": 23950
+    },
+    {
+      "epoch": 0.657795721073302,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5384950876594855e-06,
+      "loss": 0.8573,
+      "step": 23951
+    },
+    {
+      "epoch": 0.6578231853011453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.537699025880479e-06,
+      "loss": 0.8647,
+      "step": 23952
+    },
+    {
+      "epoch": 0.6578506495289885,
+      "grad_norm": 0.0,
+      "learning_rate": 5.536902999408927e-06,
+      "loss": 0.8897,
+      "step": 23953
+    },
+    {
+      "epoch": 0.6578781137568317,
+      "grad_norm": 0.0,
+      "learning_rate": 5.53610700825113e-06,
+      "loss": 0.8779,
+      "step": 23954
+    },
+    {
+      "epoch": 0.657905577984675,
+      "grad_norm": 0.0,
+      "learning_rate": 5.535311052413386e-06,
+      "loss": 0.9011,
+      "step": 23955
+    },
+    {
+      "epoch": 0.6579330422125182,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5345151319019826e-06,
+      "loss": 0.8813,
+      "step": 23956
+    },
+    {
+      "epoch": 0.6579605064403614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.533719246723228e-06,
+      "loss": 0.8171,
+      "step": 23957
+    },
+    {
+      "epoch": 0.6579879706682047,
+      "grad_norm": 0.0,
+      "learning_rate": 5.532923396883416e-06,
+      "loss": 0.8798,
+      "step": 23958
+    },
+    {
+      "epoch": 0.6580154348960479,
+      "grad_norm": 0.0,
+      "learning_rate": 5.532127582388844e-06,
+      "loss": 0.8383,
+      "step": 23959
+    },
+    {
+      "epoch": 0.6580428991238911,
+      "grad_norm": 0.0,
+      "learning_rate": 5.531331803245811e-06,
+      "loss": 0.6954,
+      "step": 23960
+    },
+    {
+      "epoch": 0.6580703633517344,
+      "grad_norm": 0.0,
+      "learning_rate": 5.530536059460614e-06,
+      "loss": 0.8715,
+      "step": 23961
+    },
+    {
+      "epoch": 0.6580978275795776,
+      "grad_norm": 0.0,
+      "learning_rate": 5.529740351039543e-06,
+      "loss": 0.8701,
+      "step": 23962
+    },
+    {
+      "epoch": 0.6581252918074209,
+      "grad_norm": 0.0,
+      "learning_rate": 5.528944677988903e-06,
+      "loss": 0.8503,
+      "step": 23963
+    },
+    {
+      "epoch": 0.658152756035264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.528149040314981e-06,
+      "loss": 0.8551,
+      "step": 23964
+    },
+    {
+      "epoch": 0.6581802202631073,
+      "grad_norm": 0.0,
+      "learning_rate": 5.527353438024075e-06,
+      "loss": 0.8022,
+      "step": 23965
+    },
+    {
+      "epoch": 0.6582076844909506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.52655787112248e-06,
+      "loss": 0.7447,
+      "step": 23966
+    },
+    {
+      "epoch": 0.6582351487187937,
+      "grad_norm": 0.0,
+      "learning_rate": 5.525762339616498e-06,
+      "loss": 0.8418,
+      "step": 23967
+    },
+    {
+      "epoch": 0.658262612946637,
+      "grad_norm": 0.0,
+      "learning_rate": 5.524966843512412e-06,
+      "loss": 0.7912,
+      "step": 23968
+    },
+    {
+      "epoch": 0.6582900771744803,
+      "grad_norm": 0.0,
+      "learning_rate": 5.524171382816523e-06,
+      "loss": 0.8282,
+      "step": 23969
+    },
+    {
+      "epoch": 0.6583175414023235,
+      "grad_norm": 0.0,
+      "learning_rate": 5.523375957535128e-06,
+      "loss": 0.8424,
+      "step": 23970
+    },
+    {
+      "epoch": 0.6583450056301667,
+      "grad_norm": 0.0,
+      "learning_rate": 5.522580567674512e-06,
+      "loss": 0.9325,
+      "step": 23971
+    },
+    {
+      "epoch": 0.6583724698580099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.521785213240972e-06,
+      "loss": 0.8266,
+      "step": 23972
+    },
+    {
+      "epoch": 0.6583999340858532,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5209898942408065e-06,
+      "loss": 0.8232,
+      "step": 23973
+    },
+    {
+      "epoch": 0.6584273983136965,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5201946106803e-06,
+      "loss": 0.7935,
+      "step": 23974
+    },
+    {
+      "epoch": 0.6584548625415396,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5193993625657495e-06,
+      "loss": 0.8965,
+      "step": 23975
+    },
+    {
+      "epoch": 0.6584823267693829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.518604149903447e-06,
+      "loss": 0.8689,
+      "step": 23976
+    },
+    {
+      "epoch": 0.6585097909972261,
+      "grad_norm": 0.0,
+      "learning_rate": 5.517808972699684e-06,
+      "loss": 0.9109,
+      "step": 23977
+    },
+    {
+      "epoch": 0.6585372552250693,
+      "grad_norm": 0.0,
+      "learning_rate": 5.517013830960757e-06,
+      "loss": 0.764,
+      "step": 23978
+    },
+    {
+      "epoch": 0.6585647194529126,
+      "grad_norm": 0.0,
+      "learning_rate": 5.516218724692952e-06,
+      "loss": 0.8965,
+      "step": 23979
+    },
+    {
+      "epoch": 0.6585921836807558,
+      "grad_norm": 0.0,
+      "learning_rate": 5.515423653902558e-06,
+      "loss": 0.8156,
+      "step": 23980
+    },
+    {
+      "epoch": 0.6586196479085991,
+      "grad_norm": 0.0,
+      "learning_rate": 5.51462861859587e-06,
+      "loss": 0.9124,
+      "step": 23981
+    },
+    {
+      "epoch": 0.6586471121364423,
+      "grad_norm": 0.0,
+      "learning_rate": 5.513833618779177e-06,
+      "loss": 0.8767,
+      "step": 23982
+    },
+    {
+      "epoch": 0.6586745763642855,
+      "grad_norm": 0.0,
+      "learning_rate": 5.513038654458771e-06,
+      "loss": 0.8597,
+      "step": 23983
+    },
+    {
+      "epoch": 0.6587020405921288,
+      "grad_norm": 0.0,
+      "learning_rate": 5.51224372564094e-06,
+      "loss": 0.7831,
+      "step": 23984
+    },
+    {
+      "epoch": 0.658729504819972,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5114488323319806e-06,
+      "loss": 0.8071,
+      "step": 23985
+    },
+    {
+      "epoch": 0.6587569690478152,
+      "grad_norm": 0.0,
+      "learning_rate": 5.510653974538176e-06,
+      "loss": 0.8288,
+      "step": 23986
+    },
+    {
+      "epoch": 0.6587844332756585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.509859152265812e-06,
+      "loss": 0.8928,
+      "step": 23987
+    },
+    {
+      "epoch": 0.6588118975035017,
+      "grad_norm": 0.0,
+      "learning_rate": 5.509064365521182e-06,
+      "loss": 0.821,
+      "step": 23988
+    },
+    {
+      "epoch": 0.6588393617313449,
+      "grad_norm": 0.0,
+      "learning_rate": 5.508269614310573e-06,
+      "loss": 0.837,
+      "step": 23989
+    },
+    {
+      "epoch": 0.6588668259591881,
+      "grad_norm": 0.0,
+      "learning_rate": 5.507474898640276e-06,
+      "loss": 0.8557,
+      "step": 23990
+    },
+    {
+      "epoch": 0.6588942901870314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.506680218516583e-06,
+      "loss": 0.7943,
+      "step": 23991
+    },
+    {
+      "epoch": 0.6589217544148747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.505885573945771e-06,
+      "loss": 0.8434,
+      "step": 23992
+    },
+    {
+      "epoch": 0.6589492186427178,
+      "grad_norm": 0.0,
+      "learning_rate": 5.505090964934132e-06,
+      "loss": 0.8698,
+      "step": 23993
+    },
+    {
+      "epoch": 0.6589766828705611,
+      "grad_norm": 0.0,
+      "learning_rate": 5.50429639148796e-06,
+      "loss": 0.8399,
+      "step": 23994
+    },
+    {
+      "epoch": 0.6590041470984044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.503501853613533e-06,
+      "loss": 0.8307,
+      "step": 23995
+    },
+    {
+      "epoch": 0.6590316113262475,
+      "grad_norm": 0.0,
+      "learning_rate": 5.502707351317144e-06,
+      "loss": 0.8162,
+      "step": 23996
+    },
+    {
+      "epoch": 0.6590590755540908,
+      "grad_norm": 0.0,
+      "learning_rate": 5.501912884605073e-06,
+      "loss": 0.8799,
+      "step": 23997
+    },
+    {
+      "epoch": 0.659086539781934,
+      "grad_norm": 0.0,
+      "learning_rate": 5.501118453483609e-06,
+      "loss": 0.8646,
+      "step": 23998
+    },
+    {
+      "epoch": 0.6591140040097773,
+      "grad_norm": 0.0,
+      "learning_rate": 5.50032405795904e-06,
+      "loss": 0.7831,
+      "step": 23999
+    },
+    {
+      "epoch": 0.6591414682376205,
+      "grad_norm": 0.0,
+      "learning_rate": 5.499529698037649e-06,
+      "loss": 0.904,
+      "step": 24000
+    },
+    {
+      "epoch": 0.6591689324654637,
+      "grad_norm": 0.0,
+      "learning_rate": 5.498735373725723e-06,
+      "loss": 0.8353,
+      "step": 24001
+    },
+    {
+      "epoch": 0.659196396693307,
+      "grad_norm": 0.0,
+      "learning_rate": 5.49794108502955e-06,
+      "loss": 0.8555,
+      "step": 24002
+    },
+    {
+      "epoch": 0.6592238609211502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.497146831955411e-06,
+      "loss": 0.8193,
+      "step": 24003
+    },
+    {
+      "epoch": 0.6592513251489934,
+      "grad_norm": 0.0,
+      "learning_rate": 5.496352614509586e-06,
+      "loss": 0.9129,
+      "step": 24004
+    },
+    {
+      "epoch": 0.6592787893768367,
+      "grad_norm": 0.0,
+      "learning_rate": 5.495558432698365e-06,
+      "loss": 0.8724,
+      "step": 24005
+    },
+    {
+      "epoch": 0.6593062536046799,
+      "grad_norm": 0.0,
+      "learning_rate": 5.494764286528029e-06,
+      "loss": 0.8296,
+      "step": 24006
+    },
+    {
+      "epoch": 0.6593337178325231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.493970176004863e-06,
+      "loss": 0.859,
+      "step": 24007
+    },
+    {
+      "epoch": 0.6593611820603664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.493176101135156e-06,
+      "loss": 0.8493,
+      "step": 24008
+    },
+    {
+      "epoch": 0.6593886462882096,
+      "grad_norm": 0.0,
+      "learning_rate": 5.49238206192518e-06,
+      "loss": 0.8134,
+      "step": 24009
+    },
+    {
+      "epoch": 0.6594161105160529,
+      "grad_norm": 0.0,
+      "learning_rate": 5.491588058381227e-06,
+      "loss": 0.815,
+      "step": 24010
+    },
+    {
+      "epoch": 0.659443574743896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.490794090509574e-06,
+      "loss": 0.8743,
+      "step": 24011
+    },
+    {
+      "epoch": 0.6594710389717393,
+      "grad_norm": 0.0,
+      "learning_rate": 5.490000158316503e-06,
+      "loss": 0.9306,
+      "step": 24012
+    },
+    {
+      "epoch": 0.6594985031995826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.489206261808298e-06,
+      "loss": 0.8778,
+      "step": 24013
+    },
+    {
+      "epoch": 0.6595259674274258,
+      "grad_norm": 0.0,
+      "learning_rate": 5.488412400991245e-06,
+      "loss": 0.7282,
+      "step": 24014
+    },
+    {
+      "epoch": 0.659553431655269,
+      "grad_norm": 0.0,
+      "learning_rate": 5.487618575871616e-06,
+      "loss": 0.8269,
+      "step": 24015
+    },
+    {
+      "epoch": 0.6595808958831122,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4868247864556966e-06,
+      "loss": 0.9022,
+      "step": 24016
+    },
+    {
+      "epoch": 0.6596083601109555,
+      "grad_norm": 0.0,
+      "learning_rate": 5.486031032749768e-06,
+      "loss": 0.6921,
+      "step": 24017
+    },
+    {
+      "epoch": 0.6596358243387987,
+      "grad_norm": 0.0,
+      "learning_rate": 5.485237314760115e-06,
+      "loss": 0.9032,
+      "step": 24018
+    },
+    {
+      "epoch": 0.6596632885666419,
+      "grad_norm": 0.0,
+      "learning_rate": 5.484443632493008e-06,
+      "loss": 0.8454,
+      "step": 24019
+    },
+    {
+      "epoch": 0.6596907527944852,
+      "grad_norm": 0.0,
+      "learning_rate": 5.483649985954737e-06,
+      "loss": 0.8456,
+      "step": 24020
+    },
+    {
+      "epoch": 0.6597182170223285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.482856375151573e-06,
+      "loss": 0.809,
+      "step": 24021
+    },
+    {
+      "epoch": 0.6597456812501716,
+      "grad_norm": 0.0,
+      "learning_rate": 5.482062800089799e-06,
+      "loss": 0.8378,
+      "step": 24022
+    },
+    {
+      "epoch": 0.6597731454780149,
+      "grad_norm": 0.0,
+      "learning_rate": 5.481269260775694e-06,
+      "loss": 0.8185,
+      "step": 24023
+    },
+    {
+      "epoch": 0.6598006097058581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.480475757215537e-06,
+      "loss": 0.9426,
+      "step": 24024
+    },
+    {
+      "epoch": 0.6598280739337014,
+      "grad_norm": 0.0,
+      "learning_rate": 5.479682289415611e-06,
+      "loss": 0.8549,
+      "step": 24025
+    },
+    {
+      "epoch": 0.6598555381615446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.478888857382185e-06,
+      "loss": 0.6932,
+      "step": 24026
+    },
+    {
+      "epoch": 0.6598830023893878,
+      "grad_norm": 0.0,
+      "learning_rate": 5.478095461121546e-06,
+      "loss": 0.9641,
+      "step": 24027
+    },
+    {
+      "epoch": 0.6599104666172311,
+      "grad_norm": 0.0,
+      "learning_rate": 5.477302100639963e-06,
+      "loss": 0.7991,
+      "step": 24028
+    },
+    {
+      "epoch": 0.6599379308450742,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4765087759437185e-06,
+      "loss": 0.8011,
+      "step": 24029
+    },
+    {
+      "epoch": 0.6599653950729175,
+      "grad_norm": 0.0,
+      "learning_rate": 5.475715487039088e-06,
+      "loss": 0.965,
+      "step": 24030
+    },
+    {
+      "epoch": 0.6599928593007608,
+      "grad_norm": 0.0,
+      "learning_rate": 5.474922233932349e-06,
+      "loss": 0.8315,
+      "step": 24031
+    },
+    {
+      "epoch": 0.660020323528604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.474129016629783e-06,
+      "loss": 0.7749,
+      "step": 24032
+    },
+    {
+      "epoch": 0.6600477877564472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.473335835137657e-06,
+      "loss": 0.8736,
+      "step": 24033
+    },
+    {
+      "epoch": 0.6600752519842905,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4725426894622545e-06,
+      "loss": 0.8443,
+      "step": 24034
+    },
+    {
+      "epoch": 0.6601027162121337,
+      "grad_norm": 0.0,
+      "learning_rate": 5.471749579609844e-06,
+      "loss": 0.9482,
+      "step": 24035
+    },
+    {
+      "epoch": 0.660130180439977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4709565055867066e-06,
+      "loss": 0.8759,
+      "step": 24036
+    },
+    {
+      "epoch": 0.6601576446678201,
+      "grad_norm": 0.0,
+      "learning_rate": 5.470163467399115e-06,
+      "loss": 0.7769,
+      "step": 24037
+    },
+    {
+      "epoch": 0.6601851088956634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.469370465053347e-06,
+      "loss": 0.7259,
+      "step": 24038
+    },
+    {
+      "epoch": 0.6602125731235067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.468577498555674e-06,
+      "loss": 0.8062,
+      "step": 24039
+    },
+    {
+      "epoch": 0.6602400373513498,
+      "grad_norm": 0.0,
+      "learning_rate": 5.467784567912369e-06,
+      "loss": 0.8959,
+      "step": 24040
+    },
+    {
+      "epoch": 0.6602675015791931,
+      "grad_norm": 0.0,
+      "learning_rate": 5.466991673129709e-06,
+      "loss": 0.7817,
+      "step": 24041
+    },
+    {
+      "epoch": 0.6602949658070363,
+      "grad_norm": 0.0,
+      "learning_rate": 5.46619881421397e-06,
+      "loss": 0.86,
+      "step": 24042
+    },
+    {
+      "epoch": 0.6603224300348796,
+      "grad_norm": 0.0,
+      "learning_rate": 5.465405991171417e-06,
+      "loss": 0.8835,
+      "step": 24043
+    },
+    {
+      "epoch": 0.6603498942627228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.464613204008336e-06,
+      "loss": 0.8082,
+      "step": 24044
+    },
+    {
+      "epoch": 0.660377358490566,
+      "grad_norm": 0.0,
+      "learning_rate": 5.463820452730985e-06,
+      "loss": 0.8366,
+      "step": 24045
+    },
+    {
+      "epoch": 0.6604048227184093,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4630277373456455e-06,
+      "loss": 0.784,
+      "step": 24046
+    },
+    {
+      "epoch": 0.6604322869462526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.462235057858587e-06,
+      "loss": 0.8535,
+      "step": 24047
+    },
+    {
+      "epoch": 0.6604597511740957,
+      "grad_norm": 0.0,
+      "learning_rate": 5.461442414276083e-06,
+      "loss": 0.8411,
+      "step": 24048
+    },
+    {
+      "epoch": 0.660487215401939,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4606498066044075e-06,
+      "loss": 0.8158,
+      "step": 24049
+    },
+    {
+      "epoch": 0.6605146796297822,
+      "grad_norm": 0.0,
+      "learning_rate": 5.459857234849827e-06,
+      "loss": 0.977,
+      "step": 24050
+    },
+    {
+      "epoch": 0.6605421438576254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.459064699018617e-06,
+      "loss": 0.8396,
+      "step": 24051
+    },
+    {
+      "epoch": 0.6605696080854687,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4582721991170425e-06,
+      "loss": 0.8905,
+      "step": 24052
+    },
+    {
+      "epoch": 0.6605970723133119,
+      "grad_norm": 0.0,
+      "learning_rate": 5.457479735151379e-06,
+      "loss": 0.9068,
+      "step": 24053
+    },
+    {
+      "epoch": 0.6606245365411552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.456687307127895e-06,
+      "loss": 0.8642,
+      "step": 24054
+    },
+    {
+      "epoch": 0.6606520007689983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.455894915052859e-06,
+      "loss": 0.8856,
+      "step": 24055
+    },
+    {
+      "epoch": 0.6606794649968416,
+      "grad_norm": 0.0,
+      "learning_rate": 5.455102558932549e-06,
+      "loss": 0.8315,
+      "step": 24056
+    },
+    {
+      "epoch": 0.6607069292246849,
+      "grad_norm": 0.0,
+      "learning_rate": 5.454310238773224e-06,
+      "loss": 0.8541,
+      "step": 24057
+    },
+    {
+      "epoch": 0.660734393452528,
+      "grad_norm": 0.0,
+      "learning_rate": 5.45351795458116e-06,
+      "loss": 0.8905,
+      "step": 24058
+    },
+    {
+      "epoch": 0.6607618576803713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.452725706362619e-06,
+      "loss": 0.825,
+      "step": 24059
+    },
+    {
+      "epoch": 0.6607893219082146,
+      "grad_norm": 0.0,
+      "learning_rate": 5.451933494123874e-06,
+      "loss": 0.7829,
+      "step": 24060
+    },
+    {
+      "epoch": 0.6608167861360578,
+      "grad_norm": 0.0,
+      "learning_rate": 5.451141317871198e-06,
+      "loss": 0.8125,
+      "step": 24061
+    },
+    {
+      "epoch": 0.660844250363901,
+      "grad_norm": 0.0,
+      "learning_rate": 5.450349177610849e-06,
+      "loss": 0.8303,
+      "step": 24062
+    },
+    {
+      "epoch": 0.6608717145917442,
+      "grad_norm": 0.0,
+      "learning_rate": 5.449557073349099e-06,
+      "loss": 0.8721,
+      "step": 24063
+    },
+    {
+      "epoch": 0.6608991788195875,
+      "grad_norm": 0.0,
+      "learning_rate": 5.448765005092217e-06,
+      "loss": 0.8962,
+      "step": 24064
+    },
+    {
+      "epoch": 0.6609266430474308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.447972972846469e-06,
+      "loss": 0.8708,
+      "step": 24065
+    },
+    {
+      "epoch": 0.6609541072752739,
+      "grad_norm": 0.0,
+      "learning_rate": 5.447180976618127e-06,
+      "loss": 0.899,
+      "step": 24066
+    },
+    {
+      "epoch": 0.6609815715031172,
+      "grad_norm": 0.0,
+      "learning_rate": 5.446389016413451e-06,
+      "loss": 0.8349,
+      "step": 24067
+    },
+    {
+      "epoch": 0.6610090357309604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.445597092238706e-06,
+      "loss": 0.8063,
+      "step": 24068
+    },
+    {
+      "epoch": 0.6610364999588036,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4448052041001606e-06,
+      "loss": 0.8604,
+      "step": 24069
+    },
+    {
+      "epoch": 0.6610639641866469,
+      "grad_norm": 0.0,
+      "learning_rate": 5.444013352004082e-06,
+      "loss": 0.942,
+      "step": 24070
+    },
+    {
+      "epoch": 0.6610914284144901,
+      "grad_norm": 0.0,
+      "learning_rate": 5.443221535956733e-06,
+      "loss": 0.7635,
+      "step": 24071
+    },
+    {
+      "epoch": 0.6611188926423334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.442429755964382e-06,
+      "loss": 0.8744,
+      "step": 24072
+    },
+    {
+      "epoch": 0.6611463568701766,
+      "grad_norm": 0.0,
+      "learning_rate": 5.441638012033294e-06,
+      "loss": 0.7388,
+      "step": 24073
+    },
+    {
+      "epoch": 0.6611738210980198,
+      "grad_norm": 0.0,
+      "learning_rate": 5.440846304169731e-06,
+      "loss": 0.8248,
+      "step": 24074
+    },
+    {
+      "epoch": 0.6612012853258631,
+      "grad_norm": 0.0,
+      "learning_rate": 5.44005463237996e-06,
+      "loss": 0.7283,
+      "step": 24075
+    },
+    {
+      "epoch": 0.6612287495537063,
+      "grad_norm": 0.0,
+      "learning_rate": 5.439262996670239e-06,
+      "loss": 0.8506,
+      "step": 24076
+    },
+    {
+      "epoch": 0.6612562137815495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.438471397046836e-06,
+      "loss": 0.8301,
+      "step": 24077
+    },
+    {
+      "epoch": 0.6612836780093928,
+      "grad_norm": 0.0,
+      "learning_rate": 5.437679833516014e-06,
+      "loss": 0.9251,
+      "step": 24078
+    },
+    {
+      "epoch": 0.661311142237236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.436888306084042e-06,
+      "loss": 0.874,
+      "step": 24079
+    },
+    {
+      "epoch": 0.6613386064650792,
+      "grad_norm": 0.0,
+      "learning_rate": 5.436096814757172e-06,
+      "loss": 0.9265,
+      "step": 24080
+    },
+    {
+      "epoch": 0.6613660706929224,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4353053595416716e-06,
+      "loss": 0.8628,
+      "step": 24081
+    },
+    {
+      "epoch": 0.6613935349207657,
+      "grad_norm": 0.0,
+      "learning_rate": 5.434513940443808e-06,
+      "loss": 0.8252,
+      "step": 24082
+    },
+    {
+      "epoch": 0.661420999148609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4337225574698345e-06,
+      "loss": 0.7381,
+      "step": 24083
+    },
+    {
+      "epoch": 0.6614484633764521,
+      "grad_norm": 0.0,
+      "learning_rate": 5.432931210626018e-06,
+      "loss": 0.9107,
+      "step": 24084
+    },
+    {
+      "epoch": 0.6614759276042954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.43213989991862e-06,
+      "loss": 0.9054,
+      "step": 24085
+    },
+    {
+      "epoch": 0.6615033918321387,
+      "grad_norm": 0.0,
+      "learning_rate": 5.431348625353899e-06,
+      "loss": 0.8065,
+      "step": 24086
+    },
+    {
+      "epoch": 0.6615308560599819,
+      "grad_norm": 0.0,
+      "learning_rate": 5.430557386938118e-06,
+      "loss": 0.8549,
+      "step": 24087
+    },
+    {
+      "epoch": 0.6615583202878251,
+      "grad_norm": 0.0,
+      "learning_rate": 5.429766184677534e-06,
+      "loss": 0.7699,
+      "step": 24088
+    },
+    {
+      "epoch": 0.6615857845156683,
+      "grad_norm": 0.0,
+      "learning_rate": 5.428975018578413e-06,
+      "loss": 0.8571,
+      "step": 24089
+    },
+    {
+      "epoch": 0.6616132487435116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.428183888647015e-06,
+      "loss": 0.8201,
+      "step": 24090
+    },
+    {
+      "epoch": 0.6616407129713548,
+      "grad_norm": 0.0,
+      "learning_rate": 5.427392794889598e-06,
+      "loss": 0.9091,
+      "step": 24091
+    },
+    {
+      "epoch": 0.661668177199198,
+      "grad_norm": 0.0,
+      "learning_rate": 5.426601737312416e-06,
+      "loss": 0.9091,
+      "step": 24092
+    },
+    {
+      "epoch": 0.6616956414270413,
+      "grad_norm": 0.0,
+      "learning_rate": 5.425810715921731e-06,
+      "loss": 0.8153,
+      "step": 24093
+    },
+    {
+      "epoch": 0.6617231056548845,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4250197307238065e-06,
+      "loss": 0.8169,
+      "step": 24094
+    },
+    {
+      "epoch": 0.6617505698827277,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4242287817248965e-06,
+      "loss": 0.7502,
+      "step": 24095
+    },
+    {
+      "epoch": 0.661778034110571,
+      "grad_norm": 0.0,
+      "learning_rate": 5.42343786893126e-06,
+      "loss": 0.8921,
+      "step": 24096
+    },
+    {
+      "epoch": 0.6618054983384142,
+      "grad_norm": 0.0,
+      "learning_rate": 5.422646992349161e-06,
+      "loss": 0.797,
+      "step": 24097
+    },
+    {
+      "epoch": 0.6618329625662575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.421856151984848e-06,
+      "loss": 0.8172,
+      "step": 24098
+    },
+    {
+      "epoch": 0.6618604267941007,
+      "grad_norm": 0.0,
+      "learning_rate": 5.421065347844585e-06,
+      "loss": 0.824,
+      "step": 24099
+    },
+    {
+      "epoch": 0.6618878910219439,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4202745799346244e-06,
+      "loss": 0.8736,
+      "step": 24100
+    },
+    {
+      "epoch": 0.6619153552497872,
+      "grad_norm": 0.0,
+      "learning_rate": 5.419483848261225e-06,
+      "loss": 0.7935,
+      "step": 24101
+    },
+    {
+      "epoch": 0.6619428194776303,
+      "grad_norm": 0.0,
+      "learning_rate": 5.418693152830644e-06,
+      "loss": 1.0245,
+      "step": 24102
+    },
+    {
+      "epoch": 0.6619702837054736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4179024936491405e-06,
+      "loss": 0.7933,
+      "step": 24103
+    },
+    {
+      "epoch": 0.6619977479333169,
+      "grad_norm": 0.0,
+      "learning_rate": 5.417111870722964e-06,
+      "loss": 0.879,
+      "step": 24104
+    },
+    {
+      "epoch": 0.6620252121611601,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416321284058375e-06,
+      "loss": 0.7874,
+      "step": 24105
+    },
+    {
+      "epoch": 0.6620526763890033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.415530733661629e-06,
+      "loss": 0.8355,
+      "step": 24106
+    },
+    {
+      "epoch": 0.6620801406168465,
+      "grad_norm": 0.0,
+      "learning_rate": 5.414740219538977e-06,
+      "loss": 0.7888,
+      "step": 24107
+    },
+    {
+      "epoch": 0.6621076048446898,
+      "grad_norm": 0.0,
+      "learning_rate": 5.413949741696676e-06,
+      "loss": 0.9071,
+      "step": 24108
+    },
+    {
+      "epoch": 0.6621350690725331,
+      "grad_norm": 0.0,
+      "learning_rate": 5.413159300140986e-06,
+      "loss": 0.8206,
+      "step": 24109
+    },
+    {
+      "epoch": 0.6621625333003762,
+      "grad_norm": 0.0,
+      "learning_rate": 5.412368894878151e-06,
+      "loss": 0.853,
+      "step": 24110
+    },
+    {
+      "epoch": 0.6621899975282195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.411578525914431e-06,
+      "loss": 0.9093,
+      "step": 24111
+    },
+    {
+      "epoch": 0.6622174617560628,
+      "grad_norm": 0.0,
+      "learning_rate": 5.41078819325608e-06,
+      "loss": 0.8722,
+      "step": 24112
+    },
+    {
+      "epoch": 0.6622449259839059,
+      "grad_norm": 0.0,
+      "learning_rate": 5.40999789690935e-06,
+      "loss": 0.8474,
+      "step": 24113
+    },
+    {
+      "epoch": 0.6622723902117492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.409207636880498e-06,
+      "loss": 0.8503,
+      "step": 24114
+    },
+    {
+      "epoch": 0.6622998544395924,
+      "grad_norm": 0.0,
+      "learning_rate": 5.408417413175775e-06,
+      "loss": 0.8918,
+      "step": 24115
+    },
+    {
+      "epoch": 0.6623273186674357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.407627225801426e-06,
+      "loss": 0.8835,
+      "step": 24116
+    },
+    {
+      "epoch": 0.6623547828952789,
+      "grad_norm": 0.0,
+      "learning_rate": 5.406837074763711e-06,
+      "loss": 0.8151,
+      "step": 24117
+    },
+    {
+      "epoch": 0.6623822471231221,
+      "grad_norm": 0.0,
+      "learning_rate": 5.40604696006888e-06,
+      "loss": 0.8876,
+      "step": 24118
+    },
+    {
+      "epoch": 0.6624097113509654,
+      "grad_norm": 0.0,
+      "learning_rate": 5.405256881723184e-06,
+      "loss": 0.8458,
+      "step": 24119
+    },
+    {
+      "epoch": 0.6624371755788087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.40446683973288e-06,
+      "loss": 0.7667,
+      "step": 24120
+    },
+    {
+      "epoch": 0.6624646398066518,
+      "grad_norm": 0.0,
+      "learning_rate": 5.40367683410421e-06,
+      "loss": 0.848,
+      "step": 24121
+    },
+    {
+      "epoch": 0.6624921040344951,
+      "grad_norm": 0.0,
+      "learning_rate": 5.402886864843431e-06,
+      "loss": 0.8172,
+      "step": 24122
+    },
+    {
+      "epoch": 0.6625195682623383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.402096931956794e-06,
+      "loss": 0.9575,
+      "step": 24123
+    },
+    {
+      "epoch": 0.6625470324901815,
+      "grad_norm": 0.0,
+      "learning_rate": 5.401307035450544e-06,
+      "loss": 0.8232,
+      "step": 24124
+    },
+    {
+      "epoch": 0.6625744967180248,
+      "grad_norm": 0.0,
+      "learning_rate": 5.400517175330934e-06,
+      "loss": 0.7854,
+      "step": 24125
+    },
+    {
+      "epoch": 0.662601960945868,
+      "grad_norm": 0.0,
+      "learning_rate": 5.399727351604217e-06,
+      "loss": 0.795,
+      "step": 24126
+    },
+    {
+      "epoch": 0.6626294251737113,
+      "grad_norm": 0.0,
+      "learning_rate": 5.398937564276636e-06,
+      "loss": 0.79,
+      "step": 24127
+    },
+    {
+      "epoch": 0.6626568894015544,
+      "grad_norm": 0.0,
+      "learning_rate": 5.398147813354444e-06,
+      "loss": 0.8536,
+      "step": 24128
+    },
+    {
+      "epoch": 0.6626843536293977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3973580988438875e-06,
+      "loss": 0.8422,
+      "step": 24129
+    },
+    {
+      "epoch": 0.662711817857241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.396568420751221e-06,
+      "loss": 0.9261,
+      "step": 24130
+    },
+    {
+      "epoch": 0.6627392820850841,
+      "grad_norm": 0.0,
+      "learning_rate": 5.395778779082685e-06,
+      "loss": 0.8273,
+      "step": 24131
+    },
+    {
+      "epoch": 0.6627667463129274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.394989173844534e-06,
+      "loss": 0.8818,
+      "step": 24132
+    },
+    {
+      "epoch": 0.6627942105407707,
+      "grad_norm": 0.0,
+      "learning_rate": 5.394199605043009e-06,
+      "loss": 0.7948,
+      "step": 24133
+    },
+    {
+      "epoch": 0.6628216747686139,
+      "grad_norm": 0.0,
+      "learning_rate": 5.393410072684362e-06,
+      "loss": 0.8588,
+      "step": 24134
+    },
+    {
+      "epoch": 0.6628491389964571,
+      "grad_norm": 0.0,
+      "learning_rate": 5.392620576774836e-06,
+      "loss": 0.8235,
+      "step": 24135
+    },
+    {
+      "epoch": 0.6628766032243003,
+      "grad_norm": 0.0,
+      "learning_rate": 5.391831117320683e-06,
+      "loss": 0.8828,
+      "step": 24136
+    },
+    {
+      "epoch": 0.6629040674521436,
+      "grad_norm": 0.0,
+      "learning_rate": 5.391041694328148e-06,
+      "loss": 0.7737,
+      "step": 24137
+    },
+    {
+      "epoch": 0.6629315316799869,
+      "grad_norm": 0.0,
+      "learning_rate": 5.390252307803478e-06,
+      "loss": 0.9456,
+      "step": 24138
+    },
+    {
+      "epoch": 0.66295899590783,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3894629577529174e-06,
+      "loss": 0.8021,
+      "step": 24139
+    },
+    {
+      "epoch": 0.6629864601356733,
+      "grad_norm": 0.0,
+      "learning_rate": 5.38867364418271e-06,
+      "loss": 0.784,
+      "step": 24140
+    },
+    {
+      "epoch": 0.6630139243635165,
+      "grad_norm": 0.0,
+      "learning_rate": 5.387884367099102e-06,
+      "loss": 0.9451,
+      "step": 24141
+    },
+    {
+      "epoch": 0.6630413885913597,
+      "grad_norm": 0.0,
+      "learning_rate": 5.387095126508339e-06,
+      "loss": 0.8768,
+      "step": 24142
+    },
+    {
+      "epoch": 0.663068852819203,
+      "grad_norm": 0.0,
+      "learning_rate": 5.386305922416667e-06,
+      "loss": 0.7547,
+      "step": 24143
+    },
+    {
+      "epoch": 0.6630963170470462,
+      "grad_norm": 0.0,
+      "learning_rate": 5.385516754830334e-06,
+      "loss": 0.8228,
+      "step": 24144
+    },
+    {
+      "epoch": 0.6631237812748895,
+      "grad_norm": 0.0,
+      "learning_rate": 5.384727623755575e-06,
+      "loss": 0.799,
+      "step": 24145
+    },
+    {
+      "epoch": 0.6631512455027327,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3839385291986425e-06,
+      "loss": 0.919,
+      "step": 24146
+    },
+    {
+      "epoch": 0.6631787097305759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.383149471165774e-06,
+      "loss": 0.8444,
+      "step": 24147
+    },
+    {
+      "epoch": 0.6632061739584192,
+      "grad_norm": 0.0,
+      "learning_rate": 5.382360449663213e-06,
+      "loss": 0.8239,
+      "step": 24148
+    },
+    {
+      "epoch": 0.6632336381862624,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3815714646972065e-06,
+      "loss": 0.9059,
+      "step": 24149
+    },
+    {
+      "epoch": 0.6632611024141056,
+      "grad_norm": 0.0,
+      "learning_rate": 5.380782516273999e-06,
+      "loss": 0.8774,
+      "step": 24150
+    },
+    {
+      "epoch": 0.6632885666419489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.379993604399826e-06,
+      "loss": 0.7962,
+      "step": 24151
+    },
+    {
+      "epoch": 0.6633160308697921,
+      "grad_norm": 0.0,
+      "learning_rate": 5.379204729080934e-06,
+      "loss": 0.8161,
+      "step": 24152
+    },
+    {
+      "epoch": 0.6633434950976353,
+      "grad_norm": 0.0,
+      "learning_rate": 5.378415890323563e-06,
+      "loss": 0.8207,
+      "step": 24153
+    },
+    {
+      "epoch": 0.6633709593254785,
+      "grad_norm": 0.0,
+      "learning_rate": 5.377627088133959e-06,
+      "loss": 0.8537,
+      "step": 24154
+    },
+    {
+      "epoch": 0.6633984235533218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.376838322518356e-06,
+      "loss": 0.8126,
+      "step": 24155
+    },
+    {
+      "epoch": 0.6634258877811651,
+      "grad_norm": 0.0,
+      "learning_rate": 5.376049593483004e-06,
+      "loss": 0.8795,
+      "step": 24156
+    },
+    {
+      "epoch": 0.6634533520090082,
+      "grad_norm": 0.0,
+      "learning_rate": 5.375260901034135e-06,
+      "loss": 0.8787,
+      "step": 24157
+    },
+    {
+      "epoch": 0.6634808162368515,
+      "grad_norm": 0.0,
+      "learning_rate": 5.374472245177993e-06,
+      "loss": 0.8332,
+      "step": 24158
+    },
+    {
+      "epoch": 0.6635082804646948,
+      "grad_norm": 0.0,
+      "learning_rate": 5.373683625920818e-06,
+      "loss": 0.8111,
+      "step": 24159
+    },
+    {
+      "epoch": 0.663535744692538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3728950432688505e-06,
+      "loss": 0.7987,
+      "step": 24160
+    },
+    {
+      "epoch": 0.6635632089203812,
+      "grad_norm": 0.0,
+      "learning_rate": 5.372106497228329e-06,
+      "loss": 0.8226,
+      "step": 24161
+    },
+    {
+      "epoch": 0.6635906731482244,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3713179878054985e-06,
+      "loss": 0.7713,
+      "step": 24162
+    },
+    {
+      "epoch": 0.6636181373760677,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3705295150065926e-06,
+      "loss": 0.8949,
+      "step": 24163
+    },
+    {
+      "epoch": 0.663645601603911,
+      "grad_norm": 0.0,
+      "learning_rate": 5.369741078837847e-06,
+      "loss": 0.8788,
+      "step": 24164
+    },
+    {
+      "epoch": 0.6636730658317541,
+      "grad_norm": 0.0,
+      "learning_rate": 5.368952679305503e-06,
+      "loss": 0.8482,
+      "step": 24165
+    },
+    {
+      "epoch": 0.6637005300595974,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3681643164158e-06,
+      "loss": 0.9884,
+      "step": 24166
+    },
+    {
+      "epoch": 0.6637279942874406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.367375990174974e-06,
+      "loss": 0.8581,
+      "step": 24167
+    },
+    {
+      "epoch": 0.6637554585152838,
+      "grad_norm": 0.0,
+      "learning_rate": 5.366587700589268e-06,
+      "loss": 0.8284,
+      "step": 24168
+    },
+    {
+      "epoch": 0.6637829227431271,
+      "grad_norm": 0.0,
+      "learning_rate": 5.365799447664912e-06,
+      "loss": 0.8972,
+      "step": 24169
+    },
+    {
+      "epoch": 0.6638103869709703,
+      "grad_norm": 0.0,
+      "learning_rate": 5.36501123140815e-06,
+      "loss": 0.8365,
+      "step": 24170
+    },
+    {
+      "epoch": 0.6638378511988136,
+      "grad_norm": 0.0,
+      "learning_rate": 5.364223051825208e-06,
+      "loss": 0.9704,
+      "step": 24171
+    },
+    {
+      "epoch": 0.6638653154266568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.363434908922332e-06,
+      "loss": 0.9812,
+      "step": 24172
+    },
+    {
+      "epoch": 0.6638927796545,
+      "grad_norm": 0.0,
+      "learning_rate": 5.362646802705753e-06,
+      "loss": 0.9789,
+      "step": 24173
+    },
+    {
+      "epoch": 0.6639202438823433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.361858733181713e-06,
+      "loss": 0.8185,
+      "step": 24174
+    },
+    {
+      "epoch": 0.6639477081101864,
+      "grad_norm": 0.0,
+      "learning_rate": 5.361070700356441e-06,
+      "loss": 0.9502,
+      "step": 24175
+    },
+    {
+      "epoch": 0.6639751723380297,
+      "grad_norm": 0.0,
+      "learning_rate": 5.360282704236172e-06,
+      "loss": 0.8468,
+      "step": 24176
+    },
+    {
+      "epoch": 0.664002636565873,
+      "grad_norm": 0.0,
+      "learning_rate": 5.359494744827145e-06,
+      "loss": 0.8791,
+      "step": 24177
+    },
+    {
+      "epoch": 0.6640301007937162,
+      "grad_norm": 0.0,
+      "learning_rate": 5.358706822135598e-06,
+      "loss": 0.8932,
+      "step": 24178
+    },
+    {
+      "epoch": 0.6640575650215594,
+      "grad_norm": 0.0,
+      "learning_rate": 5.357918936167758e-06,
+      "loss": 0.8084,
+      "step": 24179
+    },
+    {
+      "epoch": 0.6640850292494026,
+      "grad_norm": 0.0,
+      "learning_rate": 5.357131086929861e-06,
+      "loss": 0.8698,
+      "step": 24180
+    },
+    {
+      "epoch": 0.6641124934772459,
+      "grad_norm": 0.0,
+      "learning_rate": 5.356343274428139e-06,
+      "loss": 0.8517,
+      "step": 24181
+    },
+    {
+      "epoch": 0.6641399577050892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3555554986688275e-06,
+      "loss": 0.7989,
+      "step": 24182
+    },
+    {
+      "epoch": 0.6641674219329323,
+      "grad_norm": 0.0,
+      "learning_rate": 5.354767759658161e-06,
+      "loss": 0.8849,
+      "step": 24183
+    },
+    {
+      "epoch": 0.6641948861607756,
+      "grad_norm": 0.0,
+      "learning_rate": 5.353980057402372e-06,
+      "loss": 0.8239,
+      "step": 24184
+    },
+    {
+      "epoch": 0.6642223503886189,
+      "grad_norm": 0.0,
+      "learning_rate": 5.353192391907695e-06,
+      "loss": 0.9031,
+      "step": 24185
+    },
+    {
+      "epoch": 0.664249814616462,
+      "grad_norm": 0.0,
+      "learning_rate": 5.352404763180356e-06,
+      "loss": 0.8641,
+      "step": 24186
+    },
+    {
+      "epoch": 0.6642772788443053,
+      "grad_norm": 0.0,
+      "learning_rate": 5.351617171226595e-06,
+      "loss": 0.8396,
+      "step": 24187
+    },
+    {
+      "epoch": 0.6643047430721485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.350829616052635e-06,
+      "loss": 0.827,
+      "step": 24188
+    },
+    {
+      "epoch": 0.6643322072999918,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3500420976647135e-06,
+      "loss": 0.825,
+      "step": 24189
+    },
+    {
+      "epoch": 0.664359671527835,
+      "grad_norm": 0.0,
+      "learning_rate": 5.34925461606906e-06,
+      "loss": 0.8892,
+      "step": 24190
+    },
+    {
+      "epoch": 0.6643871357556782,
+      "grad_norm": 0.0,
+      "learning_rate": 5.348467171271908e-06,
+      "loss": 0.8099,
+      "step": 24191
+    },
+    {
+      "epoch": 0.6644145999835215,
+      "grad_norm": 0.0,
+      "learning_rate": 5.347679763279483e-06,
+      "loss": 0.8385,
+      "step": 24192
+    },
+    {
+      "epoch": 0.6644420642113646,
+      "grad_norm": 0.0,
+      "learning_rate": 5.346892392098018e-06,
+      "loss": 0.7667,
+      "step": 24193
+    },
+    {
+      "epoch": 0.6644695284392079,
+      "grad_norm": 0.0,
+      "learning_rate": 5.346105057733746e-06,
+      "loss": 0.8707,
+      "step": 24194
+    },
+    {
+      "epoch": 0.6644969926670512,
+      "grad_norm": 0.0,
+      "learning_rate": 5.345317760192888e-06,
+      "loss": 0.9146,
+      "step": 24195
+    },
+    {
+      "epoch": 0.6645244568948944,
+      "grad_norm": 0.0,
+      "learning_rate": 5.344530499481681e-06,
+      "loss": 0.9239,
+      "step": 24196
+    },
+    {
+      "epoch": 0.6645519211227376,
+      "grad_norm": 0.0,
+      "learning_rate": 5.343743275606356e-06,
+      "loss": 0.8374,
+      "step": 24197
+    },
+    {
+      "epoch": 0.6645793853505809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.342956088573133e-06,
+      "loss": 0.878,
+      "step": 24198
+    },
+    {
+      "epoch": 0.6646068495784241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.342168938388246e-06,
+      "loss": 0.9135,
+      "step": 24199
+    },
+    {
+      "epoch": 0.6646343138062674,
+      "grad_norm": 0.0,
+      "learning_rate": 5.341381825057923e-06,
+      "loss": 0.7663,
+      "step": 24200
+    },
+    {
+      "epoch": 0.6646617780341105,
+      "grad_norm": 0.0,
+      "learning_rate": 5.34059474858839e-06,
+      "loss": 0.8649,
+      "step": 24201
+    },
+    {
+      "epoch": 0.6646892422619538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3398077089858825e-06,
+      "loss": 0.8956,
+      "step": 24202
+    },
+    {
+      "epoch": 0.6647167064897971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.33902070625662e-06,
+      "loss": 0.8121,
+      "step": 24203
+    },
+    {
+      "epoch": 0.6647441707176402,
+      "grad_norm": 0.0,
+      "learning_rate": 5.338233740406827e-06,
+      "loss": 0.9172,
+      "step": 24204
+    },
+    {
+      "epoch": 0.6647716349454835,
+      "grad_norm": 0.0,
+      "learning_rate": 5.337446811442736e-06,
+      "loss": 0.8906,
+      "step": 24205
+    },
+    {
+      "epoch": 0.6647990991733267,
+      "grad_norm": 0.0,
+      "learning_rate": 5.33665991937057e-06,
+      "loss": 0.8415,
+      "step": 24206
+    },
+    {
+      "epoch": 0.66482656340117,
+      "grad_norm": 0.0,
+      "learning_rate": 5.335873064196558e-06,
+      "loss": 0.8448,
+      "step": 24207
+    },
+    {
+      "epoch": 0.6648540276290132,
+      "grad_norm": 0.0,
+      "learning_rate": 5.335086245926926e-06,
+      "loss": 0.8241,
+      "step": 24208
+    },
+    {
+      "epoch": 0.6648814918568564,
+      "grad_norm": 0.0,
+      "learning_rate": 5.334299464567901e-06,
+      "loss": 0.8126,
+      "step": 24209
+    },
+    {
+      "epoch": 0.6649089560846997,
+      "grad_norm": 0.0,
+      "learning_rate": 5.333512720125703e-06,
+      "loss": 0.844,
+      "step": 24210
+    },
+    {
+      "epoch": 0.664936420312543,
+      "grad_norm": 0.0,
+      "learning_rate": 5.332726012606564e-06,
+      "loss": 0.8083,
+      "step": 24211
+    },
+    {
+      "epoch": 0.6649638845403861,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3319393420167e-06,
+      "loss": 0.7985,
+      "step": 24212
+    },
+    {
+      "epoch": 0.6649913487682294,
+      "grad_norm": 0.0,
+      "learning_rate": 5.33115270836234e-06,
+      "loss": 0.7646,
+      "step": 24213
+    },
+    {
+      "epoch": 0.6650188129960726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.330366111649709e-06,
+      "loss": 0.9753,
+      "step": 24214
+    },
+    {
+      "epoch": 0.6650462772239158,
+      "grad_norm": 0.0,
+      "learning_rate": 5.329579551885035e-06,
+      "loss": 0.8052,
+      "step": 24215
+    },
+    {
+      "epoch": 0.6650737414517591,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32879302907453e-06,
+      "loss": 0.9282,
+      "step": 24216
+    },
+    {
+      "epoch": 0.6651012056796023,
+      "grad_norm": 0.0,
+      "learning_rate": 5.328006543224427e-06,
+      "loss": 0.8647,
+      "step": 24217
+    },
+    {
+      "epoch": 0.6651286699074456,
+      "grad_norm": 0.0,
+      "learning_rate": 5.327220094340948e-06,
+      "loss": 0.9326,
+      "step": 24218
+    },
+    {
+      "epoch": 0.6651561341352887,
+      "grad_norm": 0.0,
+      "learning_rate": 5.326433682430311e-06,
+      "loss": 0.8148,
+      "step": 24219
+    },
+    {
+      "epoch": 0.665183598363132,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32564730749874e-06,
+      "loss": 0.8042,
+      "step": 24220
+    },
+    {
+      "epoch": 0.6652110625909753,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3248609695524635e-06,
+      "loss": 0.8466,
+      "step": 24221
+    },
+    {
+      "epoch": 0.6652385268188185,
+      "grad_norm": 0.0,
+      "learning_rate": 5.324074668597693e-06,
+      "loss": 0.8785,
+      "step": 24222
+    },
+    {
+      "epoch": 0.6652659910466617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.323288404640655e-06,
+      "loss": 0.8237,
+      "step": 24223
+    },
+    {
+      "epoch": 0.665293455274505,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3225021776875706e-06,
+      "loss": 0.7861,
+      "step": 24224
+    },
+    {
+      "epoch": 0.6653209195023482,
+      "grad_norm": 0.0,
+      "learning_rate": 5.321715987744662e-06,
+      "loss": 0.8094,
+      "step": 24225
+    },
+    {
+      "epoch": 0.6653483837301915,
+      "grad_norm": 0.0,
+      "learning_rate": 5.320929834818151e-06,
+      "loss": 0.8515,
+      "step": 24226
+    },
+    {
+      "epoch": 0.6653758479580346,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3201437189142565e-06,
+      "loss": 0.8648,
+      "step": 24227
+    },
+    {
+      "epoch": 0.6654033121858779,
+      "grad_norm": 0.0,
+      "learning_rate": 5.319357640039193e-06,
+      "loss": 0.8329,
+      "step": 24228
+    },
+    {
+      "epoch": 0.6654307764137212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.318571598199187e-06,
+      "loss": 0.9262,
+      "step": 24229
+    },
+    {
+      "epoch": 0.6654582406415643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.317785593400454e-06,
+      "loss": 0.8084,
+      "step": 24230
+    },
+    {
+      "epoch": 0.6654857048694076,
+      "grad_norm": 0.0,
+      "learning_rate": 5.316999625649215e-06,
+      "loss": 0.7659,
+      "step": 24231
+    },
+    {
+      "epoch": 0.6655131690972508,
+      "grad_norm": 0.0,
+      "learning_rate": 5.316213694951691e-06,
+      "loss": 0.8518,
+      "step": 24232
+    },
+    {
+      "epoch": 0.665540633325094,
+      "grad_norm": 0.0,
+      "learning_rate": 5.315427801314101e-06,
+      "loss": 0.8589,
+      "step": 24233
+    },
+    {
+      "epoch": 0.6655680975529373,
+      "grad_norm": 0.0,
+      "learning_rate": 5.314641944742657e-06,
+      "loss": 0.8599,
+      "step": 24234
+    },
+    {
+      "epoch": 0.6655955617807805,
+      "grad_norm": 0.0,
+      "learning_rate": 5.313856125243585e-06,
+      "loss": 0.8706,
+      "step": 24235
+    },
+    {
+      "epoch": 0.6656230260086238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.313070342823094e-06,
+      "loss": 0.7769,
+      "step": 24236
+    },
+    {
+      "epoch": 0.665650490236467,
+      "grad_norm": 0.0,
+      "learning_rate": 5.312284597487407e-06,
+      "loss": 0.7234,
+      "step": 24237
+    },
+    {
+      "epoch": 0.6656779544643102,
+      "grad_norm": 0.0,
+      "learning_rate": 5.311498889242743e-06,
+      "loss": 0.8382,
+      "step": 24238
+    },
+    {
+      "epoch": 0.6657054186921535,
+      "grad_norm": 0.0,
+      "learning_rate": 5.310713218095313e-06,
+      "loss": 0.9455,
+      "step": 24239
+    },
+    {
+      "epoch": 0.6657328829199967,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309927584051335e-06,
+      "loss": 0.7691,
+      "step": 24240
+    },
+    {
+      "epoch": 0.6657603471478399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309141987117028e-06,
+      "loss": 0.8312,
+      "step": 24241
+    },
+    {
+      "epoch": 0.6657878113756832,
+      "grad_norm": 0.0,
+      "learning_rate": 5.308356427298611e-06,
+      "loss": 0.8333,
+      "step": 24242
+    },
+    {
+      "epoch": 0.6658152756035264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3075709046022905e-06,
+      "loss": 0.8891,
+      "step": 24243
+    },
+    {
+      "epoch": 0.6658427398313697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.306785419034291e-06,
+      "loss": 0.8476,
+      "step": 24244
+    },
+    {
+      "epoch": 0.6658702040592128,
+      "grad_norm": 0.0,
+      "learning_rate": 5.305999970600819e-06,
+      "loss": 0.808,
+      "step": 24245
+    },
+    {
+      "epoch": 0.6658976682870561,
+      "grad_norm": 0.0,
+      "learning_rate": 5.305214559308094e-06,
+      "loss": 0.8524,
+      "step": 24246
+    },
+    {
+      "epoch": 0.6659251325148994,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3044291851623295e-06,
+      "loss": 0.8557,
+      "step": 24247
+    },
+    {
+      "epoch": 0.6659525967427425,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3036438481697405e-06,
+      "loss": 0.901,
+      "step": 24248
+    },
+    {
+      "epoch": 0.6659800609705858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.30285854833654e-06,
+      "loss": 0.8543,
+      "step": 24249
+    },
+    {
+      "epoch": 0.6660075251984291,
+      "grad_norm": 0.0,
+      "learning_rate": 5.302073285668947e-06,
+      "loss": 0.808,
+      "step": 24250
+    },
+    {
+      "epoch": 0.6660349894262723,
+      "grad_norm": 0.0,
+      "learning_rate": 5.30128806017317e-06,
+      "loss": 0.8828,
+      "step": 24251
+    },
+    {
+      "epoch": 0.6660624536541155,
+      "grad_norm": 0.0,
+      "learning_rate": 5.300502871855417e-06,
+      "loss": 0.8225,
+      "step": 24252
+    },
+    {
+      "epoch": 0.6660899178819587,
+      "grad_norm": 0.0,
+      "learning_rate": 5.299717720721906e-06,
+      "loss": 0.8216,
+      "step": 24253
+    },
+    {
+      "epoch": 0.666117382109802,
+      "grad_norm": 0.0,
+      "learning_rate": 5.29893260677885e-06,
+      "loss": 0.8504,
+      "step": 24254
+    },
+    {
+      "epoch": 0.6661448463376453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2981475300324596e-06,
+      "loss": 0.8672,
+      "step": 24255
+    },
+    {
+      "epoch": 0.6661723105654884,
+      "grad_norm": 0.0,
+      "learning_rate": 5.297362490488951e-06,
+      "loss": 0.8436,
+      "step": 24256
+    },
+    {
+      "epoch": 0.6661997747933317,
+      "grad_norm": 0.0,
+      "learning_rate": 5.296577488154528e-06,
+      "loss": 0.8251,
+      "step": 24257
+    },
+    {
+      "epoch": 0.6662272390211749,
+      "grad_norm": 0.0,
+      "learning_rate": 5.295792523035405e-06,
+      "loss": 0.8476,
+      "step": 24258
+    },
+    {
+      "epoch": 0.6662547032490181,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2950075951378e-06,
+      "loss": 0.8706,
+      "step": 24259
+    },
+    {
+      "epoch": 0.6662821674768614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.294222704467913e-06,
+      "loss": 0.7319,
+      "step": 24260
+    },
+    {
+      "epoch": 0.6663096317047046,
+      "grad_norm": 0.0,
+      "learning_rate": 5.293437851031957e-06,
+      "loss": 0.8422,
+      "step": 24261
+    },
+    {
+      "epoch": 0.6663370959325479,
+      "grad_norm": 0.0,
+      "learning_rate": 5.292653034836149e-06,
+      "loss": 0.8345,
+      "step": 24262
+    },
+    {
+      "epoch": 0.6663645601603911,
+      "grad_norm": 0.0,
+      "learning_rate": 5.29186825588669e-06,
+      "loss": 0.8826,
+      "step": 24263
+    },
+    {
+      "epoch": 0.6663920243882343,
+      "grad_norm": 0.0,
+      "learning_rate": 5.291083514189792e-06,
+      "loss": 0.7743,
+      "step": 24264
+    },
+    {
+      "epoch": 0.6664194886160776,
+      "grad_norm": 0.0,
+      "learning_rate": 5.290298809751666e-06,
+      "loss": 0.8616,
+      "step": 24265
+    },
+    {
+      "epoch": 0.6664469528439207,
+      "grad_norm": 0.0,
+      "learning_rate": 5.289514142578525e-06,
+      "loss": 0.8376,
+      "step": 24266
+    },
+    {
+      "epoch": 0.666474417071764,
+      "grad_norm": 0.0,
+      "learning_rate": 5.288729512676568e-06,
+      "loss": 0.7987,
+      "step": 24267
+    },
+    {
+      "epoch": 0.6665018812996073,
+      "grad_norm": 0.0,
+      "learning_rate": 5.287944920052011e-06,
+      "loss": 0.8914,
+      "step": 24268
+    },
+    {
+      "epoch": 0.6665293455274505,
+      "grad_norm": 0.0,
+      "learning_rate": 5.287160364711057e-06,
+      "loss": 0.807,
+      "step": 24269
+    },
+    {
+      "epoch": 0.6665568097552937,
+      "grad_norm": 0.0,
+      "learning_rate": 5.286375846659915e-06,
+      "loss": 0.7683,
+      "step": 24270
+    },
+    {
+      "epoch": 0.6665842739831369,
+      "grad_norm": 0.0,
+      "learning_rate": 5.285591365904793e-06,
+      "loss": 0.8415,
+      "step": 24271
+    },
+    {
+      "epoch": 0.6666117382109802,
+      "grad_norm": 0.0,
+      "learning_rate": 5.284806922451898e-06,
+      "loss": 0.7799,
+      "step": 24272
+    },
+    {
+      "epoch": 0.6666392024388235,
+      "grad_norm": 0.0,
+      "learning_rate": 5.284022516307439e-06,
+      "loss": 0.7563,
+      "step": 24273
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2832381474776225e-06,
+      "loss": 0.978,
+      "step": 24274
+    },
+    {
+      "epoch": 0.6666941308945099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.282453815968653e-06,
+      "loss": 0.7947,
+      "step": 24275
+    },
+    {
+      "epoch": 0.6667215951223532,
+      "grad_norm": 0.0,
+      "learning_rate": 5.281669521786733e-06,
+      "loss": 0.8013,
+      "step": 24276
+    },
+    {
+      "epoch": 0.6667490593501963,
+      "grad_norm": 0.0,
+      "learning_rate": 5.280885264938072e-06,
+      "loss": 0.8191,
+      "step": 24277
+    },
+    {
+      "epoch": 0.6667765235780396,
+      "grad_norm": 0.0,
+      "learning_rate": 5.280101045428874e-06,
+      "loss": 0.8614,
+      "step": 24278
+    },
+    {
+      "epoch": 0.6668039878058828,
+      "grad_norm": 0.0,
+      "learning_rate": 5.279316863265344e-06,
+      "loss": 0.8567,
+      "step": 24279
+    },
+    {
+      "epoch": 0.6668314520337261,
+      "grad_norm": 0.0,
+      "learning_rate": 5.278532718453693e-06,
+      "loss": 0.7909,
+      "step": 24280
+    },
+    {
+      "epoch": 0.6668589162615693,
+      "grad_norm": 0.0,
+      "learning_rate": 5.277748611000115e-06,
+      "loss": 0.8037,
+      "step": 24281
+    },
+    {
+      "epoch": 0.6668863804894125,
+      "grad_norm": 0.0,
+      "learning_rate": 5.276964540910821e-06,
+      "loss": 0.8148,
+      "step": 24282
+    },
+    {
+      "epoch": 0.6669138447172558,
+      "grad_norm": 0.0,
+      "learning_rate": 5.276180508192015e-06,
+      "loss": 0.9029,
+      "step": 24283
+    },
+    {
+      "epoch": 0.666941308945099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2753965128498955e-06,
+      "loss": 0.7473,
+      "step": 24284
+    },
+    {
+      "epoch": 0.6669687731729422,
+      "grad_norm": 0.0,
+      "learning_rate": 5.274612554890669e-06,
+      "loss": 0.8297,
+      "step": 24285
+    },
+    {
+      "epoch": 0.6669962374007855,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2738286343205415e-06,
+      "loss": 0.7933,
+      "step": 24286
+    },
+    {
+      "epoch": 0.6670237016286287,
+      "grad_norm": 0.0,
+      "learning_rate": 5.27304475114571e-06,
+      "loss": 0.8124,
+      "step": 24287
+    },
+    {
+      "epoch": 0.667051165856472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.27226090537238e-06,
+      "loss": 0.8266,
+      "step": 24288
+    },
+    {
+      "epoch": 0.6670786300843152,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2714770970067505e-06,
+      "loss": 0.7795,
+      "step": 24289
+    },
+    {
+      "epoch": 0.6671060943121584,
+      "grad_norm": 0.0,
+      "learning_rate": 5.270693326055032e-06,
+      "loss": 0.9232,
+      "step": 24290
+    },
+    {
+      "epoch": 0.6671335585400017,
+      "grad_norm": 0.0,
+      "learning_rate": 5.269909592523415e-06,
+      "loss": 0.886,
+      "step": 24291
+    },
+    {
+      "epoch": 0.6671610227678448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.26912589641811e-06,
+      "loss": 0.8457,
+      "step": 24292
+    },
+    {
+      "epoch": 0.6671884869956881,
+      "grad_norm": 0.0,
+      "learning_rate": 5.26834223774531e-06,
+      "loss": 0.7789,
+      "step": 24293
+    },
+    {
+      "epoch": 0.6672159512235314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2675586165112166e-06,
+      "loss": 0.9431,
+      "step": 24294
+    },
+    {
+      "epoch": 0.6672434154513746,
+      "grad_norm": 0.0,
+      "learning_rate": 5.266775032722036e-06,
+      "loss": 0.8588,
+      "step": 24295
+    },
+    {
+      "epoch": 0.6672708796792178,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2659914863839625e-06,
+      "loss": 0.9028,
+      "step": 24296
+    },
+    {
+      "epoch": 0.667298343907061,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2652079775032016e-06,
+      "loss": 0.7586,
+      "step": 24297
+    },
+    {
+      "epoch": 0.6673258081349043,
+      "grad_norm": 0.0,
+      "learning_rate": 5.264424506085948e-06,
+      "loss": 0.9156,
+      "step": 24298
+    },
+    {
+      "epoch": 0.6673532723627476,
+      "grad_norm": 0.0,
+      "learning_rate": 5.263641072138403e-06,
+      "loss": 0.7805,
+      "step": 24299
+    },
+    {
+      "epoch": 0.6673807365905907,
+      "grad_norm": 0.0,
+      "learning_rate": 5.262857675666762e-06,
+      "loss": 0.7858,
+      "step": 24300
+    },
+    {
+      "epoch": 0.667408200818434,
+      "grad_norm": 0.0,
+      "learning_rate": 5.262074316677227e-06,
+      "loss": 0.8751,
+      "step": 24301
+    },
+    {
+      "epoch": 0.6674356650462773,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2612909951759935e-06,
+      "loss": 0.942,
+      "step": 24302
+    },
+    {
+      "epoch": 0.6674631292741204,
+      "grad_norm": 0.0,
+      "learning_rate": 5.260507711169267e-06,
+      "loss": 0.8773,
+      "step": 24303
+    },
+    {
+      "epoch": 0.6674905935019637,
+      "grad_norm": 0.0,
+      "learning_rate": 5.259724464663234e-06,
+      "loss": 0.8004,
+      "step": 24304
+    },
+    {
+      "epoch": 0.6675180577298069,
+      "grad_norm": 0.0,
+      "learning_rate": 5.258941255664098e-06,
+      "loss": 0.7769,
+      "step": 24305
+    },
+    {
+      "epoch": 0.6675455219576502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.25815808417806e-06,
+      "loss": 0.91,
+      "step": 24306
+    },
+    {
+      "epoch": 0.6675729861854934,
+      "grad_norm": 0.0,
+      "learning_rate": 5.257374950211307e-06,
+      "loss": 0.8341,
+      "step": 24307
+    },
+    {
+      "epoch": 0.6676004504133366,
+      "grad_norm": 0.0,
+      "learning_rate": 5.256591853770041e-06,
+      "loss": 0.8145,
+      "step": 24308
+    },
+    {
+      "epoch": 0.6676279146411799,
+      "grad_norm": 0.0,
+      "learning_rate": 5.255808794860461e-06,
+      "loss": 0.8485,
+      "step": 24309
+    },
+    {
+      "epoch": 0.6676553788690232,
+      "grad_norm": 0.0,
+      "learning_rate": 5.255025773488756e-06,
+      "loss": 0.7604,
+      "step": 24310
+    },
+    {
+      "epoch": 0.6676828430968663,
+      "grad_norm": 0.0,
+      "learning_rate": 5.254242789661125e-06,
+      "loss": 0.8383,
+      "step": 24311
+    },
+    {
+      "epoch": 0.6677103073247096,
+      "grad_norm": 0.0,
+      "learning_rate": 5.253459843383763e-06,
+      "loss": 0.809,
+      "step": 24312
+    },
+    {
+      "epoch": 0.6677377715525528,
+      "grad_norm": 0.0,
+      "learning_rate": 5.252676934662866e-06,
+      "loss": 0.8681,
+      "step": 24313
+    },
+    {
+      "epoch": 0.667765235780396,
+      "grad_norm": 0.0,
+      "learning_rate": 5.251894063504632e-06,
+      "loss": 0.7876,
+      "step": 24314
+    },
+    {
+      "epoch": 0.6677927000082393,
+      "grad_norm": 0.0,
+      "learning_rate": 5.251111229915251e-06,
+      "loss": 0.8506,
+      "step": 24315
+    },
+    {
+      "epoch": 0.6678201642360825,
+      "grad_norm": 0.0,
+      "learning_rate": 5.250328433900913e-06,
+      "loss": 0.8602,
+      "step": 24316
+    },
+    {
+      "epoch": 0.6678476284639258,
+      "grad_norm": 0.0,
+      "learning_rate": 5.249545675467817e-06,
+      "loss": 0.8239,
+      "step": 24317
+    },
+    {
+      "epoch": 0.6678750926917689,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2487629546221556e-06,
+      "loss": 0.9934,
+      "step": 24318
+    },
+    {
+      "epoch": 0.6679025569196122,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2479802713701234e-06,
+      "loss": 0.8674,
+      "step": 24319
+    },
+    {
+      "epoch": 0.6679300211474555,
+      "grad_norm": 0.0,
+      "learning_rate": 5.24719762571791e-06,
+      "loss": 0.7707,
+      "step": 24320
+    },
+    {
+      "epoch": 0.6679574853752986,
+      "grad_norm": 0.0,
+      "learning_rate": 5.246415017671716e-06,
+      "loss": 0.796,
+      "step": 24321
+    },
+    {
+      "epoch": 0.6679849496031419,
+      "grad_norm": 0.0,
+      "learning_rate": 5.245632447237722e-06,
+      "loss": 0.8562,
+      "step": 24322
+    },
+    {
+      "epoch": 0.6680124138309852,
+      "grad_norm": 0.0,
+      "learning_rate": 5.24484991442213e-06,
+      "loss": 0.887,
+      "step": 24323
+    },
+    {
+      "epoch": 0.6680398780588284,
+      "grad_norm": 0.0,
+      "learning_rate": 5.244067419231125e-06,
+      "loss": 0.822,
+      "step": 24324
+    },
+    {
+      "epoch": 0.6680673422866716,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2432849616708995e-06,
+      "loss": 0.7863,
+      "step": 24325
+    },
+    {
+      "epoch": 0.6680948065145148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.242502541747645e-06,
+      "loss": 0.8786,
+      "step": 24326
+    },
+    {
+      "epoch": 0.6681222707423581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.241720159467559e-06,
+      "loss": 0.8292,
+      "step": 24327
+    },
+    {
+      "epoch": 0.6681497349702014,
+      "grad_norm": 0.0,
+      "learning_rate": 5.240937814836821e-06,
+      "loss": 0.895,
+      "step": 24328
+    },
+    {
+      "epoch": 0.6681771991980445,
+      "grad_norm": 0.0,
+      "learning_rate": 5.240155507861628e-06,
+      "loss": 0.8704,
+      "step": 24329
+    },
+    {
+      "epoch": 0.6682046634258878,
+      "grad_norm": 0.0,
+      "learning_rate": 5.239373238548171e-06,
+      "loss": 0.7817,
+      "step": 24330
+    },
+    {
+      "epoch": 0.668232127653731,
+      "grad_norm": 0.0,
+      "learning_rate": 5.238591006902634e-06,
+      "loss": 0.8245,
+      "step": 24331
+    },
+    {
+      "epoch": 0.6682595918815742,
+      "grad_norm": 0.0,
+      "learning_rate": 5.237808812931209e-06,
+      "loss": 0.815,
+      "step": 24332
+    },
+    {
+      "epoch": 0.6682870561094175,
+      "grad_norm": 0.0,
+      "learning_rate": 5.237026656640089e-06,
+      "loss": 0.7573,
+      "step": 24333
+    },
+    {
+      "epoch": 0.6683145203372607,
+      "grad_norm": 0.0,
+      "learning_rate": 5.236244538035456e-06,
+      "loss": 0.6907,
+      "step": 24334
+    },
+    {
+      "epoch": 0.668341984565104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.235462457123501e-06,
+      "loss": 0.8548,
+      "step": 24335
+    },
+    {
+      "epoch": 0.6683694487929472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.234680413910413e-06,
+      "loss": 0.9024,
+      "step": 24336
+    },
+    {
+      "epoch": 0.6683969130207904,
+      "grad_norm": 0.0,
+      "learning_rate": 5.233898408402379e-06,
+      "loss": 0.798,
+      "step": 24337
+    },
+    {
+      "epoch": 0.6684243772486337,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2331164406055916e-06,
+      "loss": 0.8122,
+      "step": 24338
+    },
+    {
+      "epoch": 0.6684518414764768,
+      "grad_norm": 0.0,
+      "learning_rate": 5.232334510526233e-06,
+      "loss": 0.8053,
+      "step": 24339
+    },
+    {
+      "epoch": 0.6684793057043201,
+      "grad_norm": 0.0,
+      "learning_rate": 5.231552618170486e-06,
+      "loss": 0.817,
+      "step": 24340
+    },
+    {
+      "epoch": 0.6685067699321634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.230770763544541e-06,
+      "loss": 0.8451,
+      "step": 24341
+    },
+    {
+      "epoch": 0.6685342341600066,
+      "grad_norm": 0.0,
+      "learning_rate": 5.229988946654587e-06,
+      "loss": 0.8015,
+      "step": 24342
+    },
+    {
+      "epoch": 0.6685616983878498,
+      "grad_norm": 0.0,
+      "learning_rate": 5.229207167506809e-06,
+      "loss": 0.8932,
+      "step": 24343
+    },
+    {
+      "epoch": 0.668589162615693,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2284254261073905e-06,
+      "loss": 0.7571,
+      "step": 24344
+    },
+    {
+      "epoch": 0.6686166268435363,
+      "grad_norm": 0.0,
+      "learning_rate": 5.227643722462522e-06,
+      "loss": 0.7588,
+      "step": 24345
+    },
+    {
+      "epoch": 0.6686440910713796,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2268620565783815e-06,
+      "loss": 0.8355,
+      "step": 24346
+    },
+    {
+      "epoch": 0.6686715552992227,
+      "grad_norm": 0.0,
+      "learning_rate": 5.226080428461161e-06,
+      "loss": 0.921,
+      "step": 24347
+    },
+    {
+      "epoch": 0.668699019527066,
+      "grad_norm": 0.0,
+      "learning_rate": 5.225298838117038e-06,
+      "loss": 0.8586,
+      "step": 24348
+    },
+    {
+      "epoch": 0.6687264837549093,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2245172855522e-06,
+      "loss": 0.9173,
+      "step": 24349
+    },
+    {
+      "epoch": 0.6687539479827525,
+      "grad_norm": 0.0,
+      "learning_rate": 5.223735770772831e-06,
+      "loss": 0.8959,
+      "step": 24350
+    },
+    {
+      "epoch": 0.6687814122105957,
+      "grad_norm": 0.0,
+      "learning_rate": 5.222954293785118e-06,
+      "loss": 0.7599,
+      "step": 24351
+    },
+    {
+      "epoch": 0.6688088764384389,
+      "grad_norm": 0.0,
+      "learning_rate": 5.222172854595238e-06,
+      "loss": 0.8155,
+      "step": 24352
+    },
+    {
+      "epoch": 0.6688363406662822,
+      "grad_norm": 0.0,
+      "learning_rate": 5.221391453209377e-06,
+      "loss": 0.7999,
+      "step": 24353
+    },
+    {
+      "epoch": 0.6688638048941254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.220610089633722e-06,
+      "loss": 0.9122,
+      "step": 24354
+    },
+    {
+      "epoch": 0.6688912691219686,
+      "grad_norm": 0.0,
+      "learning_rate": 5.219828763874446e-06,
+      "loss": 0.896,
+      "step": 24355
+    },
+    {
+      "epoch": 0.6689187333498119,
+      "grad_norm": 0.0,
+      "learning_rate": 5.219047475937737e-06,
+      "loss": 0.8768,
+      "step": 24356
+    },
+    {
+      "epoch": 0.668946197577655,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2182662258297826e-06,
+      "loss": 0.8622,
+      "step": 24357
+    },
+    {
+      "epoch": 0.6689736618054983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2174850135567516e-06,
+      "loss": 0.8028,
+      "step": 24358
+    },
+    {
+      "epoch": 0.6690011260333416,
+      "grad_norm": 0.0,
+      "learning_rate": 5.216703839124833e-06,
+      "loss": 0.8524,
+      "step": 24359
+    },
+    {
+      "epoch": 0.6690285902611848,
+      "grad_norm": 0.0,
+      "learning_rate": 5.215922702540206e-06,
+      "loss": 0.9636,
+      "step": 24360
+    },
+    {
+      "epoch": 0.669056054489028,
+      "grad_norm": 0.0,
+      "learning_rate": 5.215141603809051e-06,
+      "loss": 0.8182,
+      "step": 24361
+    },
+    {
+      "epoch": 0.6690835187168713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2143605429375535e-06,
+      "loss": 0.8949,
+      "step": 24362
+    },
+    {
+      "epoch": 0.6691109829447145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.21357951993189e-06,
+      "loss": 0.794,
+      "step": 24363
+    },
+    {
+      "epoch": 0.6691384471725578,
+      "grad_norm": 0.0,
+      "learning_rate": 5.212798534798235e-06,
+      "loss": 0.8302,
+      "step": 24364
+    },
+    {
+      "epoch": 0.6691659114004009,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2120175875427726e-06,
+      "loss": 0.8277,
+      "step": 24365
+    },
+    {
+      "epoch": 0.6691933756282442,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2112366781716805e-06,
+      "loss": 0.7338,
+      "step": 24366
+    },
+    {
+      "epoch": 0.6692208398560875,
+      "grad_norm": 0.0,
+      "learning_rate": 5.21045580669114e-06,
+      "loss": 0.8423,
+      "step": 24367
+    },
+    {
+      "epoch": 0.6692483040839307,
+      "grad_norm": 0.0,
+      "learning_rate": 5.209674973107333e-06,
+      "loss": 0.7932,
+      "step": 24368
+    },
+    {
+      "epoch": 0.6692757683117739,
+      "grad_norm": 0.0,
+      "learning_rate": 5.208894177426427e-06,
+      "loss": 0.7794,
+      "step": 24369
+    },
+    {
+      "epoch": 0.6693032325396171,
+      "grad_norm": 0.0,
+      "learning_rate": 5.208113419654608e-06,
+      "loss": 0.8698,
+      "step": 24370
+    },
+    {
+      "epoch": 0.6693306967674604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.207332699798055e-06,
+      "loss": 0.9226,
+      "step": 24371
+    },
+    {
+      "epoch": 0.6693581609953037,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2065520178629395e-06,
+      "loss": 0.8256,
+      "step": 24372
+    },
+    {
+      "epoch": 0.6693856252231468,
+      "grad_norm": 0.0,
+      "learning_rate": 5.205771373855439e-06,
+      "loss": 0.792,
+      "step": 24373
+    },
+    {
+      "epoch": 0.6694130894509901,
+      "grad_norm": 0.0,
+      "learning_rate": 5.204990767781739e-06,
+      "loss": 0.924,
+      "step": 24374
+    },
+    {
+      "epoch": 0.6694405536788334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.204210199648003e-06,
+      "loss": 0.8865,
+      "step": 24375
+    },
+    {
+      "epoch": 0.6694680179066765,
+      "grad_norm": 0.0,
+      "learning_rate": 5.203429669460416e-06,
+      "loss": 0.878,
+      "step": 24376
+    },
+    {
+      "epoch": 0.6694954821345198,
+      "grad_norm": 0.0,
+      "learning_rate": 5.20264917722515e-06,
+      "loss": 0.903,
+      "step": 24377
+    },
+    {
+      "epoch": 0.669522946362363,
+      "grad_norm": 0.0,
+      "learning_rate": 5.201868722948386e-06,
+      "loss": 0.8968,
+      "step": 24378
+    },
+    {
+      "epoch": 0.6695504105902063,
+      "grad_norm": 0.0,
+      "learning_rate": 5.201088306636293e-06,
+      "loss": 0.7956,
+      "step": 24379
+    },
+    {
+      "epoch": 0.6695778748180495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2003079282950495e-06,
+      "loss": 0.7433,
+      "step": 24380
+    },
+    {
+      "epoch": 0.6696053390458927,
+      "grad_norm": 0.0,
+      "learning_rate": 5.199527587930826e-06,
+      "loss": 0.8513,
+      "step": 24381
+    },
+    {
+      "epoch": 0.669632803273736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.198747285549801e-06,
+      "loss": 0.7914,
+      "step": 24382
+    },
+    {
+      "epoch": 0.6696602675015791,
+      "grad_norm": 0.0,
+      "learning_rate": 5.197967021158146e-06,
+      "loss": 0.8379,
+      "step": 24383
+    },
+    {
+      "epoch": 0.6696877317294224,
+      "grad_norm": 0.0,
+      "learning_rate": 5.197186794762037e-06,
+      "loss": 0.9076,
+      "step": 24384
+    },
+    {
+      "epoch": 0.6697151959572657,
+      "grad_norm": 0.0,
+      "learning_rate": 5.196406606367646e-06,
+      "loss": 0.8549,
+      "step": 24385
+    },
+    {
+      "epoch": 0.6697426601851089,
+      "grad_norm": 0.0,
+      "learning_rate": 5.195626455981149e-06,
+      "loss": 0.7742,
+      "step": 24386
+    },
+    {
+      "epoch": 0.6697701244129521,
+      "grad_norm": 0.0,
+      "learning_rate": 5.194846343608719e-06,
+      "loss": 0.8537,
+      "step": 24387
+    },
+    {
+      "epoch": 0.6697975886407954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.194066269256522e-06,
+      "loss": 0.8911,
+      "step": 24388
+    },
+    {
+      "epoch": 0.6698250528686386,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1932862329307324e-06,
+      "loss": 0.8054,
+      "step": 24389
+    },
+    {
+      "epoch": 0.6698525170964819,
+      "grad_norm": 0.0,
+      "learning_rate": 5.192506234637525e-06,
+      "loss": 0.853,
+      "step": 24390
+    },
+    {
+      "epoch": 0.669879981324325,
+      "grad_norm": 0.0,
+      "learning_rate": 5.191726274383071e-06,
+      "loss": 0.8584,
+      "step": 24391
+    },
+    {
+      "epoch": 0.6699074455521683,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1909463521735445e-06,
+      "loss": 0.7687,
+      "step": 24392
+    },
+    {
+      "epoch": 0.6699349097800116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.19016646801511e-06,
+      "loss": 0.8368,
+      "step": 24393
+    },
+    {
+      "epoch": 0.6699623740078547,
+      "grad_norm": 0.0,
+      "learning_rate": 5.189386621913941e-06,
+      "loss": 0.8502,
+      "step": 24394
+    },
+    {
+      "epoch": 0.669989838235698,
+      "grad_norm": 0.0,
+      "learning_rate": 5.188606813876212e-06,
+      "loss": 0.8676,
+      "step": 24395
+    },
+    {
+      "epoch": 0.6700173024635412,
+      "grad_norm": 0.0,
+      "learning_rate": 5.187827043908087e-06,
+      "loss": 0.9104,
+      "step": 24396
+    },
+    {
+      "epoch": 0.6700447666913845,
+      "grad_norm": 0.0,
+      "learning_rate": 5.187047312015738e-06,
+      "loss": 0.8766,
+      "step": 24397
+    },
+    {
+      "epoch": 0.6700722309192277,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1862676182053384e-06,
+      "loss": 0.7983,
+      "step": 24398
+    },
+    {
+      "epoch": 0.6700996951470709,
+      "grad_norm": 0.0,
+      "learning_rate": 5.185487962483051e-06,
+      "loss": 0.8349,
+      "step": 24399
+    },
+    {
+      "epoch": 0.6701271593749142,
+      "grad_norm": 0.0,
+      "learning_rate": 5.184708344855047e-06,
+      "loss": 0.8533,
+      "step": 24400
+    },
+    {
+      "epoch": 0.6701546236027575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.183928765327497e-06,
+      "loss": 0.7288,
+      "step": 24401
+    },
+    {
+      "epoch": 0.6701820878306006,
+      "grad_norm": 0.0,
+      "learning_rate": 5.183149223906572e-06,
+      "loss": 0.7829,
+      "step": 24402
+    },
+    {
+      "epoch": 0.6702095520584439,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1823697205984305e-06,
+      "loss": 0.8882,
+      "step": 24403
+    },
+    {
+      "epoch": 0.6702370162862871,
+      "grad_norm": 0.0,
+      "learning_rate": 5.181590255409251e-06,
+      "loss": 0.8511,
+      "step": 24404
+    },
+    {
+      "epoch": 0.6702644805141303,
+      "grad_norm": 0.0,
+      "learning_rate": 5.180810828345193e-06,
+      "loss": 0.8665,
+      "step": 24405
+    },
+    {
+      "epoch": 0.6702919447419736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.180031439412425e-06,
+      "loss": 0.8802,
+      "step": 24406
+    },
+    {
+      "epoch": 0.6703194089698168,
+      "grad_norm": 0.0,
+      "learning_rate": 5.179252088617115e-06,
+      "loss": 0.8099,
+      "step": 24407
+    },
+    {
+      "epoch": 0.6703468731976601,
+      "grad_norm": 0.0,
+      "learning_rate": 5.178472775965431e-06,
+      "loss": 0.8504,
+      "step": 24408
+    },
+    {
+      "epoch": 0.6703743374255032,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1776935014635375e-06,
+      "loss": 0.8572,
+      "step": 24409
+    },
+    {
+      "epoch": 0.6704018016533465,
+      "grad_norm": 0.0,
+      "learning_rate": 5.176914265117605e-06,
+      "loss": 0.7832,
+      "step": 24410
+    },
+    {
+      "epoch": 0.6704292658811898,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1761350669337965e-06,
+      "loss": 0.8873,
+      "step": 24411
+    },
+    {
+      "epoch": 0.670456730109033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1753559069182705e-06,
+      "loss": 0.8518,
+      "step": 24412
+    },
+    {
+      "epoch": 0.6704841943368762,
+      "grad_norm": 0.0,
+      "learning_rate": 5.174576785077196e-06,
+      "loss": 0.8111,
+      "step": 24413
+    },
+    {
+      "epoch": 0.6705116585647195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.173797701416743e-06,
+      "loss": 0.828,
+      "step": 24414
+    },
+    {
+      "epoch": 0.6705391227925627,
+      "grad_norm": 0.0,
+      "learning_rate": 5.173018655943069e-06,
+      "loss": 0.8388,
+      "step": 24415
+    },
+    {
+      "epoch": 0.670566587020406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.172239648662346e-06,
+      "loss": 0.8151,
+      "step": 24416
+    },
+    {
+      "epoch": 0.6705940512482491,
+      "grad_norm": 0.0,
+      "learning_rate": 5.171460679580729e-06,
+      "loss": 0.8937,
+      "step": 24417
+    },
+    {
+      "epoch": 0.6706215154760924,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1706817487043856e-06,
+      "loss": 0.82,
+      "step": 24418
+    },
+    {
+      "epoch": 0.6706489797039357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.169902856039484e-06,
+      "loss": 0.8576,
+      "step": 24419
+    },
+    {
+      "epoch": 0.6706764439317788,
+      "grad_norm": 0.0,
+      "learning_rate": 5.169124001592176e-06,
+      "loss": 0.8857,
+      "step": 24420
+    },
+    {
+      "epoch": 0.6707039081596221,
+      "grad_norm": 0.0,
+      "learning_rate": 5.168345185368635e-06,
+      "loss": 0.9073,
+      "step": 24421
+    },
+    {
+      "epoch": 0.6707313723874653,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1675664073750155e-06,
+      "loss": 0.8307,
+      "step": 24422
+    },
+    {
+      "epoch": 0.6707588366153086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.166787667617481e-06,
+      "loss": 0.8833,
+      "step": 24423
+    },
+    {
+      "epoch": 0.6707863008431518,
+      "grad_norm": 0.0,
+      "learning_rate": 5.166008966102196e-06,
+      "loss": 0.8188,
+      "step": 24424
+    },
+    {
+      "epoch": 0.670813765070995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.165230302835319e-06,
+      "loss": 0.843,
+      "step": 24425
+    },
+    {
+      "epoch": 0.6708412292988383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.164451677823017e-06,
+      "loss": 0.8727,
+      "step": 24426
+    },
+    {
+      "epoch": 0.6708686935266815,
+      "grad_norm": 0.0,
+      "learning_rate": 5.163673091071448e-06,
+      "loss": 0.8659,
+      "step": 24427
+    },
+    {
+      "epoch": 0.6708961577545247,
+      "grad_norm": 0.0,
+      "learning_rate": 5.162894542586765e-06,
+      "loss": 0.7893,
+      "step": 24428
+    },
+    {
+      "epoch": 0.670923621982368,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1621160323751355e-06,
+      "loss": 0.9391,
+      "step": 24429
+    },
+    {
+      "epoch": 0.6709510862102112,
+      "grad_norm": 0.0,
+      "learning_rate": 5.161337560442718e-06,
+      "loss": 0.8787,
+      "step": 24430
+    },
+    {
+      "epoch": 0.6709785504380544,
+      "grad_norm": 0.0,
+      "learning_rate": 5.160559126795671e-06,
+      "loss": 0.8128,
+      "step": 24431
+    },
+    {
+      "epoch": 0.6710060146658977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.159780731440156e-06,
+      "loss": 0.8535,
+      "step": 24432
+    },
+    {
+      "epoch": 0.6710334788937409,
+      "grad_norm": 0.0,
+      "learning_rate": 5.159002374382333e-06,
+      "loss": 0.8228,
+      "step": 24433
+    },
+    {
+      "epoch": 0.6710609431215842,
+      "grad_norm": 0.0,
+      "learning_rate": 5.158224055628356e-06,
+      "loss": 0.7936,
+      "step": 24434
+    },
+    {
+      "epoch": 0.6710884073494273,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1574457751843886e-06,
+      "loss": 0.9166,
+      "step": 24435
+    },
+    {
+      "epoch": 0.6711158715772706,
+      "grad_norm": 0.0,
+      "learning_rate": 5.156667533056583e-06,
+      "loss": 0.8864,
+      "step": 24436
+    },
+    {
+      "epoch": 0.6711433358051139,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1558893292511e-06,
+      "loss": 0.8804,
+      "step": 24437
+    },
+    {
+      "epoch": 0.671170800032957,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1551111637740966e-06,
+      "loss": 0.9355,
+      "step": 24438
+    },
+    {
+      "epoch": 0.6711982642608003,
+      "grad_norm": 0.0,
+      "learning_rate": 5.154333036631736e-06,
+      "loss": 0.8062,
+      "step": 24439
+    },
+    {
+      "epoch": 0.6712257284886436,
+      "grad_norm": 0.0,
+      "learning_rate": 5.153554947830164e-06,
+      "loss": 0.8449,
+      "step": 24440
+    },
+    {
+      "epoch": 0.6712531927164868,
+      "grad_norm": 0.0,
+      "learning_rate": 5.152776897375542e-06,
+      "loss": 0.8112,
+      "step": 24441
+    },
+    {
+      "epoch": 0.67128065694433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.151998885274028e-06,
+      "loss": 0.8963,
+      "step": 24442
+    },
+    {
+      "epoch": 0.6713081211721732,
+      "grad_norm": 0.0,
+      "learning_rate": 5.15122091153178e-06,
+      "loss": 0.8359,
+      "step": 24443
+    },
+    {
+      "epoch": 0.6713355854000165,
+      "grad_norm": 0.0,
+      "learning_rate": 5.150442976154946e-06,
+      "loss": 0.8285,
+      "step": 24444
+    },
+    {
+      "epoch": 0.6713630496278598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.14966507914969e-06,
+      "loss": 0.9212,
+      "step": 24445
+    },
+    {
+      "epoch": 0.6713905138557029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.14888722052216e-06,
+      "loss": 0.8978,
+      "step": 24446
+    },
+    {
+      "epoch": 0.6714179780835462,
+      "grad_norm": 0.0,
+      "learning_rate": 5.148109400278511e-06,
+      "loss": 0.8074,
+      "step": 24447
+    },
+    {
+      "epoch": 0.6714454423113894,
+      "grad_norm": 0.0,
+      "learning_rate": 5.147331618424902e-06,
+      "loss": 0.8091,
+      "step": 24448
+    },
+    {
+      "epoch": 0.6714729065392326,
+      "grad_norm": 0.0,
+      "learning_rate": 5.146553874967485e-06,
+      "loss": 0.8449,
+      "step": 24449
+    },
+    {
+      "epoch": 0.6715003707670759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.145776169912417e-06,
+      "loss": 0.8354,
+      "step": 24450
+    },
+    {
+      "epoch": 0.6715278349949191,
+      "grad_norm": 0.0,
+      "learning_rate": 5.144998503265848e-06,
+      "loss": 0.8934,
+      "step": 24451
+    },
+    {
+      "epoch": 0.6715552992227624,
+      "grad_norm": 0.0,
+      "learning_rate": 5.144220875033928e-06,
+      "loss": 0.895,
+      "step": 24452
+    },
+    {
+      "epoch": 0.6715827634506056,
+      "grad_norm": 0.0,
+      "learning_rate": 5.143443285222814e-06,
+      "loss": 0.8633,
+      "step": 24453
+    },
+    {
+      "epoch": 0.6716102276784488,
+      "grad_norm": 0.0,
+      "learning_rate": 5.142665733838657e-06,
+      "loss": 0.8679,
+      "step": 24454
+    },
+    {
+      "epoch": 0.6716376919062921,
+      "grad_norm": 0.0,
+      "learning_rate": 5.14188822088761e-06,
+      "loss": 0.9017,
+      "step": 24455
+    },
+    {
+      "epoch": 0.6716651561341352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1411107463758256e-06,
+      "loss": 0.8282,
+      "step": 24456
+    },
+    {
+      "epoch": 0.6716926203619785,
+      "grad_norm": 0.0,
+      "learning_rate": 5.14033331030946e-06,
+      "loss": 0.8139,
+      "step": 24457
+    },
+    {
+      "epoch": 0.6717200845898218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.139555912694655e-06,
+      "loss": 0.9114,
+      "step": 24458
+    },
+    {
+      "epoch": 0.671747548817665,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1387785535375714e-06,
+      "loss": 0.8024,
+      "step": 24459
+    },
+    {
+      "epoch": 0.6717750130455082,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1380012328443495e-06,
+      "loss": 0.8593,
+      "step": 24460
+    },
+    {
+      "epoch": 0.6718024772733514,
+      "grad_norm": 0.0,
+      "learning_rate": 5.137223950621145e-06,
+      "loss": 0.8542,
+      "step": 24461
+    },
+    {
+      "epoch": 0.6718299415011947,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1364467068741095e-06,
+      "loss": 0.9117,
+      "step": 24462
+    },
+    {
+      "epoch": 0.671857405729038,
+      "grad_norm": 0.0,
+      "learning_rate": 5.135669501609395e-06,
+      "loss": 0.8078,
+      "step": 24463
+    },
+    {
+      "epoch": 0.6718848699568811,
+      "grad_norm": 0.0,
+      "learning_rate": 5.134892334833143e-06,
+      "loss": 0.8714,
+      "step": 24464
+    },
+    {
+      "epoch": 0.6719123341847244,
+      "grad_norm": 0.0,
+      "learning_rate": 5.134115206551509e-06,
+      "loss": 0.847,
+      "step": 24465
+    },
+    {
+      "epoch": 0.6719397984125677,
+      "grad_norm": 0.0,
+      "learning_rate": 5.133338116770643e-06,
+      "loss": 0.76,
+      "step": 24466
+    },
+    {
+      "epoch": 0.6719672626404108,
+      "grad_norm": 0.0,
+      "learning_rate": 5.132561065496688e-06,
+      "loss": 0.8283,
+      "step": 24467
+    },
+    {
+      "epoch": 0.6719947268682541,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1317840527357945e-06,
+      "loss": 0.9177,
+      "step": 24468
+    },
+    {
+      "epoch": 0.6720221910960973,
+      "grad_norm": 0.0,
+      "learning_rate": 5.131007078494114e-06,
+      "loss": 0.868,
+      "step": 24469
+    },
+    {
+      "epoch": 0.6720496553239406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1302301427777904e-06,
+      "loss": 0.8197,
+      "step": 24470
+    },
+    {
+      "epoch": 0.6720771195517838,
+      "grad_norm": 0.0,
+      "learning_rate": 5.12945324559297e-06,
+      "loss": 0.8959,
+      "step": 24471
+    },
+    {
+      "epoch": 0.672104583779627,
+      "grad_norm": 0.0,
+      "learning_rate": 5.128676386945805e-06,
+      "loss": 0.8011,
+      "step": 24472
+    },
+    {
+      "epoch": 0.6721320480074703,
+      "grad_norm": 0.0,
+      "learning_rate": 5.127899566842437e-06,
+      "loss": 1.0026,
+      "step": 24473
+    },
+    {
+      "epoch": 0.6721595122353134,
+      "grad_norm": 0.0,
+      "learning_rate": 5.127122785289019e-06,
+      "loss": 0.899,
+      "step": 24474
+    },
+    {
+      "epoch": 0.6721869764631567,
+      "grad_norm": 0.0,
+      "learning_rate": 5.126346042291695e-06,
+      "loss": 0.8789,
+      "step": 24475
+    },
+    {
+      "epoch": 0.672214440691,
+      "grad_norm": 0.0,
+      "learning_rate": 5.125569337856602e-06,
+      "loss": 0.8031,
+      "step": 24476
+    },
+    {
+      "epoch": 0.6722419049188432,
+      "grad_norm": 0.0,
+      "learning_rate": 5.124792671989896e-06,
+      "loss": 0.8304,
+      "step": 24477
+    },
+    {
+      "epoch": 0.6722693691466864,
+      "grad_norm": 0.0,
+      "learning_rate": 5.124016044697716e-06,
+      "loss": 0.8437,
+      "step": 24478
+    },
+    {
+      "epoch": 0.6722968333745297,
+      "grad_norm": 0.0,
+      "learning_rate": 5.123239455986211e-06,
+      "loss": 0.8363,
+      "step": 24479
+    },
+    {
+      "epoch": 0.6723242976023729,
+      "grad_norm": 0.0,
+      "learning_rate": 5.122462905861528e-06,
+      "loss": 0.8263,
+      "step": 24480
+    },
+    {
+      "epoch": 0.6723517618302162,
+      "grad_norm": 0.0,
+      "learning_rate": 5.121686394329804e-06,
+      "loss": 0.8668,
+      "step": 24481
+    },
+    {
+      "epoch": 0.6723792260580593,
+      "grad_norm": 0.0,
+      "learning_rate": 5.120909921397188e-06,
+      "loss": 0.8757,
+      "step": 24482
+    },
+    {
+      "epoch": 0.6724066902859026,
+      "grad_norm": 0.0,
+      "learning_rate": 5.120133487069825e-06,
+      "loss": 0.8738,
+      "step": 24483
+    },
+    {
+      "epoch": 0.6724341545137459,
+      "grad_norm": 0.0,
+      "learning_rate": 5.119357091353854e-06,
+      "loss": 0.7991,
+      "step": 24484
+    },
+    {
+      "epoch": 0.672461618741589,
+      "grad_norm": 0.0,
+      "learning_rate": 5.118580734255419e-06,
+      "loss": 0.8748,
+      "step": 24485
+    },
+    {
+      "epoch": 0.6724890829694323,
+      "grad_norm": 0.0,
+      "learning_rate": 5.117804415780667e-06,
+      "loss": 0.8165,
+      "step": 24486
+    },
+    {
+      "epoch": 0.6725165471972755,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1170281359357355e-06,
+      "loss": 0.8962,
+      "step": 24487
+    },
+    {
+      "epoch": 0.6725440114251188,
+      "grad_norm": 0.0,
+      "learning_rate": 5.116251894726769e-06,
+      "loss": 0.8573,
+      "step": 24488
+    },
+    {
+      "epoch": 0.672571475652962,
+      "grad_norm": 0.0,
+      "learning_rate": 5.115475692159908e-06,
+      "loss": 0.8647,
+      "step": 24489
+    },
+    {
+      "epoch": 0.6725989398808052,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1146995282413005e-06,
+      "loss": 0.9282,
+      "step": 24490
+    },
+    {
+      "epoch": 0.6726264041086485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.113923402977079e-06,
+      "loss": 0.865,
+      "step": 24491
+    },
+    {
+      "epoch": 0.6726538683364918,
+      "grad_norm": 0.0,
+      "learning_rate": 5.11314731637339e-06,
+      "loss": 0.8139,
+      "step": 24492
+    },
+    {
+      "epoch": 0.6726813325643349,
+      "grad_norm": 0.0,
+      "learning_rate": 5.112371268436371e-06,
+      "loss": 0.7504,
+      "step": 24493
+    },
+    {
+      "epoch": 0.6727087967921782,
+      "grad_norm": 0.0,
+      "learning_rate": 5.111595259172162e-06,
+      "loss": 0.8654,
+      "step": 24494
+    },
+    {
+      "epoch": 0.6727362610200214,
+      "grad_norm": 0.0,
+      "learning_rate": 5.110819288586906e-06,
+      "loss": 0.8959,
+      "step": 24495
+    },
+    {
+      "epoch": 0.6727637252478647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.110043356686742e-06,
+      "loss": 0.8343,
+      "step": 24496
+    },
+    {
+      "epoch": 0.6727911894757079,
+      "grad_norm": 0.0,
+      "learning_rate": 5.109267463477807e-06,
+      "loss": 0.9157,
+      "step": 24497
+    },
+    {
+      "epoch": 0.6728186537035511,
+      "grad_norm": 0.0,
+      "learning_rate": 5.108491608966248e-06,
+      "loss": 0.7955,
+      "step": 24498
+    },
+    {
+      "epoch": 0.6728461179313944,
+      "grad_norm": 0.0,
+      "learning_rate": 5.107715793158199e-06,
+      "loss": 0.8985,
+      "step": 24499
+    },
+    {
+      "epoch": 0.6728735821592375,
+      "grad_norm": 0.0,
+      "learning_rate": 5.106940016059792e-06,
+      "loss": 0.8563,
+      "step": 24500
+    },
+    {
+      "epoch": 0.6729010463870808,
+      "grad_norm": 0.0,
+      "learning_rate": 5.106164277677173e-06,
+      "loss": 0.8352,
+      "step": 24501
+    },
+    {
+      "epoch": 0.6729285106149241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.105388578016476e-06,
+      "loss": 0.7979,
+      "step": 24502
+    },
+    {
+      "epoch": 0.6729559748427673,
+      "grad_norm": 0.0,
+      "learning_rate": 5.104612917083841e-06,
+      "loss": 0.8015,
+      "step": 24503
+    },
+    {
+      "epoch": 0.6729834390706105,
+      "grad_norm": 0.0,
+      "learning_rate": 5.103837294885409e-06,
+      "loss": 0.9677,
+      "step": 24504
+    },
+    {
+      "epoch": 0.6730109032984538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.103061711427309e-06,
+      "loss": 0.8125,
+      "step": 24505
+    },
+    {
+      "epoch": 0.673038367526297,
+      "grad_norm": 0.0,
+      "learning_rate": 5.102286166715682e-06,
+      "loss": 0.9082,
+      "step": 24506
+    },
+    {
+      "epoch": 0.6730658317541403,
+      "grad_norm": 0.0,
+      "learning_rate": 5.101510660756667e-06,
+      "loss": 0.834,
+      "step": 24507
+    },
+    {
+      "epoch": 0.6730932959819834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.100735193556394e-06,
+      "loss": 0.8426,
+      "step": 24508
+    },
+    {
+      "epoch": 0.6731207602098267,
+      "grad_norm": 0.0,
+      "learning_rate": 5.099959765121001e-06,
+      "loss": 0.7598,
+      "step": 24509
+    },
+    {
+      "epoch": 0.67314822443767,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0991843754566295e-06,
+      "loss": 0.8264,
+      "step": 24510
+    },
+    {
+      "epoch": 0.6731756886655131,
+      "grad_norm": 0.0,
+      "learning_rate": 5.098409024569405e-06,
+      "loss": 0.9261,
+      "step": 24511
+    },
+    {
+      "epoch": 0.6732031528933564,
+      "grad_norm": 0.0,
+      "learning_rate": 5.097633712465469e-06,
+      "loss": 0.844,
+      "step": 24512
+    },
+    {
+      "epoch": 0.6732306171211997,
+      "grad_norm": 0.0,
+      "learning_rate": 5.096858439150951e-06,
+      "loss": 0.8994,
+      "step": 24513
+    },
+    {
+      "epoch": 0.6732580813490429,
+      "grad_norm": 0.0,
+      "learning_rate": 5.096083204631994e-06,
+      "loss": 0.937,
+      "step": 24514
+    },
+    {
+      "epoch": 0.6732855455768861,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0953080089147215e-06,
+      "loss": 0.8148,
+      "step": 24515
+    },
+    {
+      "epoch": 0.6733130098047293,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0945328520052764e-06,
+      "loss": 0.9165,
+      "step": 24516
+    },
+    {
+      "epoch": 0.6733404740325726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.093757733909783e-06,
+      "loss": 0.9366,
+      "step": 24517
+    },
+    {
+      "epoch": 0.6733679382604159,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0929826546343796e-06,
+      "loss": 0.763,
+      "step": 24518
+    },
+    {
+      "epoch": 0.673395402488259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.092207614185199e-06,
+      "loss": 0.9418,
+      "step": 24519
+    },
+    {
+      "epoch": 0.6734228667161023,
+      "grad_norm": 0.0,
+      "learning_rate": 5.091432612568372e-06,
+      "loss": 0.8318,
+      "step": 24520
+    },
+    {
+      "epoch": 0.6734503309439455,
+      "grad_norm": 0.0,
+      "learning_rate": 5.090657649790032e-06,
+      "loss": 0.9146,
+      "step": 24521
+    },
+    {
+      "epoch": 0.6734777951717887,
+      "grad_norm": 0.0,
+      "learning_rate": 5.089882725856312e-06,
+      "loss": 0.8343,
+      "step": 24522
+    },
+    {
+      "epoch": 0.673505259399632,
+      "grad_norm": 0.0,
+      "learning_rate": 5.089107840773344e-06,
+      "loss": 0.7428,
+      "step": 24523
+    },
+    {
+      "epoch": 0.6735327236274752,
+      "grad_norm": 0.0,
+      "learning_rate": 5.088332994547254e-06,
+      "loss": 0.8002,
+      "step": 24524
+    },
+    {
+      "epoch": 0.6735601878553185,
+      "grad_norm": 0.0,
+      "learning_rate": 5.087558187184174e-06,
+      "loss": 0.7091,
+      "step": 24525
+    },
+    {
+      "epoch": 0.6735876520831617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.086783418690238e-06,
+      "loss": 0.8169,
+      "step": 24526
+    },
+    {
+      "epoch": 0.6736151163110049,
+      "grad_norm": 0.0,
+      "learning_rate": 5.086008689071573e-06,
+      "loss": 0.7659,
+      "step": 24527
+    },
+    {
+      "epoch": 0.6736425805388482,
+      "grad_norm": 0.0,
+      "learning_rate": 5.085233998334315e-06,
+      "loss": 0.9119,
+      "step": 24528
+    },
+    {
+      "epoch": 0.6736700447666913,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0844593464845874e-06,
+      "loss": 0.943,
+      "step": 24529
+    },
+    {
+      "epoch": 0.6736975089945346,
+      "grad_norm": 0.0,
+      "learning_rate": 5.08368473352852e-06,
+      "loss": 0.8655,
+      "step": 24530
+    },
+    {
+      "epoch": 0.6737249732223779,
+      "grad_norm": 0.0,
+      "learning_rate": 5.082910159472247e-06,
+      "loss": 0.8491,
+      "step": 24531
+    },
+    {
+      "epoch": 0.6737524374502211,
+      "grad_norm": 0.0,
+      "learning_rate": 5.082135624321891e-06,
+      "loss": 0.8105,
+      "step": 24532
+    },
+    {
+      "epoch": 0.6737799016780643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0813611280835824e-06,
+      "loss": 0.7932,
+      "step": 24533
+    },
+    {
+      "epoch": 0.6738073659059075,
+      "grad_norm": 0.0,
+      "learning_rate": 5.080586670763455e-06,
+      "loss": 0.7964,
+      "step": 24534
+    },
+    {
+      "epoch": 0.6738348301337508,
+      "grad_norm": 0.0,
+      "learning_rate": 5.079812252367625e-06,
+      "loss": 0.8158,
+      "step": 24535
+    },
+    {
+      "epoch": 0.6738622943615941,
+      "grad_norm": 0.0,
+      "learning_rate": 5.079037872902228e-06,
+      "loss": 0.8162,
+      "step": 24536
+    },
+    {
+      "epoch": 0.6738897585894372,
+      "grad_norm": 0.0,
+      "learning_rate": 5.078263532373389e-06,
+      "loss": 0.8843,
+      "step": 24537
+    },
+    {
+      "epoch": 0.6739172228172805,
+      "grad_norm": 0.0,
+      "learning_rate": 5.077489230787238e-06,
+      "loss": 0.9171,
+      "step": 24538
+    },
+    {
+      "epoch": 0.6739446870451238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.076714968149897e-06,
+      "loss": 0.8783,
+      "step": 24539
+    },
+    {
+      "epoch": 0.673972151272967,
+      "grad_norm": 0.0,
+      "learning_rate": 5.075940744467498e-06,
+      "loss": 0.8686,
+      "step": 24540
+    },
+    {
+      "epoch": 0.6739996155008102,
+      "grad_norm": 0.0,
+      "learning_rate": 5.075166559746159e-06,
+      "loss": 0.7974,
+      "step": 24541
+    },
+    {
+      "epoch": 0.6740270797286534,
+      "grad_norm": 0.0,
+      "learning_rate": 5.07439241399201e-06,
+      "loss": 0.7906,
+      "step": 24542
+    },
+    {
+      "epoch": 0.6740545439564967,
+      "grad_norm": 0.0,
+      "learning_rate": 5.073618307211177e-06,
+      "loss": 0.843,
+      "step": 24543
+    },
+    {
+      "epoch": 0.6740820081843399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.072844239409782e-06,
+      "loss": 0.9019,
+      "step": 24544
+    },
+    {
+      "epoch": 0.6741094724121831,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0720702105939575e-06,
+      "loss": 0.7714,
+      "step": 24545
+    },
+    {
+      "epoch": 0.6741369366400264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.071296220769818e-06,
+      "loss": 0.8745,
+      "step": 24546
+    },
+    {
+      "epoch": 0.6741644008678696,
+      "grad_norm": 0.0,
+      "learning_rate": 5.070522269943497e-06,
+      "loss": 0.8311,
+      "step": 24547
+    },
+    {
+      "epoch": 0.6741918650957128,
+      "grad_norm": 0.0,
+      "learning_rate": 5.069748358121107e-06,
+      "loss": 0.8527,
+      "step": 24548
+    },
+    {
+      "epoch": 0.6742193293235561,
+      "grad_norm": 0.0,
+      "learning_rate": 5.06897448530878e-06,
+      "loss": 0.7925,
+      "step": 24549
+    },
+    {
+      "epoch": 0.6742467935513993,
+      "grad_norm": 0.0,
+      "learning_rate": 5.068200651512636e-06,
+      "loss": 0.9152,
+      "step": 24550
+    },
+    {
+      "epoch": 0.6742742577792425,
+      "grad_norm": 0.0,
+      "learning_rate": 5.067426856738804e-06,
+      "loss": 0.8497,
+      "step": 24551
+    },
+    {
+      "epoch": 0.6743017220070858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.066653100993395e-06,
+      "loss": 0.7971,
+      "step": 24552
+    },
+    {
+      "epoch": 0.674329186234929,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0658793842825395e-06,
+      "loss": 0.9801,
+      "step": 24553
+    },
+    {
+      "epoch": 0.6743566504627723,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0651057066123576e-06,
+      "loss": 0.859,
+      "step": 24554
+    },
+    {
+      "epoch": 0.6743841146906154,
+      "grad_norm": 0.0,
+      "learning_rate": 5.064332067988975e-06,
+      "loss": 0.8584,
+      "step": 24555
+    },
+    {
+      "epoch": 0.6744115789184587,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0635584684185035e-06,
+      "loss": 0.8426,
+      "step": 24556
+    },
+    {
+      "epoch": 0.674439043146302,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0627849079070745e-06,
+      "loss": 0.9145,
+      "step": 24557
+    },
+    {
+      "epoch": 0.6744665073741452,
+      "grad_norm": 0.0,
+      "learning_rate": 5.062011386460801e-06,
+      "loss": 0.7873,
+      "step": 24558
+    },
+    {
+      "epoch": 0.6744939716019884,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0612379040858055e-06,
+      "loss": 0.8546,
+      "step": 24559
+    },
+    {
+      "epoch": 0.6745214358298316,
+      "grad_norm": 0.0,
+      "learning_rate": 5.060464460788208e-06,
+      "loss": 0.8654,
+      "step": 24560
+    },
+    {
+      "epoch": 0.6745489000576749,
+      "grad_norm": 0.0,
+      "learning_rate": 5.05969105657413e-06,
+      "loss": 0.7362,
+      "step": 24561
+    },
+    {
+      "epoch": 0.6745763642855181,
+      "grad_norm": 0.0,
+      "learning_rate": 5.058917691449693e-06,
+      "loss": 0.8402,
+      "step": 24562
+    },
+    {
+      "epoch": 0.6746038285133613,
+      "grad_norm": 0.0,
+      "learning_rate": 5.058144365421014e-06,
+      "loss": 0.8147,
+      "step": 24563
+    },
+    {
+      "epoch": 0.6746312927412046,
+      "grad_norm": 0.0,
+      "learning_rate": 5.057371078494207e-06,
+      "loss": 0.8332,
+      "step": 24564
+    },
+    {
+      "epoch": 0.6746587569690479,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0565978306753935e-06,
+      "loss": 0.7993,
+      "step": 24565
+    },
+    {
+      "epoch": 0.674686221196891,
+      "grad_norm": 0.0,
+      "learning_rate": 5.055824621970695e-06,
+      "loss": 0.8327,
+      "step": 24566
+    },
+    {
+      "epoch": 0.6747136854247343,
+      "grad_norm": 0.0,
+      "learning_rate": 5.055051452386226e-06,
+      "loss": 0.8653,
+      "step": 24567
+    },
+    {
+      "epoch": 0.6747411496525775,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0542783219281055e-06,
+      "loss": 0.8783,
+      "step": 24568
+    },
+    {
+      "epoch": 0.6747686138804208,
+      "grad_norm": 0.0,
+      "learning_rate": 5.053505230602454e-06,
+      "loss": 0.8083,
+      "step": 24569
+    },
+    {
+      "epoch": 0.674796078108264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.052732178415383e-06,
+      "loss": 0.9282,
+      "step": 24570
+    },
+    {
+      "epoch": 0.6748235423361072,
+      "grad_norm": 0.0,
+      "learning_rate": 5.051959165373014e-06,
+      "loss": 0.8156,
+      "step": 24571
+    },
+    {
+      "epoch": 0.6748510065639505,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0511861914814565e-06,
+      "loss": 0.8926,
+      "step": 24572
+    },
+    {
+      "epoch": 0.6748784707917936,
+      "grad_norm": 0.0,
+      "learning_rate": 5.050413256746831e-06,
+      "loss": 0.8548,
+      "step": 24573
+    },
+    {
+      "epoch": 0.6749059350196369,
+      "grad_norm": 0.0,
+      "learning_rate": 5.049640361175255e-06,
+      "loss": 0.9257,
+      "step": 24574
+    },
+    {
+      "epoch": 0.6749333992474802,
+      "grad_norm": 0.0,
+      "learning_rate": 5.048867504772843e-06,
+      "loss": 0.7777,
+      "step": 24575
+    },
+    {
+      "epoch": 0.6749608634753234,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0480946875457085e-06,
+      "loss": 0.8007,
+      "step": 24576
+    },
+    {
+      "epoch": 0.6749883277031666,
+      "grad_norm": 0.0,
+      "learning_rate": 5.047321909499965e-06,
+      "loss": 0.8013,
+      "step": 24577
+    },
+    {
+      "epoch": 0.6750157919310099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.046549170641728e-06,
+      "loss": 0.7706,
+      "step": 24578
+    },
+    {
+      "epoch": 0.6750432561588531,
+      "grad_norm": 0.0,
+      "learning_rate": 5.045776470977118e-06,
+      "loss": 0.9243,
+      "step": 24579
+    },
+    {
+      "epoch": 0.6750707203866964,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0450038105122415e-06,
+      "loss": 0.898,
+      "step": 24580
+    },
+    {
+      "epoch": 0.6750981846145395,
+      "grad_norm": 0.0,
+      "learning_rate": 5.044231189253215e-06,
+      "loss": 0.7516,
+      "step": 24581
+    },
+    {
+      "epoch": 0.6751256488423828,
+      "grad_norm": 0.0,
+      "learning_rate": 5.043458607206149e-06,
+      "loss": 0.9075,
+      "step": 24582
+    },
+    {
+      "epoch": 0.6751531130702261,
+      "grad_norm": 0.0,
+      "learning_rate": 5.042686064377158e-06,
+      "loss": 0.8383,
+      "step": 24583
+    },
+    {
+      "epoch": 0.6751805772980692,
+      "grad_norm": 0.0,
+      "learning_rate": 5.041913560772354e-06,
+      "loss": 0.7563,
+      "step": 24584
+    },
+    {
+      "epoch": 0.6752080415259125,
+      "grad_norm": 0.0,
+      "learning_rate": 5.041141096397852e-06,
+      "loss": 0.9205,
+      "step": 24585
+    },
+    {
+      "epoch": 0.6752355057537557,
+      "grad_norm": 0.0,
+      "learning_rate": 5.040368671259767e-06,
+      "loss": 0.8139,
+      "step": 24586
+    },
+    {
+      "epoch": 0.675262969981599,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0395962853642055e-06,
+      "loss": 0.9039,
+      "step": 24587
+    },
+    {
+      "epoch": 0.6752904342094422,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0388239387172744e-06,
+      "loss": 0.8435,
+      "step": 24588
+    },
+    {
+      "epoch": 0.6753178984372854,
+      "grad_norm": 0.0,
+      "learning_rate": 5.038051631325092e-06,
+      "loss": 0.7648,
+      "step": 24589
+    },
+    {
+      "epoch": 0.6753453626651287,
+      "grad_norm": 0.0,
+      "learning_rate": 5.037279363193765e-06,
+      "loss": 0.8605,
+      "step": 24590
+    },
+    {
+      "epoch": 0.675372826892972,
+      "grad_norm": 0.0,
+      "learning_rate": 5.036507134329409e-06,
+      "loss": 0.8455,
+      "step": 24591
+    },
+    {
+      "epoch": 0.6754002911208151,
+      "grad_norm": 0.0,
+      "learning_rate": 5.035734944738129e-06,
+      "loss": 0.8914,
+      "step": 24592
+    },
+    {
+      "epoch": 0.6754277553486584,
+      "grad_norm": 0.0,
+      "learning_rate": 5.034962794426042e-06,
+      "loss": 0.7577,
+      "step": 24593
+    },
+    {
+      "epoch": 0.6754552195765016,
+      "grad_norm": 0.0,
+      "learning_rate": 5.034190683399248e-06,
+      "loss": 0.8463,
+      "step": 24594
+    },
+    {
+      "epoch": 0.6754826838043448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.033418611663865e-06,
+      "loss": 0.7812,
+      "step": 24595
+    },
+    {
+      "epoch": 0.6755101480321881,
+      "grad_norm": 0.0,
+      "learning_rate": 5.032646579225995e-06,
+      "loss": 0.9105,
+      "step": 24596
+    },
+    {
+      "epoch": 0.6755376122600313,
+      "grad_norm": 0.0,
+      "learning_rate": 5.031874586091747e-06,
+      "loss": 0.8366,
+      "step": 24597
+    },
+    {
+      "epoch": 0.6755650764878746,
+      "grad_norm": 0.0,
+      "learning_rate": 5.031102632267234e-06,
+      "loss": 0.9521,
+      "step": 24598
+    },
+    {
+      "epoch": 0.6755925407157177,
+      "grad_norm": 0.0,
+      "learning_rate": 5.030330717758565e-06,
+      "loss": 0.8193,
+      "step": 24599
+    },
+    {
+      "epoch": 0.675620004943561,
+      "grad_norm": 0.0,
+      "learning_rate": 5.02955884257184e-06,
+      "loss": 0.7958,
+      "step": 24600
+    },
+    {
+      "epoch": 0.6756474691714043,
+      "grad_norm": 0.0,
+      "learning_rate": 5.028787006713172e-06,
+      "loss": 0.8604,
+      "step": 24601
+    },
+    {
+      "epoch": 0.6756749333992474,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0280152101886705e-06,
+      "loss": 0.8109,
+      "step": 24602
+    },
+    {
+      "epoch": 0.6757023976270907,
+      "grad_norm": 0.0,
+      "learning_rate": 5.027243453004433e-06,
+      "loss": 0.9269,
+      "step": 24603
+    },
+    {
+      "epoch": 0.675729861854934,
+      "grad_norm": 0.0,
+      "learning_rate": 5.026471735166577e-06,
+      "loss": 0.9069,
+      "step": 24604
+    },
+    {
+      "epoch": 0.6757573260827772,
+      "grad_norm": 0.0,
+      "learning_rate": 5.025700056681199e-06,
+      "loss": 0.8383,
+      "step": 24605
+    },
+    {
+      "epoch": 0.6757847903106204,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0249284175544085e-06,
+      "loss": 0.8496,
+      "step": 24606
+    },
+    {
+      "epoch": 0.6758122545384636,
+      "grad_norm": 0.0,
+      "learning_rate": 5.024156817792312e-06,
+      "loss": 0.8321,
+      "step": 24607
+    },
+    {
+      "epoch": 0.6758397187663069,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0233852574010144e-06,
+      "loss": 0.9036,
+      "step": 24608
+    },
+    {
+      "epoch": 0.6758671829941502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.022613736386621e-06,
+      "loss": 0.8445,
+      "step": 24609
+    },
+    {
+      "epoch": 0.6758946472219933,
+      "grad_norm": 0.0,
+      "learning_rate": 5.021842254755238e-06,
+      "loss": 0.8056,
+      "step": 24610
+    },
+    {
+      "epoch": 0.6759221114498366,
+      "grad_norm": 0.0,
+      "learning_rate": 5.02107081251297e-06,
+      "loss": 0.8657,
+      "step": 24611
+    },
+    {
+      "epoch": 0.6759495756776798,
+      "grad_norm": 0.0,
+      "learning_rate": 5.020299409665912e-06,
+      "loss": 0.8648,
+      "step": 24612
+    },
+    {
+      "epoch": 0.675977039905523,
+      "grad_norm": 0.0,
+      "learning_rate": 5.019528046220175e-06,
+      "loss": 0.8192,
+      "step": 24613
+    },
+    {
+      "epoch": 0.6760045041333663,
+      "grad_norm": 0.0,
+      "learning_rate": 5.018756722181861e-06,
+      "loss": 0.9124,
+      "step": 24614
+    },
+    {
+      "epoch": 0.6760319683612095,
+      "grad_norm": 0.0,
+      "learning_rate": 5.017985437557074e-06,
+      "loss": 0.9038,
+      "step": 24615
+    },
+    {
+      "epoch": 0.6760594325890528,
+      "grad_norm": 0.0,
+      "learning_rate": 5.017214192351918e-06,
+      "loss": 0.864,
+      "step": 24616
+    },
+    {
+      "epoch": 0.676086896816896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.016442986572492e-06,
+      "loss": 0.8959,
+      "step": 24617
+    },
+    {
+      "epoch": 0.6761143610447392,
+      "grad_norm": 0.0,
+      "learning_rate": 5.015671820224899e-06,
+      "loss": 0.8597,
+      "step": 24618
+    },
+    {
+      "epoch": 0.6761418252725825,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0149006933152436e-06,
+      "loss": 0.8744,
+      "step": 24619
+    },
+    {
+      "epoch": 0.6761692895004257,
+      "grad_norm": 0.0,
+      "learning_rate": 5.014129605849623e-06,
+      "loss": 0.8386,
+      "step": 24620
+    },
+    {
+      "epoch": 0.6761967537282689,
+      "grad_norm": 0.0,
+      "learning_rate": 5.013358557834139e-06,
+      "loss": 0.8251,
+      "step": 24621
+    },
+    {
+      "epoch": 0.6762242179561122,
+      "grad_norm": 0.0,
+      "learning_rate": 5.012587549274899e-06,
+      "loss": 0.9091,
+      "step": 24622
+    },
+    {
+      "epoch": 0.6762516821839554,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0118165801779926e-06,
+      "loss": 0.8004,
+      "step": 24623
+    },
+    {
+      "epoch": 0.6762791464117986,
+      "grad_norm": 0.0,
+      "learning_rate": 5.011045650549526e-06,
+      "loss": 0.8415,
+      "step": 24624
+    },
+    {
+      "epoch": 0.6763066106396418,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0102747603956e-06,
+      "loss": 0.7398,
+      "step": 24625
+    },
+    {
+      "epoch": 0.6763340748674851,
+      "grad_norm": 0.0,
+      "learning_rate": 5.009503909722316e-06,
+      "loss": 0.7543,
+      "step": 24626
+    },
+    {
+      "epoch": 0.6763615390953284,
+      "grad_norm": 0.0,
+      "learning_rate": 5.008733098535767e-06,
+      "loss": 0.9152,
+      "step": 24627
+    },
+    {
+      "epoch": 0.6763890033231715,
+      "grad_norm": 0.0,
+      "learning_rate": 5.007962326842058e-06,
+      "loss": 0.8186,
+      "step": 24628
+    },
+    {
+      "epoch": 0.6764164675510148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.007191594647283e-06,
+      "loss": 0.8895,
+      "step": 24629
+    },
+    {
+      "epoch": 0.6764439317788581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.00642090195754e-06,
+      "loss": 0.9212,
+      "step": 24630
+    },
+    {
+      "epoch": 0.6764713960067013,
+      "grad_norm": 0.0,
+      "learning_rate": 5.005650248778929e-06,
+      "loss": 0.8333,
+      "step": 24631
+    },
+    {
+      "epoch": 0.6764988602345445,
+      "grad_norm": 0.0,
+      "learning_rate": 5.004879635117549e-06,
+      "loss": 0.8742,
+      "step": 24632
+    },
+    {
+      "epoch": 0.6765263244623877,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0041090609794965e-06,
+      "loss": 0.8218,
+      "step": 24633
+    },
+    {
+      "epoch": 0.676553788690231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.003338526370871e-06,
+      "loss": 0.8398,
+      "step": 24634
+    },
+    {
+      "epoch": 0.6765812529180742,
+      "grad_norm": 0.0,
+      "learning_rate": 5.002568031297767e-06,
+      "loss": 0.8129,
+      "step": 24635
+    },
+    {
+      "epoch": 0.6766087171459174,
+      "grad_norm": 0.0,
+      "learning_rate": 5.001797575766276e-06,
+      "loss": 0.9021,
+      "step": 24636
+    },
+    {
+      "epoch": 0.6766361813737607,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0010271597825e-06,
+      "loss": 0.799,
+      "step": 24637
+    },
+    {
+      "epoch": 0.6766636456016039,
+      "grad_norm": 0.0,
+      "learning_rate": 5.000256783352533e-06,
+      "loss": 0.9193,
+      "step": 24638
+    },
+    {
+      "epoch": 0.6766911098294471,
+      "grad_norm": 0.0,
+      "learning_rate": 4.999486446482471e-06,
+      "loss": 0.9016,
+      "step": 24639
+    },
+    {
+      "epoch": 0.6767185740572904,
+      "grad_norm": 0.0,
+      "learning_rate": 4.998716149178413e-06,
+      "loss": 0.7766,
+      "step": 24640
+    },
+    {
+      "epoch": 0.6767460382851336,
+      "grad_norm": 0.0,
+      "learning_rate": 4.997945891446447e-06,
+      "loss": 0.8323,
+      "step": 24641
+    },
+    {
+      "epoch": 0.6767735025129769,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9971756732926706e-06,
+      "loss": 0.8232,
+      "step": 24642
+    },
+    {
+      "epoch": 0.6768009667408201,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9964054947231814e-06,
+      "loss": 0.8746,
+      "step": 24643
+    },
+    {
+      "epoch": 0.6768284309686633,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9956353557440665e-06,
+      "loss": 0.7719,
+      "step": 24644
+    },
+    {
+      "epoch": 0.6768558951965066,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9948652563614234e-06,
+      "loss": 0.879,
+      "step": 24645
+    },
+    {
+      "epoch": 0.6768833594243497,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9940951965813485e-06,
+      "loss": 0.8649,
+      "step": 24646
+    },
+    {
+      "epoch": 0.676910823652193,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9933251764099275e-06,
+      "loss": 0.9479,
+      "step": 24647
+    },
+    {
+      "epoch": 0.6769382878800363,
+      "grad_norm": 0.0,
+      "learning_rate": 4.992555195853258e-06,
+      "loss": 0.8048,
+      "step": 24648
+    },
+    {
+      "epoch": 0.6769657521078795,
+      "grad_norm": 0.0,
+      "learning_rate": 4.991785254917431e-06,
+      "loss": 0.8764,
+      "step": 24649
+    },
+    {
+      "epoch": 0.6769932163357227,
+      "grad_norm": 0.0,
+      "learning_rate": 4.991015353608543e-06,
+      "loss": 0.7511,
+      "step": 24650
+    },
+    {
+      "epoch": 0.6770206805635659,
+      "grad_norm": 0.0,
+      "learning_rate": 4.990245491932678e-06,
+      "loss": 0.7108,
+      "step": 24651
+    },
+    {
+      "epoch": 0.6770481447914092,
+      "grad_norm": 0.0,
+      "learning_rate": 4.989475669895935e-06,
+      "loss": 0.8884,
+      "step": 24652
+    },
+    {
+      "epoch": 0.6770756090192525,
+      "grad_norm": 0.0,
+      "learning_rate": 4.988705887504398e-06,
+      "loss": 0.8321,
+      "step": 24653
+    },
+    {
+      "epoch": 0.6771030732470956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.987936144764161e-06,
+      "loss": 0.9382,
+      "step": 24654
+    },
+    {
+      "epoch": 0.6771305374749389,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9871664416813135e-06,
+      "loss": 0.8923,
+      "step": 24655
+    },
+    {
+      "epoch": 0.6771580017027822,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9863967782619485e-06,
+      "loss": 0.7978,
+      "step": 24656
+    },
+    {
+      "epoch": 0.6771854659306253,
+      "grad_norm": 0.0,
+      "learning_rate": 4.985627154512154e-06,
+      "loss": 0.8951,
+      "step": 24657
+    },
+    {
+      "epoch": 0.6772129301584686,
+      "grad_norm": 0.0,
+      "learning_rate": 4.984857570438023e-06,
+      "loss": 0.9341,
+      "step": 24658
+    },
+    {
+      "epoch": 0.6772403943863118,
+      "grad_norm": 0.0,
+      "learning_rate": 4.984088026045642e-06,
+      "loss": 0.8022,
+      "step": 24659
+    },
+    {
+      "epoch": 0.6772678586141551,
+      "grad_norm": 0.0,
+      "learning_rate": 4.983318521341096e-06,
+      "loss": 0.8119,
+      "step": 24660
+    },
+    {
+      "epoch": 0.6772953228419983,
+      "grad_norm": 0.0,
+      "learning_rate": 4.982549056330477e-06,
+      "loss": 0.8009,
+      "step": 24661
+    },
+    {
+      "epoch": 0.6773227870698415,
+      "grad_norm": 0.0,
+      "learning_rate": 4.981779631019873e-06,
+      "loss": 0.8421,
+      "step": 24662
+    },
+    {
+      "epoch": 0.6773502512976848,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9810102454153765e-06,
+      "loss": 0.8024,
+      "step": 24663
+    },
+    {
+      "epoch": 0.677377715525528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.980240899523067e-06,
+      "loss": 0.9813,
+      "step": 24664
+    },
+    {
+      "epoch": 0.6774051797533712,
+      "grad_norm": 0.0,
+      "learning_rate": 4.979471593349037e-06,
+      "loss": 0.9172,
+      "step": 24665
+    },
+    {
+      "epoch": 0.6774326439812145,
+      "grad_norm": 0.0,
+      "learning_rate": 4.978702326899371e-06,
+      "loss": 0.8954,
+      "step": 24666
+    },
+    {
+      "epoch": 0.6774601082090577,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977933100180163e-06,
+      "loss": 0.8806,
+      "step": 24667
+    },
+    {
+      "epoch": 0.6774875724369009,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977163913197488e-06,
+      "loss": 0.8766,
+      "step": 24668
+    },
+    {
+      "epoch": 0.6775150366647442,
+      "grad_norm": 0.0,
+      "learning_rate": 4.976394765957442e-06,
+      "loss": 0.8033,
+      "step": 24669
+    },
+    {
+      "epoch": 0.6775425008925874,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9756256584661036e-06,
+      "loss": 0.8675,
+      "step": 24670
+    },
+    {
+      "epoch": 0.6775699651204307,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9748565907295595e-06,
+      "loss": 0.809,
+      "step": 24671
+    },
+    {
+      "epoch": 0.6775974293482738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.974087562753899e-06,
+      "loss": 0.8353,
+      "step": 24672
+    },
+    {
+      "epoch": 0.6776248935761171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.973318574545203e-06,
+      "loss": 0.844,
+      "step": 24673
+    },
+    {
+      "epoch": 0.6776523578039604,
+      "grad_norm": 0.0,
+      "learning_rate": 4.972549626109562e-06,
+      "loss": 0.9042,
+      "step": 24674
+    },
+    {
+      "epoch": 0.6776798220318035,
+      "grad_norm": 0.0,
+      "learning_rate": 4.971780717453051e-06,
+      "loss": 0.8075,
+      "step": 24675
+    },
+    {
+      "epoch": 0.6777072862596468,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9710118485817635e-06,
+      "loss": 0.9093,
+      "step": 24676
+    },
+    {
+      "epoch": 0.67773475048749,
+      "grad_norm": 0.0,
+      "learning_rate": 4.970243019501774e-06,
+      "loss": 0.7804,
+      "step": 24677
+    },
+    {
+      "epoch": 0.6777622147153333,
+      "grad_norm": 0.0,
+      "learning_rate": 4.96947423021917e-06,
+      "loss": 0.8119,
+      "step": 24678
+    },
+    {
+      "epoch": 0.6777896789431765,
+      "grad_norm": 0.0,
+      "learning_rate": 4.968705480740035e-06,
+      "loss": 0.8635,
+      "step": 24679
+    },
+    {
+      "epoch": 0.6778171431710197,
+      "grad_norm": 0.0,
+      "learning_rate": 4.967936771070452e-06,
+      "loss": 0.8361,
+      "step": 24680
+    },
+    {
+      "epoch": 0.677844607398863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.967168101216505e-06,
+      "loss": 0.7016,
+      "step": 24681
+    },
+    {
+      "epoch": 0.6778720716267063,
+      "grad_norm": 0.0,
+      "learning_rate": 4.96639947118427e-06,
+      "loss": 0.8772,
+      "step": 24682
+    },
+    {
+      "epoch": 0.6778995358545494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.965630880979836e-06,
+      "loss": 0.796,
+      "step": 24683
+    },
+    {
+      "epoch": 0.6779270000823927,
+      "grad_norm": 0.0,
+      "learning_rate": 4.964862330609277e-06,
+      "loss": 0.8287,
+      "step": 24684
+    },
+    {
+      "epoch": 0.6779544643102359,
+      "grad_norm": 0.0,
+      "learning_rate": 4.964093820078677e-06,
+      "loss": 0.8203,
+      "step": 24685
+    },
+    {
+      "epoch": 0.6779819285380791,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9633253493941184e-06,
+      "loss": 0.8309,
+      "step": 24686
+    },
+    {
+      "epoch": 0.6780093927659224,
+      "grad_norm": 0.0,
+      "learning_rate": 4.962556918561684e-06,
+      "loss": 0.8803,
+      "step": 24687
+    },
+    {
+      "epoch": 0.6780368569937656,
+      "grad_norm": 0.0,
+      "learning_rate": 4.961788527587447e-06,
+      "loss": 0.8651,
+      "step": 24688
+    },
+    {
+      "epoch": 0.6780643212216089,
+      "grad_norm": 0.0,
+      "learning_rate": 4.96102017647749e-06,
+      "loss": 0.8265,
+      "step": 24689
+    },
+    {
+      "epoch": 0.678091785449452,
+      "grad_norm": 0.0,
+      "learning_rate": 4.960251865237894e-06,
+      "loss": 0.765,
+      "step": 24690
+    },
+    {
+      "epoch": 0.6781192496772953,
+      "grad_norm": 0.0,
+      "learning_rate": 4.959483593874741e-06,
+      "loss": 0.8073,
+      "step": 24691
+    },
+    {
+      "epoch": 0.6781467139051386,
+      "grad_norm": 0.0,
+      "learning_rate": 4.958715362394103e-06,
+      "loss": 0.9269,
+      "step": 24692
+    },
+    {
+      "epoch": 0.6781741781329818,
+      "grad_norm": 0.0,
+      "learning_rate": 4.957947170802065e-06,
+      "loss": 0.7775,
+      "step": 24693
+    },
+    {
+      "epoch": 0.678201642360825,
+      "grad_norm": 0.0,
+      "learning_rate": 4.957179019104698e-06,
+      "loss": 0.8143,
+      "step": 24694
+    },
+    {
+      "epoch": 0.6782291065886683,
+      "grad_norm": 0.0,
+      "learning_rate": 4.956410907308085e-06,
+      "loss": 0.9638,
+      "step": 24695
+    },
+    {
+      "epoch": 0.6782565708165115,
+      "grad_norm": 0.0,
+      "learning_rate": 4.955642835418301e-06,
+      "loss": 0.7918,
+      "step": 24696
+    },
+    {
+      "epoch": 0.6782840350443547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.954874803441427e-06,
+      "loss": 0.8038,
+      "step": 24697
+    },
+    {
+      "epoch": 0.6783114992721979,
+      "grad_norm": 0.0,
+      "learning_rate": 4.954106811383541e-06,
+      "loss": 0.8398,
+      "step": 24698
+    },
+    {
+      "epoch": 0.6783389635000412,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9533388592507125e-06,
+      "loss": 0.8469,
+      "step": 24699
+    },
+    {
+      "epoch": 0.6783664277278845,
+      "grad_norm": 0.0,
+      "learning_rate": 4.952570947049026e-06,
+      "loss": 0.7991,
+      "step": 24700
+    },
+    {
+      "epoch": 0.6783938919557276,
+      "grad_norm": 0.0,
+      "learning_rate": 4.951803074784548e-06,
+      "loss": 0.7663,
+      "step": 24701
+    },
+    {
+      "epoch": 0.6784213561835709,
+      "grad_norm": 0.0,
+      "learning_rate": 4.951035242463361e-06,
+      "loss": 0.9066,
+      "step": 24702
+    },
+    {
+      "epoch": 0.6784488204114142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.950267450091539e-06,
+      "loss": 0.8571,
+      "step": 24703
+    },
+    {
+      "epoch": 0.6784762846392574,
+      "grad_norm": 0.0,
+      "learning_rate": 4.949499697675156e-06,
+      "loss": 0.9128,
+      "step": 24704
+    },
+    {
+      "epoch": 0.6785037488671006,
+      "grad_norm": 0.0,
+      "learning_rate": 4.948731985220293e-06,
+      "loss": 0.8138,
+      "step": 24705
+    },
+    {
+      "epoch": 0.6785312130949438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.947964312733015e-06,
+      "loss": 0.8307,
+      "step": 24706
+    },
+    {
+      "epoch": 0.6785586773227871,
+      "grad_norm": 0.0,
+      "learning_rate": 4.947196680219402e-06,
+      "loss": 0.8522,
+      "step": 24707
+    },
+    {
+      "epoch": 0.6785861415506304,
+      "grad_norm": 0.0,
+      "learning_rate": 4.946429087685525e-06,
+      "loss": 0.801,
+      "step": 24708
+    },
+    {
+      "epoch": 0.6786136057784735,
+      "grad_norm": 0.0,
+      "learning_rate": 4.945661535137457e-06,
+      "loss": 0.8596,
+      "step": 24709
+    },
+    {
+      "epoch": 0.6786410700063168,
+      "grad_norm": 0.0,
+      "learning_rate": 4.944894022581273e-06,
+      "loss": 0.8234,
+      "step": 24710
+    },
+    {
+      "epoch": 0.67866853423416,
+      "grad_norm": 0.0,
+      "learning_rate": 4.944126550023051e-06,
+      "loss": 0.8891,
+      "step": 24711
+    },
+    {
+      "epoch": 0.6786959984620032,
+      "grad_norm": 0.0,
+      "learning_rate": 4.943359117468854e-06,
+      "loss": 0.8767,
+      "step": 24712
+    },
+    {
+      "epoch": 0.6787234626898465,
+      "grad_norm": 0.0,
+      "learning_rate": 4.942591724924757e-06,
+      "loss": 0.8702,
+      "step": 24713
+    },
+    {
+      "epoch": 0.6787509269176897,
+      "grad_norm": 0.0,
+      "learning_rate": 4.941824372396834e-06,
+      "loss": 0.8782,
+      "step": 24714
+    },
+    {
+      "epoch": 0.678778391145533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.94105705989116e-06,
+      "loss": 0.7777,
+      "step": 24715
+    },
+    {
+      "epoch": 0.6788058553733762,
+      "grad_norm": 0.0,
+      "learning_rate": 4.940289787413798e-06,
+      "loss": 0.8049,
+      "step": 24716
+    },
+    {
+      "epoch": 0.6788333196012194,
+      "grad_norm": 0.0,
+      "learning_rate": 4.939522554970827e-06,
+      "loss": 0.8248,
+      "step": 24717
+    },
+    {
+      "epoch": 0.6788607838290627,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9387553625683105e-06,
+      "loss": 0.9119,
+      "step": 24718
+    },
+    {
+      "epoch": 0.6788882480569058,
+      "grad_norm": 0.0,
+      "learning_rate": 4.937988210212321e-06,
+      "loss": 0.7791,
+      "step": 24719
+    },
+    {
+      "epoch": 0.6789157122847491,
+      "grad_norm": 0.0,
+      "learning_rate": 4.93722109790893e-06,
+      "loss": 0.8889,
+      "step": 24720
+    },
+    {
+      "epoch": 0.6789431765125924,
+      "grad_norm": 0.0,
+      "learning_rate": 4.936454025664208e-06,
+      "loss": 0.8643,
+      "step": 24721
+    },
+    {
+      "epoch": 0.6789706407404356,
+      "grad_norm": 0.0,
+      "learning_rate": 4.935686993484226e-06,
+      "loss": 0.7914,
+      "step": 24722
+    },
+    {
+      "epoch": 0.6789981049682788,
+      "grad_norm": 0.0,
+      "learning_rate": 4.934920001375051e-06,
+      "loss": 0.8123,
+      "step": 24723
+    },
+    {
+      "epoch": 0.679025569196122,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9341530493427455e-06,
+      "loss": 0.777,
+      "step": 24724
+    },
+    {
+      "epoch": 0.6790530334239653,
+      "grad_norm": 0.0,
+      "learning_rate": 4.933386137393384e-06,
+      "loss": 0.9041,
+      "step": 24725
+    },
+    {
+      "epoch": 0.6790804976518086,
+      "grad_norm": 0.0,
+      "learning_rate": 4.932619265533035e-06,
+      "loss": 0.8448,
+      "step": 24726
+    },
+    {
+      "epoch": 0.6791079618796517,
+      "grad_norm": 0.0,
+      "learning_rate": 4.931852433767765e-06,
+      "loss": 0.843,
+      "step": 24727
+    },
+    {
+      "epoch": 0.679135426107495,
+      "grad_norm": 0.0,
+      "learning_rate": 4.931085642103645e-06,
+      "loss": 0.9036,
+      "step": 24728
+    },
+    {
+      "epoch": 0.6791628903353383,
+      "grad_norm": 0.0,
+      "learning_rate": 4.930318890546736e-06,
+      "loss": 0.8747,
+      "step": 24729
+    },
+    {
+      "epoch": 0.6791903545631814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.929552179103106e-06,
+      "loss": 0.8458,
+      "step": 24730
+    },
+    {
+      "epoch": 0.6792178187910247,
+      "grad_norm": 0.0,
+      "learning_rate": 4.928785507778828e-06,
+      "loss": 0.8279,
+      "step": 24731
+    },
+    {
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9280188765799604e-06,
+      "loss": 0.7471,
+      "step": 24732
+    },
+    {
+      "epoch": 0.6792727472467112,
+      "grad_norm": 0.0,
+      "learning_rate": 4.927252285512572e-06,
+      "loss": 0.8651,
+      "step": 24733
+    },
+    {
+      "epoch": 0.6793002114745544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.926485734582732e-06,
+      "loss": 0.8835,
+      "step": 24734
+    },
+    {
+      "epoch": 0.6793276757023976,
+      "grad_norm": 0.0,
+      "learning_rate": 4.925719223796498e-06,
+      "loss": 0.8398,
+      "step": 24735
+    },
+    {
+      "epoch": 0.6793551399302409,
+      "grad_norm": 0.0,
+      "learning_rate": 4.92495275315994e-06,
+      "loss": 0.8344,
+      "step": 24736
+    },
+    {
+      "epoch": 0.679382604158084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.92418632267912e-06,
+      "loss": 1.0033,
+      "step": 24737
+    },
+    {
+      "epoch": 0.6794100683859273,
+      "grad_norm": 0.0,
+      "learning_rate": 4.923419932360105e-06,
+      "loss": 0.7725,
+      "step": 24738
+    },
+    {
+      "epoch": 0.6794375326137706,
+      "grad_norm": 0.0,
+      "learning_rate": 4.922653582208963e-06,
+      "loss": 0.8688,
+      "step": 24739
+    },
+    {
+      "epoch": 0.6794649968416138,
+      "grad_norm": 0.0,
+      "learning_rate": 4.92188727223175e-06,
+      "loss": 0.8678,
+      "step": 24740
+    },
+    {
+      "epoch": 0.679492461069457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.92112100243453e-06,
+      "loss": 0.9474,
+      "step": 24741
+    },
+    {
+      "epoch": 0.6795199252973003,
+      "grad_norm": 0.0,
+      "learning_rate": 4.920354772823368e-06,
+      "loss": 0.8023,
+      "step": 24742
+    },
+    {
+      "epoch": 0.6795473895251435,
+      "grad_norm": 0.0,
+      "learning_rate": 4.919588583404327e-06,
+      "loss": 0.7992,
+      "step": 24743
+    },
+    {
+      "epoch": 0.6795748537529868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9188224341834674e-06,
+      "loss": 0.8561,
+      "step": 24744
+    },
+    {
+      "epoch": 0.6796023179808299,
+      "grad_norm": 0.0,
+      "learning_rate": 4.918056325166855e-06,
+      "loss": 0.8849,
+      "step": 24745
+    },
+    {
+      "epoch": 0.6796297822086732,
+      "grad_norm": 0.0,
+      "learning_rate": 4.917290256360552e-06,
+      "loss": 0.9626,
+      "step": 24746
+    },
+    {
+      "epoch": 0.6796572464365165,
+      "grad_norm": 0.0,
+      "learning_rate": 4.916524227770617e-06,
+      "loss": 0.7855,
+      "step": 24747
+    },
+    {
+      "epoch": 0.6796847106643596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.915758239403109e-06,
+      "loss": 0.8491,
+      "step": 24748
+    },
+    {
+      "epoch": 0.6797121748922029,
+      "grad_norm": 0.0,
+      "learning_rate": 4.91499229126409e-06,
+      "loss": 0.7961,
+      "step": 24749
+    },
+    {
+      "epoch": 0.6797396391200461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.914226383359623e-06,
+      "loss": 0.865,
+      "step": 24750
+    },
+    {
+      "epoch": 0.6797671033478894,
+      "grad_norm": 0.0,
+      "learning_rate": 4.913460515695766e-06,
+      "loss": 0.9091,
+      "step": 24751
+    },
+    {
+      "epoch": 0.6797945675757326,
+      "grad_norm": 0.0,
+      "learning_rate": 4.912694688278583e-06,
+      "loss": 0.868,
+      "step": 24752
+    },
+    {
+      "epoch": 0.6798220318035758,
+      "grad_norm": 0.0,
+      "learning_rate": 4.911928901114127e-06,
+      "loss": 0.8293,
+      "step": 24753
+    },
+    {
+      "epoch": 0.6798494960314191,
+      "grad_norm": 0.0,
+      "learning_rate": 4.911163154208461e-06,
+      "loss": 0.8155,
+      "step": 24754
+    },
+    {
+      "epoch": 0.6798769602592624,
+      "grad_norm": 0.0,
+      "learning_rate": 4.910397447567646e-06,
+      "loss": 0.8751,
+      "step": 24755
+    },
+    {
+      "epoch": 0.6799044244871055,
+      "grad_norm": 0.0,
+      "learning_rate": 4.909631781197736e-06,
+      "loss": 0.8654,
+      "step": 24756
+    },
+    {
+      "epoch": 0.6799318887149488,
+      "grad_norm": 0.0,
+      "learning_rate": 4.90886615510479e-06,
+      "loss": 0.8785,
+      "step": 24757
+    },
+    {
+      "epoch": 0.679959352942792,
+      "grad_norm": 0.0,
+      "learning_rate": 4.908100569294871e-06,
+      "loss": 0.8683,
+      "step": 24758
+    },
+    {
+      "epoch": 0.6799868171706352,
+      "grad_norm": 0.0,
+      "learning_rate": 4.907335023774029e-06,
+      "loss": 0.8142,
+      "step": 24759
+    },
+    {
+      "epoch": 0.6800142813984785,
+      "grad_norm": 0.0,
+      "learning_rate": 4.906569518548325e-06,
+      "loss": 0.7894,
+      "step": 24760
+    },
+    {
+      "epoch": 0.6800417456263217,
+      "grad_norm": 0.0,
+      "learning_rate": 4.905804053623815e-06,
+      "loss": 0.8653,
+      "step": 24761
+    },
+    {
+      "epoch": 0.680069209854165,
+      "grad_norm": 0.0,
+      "learning_rate": 4.905038629006562e-06,
+      "loss": 0.8469,
+      "step": 24762
+    },
+    {
+      "epoch": 0.6800966740820081,
+      "grad_norm": 0.0,
+      "learning_rate": 4.904273244702612e-06,
+      "loss": 0.8638,
+      "step": 24763
+    },
+    {
+      "epoch": 0.6801241383098514,
+      "grad_norm": 0.0,
+      "learning_rate": 4.903507900718031e-06,
+      "loss": 0.7663,
+      "step": 24764
+    },
+    {
+      "epoch": 0.6801516025376947,
+      "grad_norm": 0.0,
+      "learning_rate": 4.902742597058865e-06,
+      "loss": 0.8074,
+      "step": 24765
+    },
+    {
+      "epoch": 0.6801790667655379,
+      "grad_norm": 0.0,
+      "learning_rate": 4.901977333731174e-06,
+      "loss": 0.8739,
+      "step": 24766
+    },
+    {
+      "epoch": 0.6802065309933811,
+      "grad_norm": 0.0,
+      "learning_rate": 4.901212110741013e-06,
+      "loss": 0.8553,
+      "step": 24767
+    },
+    {
+      "epoch": 0.6802339952212244,
+      "grad_norm": 0.0,
+      "learning_rate": 4.900446928094437e-06,
+      "loss": 0.8729,
+      "step": 24768
+    },
+    {
+      "epoch": 0.6802614594490676,
+      "grad_norm": 0.0,
+      "learning_rate": 4.899681785797501e-06,
+      "loss": 0.7371,
+      "step": 24769
+    },
+    {
+      "epoch": 0.6802889236769109,
+      "grad_norm": 0.0,
+      "learning_rate": 4.898916683856262e-06,
+      "loss": 0.7859,
+      "step": 24770
+    },
+    {
+      "epoch": 0.680316387904754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.898151622276769e-06,
+      "loss": 0.8484,
+      "step": 24771
+    },
+    {
+      "epoch": 0.6803438521325973,
+      "grad_norm": 0.0,
+      "learning_rate": 4.897386601065074e-06,
+      "loss": 0.8202,
+      "step": 24772
+    },
+    {
+      "epoch": 0.6803713163604406,
+      "grad_norm": 0.0,
+      "learning_rate": 4.896621620227232e-06,
+      "loss": 0.8688,
+      "step": 24773
+    },
+    {
+      "epoch": 0.6803987805882837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8958566797692974e-06,
+      "loss": 0.7222,
+      "step": 24774
+    },
+    {
+      "epoch": 0.680426244816127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.895091779697321e-06,
+      "loss": 0.8423,
+      "step": 24775
+    },
+    {
+      "epoch": 0.6804537090439702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.89432692001736e-06,
+      "loss": 0.9102,
+      "step": 24776
+    },
+    {
+      "epoch": 0.6804811732718135,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8935621007354594e-06,
+      "loss": 0.8504,
+      "step": 24777
+    },
+    {
+      "epoch": 0.6805086374996567,
+      "grad_norm": 0.0,
+      "learning_rate": 4.892797321857672e-06,
+      "loss": 0.8479,
+      "step": 24778
+    },
+    {
+      "epoch": 0.6805361017274999,
+      "grad_norm": 0.0,
+      "learning_rate": 4.892032583390055e-06,
+      "loss": 0.8511,
+      "step": 24779
+    },
+    {
+      "epoch": 0.6805635659553432,
+      "grad_norm": 0.0,
+      "learning_rate": 4.891267885338652e-06,
+      "loss": 0.8574,
+      "step": 24780
+    },
+    {
+      "epoch": 0.6805910301831865,
+      "grad_norm": 0.0,
+      "learning_rate": 4.890503227709519e-06,
+      "loss": 0.8379,
+      "step": 24781
+    },
+    {
+      "epoch": 0.6806184944110296,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8897386105087e-06,
+      "loss": 0.9209,
+      "step": 24782
+    },
+    {
+      "epoch": 0.6806459586388729,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8889740337422485e-06,
+      "loss": 0.8259,
+      "step": 24783
+    },
+    {
+      "epoch": 0.6806734228667161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.888209497416214e-06,
+      "loss": 0.8802,
+      "step": 24784
+    },
+    {
+      "epoch": 0.6807008870945593,
+      "grad_norm": 0.0,
+      "learning_rate": 4.887445001536647e-06,
+      "loss": 0.7671,
+      "step": 24785
+    },
+    {
+      "epoch": 0.6807283513224026,
+      "grad_norm": 0.0,
+      "learning_rate": 4.886680546109599e-06,
+      "loss": 0.7736,
+      "step": 24786
+    },
+    {
+      "epoch": 0.6807558155502458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.885916131141112e-06,
+      "loss": 0.9135,
+      "step": 24787
+    },
+    {
+      "epoch": 0.6807832797780891,
+      "grad_norm": 0.0,
+      "learning_rate": 4.885151756637241e-06,
+      "loss": 0.8546,
+      "step": 24788
+    },
+    {
+      "epoch": 0.6808107440059322,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8843874226040265e-06,
+      "loss": 0.8284,
+      "step": 24789
+    },
+    {
+      "epoch": 0.6808382082337755,
+      "grad_norm": 0.0,
+      "learning_rate": 4.88362312904752e-06,
+      "loss": 0.889,
+      "step": 24790
+    },
+    {
+      "epoch": 0.6808656724616188,
+      "grad_norm": 0.0,
+      "learning_rate": 4.88285887597377e-06,
+      "loss": 0.8629,
+      "step": 24791
+    },
+    {
+      "epoch": 0.6808931366894619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.882094663388822e-06,
+      "loss": 0.805,
+      "step": 24792
+    },
+    {
+      "epoch": 0.6809206009173052,
+      "grad_norm": 0.0,
+      "learning_rate": 4.881330491298728e-06,
+      "loss": 0.8713,
+      "step": 24793
+    },
+    {
+      "epoch": 0.6809480651451485,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8805663597095256e-06,
+      "loss": 1.0116,
+      "step": 24794
+    },
+    {
+      "epoch": 0.6809755293729917,
+      "grad_norm": 0.0,
+      "learning_rate": 4.87980226862727e-06,
+      "loss": 0.8049,
+      "step": 24795
+    },
+    {
+      "epoch": 0.6810029936008349,
+      "grad_norm": 0.0,
+      "learning_rate": 4.879038218057996e-06,
+      "loss": 0.8377,
+      "step": 24796
+    },
+    {
+      "epoch": 0.6810304578286781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.878274208007758e-06,
+      "loss": 0.9736,
+      "step": 24797
+    },
+    {
+      "epoch": 0.6810579220565214,
+      "grad_norm": 0.0,
+      "learning_rate": 4.877510238482598e-06,
+      "loss": 0.8356,
+      "step": 24798
+    },
+    {
+      "epoch": 0.6810853862843647,
+      "grad_norm": 0.0,
+      "learning_rate": 4.876746309488564e-06,
+      "loss": 0.8467,
+      "step": 24799
+    },
+    {
+      "epoch": 0.6811128505122078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.875982421031694e-06,
+      "loss": 0.8453,
+      "step": 24800
+    },
+    {
+      "epoch": 0.6811403147400511,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8752185731180364e-06,
+      "loss": 0.7828,
+      "step": 24801
+    },
+    {
+      "epoch": 0.6811677789678943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.874454765753635e-06,
+      "loss": 0.8521,
+      "step": 24802
+    },
+    {
+      "epoch": 0.6811952431957375,
+      "grad_norm": 0.0,
+      "learning_rate": 4.873690998944536e-06,
+      "loss": 0.7848,
+      "step": 24803
+    },
+    {
+      "epoch": 0.6812227074235808,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8729272726967754e-06,
+      "loss": 0.8546,
+      "step": 24804
+    },
+    {
+      "epoch": 0.681250171651424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.872163587016405e-06,
+      "loss": 0.8032,
+      "step": 24805
+    },
+    {
+      "epoch": 0.6812776358792673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8713999419094594e-06,
+      "loss": 0.8573,
+      "step": 24806
+    },
+    {
+      "epoch": 0.6813051001071105,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8706363373819835e-06,
+      "loss": 0.871,
+      "step": 24807
+    },
+    {
+      "epoch": 0.6813325643349537,
+      "grad_norm": 0.0,
+      "learning_rate": 4.869872773440018e-06,
+      "loss": 0.756,
+      "step": 24808
+    },
+    {
+      "epoch": 0.681360028562797,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8691092500896085e-06,
+      "loss": 0.8812,
+      "step": 24809
+    },
+    {
+      "epoch": 0.6813874927906401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.868345767336797e-06,
+      "loss": 0.8181,
+      "step": 24810
+    },
+    {
+      "epoch": 0.6814149570184834,
+      "grad_norm": 0.0,
+      "learning_rate": 4.867582325187619e-06,
+      "loss": 0.8665,
+      "step": 24811
+    },
+    {
+      "epoch": 0.6814424212463267,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8668189236481215e-06,
+      "loss": 0.8623,
+      "step": 24812
+    },
+    {
+      "epoch": 0.6814698854741699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.866055562724335e-06,
+      "loss": 0.9122,
+      "step": 24813
+    },
+    {
+      "epoch": 0.6814973497020131,
+      "grad_norm": 0.0,
+      "learning_rate": 4.865292242422308e-06,
+      "loss": 0.8451,
+      "step": 24814
+    },
+    {
+      "epoch": 0.6815248139298563,
+      "grad_norm": 0.0,
+      "learning_rate": 4.864528962748076e-06,
+      "loss": 0.8685,
+      "step": 24815
+    },
+    {
+      "epoch": 0.6815522781576996,
+      "grad_norm": 0.0,
+      "learning_rate": 4.863765723707682e-06,
+      "loss": 0.8671,
+      "step": 24816
+    },
+    {
+      "epoch": 0.6815797423855429,
+      "grad_norm": 0.0,
+      "learning_rate": 4.863002525307166e-06,
+      "loss": 0.9405,
+      "step": 24817
+    },
+    {
+      "epoch": 0.681607206613386,
+      "grad_norm": 0.0,
+      "learning_rate": 4.862239367552559e-06,
+      "loss": 0.9367,
+      "step": 24818
+    },
+    {
+      "epoch": 0.6816346708412293,
+      "grad_norm": 0.0,
+      "learning_rate": 4.86147625044991e-06,
+      "loss": 0.8162,
+      "step": 24819
+    },
+    {
+      "epoch": 0.6816621350690726,
+      "grad_norm": 0.0,
+      "learning_rate": 4.860713174005246e-06,
+      "loss": 0.7609,
+      "step": 24820
+    },
+    {
+      "epoch": 0.6816895992969157,
+      "grad_norm": 0.0,
+      "learning_rate": 4.859950138224611e-06,
+      "loss": 0.7719,
+      "step": 24821
+    },
+    {
+      "epoch": 0.681717063524759,
+      "grad_norm": 0.0,
+      "learning_rate": 4.859187143114042e-06,
+      "loss": 0.9094,
+      "step": 24822
+    },
+    {
+      "epoch": 0.6817445277526022,
+      "grad_norm": 0.0,
+      "learning_rate": 4.858424188679579e-06,
+      "loss": 0.8669,
+      "step": 24823
+    },
+    {
+      "epoch": 0.6817719919804455,
+      "grad_norm": 0.0,
+      "learning_rate": 4.857661274927251e-06,
+      "loss": 0.7975,
+      "step": 24824
+    },
+    {
+      "epoch": 0.6817994562082887,
+      "grad_norm": 0.0,
+      "learning_rate": 4.856898401863099e-06,
+      "loss": 0.9593,
+      "step": 24825
+    },
+    {
+      "epoch": 0.6818269204361319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.85613556949316e-06,
+      "loss": 0.7733,
+      "step": 24826
+    },
+    {
+      "epoch": 0.6818543846639752,
+      "grad_norm": 0.0,
+      "learning_rate": 4.855372777823473e-06,
+      "loss": 0.8505,
+      "step": 24827
+    },
+    {
+      "epoch": 0.6818818488918184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.854610026860065e-06,
+      "loss": 0.91,
+      "step": 24828
+    },
+    {
+      "epoch": 0.6819093131196616,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8538473166089795e-06,
+      "loss": 0.8126,
+      "step": 24829
+    },
+    {
+      "epoch": 0.6819367773475049,
+      "grad_norm": 0.0,
+      "learning_rate": 4.853084647076242e-06,
+      "loss": 0.795,
+      "step": 24830
+    },
+    {
+      "epoch": 0.6819642415753481,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8523220182678955e-06,
+      "loss": 0.7861,
+      "step": 24831
+    },
+    {
+      "epoch": 0.6819917058031914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.851559430189969e-06,
+      "loss": 0.793,
+      "step": 24832
+    },
+    {
+      "epoch": 0.6820191700310346,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8507968828485e-06,
+      "loss": 0.8105,
+      "step": 24833
+    },
+    {
+      "epoch": 0.6820466342588778,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8500343762495235e-06,
+      "loss": 0.8061,
+      "step": 24834
+    },
+    {
+      "epoch": 0.6820740984867211,
+      "grad_norm": 0.0,
+      "learning_rate": 4.849271910399067e-06,
+      "loss": 0.8986,
+      "step": 24835
+    },
+    {
+      "epoch": 0.6821015627145642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.848509485303169e-06,
+      "loss": 0.7851,
+      "step": 24836
+    },
+    {
+      "epoch": 0.6821290269424075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.847747100967857e-06,
+      "loss": 0.798,
+      "step": 24837
+    },
+    {
+      "epoch": 0.6821564911702508,
+      "grad_norm": 0.0,
+      "learning_rate": 4.846984757399164e-06,
+      "loss": 0.7447,
+      "step": 24838
+    },
+    {
+      "epoch": 0.682183955398094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.846222454603125e-06,
+      "loss": 0.8557,
+      "step": 24839
+    },
+    {
+      "epoch": 0.6822114196259372,
+      "grad_norm": 0.0,
+      "learning_rate": 4.845460192585773e-06,
+      "loss": 0.7561,
+      "step": 24840
+    },
+    {
+      "epoch": 0.6822388838537804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.844697971353135e-06,
+      "loss": 0.7729,
+      "step": 24841
+    },
+    {
+      "epoch": 0.6822663480816237,
+      "grad_norm": 0.0,
+      "learning_rate": 4.843935790911241e-06,
+      "loss": 0.872,
+      "step": 24842
+    },
+    {
+      "epoch": 0.682293812309467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.84317365126613e-06,
+      "loss": 0.854,
+      "step": 24843
+    },
+    {
+      "epoch": 0.6823212765373101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.842411552423822e-06,
+      "loss": 0.8193,
+      "step": 24844
+    },
+    {
+      "epoch": 0.6823487407651534,
+      "grad_norm": 0.0,
+      "learning_rate": 4.841649494390352e-06,
+      "loss": 0.8087,
+      "step": 24845
+    },
+    {
+      "epoch": 0.6823762049929967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.840887477171753e-06,
+      "loss": 0.9374,
+      "step": 24846
+    },
+    {
+      "epoch": 0.6824036692208398,
+      "grad_norm": 0.0,
+      "learning_rate": 4.840125500774047e-06,
+      "loss": 0.9246,
+      "step": 24847
+    },
+    {
+      "epoch": 0.6824311334486831,
+      "grad_norm": 0.0,
+      "learning_rate": 4.839363565203266e-06,
+      "loss": 0.899,
+      "step": 24848
+    },
+    {
+      "epoch": 0.6824585976765263,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8386016704654404e-06,
+      "loss": 0.8088,
+      "step": 24849
+    },
+    {
+      "epoch": 0.6824860619043696,
+      "grad_norm": 0.0,
+      "learning_rate": 4.837839816566598e-06,
+      "loss": 0.7909,
+      "step": 24850
+    },
+    {
+      "epoch": 0.6825135261322128,
+      "grad_norm": 0.0,
+      "learning_rate": 4.83707800351277e-06,
+      "loss": 0.8753,
+      "step": 24851
+    },
+    {
+      "epoch": 0.682540990360056,
+      "grad_norm": 0.0,
+      "learning_rate": 4.836316231309981e-06,
+      "loss": 0.7377,
+      "step": 24852
+    },
+    {
+      "epoch": 0.6825684545878993,
+      "grad_norm": 0.0,
+      "learning_rate": 4.835554499964253e-06,
+      "loss": 0.8709,
+      "step": 24853
+    },
+    {
+      "epoch": 0.6825959188157424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.83479280948162e-06,
+      "loss": 0.8717,
+      "step": 24854
+    },
+    {
+      "epoch": 0.6826233830435857,
+      "grad_norm": 0.0,
+      "learning_rate": 4.834031159868107e-06,
+      "loss": 0.7876,
+      "step": 24855
+    },
+    {
+      "epoch": 0.682650847271429,
+      "grad_norm": 0.0,
+      "learning_rate": 4.83326955112974e-06,
+      "loss": 0.7048,
+      "step": 24856
+    },
+    {
+      "epoch": 0.6826783114992722,
+      "grad_norm": 0.0,
+      "learning_rate": 4.832507983272545e-06,
+      "loss": 0.8162,
+      "step": 24857
+    },
+    {
+      "epoch": 0.6827057757271154,
+      "grad_norm": 0.0,
+      "learning_rate": 4.831746456302553e-06,
+      "loss": 0.9086,
+      "step": 24858
+    },
+    {
+      "epoch": 0.6827332399549587,
+      "grad_norm": 0.0,
+      "learning_rate": 4.830984970225782e-06,
+      "loss": 0.8635,
+      "step": 24859
+    },
+    {
+      "epoch": 0.6827607041828019,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8302235250482636e-06,
+      "loss": 0.772,
+      "step": 24860
+    },
+    {
+      "epoch": 0.6827881684106452,
+      "grad_norm": 0.0,
+      "learning_rate": 4.829462120776015e-06,
+      "loss": 0.8737,
+      "step": 24861
+    },
+    {
+      "epoch": 0.6828156326384883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.828700757415065e-06,
+      "loss": 0.8409,
+      "step": 24862
+    },
+    {
+      "epoch": 0.6828430968663316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.827939434971438e-06,
+      "loss": 0.8238,
+      "step": 24863
+    },
+    {
+      "epoch": 0.6828705610941749,
+      "grad_norm": 0.0,
+      "learning_rate": 4.827178153451161e-06,
+      "loss": 0.8141,
+      "step": 24864
+    },
+    {
+      "epoch": 0.682898025322018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.826416912860251e-06,
+      "loss": 0.8236,
+      "step": 24865
+    },
+    {
+      "epoch": 0.6829254895498613,
+      "grad_norm": 0.0,
+      "learning_rate": 4.825655713204735e-06,
+      "loss": 0.8332,
+      "step": 24866
+    },
+    {
+      "epoch": 0.6829529537777045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.824894554490638e-06,
+      "loss": 0.8076,
+      "step": 24867
+    },
+    {
+      "epoch": 0.6829804180055478,
+      "grad_norm": 0.0,
+      "learning_rate": 4.824133436723976e-06,
+      "loss": 0.7114,
+      "step": 24868
+    },
+    {
+      "epoch": 0.683007882233391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.823372359910776e-06,
+      "loss": 0.8426,
+      "step": 24869
+    },
+    {
+      "epoch": 0.6830353464612342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.822611324057062e-06,
+      "loss": 0.8264,
+      "step": 24870
+    },
+    {
+      "epoch": 0.6830628106890775,
+      "grad_norm": 0.0,
+      "learning_rate": 4.82185032916885e-06,
+      "loss": 0.8709,
+      "step": 24871
+    },
+    {
+      "epoch": 0.6830902749169208,
+      "grad_norm": 0.0,
+      "learning_rate": 4.821089375252164e-06,
+      "loss": 0.7612,
+      "step": 24872
+    },
+    {
+      "epoch": 0.6831177391447639,
+      "grad_norm": 0.0,
+      "learning_rate": 4.820328462313024e-06,
+      "loss": 0.8085,
+      "step": 24873
+    },
+    {
+      "epoch": 0.6831452033726072,
+      "grad_norm": 0.0,
+      "learning_rate": 4.819567590357452e-06,
+      "loss": 0.7597,
+      "step": 24874
+    },
+    {
+      "epoch": 0.6831726676004504,
+      "grad_norm": 0.0,
+      "learning_rate": 4.818806759391472e-06,
+      "loss": 0.8779,
+      "step": 24875
+    },
+    {
+      "epoch": 0.6832001318282936,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8180459694211e-06,
+      "loss": 0.8179,
+      "step": 24876
+    },
+    {
+      "epoch": 0.6832275960561369,
+      "grad_norm": 0.0,
+      "learning_rate": 4.817285220452351e-06,
+      "loss": 0.8621,
+      "step": 24877
+    },
+    {
+      "epoch": 0.6832550602839801,
+      "grad_norm": 0.0,
+      "learning_rate": 4.81652451249125e-06,
+      "loss": 0.8645,
+      "step": 24878
+    },
+    {
+      "epoch": 0.6832825245118234,
+      "grad_norm": 0.0,
+      "learning_rate": 4.815763845543815e-06,
+      "loss": 0.9395,
+      "step": 24879
+    },
+    {
+      "epoch": 0.6833099887396665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8150032196160636e-06,
+      "loss": 0.8079,
+      "step": 24880
+    },
+    {
+      "epoch": 0.6833374529675098,
+      "grad_norm": 0.0,
+      "learning_rate": 4.814242634714016e-06,
+      "loss": 0.8658,
+      "step": 24881
+    },
+    {
+      "epoch": 0.6833649171953531,
+      "grad_norm": 0.0,
+      "learning_rate": 4.813482090843693e-06,
+      "loss": 0.8445,
+      "step": 24882
+    },
+    {
+      "epoch": 0.6833923814231962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8127215880111085e-06,
+      "loss": 0.9011,
+      "step": 24883
+    },
+    {
+      "epoch": 0.6834198456510395,
+      "grad_norm": 0.0,
+      "learning_rate": 4.811961126222276e-06,
+      "loss": 0.9602,
+      "step": 24884
+    },
+    {
+      "epoch": 0.6834473098788828,
+      "grad_norm": 0.0,
+      "learning_rate": 4.811200705483218e-06,
+      "loss": 0.7636,
+      "step": 24885
+    },
+    {
+      "epoch": 0.683474774106726,
+      "grad_norm": 0.0,
+      "learning_rate": 4.810440325799949e-06,
+      "loss": 0.8303,
+      "step": 24886
+    },
+    {
+      "epoch": 0.6835022383345692,
+      "grad_norm": 0.0,
+      "learning_rate": 4.809679987178487e-06,
+      "loss": 0.7876,
+      "step": 24887
+    },
+    {
+      "epoch": 0.6835297025624124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.80891968962485e-06,
+      "loss": 0.8115,
+      "step": 24888
+    },
+    {
+      "epoch": 0.6835571667902557,
+      "grad_norm": 0.0,
+      "learning_rate": 4.808159433145049e-06,
+      "loss": 0.8209,
+      "step": 24889
+    },
+    {
+      "epoch": 0.683584631018099,
+      "grad_norm": 0.0,
+      "learning_rate": 4.807399217745101e-06,
+      "loss": 0.847,
+      "step": 24890
+    },
+    {
+      "epoch": 0.6836120952459421,
+      "grad_norm": 0.0,
+      "learning_rate": 4.806639043431026e-06,
+      "loss": 0.8417,
+      "step": 24891
+    },
+    {
+      "epoch": 0.6836395594737854,
+      "grad_norm": 0.0,
+      "learning_rate": 4.80587891020883e-06,
+      "loss": 0.9247,
+      "step": 24892
+    },
+    {
+      "epoch": 0.6836670237016286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.805118818084533e-06,
+      "loss": 0.8583,
+      "step": 24893
+    },
+    {
+      "epoch": 0.6836944879294719,
+      "grad_norm": 0.0,
+      "learning_rate": 4.804358767064151e-06,
+      "loss": 0.8467,
+      "step": 24894
+    },
+    {
+      "epoch": 0.6837219521573151,
+      "grad_norm": 0.0,
+      "learning_rate": 4.803598757153692e-06,
+      "loss": 0.7414,
+      "step": 24895
+    },
+    {
+      "epoch": 0.6837494163851583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.802838788359173e-06,
+      "loss": 0.8604,
+      "step": 24896
+    },
+    {
+      "epoch": 0.6837768806130016,
+      "grad_norm": 0.0,
+      "learning_rate": 4.802078860686605e-06,
+      "loss": 0.8022,
+      "step": 24897
+    },
+    {
+      "epoch": 0.6838043448408448,
+      "grad_norm": 0.0,
+      "learning_rate": 4.801318974142004e-06,
+      "loss": 0.8673,
+      "step": 24898
+    },
+    {
+      "epoch": 0.683831809068688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.800559128731384e-06,
+      "loss": 0.9436,
+      "step": 24899
+    },
+    {
+      "epoch": 0.6838592732965313,
+      "grad_norm": 0.0,
+      "learning_rate": 4.799799324460755e-06,
+      "loss": 0.9553,
+      "step": 24900
+    },
+    {
+      "epoch": 0.6838867375243745,
+      "grad_norm": 0.0,
+      "learning_rate": 4.799039561336124e-06,
+      "loss": 0.8847,
+      "step": 24901
+    },
+    {
+      "epoch": 0.6839142017522177,
+      "grad_norm": 0.0,
+      "learning_rate": 4.798279839363506e-06,
+      "loss": 0.8282,
+      "step": 24902
+    },
+    {
+      "epoch": 0.683941665980061,
+      "grad_norm": 0.0,
+      "learning_rate": 4.797520158548914e-06,
+      "loss": 0.8163,
+      "step": 24903
+    },
+    {
+      "epoch": 0.6839691302079042,
+      "grad_norm": 0.0,
+      "learning_rate": 4.796760518898358e-06,
+      "loss": 0.9365,
+      "step": 24904
+    },
+    {
+      "epoch": 0.6839965944357475,
+      "grad_norm": 0.0,
+      "learning_rate": 4.796000920417852e-06,
+      "loss": 0.8884,
+      "step": 24905
+    },
+    {
+      "epoch": 0.6840240586635907,
+      "grad_norm": 0.0,
+      "learning_rate": 4.795241363113398e-06,
+      "loss": 0.8025,
+      "step": 24906
+    },
+    {
+      "epoch": 0.6840515228914339,
+      "grad_norm": 0.0,
+      "learning_rate": 4.794481846991013e-06,
+      "loss": 0.9309,
+      "step": 24907
+    },
+    {
+      "epoch": 0.6840789871192772,
+      "grad_norm": 0.0,
+      "learning_rate": 4.793722372056702e-06,
+      "loss": 0.822,
+      "step": 24908
+    },
+    {
+      "epoch": 0.6841064513471203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.792962938316477e-06,
+      "loss": 0.8144,
+      "step": 24909
+    },
+    {
+      "epoch": 0.6841339155749636,
+      "grad_norm": 0.0,
+      "learning_rate": 4.792203545776344e-06,
+      "loss": 0.8623,
+      "step": 24910
+    },
+    {
+      "epoch": 0.6841613798028069,
+      "grad_norm": 0.0,
+      "learning_rate": 4.791444194442319e-06,
+      "loss": 0.9067,
+      "step": 24911
+    },
+    {
+      "epoch": 0.6841888440306501,
+      "grad_norm": 0.0,
+      "learning_rate": 4.790684884320402e-06,
+      "loss": 0.7681,
+      "step": 24912
+    },
+    {
+      "epoch": 0.6842163082584933,
+      "grad_norm": 0.0,
+      "learning_rate": 4.789925615416603e-06,
+      "loss": 0.8209,
+      "step": 24913
+    },
+    {
+      "epoch": 0.6842437724863365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.789166387736931e-06,
+      "loss": 0.7681,
+      "step": 24914
+    },
+    {
+      "epoch": 0.6842712367141798,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7884072012873965e-06,
+      "loss": 0.8768,
+      "step": 24915
+    },
+    {
+      "epoch": 0.684298700942023,
+      "grad_norm": 0.0,
+      "learning_rate": 4.787648056073999e-06,
+      "loss": 0.8022,
+      "step": 24916
+    },
+    {
+      "epoch": 0.6843261651698662,
+      "grad_norm": 0.0,
+      "learning_rate": 4.786888952102754e-06,
+      "loss": 0.8195,
+      "step": 24917
+    },
+    {
+      "epoch": 0.6843536293977095,
+      "grad_norm": 0.0,
+      "learning_rate": 4.786129889379658e-06,
+      "loss": 0.8166,
+      "step": 24918
+    },
+    {
+      "epoch": 0.6843810936255528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.785370867910723e-06,
+      "loss": 0.8878,
+      "step": 24919
+    },
+    {
+      "epoch": 0.6844085578533959,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7846118877019535e-06,
+      "loss": 0.9279,
+      "step": 24920
+    },
+    {
+      "epoch": 0.6844360220812392,
+      "grad_norm": 0.0,
+      "learning_rate": 4.783852948759354e-06,
+      "loss": 0.8616,
+      "step": 24921
+    },
+    {
+      "epoch": 0.6844634863090824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.783094051088936e-06,
+      "loss": 0.8179,
+      "step": 24922
+    },
+    {
+      "epoch": 0.6844909505369257,
+      "grad_norm": 0.0,
+      "learning_rate": 4.782335194696693e-06,
+      "loss": 0.8068,
+      "step": 24923
+    },
+    {
+      "epoch": 0.6845184147647689,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7815763795886405e-06,
+      "loss": 0.8125,
+      "step": 24924
+    },
+    {
+      "epoch": 0.6845458789926121,
+      "grad_norm": 0.0,
+      "learning_rate": 4.780817605770772e-06,
+      "loss": 0.8101,
+      "step": 24925
+    },
+    {
+      "epoch": 0.6845733432204554,
+      "grad_norm": 0.0,
+      "learning_rate": 4.780058873249098e-06,
+      "loss": 0.8627,
+      "step": 24926
+    },
+    {
+      "epoch": 0.6846008074482985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.779300182029619e-06,
+      "loss": 0.8619,
+      "step": 24927
+    },
+    {
+      "epoch": 0.6846282716761418,
+      "grad_norm": 0.0,
+      "learning_rate": 4.778541532118339e-06,
+      "loss": 0.9272,
+      "step": 24928
+    },
+    {
+      "epoch": 0.6846557359039851,
+      "grad_norm": 0.0,
+      "learning_rate": 4.777782923521266e-06,
+      "loss": 0.8906,
+      "step": 24929
+    },
+    {
+      "epoch": 0.6846832001318283,
+      "grad_norm": 0.0,
+      "learning_rate": 4.777024356244393e-06,
+      "loss": 0.7367,
+      "step": 24930
+    },
+    {
+      "epoch": 0.6847106643596715,
+      "grad_norm": 0.0,
+      "learning_rate": 4.776265830293729e-06,
+      "loss": 0.7724,
+      "step": 24931
+    },
+    {
+      "epoch": 0.6847381285875148,
+      "grad_norm": 0.0,
+      "learning_rate": 4.775507345675271e-06,
+      "loss": 0.784,
+      "step": 24932
+    },
+    {
+      "epoch": 0.684765592815358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774748902395022e-06,
+      "loss": 0.7693,
+      "step": 24933
+    },
+    {
+      "epoch": 0.6847930570432013,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7739905004589845e-06,
+      "loss": 0.7493,
+      "step": 24934
+    },
+    {
+      "epoch": 0.6848205212710444,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7732321398731615e-06,
+      "loss": 0.7825,
+      "step": 24935
+    },
+    {
+      "epoch": 0.6848479854988877,
+      "grad_norm": 0.0,
+      "learning_rate": 4.772473820643548e-06,
+      "loss": 0.8933,
+      "step": 24936
+    },
+    {
+      "epoch": 0.684875449726731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.771715542776144e-06,
+      "loss": 0.7877,
+      "step": 24937
+    },
+    {
+      "epoch": 0.6849029139545741,
+      "grad_norm": 0.0,
+      "learning_rate": 4.770957306276954e-06,
+      "loss": 0.8763,
+      "step": 24938
+    },
+    {
+      "epoch": 0.6849303781824174,
+      "grad_norm": 0.0,
+      "learning_rate": 4.770199111151978e-06,
+      "loss": 0.8634,
+      "step": 24939
+    },
+    {
+      "epoch": 0.6849578424102606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.76944095740721e-06,
+      "loss": 0.7688,
+      "step": 24940
+    },
+    {
+      "epoch": 0.6849853066381039,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7686828450486545e-06,
+      "loss": 0.7972,
+      "step": 24941
+    },
+    {
+      "epoch": 0.6850127708659471,
+      "grad_norm": 0.0,
+      "learning_rate": 4.767924774082302e-06,
+      "loss": 0.838,
+      "step": 24942
+    },
+    {
+      "epoch": 0.6850402350937903,
+      "grad_norm": 0.0,
+      "learning_rate": 4.767166744514157e-06,
+      "loss": 0.864,
+      "step": 24943
+    },
+    {
+      "epoch": 0.6850676993216336,
+      "grad_norm": 0.0,
+      "learning_rate": 4.766408756350214e-06,
+      "loss": 0.8549,
+      "step": 24944
+    },
+    {
+      "epoch": 0.6850951635494769,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7656508095964735e-06,
+      "loss": 0.7849,
+      "step": 24945
+    },
+    {
+      "epoch": 0.68512262777732,
+      "grad_norm": 0.0,
+      "learning_rate": 4.764892904258934e-06,
+      "loss": 0.8148,
+      "step": 24946
+    },
+    {
+      "epoch": 0.6851500920051633,
+      "grad_norm": 0.0,
+      "learning_rate": 4.764135040343587e-06,
+      "loss": 0.8464,
+      "step": 24947
+    },
+    {
+      "epoch": 0.6851775562330065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.763377217856435e-06,
+      "loss": 0.9023,
+      "step": 24948
+    },
+    {
+      "epoch": 0.6852050204608497,
+      "grad_norm": 0.0,
+      "learning_rate": 4.762619436803467e-06,
+      "loss": 0.8815,
+      "step": 24949
+    },
+    {
+      "epoch": 0.685232484688693,
+      "grad_norm": 0.0,
+      "learning_rate": 4.761861697190684e-06,
+      "loss": 0.8155,
+      "step": 24950
+    },
+    {
+      "epoch": 0.6852599489165362,
+      "grad_norm": 0.0,
+      "learning_rate": 4.761103999024079e-06,
+      "loss": 0.8213,
+      "step": 24951
+    },
+    {
+      "epoch": 0.6852874131443795,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7603463423096496e-06,
+      "loss": 0.8871,
+      "step": 24952
+    },
+    {
+      "epoch": 0.6853148773722226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.759588727053392e-06,
+      "loss": 0.9236,
+      "step": 24953
+    },
+    {
+      "epoch": 0.6853423416000659,
+      "grad_norm": 0.0,
+      "learning_rate": 4.758831153261295e-06,
+      "loss": 0.8454,
+      "step": 24954
+    },
+    {
+      "epoch": 0.6853698058279092,
+      "grad_norm": 0.0,
+      "learning_rate": 4.758073620939361e-06,
+      "loss": 0.7012,
+      "step": 24955
+    },
+    {
+      "epoch": 0.6853972700557524,
+      "grad_norm": 0.0,
+      "learning_rate": 4.757316130093574e-06,
+      "loss": 0.885,
+      "step": 24956
+    },
+    {
+      "epoch": 0.6854247342835956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7565586807299336e-06,
+      "loss": 0.7341,
+      "step": 24957
+    },
+    {
+      "epoch": 0.6854521985114389,
+      "grad_norm": 0.0,
+      "learning_rate": 4.755801272854431e-06,
+      "loss": 0.844,
+      "step": 24958
+    },
+    {
+      "epoch": 0.6854796627392821,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7550439064730645e-06,
+      "loss": 0.8616,
+      "step": 24959
+    },
+    {
+      "epoch": 0.6855071269671253,
+      "grad_norm": 0.0,
+      "learning_rate": 4.754286581591818e-06,
+      "loss": 0.9367,
+      "step": 24960
+    },
+    {
+      "epoch": 0.6855345911949685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.753529298216689e-06,
+      "loss": 0.8589,
+      "step": 24961
+    },
+    {
+      "epoch": 0.6855620554228118,
+      "grad_norm": 0.0,
+      "learning_rate": 4.752772056353668e-06,
+      "loss": 0.802,
+      "step": 24962
+    },
+    {
+      "epoch": 0.6855895196506551,
+      "grad_norm": 0.0,
+      "learning_rate": 4.752014856008751e-06,
+      "loss": 0.8041,
+      "step": 24963
+    },
+    {
+      "epoch": 0.6856169838784982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.751257697187925e-06,
+      "loss": 0.7921,
+      "step": 24964
+    },
+    {
+      "epoch": 0.6856444481063415,
+      "grad_norm": 0.0,
+      "learning_rate": 4.750500579897178e-06,
+      "loss": 0.7562,
+      "step": 24965
+    },
+    {
+      "epoch": 0.6856719123341847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.749743504142503e-06,
+      "loss": 0.8278,
+      "step": 24966
+    },
+    {
+      "epoch": 0.685699376562028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.748986469929893e-06,
+      "loss": 0.8259,
+      "step": 24967
+    },
+    {
+      "epoch": 0.6857268407898712,
+      "grad_norm": 0.0,
+      "learning_rate": 4.748229477265334e-06,
+      "loss": 0.8401,
+      "step": 24968
+    },
+    {
+      "epoch": 0.6857543050177144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7474725261548185e-06,
+      "loss": 0.7798,
+      "step": 24969
+    },
+    {
+      "epoch": 0.6857817692455577,
+      "grad_norm": 0.0,
+      "learning_rate": 4.746715616604339e-06,
+      "loss": 0.8,
+      "step": 24970
+    },
+    {
+      "epoch": 0.685809233473401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.745958748619876e-06,
+      "loss": 0.8266,
+      "step": 24971
+    },
+    {
+      "epoch": 0.6858366977012441,
+      "grad_norm": 0.0,
+      "learning_rate": 4.745201922207427e-06,
+      "loss": 0.9044,
+      "step": 24972
+    },
+    {
+      "epoch": 0.6858641619290874,
+      "grad_norm": 0.0,
+      "learning_rate": 4.744445137372972e-06,
+      "loss": 0.8616,
+      "step": 24973
+    },
+    {
+      "epoch": 0.6858916261569306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.743688394122503e-06,
+      "loss": 0.8622,
+      "step": 24974
+    },
+    {
+      "epoch": 0.6859190903847738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7429316924620075e-06,
+      "loss": 0.8007,
+      "step": 24975
+    },
+    {
+      "epoch": 0.6859465546126171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.742175032397477e-06,
+      "loss": 0.9507,
+      "step": 24976
+    },
+    {
+      "epoch": 0.6859740188404603,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7414184139348896e-06,
+      "loss": 0.8662,
+      "step": 24977
+    },
+    {
+      "epoch": 0.6860014830683036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7406618370802375e-06,
+      "loss": 0.7704,
+      "step": 24978
+    },
+    {
+      "epoch": 0.6860289472961467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.739905301839509e-06,
+      "loss": 0.8052,
+      "step": 24979
+    },
+    {
+      "epoch": 0.68605641152399,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7391488082186845e-06,
+      "loss": 0.7524,
+      "step": 24980
+    },
+    {
+      "epoch": 0.6860838757518333,
+      "grad_norm": 0.0,
+      "learning_rate": 4.738392356223752e-06,
+      "loss": 0.9083,
+      "step": 24981
+    },
+    {
+      "epoch": 0.6861113399796764,
+      "grad_norm": 0.0,
+      "learning_rate": 4.737635945860701e-06,
+      "loss": 0.9073,
+      "step": 24982
+    },
+    {
+      "epoch": 0.6861388042075197,
+      "grad_norm": 0.0,
+      "learning_rate": 4.73687957713551e-06,
+      "loss": 0.7896,
+      "step": 24983
+    },
+    {
+      "epoch": 0.686166268435363,
+      "grad_norm": 0.0,
+      "learning_rate": 4.736123250054166e-06,
+      "loss": 0.8795,
+      "step": 24984
+    },
+    {
+      "epoch": 0.6861937326632062,
+      "grad_norm": 0.0,
+      "learning_rate": 4.735366964622653e-06,
+      "loss": 0.8261,
+      "step": 24985
+    },
+    {
+      "epoch": 0.6862211968910494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.734610720846957e-06,
+      "loss": 0.823,
+      "step": 24986
+    },
+    {
+      "epoch": 0.6862486611188926,
+      "grad_norm": 0.0,
+      "learning_rate": 4.733854518733064e-06,
+      "loss": 0.7761,
+      "step": 24987
+    },
+    {
+      "epoch": 0.6862761253467359,
+      "grad_norm": 0.0,
+      "learning_rate": 4.733098358286954e-06,
+      "loss": 0.7909,
+      "step": 24988
+    },
+    {
+      "epoch": 0.6863035895745792,
+      "grad_norm": 0.0,
+      "learning_rate": 4.732342239514606e-06,
+      "loss": 0.7775,
+      "step": 24989
+    },
+    {
+      "epoch": 0.6863310538024223,
+      "grad_norm": 0.0,
+      "learning_rate": 4.731586162422007e-06,
+      "loss": 0.916,
+      "step": 24990
+    },
+    {
+      "epoch": 0.6863585180302656,
+      "grad_norm": 0.0,
+      "learning_rate": 4.730830127015139e-06,
+      "loss": 0.9096,
+      "step": 24991
+    },
+    {
+      "epoch": 0.6863859822581088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.730074133299984e-06,
+      "loss": 0.764,
+      "step": 24992
+    },
+    {
+      "epoch": 0.686413446485952,
+      "grad_norm": 0.0,
+      "learning_rate": 4.729318181282524e-06,
+      "loss": 0.8557,
+      "step": 24993
+    },
+    {
+      "epoch": 0.6864409107137953,
+      "grad_norm": 0.0,
+      "learning_rate": 4.728562270968744e-06,
+      "loss": 0.8796,
+      "step": 24994
+    },
+    {
+      "epoch": 0.6864683749416385,
+      "grad_norm": 0.0,
+      "learning_rate": 4.727806402364617e-06,
+      "loss": 0.8614,
+      "step": 24995
+    },
+    {
+      "epoch": 0.6864958391694818,
+      "grad_norm": 0.0,
+      "learning_rate": 4.727050575476133e-06,
+      "loss": 0.8163,
+      "step": 24996
+    },
+    {
+      "epoch": 0.686523303397325,
+      "grad_norm": 0.0,
+      "learning_rate": 4.726294790309263e-06,
+      "loss": 0.8196,
+      "step": 24997
+    },
+    {
+      "epoch": 0.6865507676251682,
+      "grad_norm": 0.0,
+      "learning_rate": 4.72553904686999e-06,
+      "loss": 0.7875,
+      "step": 24998
+    },
+    {
+      "epoch": 0.6865782318530115,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7247833451642965e-06,
+      "loss": 0.7324,
+      "step": 24999
+    },
+    {
+      "epoch": 0.6866056960808546,
+      "grad_norm": 0.0,
+      "learning_rate": 4.724027685198165e-06,
+      "loss": 0.846,
+      "step": 25000
+    },
+    {
+      "epoch": 0.6866331603086979,
+      "grad_norm": 0.0,
+      "learning_rate": 4.723272066977565e-06,
+      "loss": 0.8647,
+      "step": 25001
+    },
+    {
+      "epoch": 0.6866606245365412,
+      "grad_norm": 0.0,
+      "learning_rate": 4.722516490508482e-06,
+      "loss": 0.8766,
+      "step": 25002
+    },
+    {
+      "epoch": 0.6866880887643844,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7217609557968956e-06,
+      "loss": 0.8756,
+      "step": 25003
+    },
+    {
+      "epoch": 0.6867155529922276,
+      "grad_norm": 0.0,
+      "learning_rate": 4.721005462848778e-06,
+      "loss": 0.8535,
+      "step": 25004
+    },
+    {
+      "epoch": 0.6867430172200708,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7202500116701085e-06,
+      "loss": 0.905,
+      "step": 25005
+    },
+    {
+      "epoch": 0.6867704814479141,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7194946022668704e-06,
+      "loss": 0.9427,
+      "step": 25006
+    },
+    {
+      "epoch": 0.6867979456757574,
+      "grad_norm": 0.0,
+      "learning_rate": 4.718739234645034e-06,
+      "loss": 0.758,
+      "step": 25007
+    },
+    {
+      "epoch": 0.6868254099036005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.717983908810578e-06,
+      "loss": 0.8079,
+      "step": 25008
+    },
+    {
+      "epoch": 0.6868528741314438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.717228624769479e-06,
+      "loss": 0.7559,
+      "step": 25009
+    },
+    {
+      "epoch": 0.6868803383592871,
+      "grad_norm": 0.0,
+      "learning_rate": 4.716473382527713e-06,
+      "loss": 0.891,
+      "step": 25010
+    },
+    {
+      "epoch": 0.6869078025871302,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7157181820912616e-06,
+      "loss": 0.8693,
+      "step": 25011
+    },
+    {
+      "epoch": 0.6869352668149735,
+      "grad_norm": 0.0,
+      "learning_rate": 4.714963023466093e-06,
+      "loss": 0.8177,
+      "step": 25012
+    },
+    {
+      "epoch": 0.6869627310428167,
+      "grad_norm": 0.0,
+      "learning_rate": 4.714207906658181e-06,
+      "loss": 0.876,
+      "step": 25013
+    },
+    {
+      "epoch": 0.68699019527066,
+      "grad_norm": 0.0,
+      "learning_rate": 4.713452831673504e-06,
+      "loss": 0.8372,
+      "step": 25014
+    },
+    {
+      "epoch": 0.6870176594985032,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7126977985180365e-06,
+      "loss": 0.9138,
+      "step": 25015
+    },
+    {
+      "epoch": 0.6870451237263464,
+      "grad_norm": 0.0,
+      "learning_rate": 4.711942807197751e-06,
+      "loss": 0.8184,
+      "step": 25016
+    },
+    {
+      "epoch": 0.6870725879541897,
+      "grad_norm": 0.0,
+      "learning_rate": 4.711187857718623e-06,
+      "loss": 0.7835,
+      "step": 25017
+    },
+    {
+      "epoch": 0.6871000521820328,
+      "grad_norm": 0.0,
+      "learning_rate": 4.71043295008663e-06,
+      "loss": 0.9023,
+      "step": 25018
+    },
+    {
+      "epoch": 0.6871275164098761,
+      "grad_norm": 0.0,
+      "learning_rate": 4.709678084307735e-06,
+      "loss": 0.9438,
+      "step": 25019
+    },
+    {
+      "epoch": 0.6871549806377194,
+      "grad_norm": 0.0,
+      "learning_rate": 4.708923260387921e-06,
+      "loss": 0.823,
+      "step": 25020
+    },
+    {
+      "epoch": 0.6871824448655626,
+      "grad_norm": 0.0,
+      "learning_rate": 4.708168478333152e-06,
+      "loss": 0.8147,
+      "step": 25021
+    },
+    {
+      "epoch": 0.6872099090934058,
+      "grad_norm": 0.0,
+      "learning_rate": 4.707413738149404e-06,
+      "loss": 0.7722,
+      "step": 25022
+    },
+    {
+      "epoch": 0.6872373733212491,
+      "grad_norm": 0.0,
+      "learning_rate": 4.706659039842652e-06,
+      "loss": 0.8674,
+      "step": 25023
+    },
+    {
+      "epoch": 0.6872648375490923,
+      "grad_norm": 0.0,
+      "learning_rate": 4.705904383418859e-06,
+      "loss": 0.8051,
+      "step": 25024
+    },
+    {
+      "epoch": 0.6872923017769356,
+      "grad_norm": 0.0,
+      "learning_rate": 4.705149768884003e-06,
+      "loss": 0.8405,
+      "step": 25025
+    },
+    {
+      "epoch": 0.6873197660047787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.704395196244052e-06,
+      "loss": 0.8656,
+      "step": 25026
+    },
+    {
+      "epoch": 0.687347230232622,
+      "grad_norm": 0.0,
+      "learning_rate": 4.703640665504983e-06,
+      "loss": 0.7847,
+      "step": 25027
+    },
+    {
+      "epoch": 0.6873746944604653,
+      "grad_norm": 0.0,
+      "learning_rate": 4.702886176672755e-06,
+      "loss": 0.8199,
+      "step": 25028
+    },
+    {
+      "epoch": 0.6874021586883085,
+      "grad_norm": 0.0,
+      "learning_rate": 4.702131729753349e-06,
+      "loss": 0.8405,
+      "step": 25029
+    },
+    {
+      "epoch": 0.6874296229161517,
+      "grad_norm": 0.0,
+      "learning_rate": 4.701377324752724e-06,
+      "loss": 0.7946,
+      "step": 25030
+    },
+    {
+      "epoch": 0.6874570871439949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.700622961676854e-06,
+      "loss": 0.8738,
+      "step": 25031
+    },
+    {
+      "epoch": 0.6874845513718382,
+      "grad_norm": 0.0,
+      "learning_rate": 4.699868640531708e-06,
+      "loss": 0.8298,
+      "step": 25032
+    },
+    {
+      "epoch": 0.6875120155996814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.699114361323256e-06,
+      "loss": 0.8691,
+      "step": 25033
+    },
+    {
+      "epoch": 0.6875394798275246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.698360124057463e-06,
+      "loss": 0.786,
+      "step": 25034
+    },
+    {
+      "epoch": 0.6875669440553679,
+      "grad_norm": 0.0,
+      "learning_rate": 4.697605928740302e-06,
+      "loss": 0.9417,
+      "step": 25035
+    },
+    {
+      "epoch": 0.6875944082832112,
+      "grad_norm": 0.0,
+      "learning_rate": 4.696851775377738e-06,
+      "loss": 0.7706,
+      "step": 25036
+    },
+    {
+      "epoch": 0.6876218725110543,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6960976639757315e-06,
+      "loss": 0.8733,
+      "step": 25037
+    },
+    {
+      "epoch": 0.6876493367388976,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6953435945402555e-06,
+      "loss": 0.8405,
+      "step": 25038
+    },
+    {
+      "epoch": 0.6876768009667408,
+      "grad_norm": 0.0,
+      "learning_rate": 4.694589567077276e-06,
+      "loss": 0.8276,
+      "step": 25039
+    },
+    {
+      "epoch": 0.687704265194584,
+      "grad_norm": 0.0,
+      "learning_rate": 4.69383558159276e-06,
+      "loss": 0.8845,
+      "step": 25040
+    },
+    {
+      "epoch": 0.6877317294224273,
+      "grad_norm": 0.0,
+      "learning_rate": 4.693081638092674e-06,
+      "loss": 0.7761,
+      "step": 25041
+    },
+    {
+      "epoch": 0.6877591936502705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.69232773658298e-06,
+      "loss": 0.9334,
+      "step": 25042
+    },
+    {
+      "epoch": 0.6877866578781138,
+      "grad_norm": 0.0,
+      "learning_rate": 4.691573877069647e-06,
+      "loss": 0.8716,
+      "step": 25043
+    },
+    {
+      "epoch": 0.6878141221059569,
+      "grad_norm": 0.0,
+      "learning_rate": 4.690820059558635e-06,
+      "loss": 0.8208,
+      "step": 25044
+    },
+    {
+      "epoch": 0.6878415863338002,
+      "grad_norm": 0.0,
+      "learning_rate": 4.690066284055912e-06,
+      "loss": 0.9056,
+      "step": 25045
+    },
+    {
+      "epoch": 0.6878690505616435,
+      "grad_norm": 0.0,
+      "learning_rate": 4.689312550567439e-06,
+      "loss": 0.8522,
+      "step": 25046
+    },
+    {
+      "epoch": 0.6878965147894867,
+      "grad_norm": 0.0,
+      "learning_rate": 4.688558859099188e-06,
+      "loss": 0.8509,
+      "step": 25047
+    },
+    {
+      "epoch": 0.6879239790173299,
+      "grad_norm": 0.0,
+      "learning_rate": 4.687805209657113e-06,
+      "loss": 0.7918,
+      "step": 25048
+    },
+    {
+      "epoch": 0.6879514432451732,
+      "grad_norm": 0.0,
+      "learning_rate": 4.68705160224718e-06,
+      "loss": 0.8343,
+      "step": 25049
+    },
+    {
+      "epoch": 0.6879789074730164,
+      "grad_norm": 0.0,
+      "learning_rate": 4.686298036875354e-06,
+      "loss": 0.8516,
+      "step": 25050
+    },
+    {
+      "epoch": 0.6880063717008597,
+      "grad_norm": 0.0,
+      "learning_rate": 4.685544513547599e-06,
+      "loss": 0.775,
+      "step": 25051
+    },
+    {
+      "epoch": 0.6880338359287028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.684791032269871e-06,
+      "loss": 0.8719,
+      "step": 25052
+    },
+    {
+      "epoch": 0.6880613001565461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6840375930481384e-06,
+      "loss": 0.8036,
+      "step": 25053
+    },
+    {
+      "epoch": 0.6880887643843894,
+      "grad_norm": 0.0,
+      "learning_rate": 4.683284195888357e-06,
+      "loss": 0.8556,
+      "step": 25054
+    },
+    {
+      "epoch": 0.6881162286122325,
+      "grad_norm": 0.0,
+      "learning_rate": 4.68253084079649e-06,
+      "loss": 0.8473,
+      "step": 25055
+    },
+    {
+      "epoch": 0.6881436928400758,
+      "grad_norm": 0.0,
+      "learning_rate": 4.681777527778498e-06,
+      "loss": 0.771,
+      "step": 25056
+    },
+    {
+      "epoch": 0.688171157067919,
+      "grad_norm": 0.0,
+      "learning_rate": 4.681024256840343e-06,
+      "loss": 0.7269,
+      "step": 25057
+    },
+    {
+      "epoch": 0.6881986212957623,
+      "grad_norm": 0.0,
+      "learning_rate": 4.680271027987985e-06,
+      "loss": 0.7367,
+      "step": 25058
+    },
+    {
+      "epoch": 0.6882260855236055,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6795178412273865e-06,
+      "loss": 0.8449,
+      "step": 25059
+    },
+    {
+      "epoch": 0.6882535497514487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.678764696564504e-06,
+      "loss": 0.8385,
+      "step": 25060
+    },
+    {
+      "epoch": 0.688281013979292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.678011594005292e-06,
+      "loss": 0.7817,
+      "step": 25061
+    },
+    {
+      "epoch": 0.6883084782071353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.677258533555714e-06,
+      "loss": 0.86,
+      "step": 25062
+    },
+    {
+      "epoch": 0.6883359424349784,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6765055152217285e-06,
+      "loss": 0.9576,
+      "step": 25063
+    },
+    {
+      "epoch": 0.6883634066628217,
+      "grad_norm": 0.0,
+      "learning_rate": 4.675752539009294e-06,
+      "loss": 0.8304,
+      "step": 25064
+    },
+    {
+      "epoch": 0.6883908708906649,
+      "grad_norm": 0.0,
+      "learning_rate": 4.674999604924372e-06,
+      "loss": 0.8216,
+      "step": 25065
+    },
+    {
+      "epoch": 0.6884183351185081,
+      "grad_norm": 0.0,
+      "learning_rate": 4.674246712972911e-06,
+      "loss": 0.9061,
+      "step": 25066
+    },
+    {
+      "epoch": 0.6884457993463514,
+      "grad_norm": 0.0,
+      "learning_rate": 4.673493863160876e-06,
+      "loss": 0.9265,
+      "step": 25067
+    },
+    {
+      "epoch": 0.6884732635741946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.672741055494219e-06,
+      "loss": 0.9421,
+      "step": 25068
+    },
+    {
+      "epoch": 0.6885007278020379,
+      "grad_norm": 0.0,
+      "learning_rate": 4.671988289978897e-06,
+      "loss": 0.8745,
+      "step": 25069
+    },
+    {
+      "epoch": 0.688528192029881,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6712355666208674e-06,
+      "loss": 0.853,
+      "step": 25070
+    },
+    {
+      "epoch": 0.6885556562577243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.670482885426091e-06,
+      "loss": 0.9018,
+      "step": 25071
+    },
+    {
+      "epoch": 0.6885831204855676,
+      "grad_norm": 0.0,
+      "learning_rate": 4.669730246400515e-06,
+      "loss": 0.8049,
+      "step": 25072
+    },
+    {
+      "epoch": 0.6886105847134107,
+      "grad_norm": 0.0,
+      "learning_rate": 4.668977649550097e-06,
+      "loss": 0.9003,
+      "step": 25073
+    },
+    {
+      "epoch": 0.688638048941254,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6682250948807925e-06,
+      "loss": 0.8433,
+      "step": 25074
+    },
+    {
+      "epoch": 0.6886655131690973,
+      "grad_norm": 0.0,
+      "learning_rate": 4.667472582398561e-06,
+      "loss": 0.7355,
+      "step": 25075
+    },
+    {
+      "epoch": 0.6886929773969405,
+      "grad_norm": 0.0,
+      "learning_rate": 4.666720112109348e-06,
+      "loss": 0.7773,
+      "step": 25076
+    },
+    {
+      "epoch": 0.6887204416247837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.665967684019115e-06,
+      "loss": 0.7691,
+      "step": 25077
+    },
+    {
+      "epoch": 0.6887479058526269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.665215298133808e-06,
+      "loss": 0.8249,
+      "step": 25078
+    },
+    {
+      "epoch": 0.6887753700804702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.664462954459386e-06,
+      "loss": 0.9016,
+      "step": 25079
+    },
+    {
+      "epoch": 0.6888028343083135,
+      "grad_norm": 0.0,
+      "learning_rate": 4.663710653001797e-06,
+      "loss": 0.8949,
+      "step": 25080
+    },
+    {
+      "epoch": 0.6888302985361566,
+      "grad_norm": 0.0,
+      "learning_rate": 4.662958393766998e-06,
+      "loss": 0.7504,
+      "step": 25081
+    },
+    {
+      "epoch": 0.6888577627639999,
+      "grad_norm": 0.0,
+      "learning_rate": 4.662206176760943e-06,
+      "loss": 0.8625,
+      "step": 25082
+    },
+    {
+      "epoch": 0.6888852269918431,
+      "grad_norm": 0.0,
+      "learning_rate": 4.661454001989577e-06,
+      "loss": 0.8485,
+      "step": 25083
+    },
+    {
+      "epoch": 0.6889126912196863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.660701869458858e-06,
+      "loss": 0.8036,
+      "step": 25084
+    },
+    {
+      "epoch": 0.6889401554475296,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6599497791747305e-06,
+      "loss": 0.8561,
+      "step": 25085
+    },
+    {
+      "epoch": 0.6889676196753728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.659197731143149e-06,
+      "loss": 0.8824,
+      "step": 25086
+    },
+    {
+      "epoch": 0.6889950839032161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.658445725370064e-06,
+      "loss": 0.8729,
+      "step": 25087
+    },
+    {
+      "epoch": 0.6890225481310593,
+      "grad_norm": 0.0,
+      "learning_rate": 4.657693761861429e-06,
+      "loss": 0.8707,
+      "step": 25088
+    },
+    {
+      "epoch": 0.6890500123589025,
+      "grad_norm": 0.0,
+      "learning_rate": 4.656941840623187e-06,
+      "loss": 0.7777,
+      "step": 25089
+    },
+    {
+      "epoch": 0.6890774765867458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.65618996166129e-06,
+      "loss": 0.9553,
+      "step": 25090
+    },
+    {
+      "epoch": 0.689104940814589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.655438124981693e-06,
+      "loss": 0.8322,
+      "step": 25091
+    },
+    {
+      "epoch": 0.6891324050424322,
+      "grad_norm": 0.0,
+      "learning_rate": 4.654686330590335e-06,
+      "loss": 0.8705,
+      "step": 25092
+    },
+    {
+      "epoch": 0.6891598692702755,
+      "grad_norm": 0.0,
+      "learning_rate": 4.65393457849317e-06,
+      "loss": 0.7953,
+      "step": 25093
+    },
+    {
+      "epoch": 0.6891873334981187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.653182868696149e-06,
+      "loss": 0.8065,
+      "step": 25094
+    },
+    {
+      "epoch": 0.689214797725962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.652431201205213e-06,
+      "loss": 0.8455,
+      "step": 25095
+    },
+    {
+      "epoch": 0.6892422619538052,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6516795760263125e-06,
+      "loss": 0.7964,
+      "step": 25096
+    },
+    {
+      "epoch": 0.6892697261816484,
+      "grad_norm": 0.0,
+      "learning_rate": 4.650927993165395e-06,
+      "loss": 0.9007,
+      "step": 25097
+    },
+    {
+      "epoch": 0.6892971904094917,
+      "grad_norm": 0.0,
+      "learning_rate": 4.650176452628407e-06,
+      "loss": 0.8766,
+      "step": 25098
+    },
+    {
+      "epoch": 0.6893246546373348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6494249544212985e-06,
+      "loss": 0.8238,
+      "step": 25099
+    },
+    {
+      "epoch": 0.6893521188651781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6486734985500135e-06,
+      "loss": 0.7747,
+      "step": 25100
+    },
+    {
+      "epoch": 0.6893795830930214,
+      "grad_norm": 0.0,
+      "learning_rate": 4.647922085020492e-06,
+      "loss": 0.902,
+      "step": 25101
+    },
+    {
+      "epoch": 0.6894070473208646,
+      "grad_norm": 0.0,
+      "learning_rate": 4.647170713838685e-06,
+      "loss": 0.8306,
+      "step": 25102
+    },
+    {
+      "epoch": 0.6894345115487078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.646419385010536e-06,
+      "loss": 0.8075,
+      "step": 25103
+    },
+    {
+      "epoch": 0.689461975776551,
+      "grad_norm": 0.0,
+      "learning_rate": 4.645668098541991e-06,
+      "loss": 0.8253,
+      "step": 25104
+    },
+    {
+      "epoch": 0.6894894400043943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.644916854438993e-06,
+      "loss": 0.8289,
+      "step": 25105
+    },
+    {
+      "epoch": 0.6895169042322375,
+      "grad_norm": 0.0,
+      "learning_rate": 4.644165652707493e-06,
+      "loss": 0.844,
+      "step": 25106
+    },
+    {
+      "epoch": 0.6895443684600807,
+      "grad_norm": 0.0,
+      "learning_rate": 4.643414493353424e-06,
+      "loss": 0.8734,
+      "step": 25107
+    },
+    {
+      "epoch": 0.689571832687924,
+      "grad_norm": 0.0,
+      "learning_rate": 4.642663376382739e-06,
+      "loss": 0.8593,
+      "step": 25108
+    },
+    {
+      "epoch": 0.6895992969157673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.641912301801371e-06,
+      "loss": 0.8261,
+      "step": 25109
+    },
+    {
+      "epoch": 0.6896267611436104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.641161269615271e-06,
+      "loss": 0.8335,
+      "step": 25110
+    },
+    {
+      "epoch": 0.6896542253714537,
+      "grad_norm": 0.0,
+      "learning_rate": 4.640410279830377e-06,
+      "loss": 0.8648,
+      "step": 25111
+    },
+    {
+      "epoch": 0.6896816895992969,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6396593324526376e-06,
+      "loss": 0.8204,
+      "step": 25112
+    },
+    {
+      "epoch": 0.6897091538271402,
+      "grad_norm": 0.0,
+      "learning_rate": 4.638908427487986e-06,
+      "loss": 0.8374,
+      "step": 25113
+    },
+    {
+      "epoch": 0.6897366180549834,
+      "grad_norm": 0.0,
+      "learning_rate": 4.638157564942367e-06,
+      "loss": 0.8513,
+      "step": 25114
+    },
+    {
+      "epoch": 0.6897640822828266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.637406744821727e-06,
+      "loss": 0.8424,
+      "step": 25115
+    },
+    {
+      "epoch": 0.6897915465106699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6366559671319974e-06,
+      "loss": 0.8554,
+      "step": 25116
+    },
+    {
+      "epoch": 0.689819010738513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6359052318791245e-06,
+      "loss": 0.8135,
+      "step": 25117
+    },
+    {
+      "epoch": 0.6898464749663563,
+      "grad_norm": 0.0,
+      "learning_rate": 4.63515453906905e-06,
+      "loss": 0.8659,
+      "step": 25118
+    },
+    {
+      "epoch": 0.6898739391941996,
+      "grad_norm": 0.0,
+      "learning_rate": 4.634403888707707e-06,
+      "loss": 0.8585,
+      "step": 25119
+    },
+    {
+      "epoch": 0.6899014034220428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.633653280801039e-06,
+      "loss": 0.8605,
+      "step": 25120
+    },
+    {
+      "epoch": 0.689928867649886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.632902715354986e-06,
+      "loss": 0.8155,
+      "step": 25121
+    },
+    {
+      "epoch": 0.6899563318777293,
+      "grad_norm": 0.0,
+      "learning_rate": 4.632152192375485e-06,
+      "loss": 0.8854,
+      "step": 25122
+    },
+    {
+      "epoch": 0.6899837961055725,
+      "grad_norm": 0.0,
+      "learning_rate": 4.631401711868479e-06,
+      "loss": 0.8801,
+      "step": 25123
+    },
+    {
+      "epoch": 0.6900112603334158,
+      "grad_norm": 0.0,
+      "learning_rate": 4.630651273839902e-06,
+      "loss": 0.8715,
+      "step": 25124
+    },
+    {
+      "epoch": 0.6900387245612589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6299008782956885e-06,
+      "loss": 0.8351,
+      "step": 25125
+    },
+    {
+      "epoch": 0.6900661887891022,
+      "grad_norm": 0.0,
+      "learning_rate": 4.62915052524178e-06,
+      "loss": 0.9158,
+      "step": 25126
+    },
+    {
+      "epoch": 0.6900936530169455,
+      "grad_norm": 0.0,
+      "learning_rate": 4.628400214684113e-06,
+      "loss": 0.8429,
+      "step": 25127
+    },
+    {
+      "epoch": 0.6901211172447886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.627649946628623e-06,
+      "loss": 0.7743,
+      "step": 25128
+    },
+    {
+      "epoch": 0.6901485814726319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.626899721081248e-06,
+      "loss": 0.7153,
+      "step": 25129
+    },
+    {
+      "epoch": 0.6901760457004751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.626149538047928e-06,
+      "loss": 0.9181,
+      "step": 25130
+    },
+    {
+      "epoch": 0.6902035099283184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.625399397534591e-06,
+      "loss": 0.8273,
+      "step": 25131
+    },
+    {
+      "epoch": 0.6902309741561616,
+      "grad_norm": 0.0,
+      "learning_rate": 4.624649299547179e-06,
+      "loss": 0.7796,
+      "step": 25132
+    },
+    {
+      "epoch": 0.6902584383840048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.623899244091621e-06,
+      "loss": 0.8947,
+      "step": 25133
+    },
+    {
+      "epoch": 0.6902859026118481,
+      "grad_norm": 0.0,
+      "learning_rate": 4.623149231173854e-06,
+      "loss": 0.8234,
+      "step": 25134
+    },
+    {
+      "epoch": 0.6903133668396914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.622399260799814e-06,
+      "loss": 0.8068,
+      "step": 25135
+    },
+    {
+      "epoch": 0.6903408310675345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.621649332975436e-06,
+      "loss": 0.8858,
+      "step": 25136
+    },
+    {
+      "epoch": 0.6903682952953778,
+      "grad_norm": 0.0,
+      "learning_rate": 4.620899447706649e-06,
+      "loss": 0.7571,
+      "step": 25137
+    },
+    {
+      "epoch": 0.690395759523221,
+      "grad_norm": 0.0,
+      "learning_rate": 4.62014960499939e-06,
+      "loss": 0.8554,
+      "step": 25138
+    },
+    {
+      "epoch": 0.6904232237510642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.619399804859594e-06,
+      "loss": 0.876,
+      "step": 25139
+    },
+    {
+      "epoch": 0.6904506879789075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.618650047293189e-06,
+      "loss": 0.8434,
+      "step": 25140
+    },
+    {
+      "epoch": 0.6904781522067507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.617900332306111e-06,
+      "loss": 0.8371,
+      "step": 25141
+    },
+    {
+      "epoch": 0.690505616434594,
+      "grad_norm": 0.0,
+      "learning_rate": 4.617150659904287e-06,
+      "loss": 0.8039,
+      "step": 25142
+    },
+    {
+      "epoch": 0.6905330806624371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.616401030093653e-06,
+      "loss": 0.8954,
+      "step": 25143
+    },
+    {
+      "epoch": 0.6905605448902804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.615651442880139e-06,
+      "loss": 0.8922,
+      "step": 25144
+    },
+    {
+      "epoch": 0.6905880091181237,
+      "grad_norm": 0.0,
+      "learning_rate": 4.614901898269677e-06,
+      "loss": 0.8335,
+      "step": 25145
+    },
+    {
+      "epoch": 0.6906154733459668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.614152396268196e-06,
+      "loss": 0.7472,
+      "step": 25146
+    },
+    {
+      "epoch": 0.6906429375738101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.613402936881633e-06,
+      "loss": 0.7526,
+      "step": 25147
+    },
+    {
+      "epoch": 0.6906704018016534,
+      "grad_norm": 0.0,
+      "learning_rate": 4.61265352011591e-06,
+      "loss": 0.7494,
+      "step": 25148
+    },
+    {
+      "epoch": 0.6906978660294966,
+      "grad_norm": 0.0,
+      "learning_rate": 4.611904145976958e-06,
+      "loss": 0.8988,
+      "step": 25149
+    },
+    {
+      "epoch": 0.6907253302573398,
+      "grad_norm": 0.0,
+      "learning_rate": 4.611154814470706e-06,
+      "loss": 0.8755,
+      "step": 25150
+    },
+    {
+      "epoch": 0.690752794485183,
+      "grad_norm": 0.0,
+      "learning_rate": 4.610405525603085e-06,
+      "loss": 0.9513,
+      "step": 25151
+    },
+    {
+      "epoch": 0.6907802587130263,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6096562793800234e-06,
+      "loss": 0.8825,
+      "step": 25152
+    },
+    {
+      "epoch": 0.6908077229408696,
+      "grad_norm": 0.0,
+      "learning_rate": 4.608907075807453e-06,
+      "loss": 0.8613,
+      "step": 25153
+    },
+    {
+      "epoch": 0.6908351871687127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.608157914891294e-06,
+      "loss": 0.8125,
+      "step": 25154
+    },
+    {
+      "epoch": 0.690862651396556,
+      "grad_norm": 0.0,
+      "learning_rate": 4.607408796637478e-06,
+      "loss": 0.8317,
+      "step": 25155
+    },
+    {
+      "epoch": 0.6908901156243992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.606659721051937e-06,
+      "loss": 0.8567,
+      "step": 25156
+    },
+    {
+      "epoch": 0.6909175798522424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6059106881405906e-06,
+      "loss": 0.8148,
+      "step": 25157
+    },
+    {
+      "epoch": 0.6909450440800857,
+      "grad_norm": 0.0,
+      "learning_rate": 4.605161697909366e-06,
+      "loss": 0.8033,
+      "step": 25158
+    },
+    {
+      "epoch": 0.6909725083079289,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6044127503641964e-06,
+      "loss": 0.8745,
+      "step": 25159
+    },
+    {
+      "epoch": 0.6909999725357722,
+      "grad_norm": 0.0,
+      "learning_rate": 4.603663845511e-06,
+      "loss": 0.7976,
+      "step": 25160
+    },
+    {
+      "epoch": 0.6910274367636154,
+      "grad_norm": 0.0,
+      "learning_rate": 4.602914983355705e-06,
+      "loss": 0.7772,
+      "step": 25161
+    },
+    {
+      "epoch": 0.6910549009914586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.602166163904238e-06,
+      "loss": 0.9148,
+      "step": 25162
+    },
+    {
+      "epoch": 0.6910823652193019,
+      "grad_norm": 0.0,
+      "learning_rate": 4.601417387162526e-06,
+      "loss": 0.9504,
+      "step": 25163
+    },
+    {
+      "epoch": 0.691109829447145,
+      "grad_norm": 0.0,
+      "learning_rate": 4.600668653136489e-06,
+      "loss": 0.8829,
+      "step": 25164
+    },
+    {
+      "epoch": 0.6911372936749883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5999199618320545e-06,
+      "loss": 0.7997,
+      "step": 25165
+    },
+    {
+      "epoch": 0.6911647579028316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.599171313255143e-06,
+      "loss": 0.7999,
+      "step": 25166
+    },
+    {
+      "epoch": 0.6911922221306748,
+      "grad_norm": 0.0,
+      "learning_rate": 4.598422707411679e-06,
+      "loss": 0.8355,
+      "step": 25167
+    },
+    {
+      "epoch": 0.691219686358518,
+      "grad_norm": 0.0,
+      "learning_rate": 4.597674144307588e-06,
+      "loss": 0.7997,
+      "step": 25168
+    },
+    {
+      "epoch": 0.6912471505863612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5969256239487915e-06,
+      "loss": 0.8526,
+      "step": 25169
+    },
+    {
+      "epoch": 0.6912746148142045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.596177146341212e-06,
+      "loss": 0.8908,
+      "step": 25170
+    },
+    {
+      "epoch": 0.6913020790420478,
+      "grad_norm": 0.0,
+      "learning_rate": 4.595428711490776e-06,
+      "loss": 0.8219,
+      "step": 25171
+    },
+    {
+      "epoch": 0.6913295432698909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.594680319403401e-06,
+      "loss": 0.7178,
+      "step": 25172
+    },
+    {
+      "epoch": 0.6913570074977342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5939319700850064e-06,
+      "loss": 0.8404,
+      "step": 25173
+    },
+    {
+      "epoch": 0.6913844717255775,
+      "grad_norm": 0.0,
+      "learning_rate": 4.593183663541516e-06,
+      "loss": 0.7824,
+      "step": 25174
+    },
+    {
+      "epoch": 0.6914119359534207,
+      "grad_norm": 0.0,
+      "learning_rate": 4.592435399778851e-06,
+      "loss": 0.8641,
+      "step": 25175
+    },
+    {
+      "epoch": 0.6914394001812639,
+      "grad_norm": 0.0,
+      "learning_rate": 4.591687178802931e-06,
+      "loss": 0.7533,
+      "step": 25176
+    },
+    {
+      "epoch": 0.6914668644091071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.59093900061968e-06,
+      "loss": 0.8065,
+      "step": 25177
+    },
+    {
+      "epoch": 0.6914943286369504,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5901908652350134e-06,
+      "loss": 0.8094,
+      "step": 25178
+    },
+    {
+      "epoch": 0.6915217928647936,
+      "grad_norm": 0.0,
+      "learning_rate": 4.589442772654851e-06,
+      "loss": 0.7628,
+      "step": 25179
+    },
+    {
+      "epoch": 0.6915492570926368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.588694722885117e-06,
+      "loss": 0.953,
+      "step": 25180
+    },
+    {
+      "epoch": 0.6915767213204801,
+      "grad_norm": 0.0,
+      "learning_rate": 4.587946715931723e-06,
+      "loss": 0.8325,
+      "step": 25181
+    },
+    {
+      "epoch": 0.6916041855483233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.587198751800591e-06,
+      "loss": 0.7254,
+      "step": 25182
+    },
+    {
+      "epoch": 0.6916316497761665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.586450830497643e-06,
+      "loss": 0.8428,
+      "step": 25183
+    },
+    {
+      "epoch": 0.6916591140040098,
+      "grad_norm": 0.0,
+      "learning_rate": 4.585702952028789e-06,
+      "loss": 0.7306,
+      "step": 25184
+    },
+    {
+      "epoch": 0.691686578231853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.584955116399951e-06,
+      "loss": 0.7814,
+      "step": 25185
+    },
+    {
+      "epoch": 0.6917140424596963,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5842073236170445e-06,
+      "loss": 0.7799,
+      "step": 25186
+    },
+    {
+      "epoch": 0.6917415066875395,
+      "grad_norm": 0.0,
+      "learning_rate": 4.583459573685993e-06,
+      "loss": 0.867,
+      "step": 25187
+    },
+    {
+      "epoch": 0.6917689709153827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.582711866612703e-06,
+      "loss": 0.9473,
+      "step": 25188
+    },
+    {
+      "epoch": 0.691796435143226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581964202403099e-06,
+      "loss": 0.8666,
+      "step": 25189
+    },
+    {
+      "epoch": 0.6918238993710691,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581216581063088e-06,
+      "loss": 0.8005,
+      "step": 25190
+    },
+    {
+      "epoch": 0.6918513635989124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.580469002598592e-06,
+      "loss": 0.833,
+      "step": 25191
+    },
+    {
+      "epoch": 0.6918788278267557,
+      "grad_norm": 0.0,
+      "learning_rate": 4.579721467015524e-06,
+      "loss": 0.8864,
+      "step": 25192
+    },
+    {
+      "epoch": 0.6919062920545989,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5789739743198e-06,
+      "loss": 0.8882,
+      "step": 25193
+    },
+    {
+      "epoch": 0.6919337562824421,
+      "grad_norm": 0.0,
+      "learning_rate": 4.578226524517333e-06,
+      "loss": 0.8898,
+      "step": 25194
+    },
+    {
+      "epoch": 0.6919612205102853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.577479117614042e-06,
+      "loss": 0.8491,
+      "step": 25195
+    },
+    {
+      "epoch": 0.6919886847381286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.576731753615836e-06,
+      "loss": 0.8113,
+      "step": 25196
+    },
+    {
+      "epoch": 0.6920161489659719,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5759844325286275e-06,
+      "loss": 0.8863,
+      "step": 25197
+    },
+    {
+      "epoch": 0.692043613193815,
+      "grad_norm": 0.0,
+      "learning_rate": 4.575237154358329e-06,
+      "loss": 0.8438,
+      "step": 25198
+    },
+    {
+      "epoch": 0.6920710774216583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.574489919110856e-06,
+      "loss": 0.771,
+      "step": 25199
+    },
+    {
+      "epoch": 0.6920985416495016,
+      "grad_norm": 0.0,
+      "learning_rate": 4.573742726792121e-06,
+      "loss": 0.8701,
+      "step": 25200
+    },
+    {
+      "epoch": 0.6921260058773447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.572995577408039e-06,
+      "loss": 0.7477,
+      "step": 25201
+    },
+    {
+      "epoch": 0.692153470105188,
+      "grad_norm": 0.0,
+      "learning_rate": 4.572248470964515e-06,
+      "loss": 0.875,
+      "step": 25202
+    },
+    {
+      "epoch": 0.6921809343330312,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5715014074674655e-06,
+      "loss": 0.8949,
+      "step": 25203
+    },
+    {
+      "epoch": 0.6922083985608745,
+      "grad_norm": 0.0,
+      "learning_rate": 4.570754386922797e-06,
+      "loss": 0.8236,
+      "step": 25204
+    },
+    {
+      "epoch": 0.6922358627887177,
+      "grad_norm": 0.0,
+      "learning_rate": 4.570007409336424e-06,
+      "loss": 0.7753,
+      "step": 25205
+    },
+    {
+      "epoch": 0.6922633270165609,
+      "grad_norm": 0.0,
+      "learning_rate": 4.569260474714259e-06,
+      "loss": 0.9035,
+      "step": 25206
+    },
+    {
+      "epoch": 0.6922907912444042,
+      "grad_norm": 0.0,
+      "learning_rate": 4.568513583062204e-06,
+      "loss": 0.7902,
+      "step": 25207
+    },
+    {
+      "epoch": 0.6923182554722473,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5677667343861745e-06,
+      "loss": 0.8594,
+      "step": 25208
+    },
+    {
+      "epoch": 0.6923457197000906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.567019928692078e-06,
+      "loss": 0.9328,
+      "step": 25209
+    },
+    {
+      "epoch": 0.6923731839279339,
+      "grad_norm": 0.0,
+      "learning_rate": 4.566273165985825e-06,
+      "loss": 0.8614,
+      "step": 25210
+    },
+    {
+      "epoch": 0.6924006481557771,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5655264462733275e-06,
+      "loss": 0.9501,
+      "step": 25211
+    },
+    {
+      "epoch": 0.6924281123836203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.564779769560489e-06,
+      "loss": 0.8241,
+      "step": 25212
+    },
+    {
+      "epoch": 0.6924555766114636,
+      "grad_norm": 0.0,
+      "learning_rate": 4.564033135853215e-06,
+      "loss": 0.7831,
+      "step": 25213
+    },
+    {
+      "epoch": 0.6924830408393068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.563286545157416e-06,
+      "loss": 0.8473,
+      "step": 25214
+    },
+    {
+      "epoch": 0.6925105050671501,
+      "grad_norm": 0.0,
+      "learning_rate": 4.562539997479001e-06,
+      "loss": 0.8286,
+      "step": 25215
+    },
+    {
+      "epoch": 0.6925379692949932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.561793492823875e-06,
+      "loss": 0.9422,
+      "step": 25216
+    },
+    {
+      "epoch": 0.6925654335228365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.561047031197946e-06,
+      "loss": 0.8623,
+      "step": 25217
+    },
+    {
+      "epoch": 0.6925928977506798,
+      "grad_norm": 0.0,
+      "learning_rate": 4.560300612607122e-06,
+      "loss": 0.8907,
+      "step": 25218
+    },
+    {
+      "epoch": 0.692620361978523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5595542370573054e-06,
+      "loss": 0.835,
+      "step": 25219
+    },
+    {
+      "epoch": 0.6926478262063662,
+      "grad_norm": 0.0,
+      "learning_rate": 4.558807904554405e-06,
+      "loss": 0.8257,
+      "step": 25220
+    },
+    {
+      "epoch": 0.6926752904342094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.558061615104321e-06,
+      "loss": 0.9907,
+      "step": 25221
+    },
+    {
+      "epoch": 0.6927027546620527,
+      "grad_norm": 0.0,
+      "learning_rate": 4.557315368712961e-06,
+      "loss": 0.9155,
+      "step": 25222
+    },
+    {
+      "epoch": 0.6927302188898959,
+      "grad_norm": 0.0,
+      "learning_rate": 4.556569165386231e-06,
+      "loss": 0.8264,
+      "step": 25223
+    },
+    {
+      "epoch": 0.6927576831177391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.555823005130038e-06,
+      "loss": 0.8473,
+      "step": 25224
+    },
+    {
+      "epoch": 0.6927851473455824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.555076887950278e-06,
+      "loss": 0.7393,
+      "step": 25225
+    },
+    {
+      "epoch": 0.6928126115734257,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5543308138528584e-06,
+      "loss": 0.847,
+      "step": 25226
+    },
+    {
+      "epoch": 0.6928400758012688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.553584782843687e-06,
+      "loss": 0.8538,
+      "step": 25227
+    },
+    {
+      "epoch": 0.6928675400291121,
+      "grad_norm": 0.0,
+      "learning_rate": 4.55283879492866e-06,
+      "loss": 0.7799,
+      "step": 25228
+    },
+    {
+      "epoch": 0.6928950042569553,
+      "grad_norm": 0.0,
+      "learning_rate": 4.552092850113682e-06,
+      "loss": 0.9062,
+      "step": 25229
+    },
+    {
+      "epoch": 0.6929224684847985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.551346948404658e-06,
+      "loss": 0.8362,
+      "step": 25230
+    },
+    {
+      "epoch": 0.6929499327126418,
+      "grad_norm": 0.0,
+      "learning_rate": 4.550601089807485e-06,
+      "loss": 0.8434,
+      "step": 25231
+    },
+    {
+      "epoch": 0.692977396940485,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5498552743280665e-06,
+      "loss": 0.7766,
+      "step": 25232
+    },
+    {
+      "epoch": 0.6930048611683283,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5491095019723056e-06,
+      "loss": 0.8221,
+      "step": 25233
+    },
+    {
+      "epoch": 0.6930323253961714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5483637727460995e-06,
+      "loss": 0.8511,
+      "step": 25234
+    },
+    {
+      "epoch": 0.6930597896240147,
+      "grad_norm": 0.0,
+      "learning_rate": 4.547618086655356e-06,
+      "loss": 0.7747,
+      "step": 25235
+    },
+    {
+      "epoch": 0.693087253851858,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546872443705971e-06,
+      "loss": 0.8233,
+      "step": 25236
+    },
+    {
+      "epoch": 0.6931147180797012,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546126843903839e-06,
+      "loss": 0.9416,
+      "step": 25237
+    },
+    {
+      "epoch": 0.6931421823075444,
+      "grad_norm": 0.0,
+      "learning_rate": 4.545381287254864e-06,
+      "loss": 0.9114,
+      "step": 25238
+    },
+    {
+      "epoch": 0.6931696465353877,
+      "grad_norm": 0.0,
+      "learning_rate": 4.544635773764945e-06,
+      "loss": 0.8787,
+      "step": 25239
+    },
+    {
+      "epoch": 0.6931971107632309,
+      "grad_norm": 0.0,
+      "learning_rate": 4.543890303439982e-06,
+      "loss": 0.9216,
+      "step": 25240
+    },
+    {
+      "epoch": 0.6932245749910741,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5431448762858706e-06,
+      "loss": 0.809,
+      "step": 25241
+    },
+    {
+      "epoch": 0.6932520392189173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.542399492308516e-06,
+      "loss": 0.7467,
+      "step": 25242
+    },
+    {
+      "epoch": 0.6932795034467606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.541654151513805e-06,
+      "loss": 0.7687,
+      "step": 25243
+    },
+    {
+      "epoch": 0.6933069676746039,
+      "grad_norm": 0.0,
+      "learning_rate": 4.540908853907647e-06,
+      "loss": 0.7516,
+      "step": 25244
+    },
+    {
+      "epoch": 0.693334431902447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.540163599495928e-06,
+      "loss": 0.8189,
+      "step": 25245
+    },
+    {
+      "epoch": 0.6933618961302903,
+      "grad_norm": 0.0,
+      "learning_rate": 4.53941838828455e-06,
+      "loss": 0.9724,
+      "step": 25246
+    },
+    {
+      "epoch": 0.6933893603581335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.538673220279408e-06,
+      "loss": 0.7856,
+      "step": 25247
+    },
+    {
+      "epoch": 0.6934168245859768,
+      "grad_norm": 0.0,
+      "learning_rate": 4.537928095486404e-06,
+      "loss": 0.7947,
+      "step": 25248
+    },
+    {
+      "epoch": 0.69344428881382,
+      "grad_norm": 0.0,
+      "learning_rate": 4.537183013911424e-06,
+      "loss": 0.7383,
+      "step": 25249
+    },
+    {
+      "epoch": 0.6934717530416632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.536437975560367e-06,
+      "loss": 0.8397,
+      "step": 25250
+    },
+    {
+      "epoch": 0.6934992172695065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.535692980439135e-06,
+      "loss": 0.7989,
+      "step": 25251
+    },
+    {
+      "epoch": 0.6935266814973498,
+      "grad_norm": 0.0,
+      "learning_rate": 4.534948028553612e-06,
+      "loss": 0.8438,
+      "step": 25252
+    },
+    {
+      "epoch": 0.6935541457251929,
+      "grad_norm": 0.0,
+      "learning_rate": 4.534203119909697e-06,
+      "loss": 0.8359,
+      "step": 25253
+    },
+    {
+      "epoch": 0.6935816099530362,
+      "grad_norm": 0.0,
+      "learning_rate": 4.533458254513289e-06,
+      "loss": 0.9283,
+      "step": 25254
+    },
+    {
+      "epoch": 0.6936090741808794,
+      "grad_norm": 0.0,
+      "learning_rate": 4.532713432370271e-06,
+      "loss": 0.8499,
+      "step": 25255
+    },
+    {
+      "epoch": 0.6936365384087226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.531968653486544e-06,
+      "loss": 0.9439,
+      "step": 25256
+    },
+    {
+      "epoch": 0.6936640026365659,
+      "grad_norm": 0.0,
+      "learning_rate": 4.531223917867998e-06,
+      "loss": 0.8514,
+      "step": 25257
+    },
+    {
+      "epoch": 0.6936914668644091,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5304792255205276e-06,
+      "loss": 0.8812,
+      "step": 25258
+    },
+    {
+      "epoch": 0.6937189310922524,
+      "grad_norm": 0.0,
+      "learning_rate": 4.529734576450028e-06,
+      "loss": 0.7707,
+      "step": 25259
+    },
+    {
+      "epoch": 0.6937463953200955,
+      "grad_norm": 0.0,
+      "learning_rate": 4.528989970662387e-06,
+      "loss": 0.7782,
+      "step": 25260
+    },
+    {
+      "epoch": 0.6937738595479388,
+      "grad_norm": 0.0,
+      "learning_rate": 4.528245408163493e-06,
+      "loss": 0.8319,
+      "step": 25261
+    },
+    {
+      "epoch": 0.6938013237757821,
+      "grad_norm": 0.0,
+      "learning_rate": 4.527500888959241e-06,
+      "loss": 0.8485,
+      "step": 25262
+    },
+    {
+      "epoch": 0.6938287880036252,
+      "grad_norm": 0.0,
+      "learning_rate": 4.526756413055522e-06,
+      "loss": 0.8508,
+      "step": 25263
+    },
+    {
+      "epoch": 0.6938562522314685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5260119804582256e-06,
+      "loss": 0.8155,
+      "step": 25264
+    },
+    {
+      "epoch": 0.6938837164593118,
+      "grad_norm": 0.0,
+      "learning_rate": 4.525267591173248e-06,
+      "loss": 0.8368,
+      "step": 25265
+    },
+    {
+      "epoch": 0.693911180687155,
+      "grad_norm": 0.0,
+      "learning_rate": 4.524523245206469e-06,
+      "loss": 0.8278,
+      "step": 25266
+    },
+    {
+      "epoch": 0.6939386449149982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.523778942563783e-06,
+      "loss": 0.8817,
+      "step": 25267
+    },
+    {
+      "epoch": 0.6939661091428414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.523034683251083e-06,
+      "loss": 0.8917,
+      "step": 25268
+    },
+    {
+      "epoch": 0.6939935733706847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.522290467274252e-06,
+      "loss": 0.8452,
+      "step": 25269
+    },
+    {
+      "epoch": 0.694021037598528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.521546294639179e-06,
+      "loss": 0.7591,
+      "step": 25270
+    },
+    {
+      "epoch": 0.6940485018263711,
+      "grad_norm": 0.0,
+      "learning_rate": 4.520802165351758e-06,
+      "loss": 0.7513,
+      "step": 25271
+    },
+    {
+      "epoch": 0.6940759660542144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5200580794178695e-06,
+      "loss": 0.7709,
+      "step": 25272
+    },
+    {
+      "epoch": 0.6941034302820576,
+      "grad_norm": 0.0,
+      "learning_rate": 4.519314036843403e-06,
+      "loss": 0.8752,
+      "step": 25273
+    },
+    {
+      "epoch": 0.6941308945099008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.518570037634248e-06,
+      "loss": 0.8715,
+      "step": 25274
+    },
+    {
+      "epoch": 0.6941583587377441,
+      "grad_norm": 0.0,
+      "learning_rate": 4.517826081796293e-06,
+      "loss": 0.816,
+      "step": 25275
+    },
+    {
+      "epoch": 0.6941858229655873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.517082169335419e-06,
+      "loss": 0.8298,
+      "step": 25276
+    },
+    {
+      "epoch": 0.6942132871934306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.516338300257518e-06,
+      "loss": 0.8736,
+      "step": 25277
+    },
+    {
+      "epoch": 0.6942407514212738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.515594474568469e-06,
+      "loss": 0.8518,
+      "step": 25278
+    },
+    {
+      "epoch": 0.694268215649117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.51485069227416e-06,
+      "loss": 0.8988,
+      "step": 25279
+    },
+    {
+      "epoch": 0.6942956798769603,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5141069533804785e-06,
+      "loss": 0.8501,
+      "step": 25280
+    },
+    {
+      "epoch": 0.6943231441048034,
+      "grad_norm": 0.0,
+      "learning_rate": 4.513363257893306e-06,
+      "loss": 0.8142,
+      "step": 25281
+    },
+    {
+      "epoch": 0.6943506083326467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.51261960581853e-06,
+      "loss": 0.8683,
+      "step": 25282
+    },
+    {
+      "epoch": 0.69437807256049,
+      "grad_norm": 0.0,
+      "learning_rate": 4.511875997162037e-06,
+      "loss": 0.857,
+      "step": 25283
+    },
+    {
+      "epoch": 0.6944055367883332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.511132431929707e-06,
+      "loss": 0.7326,
+      "step": 25284
+    },
+    {
+      "epoch": 0.6944330010161764,
+      "grad_norm": 0.0,
+      "learning_rate": 4.510388910127419e-06,
+      "loss": 0.7787,
+      "step": 25285
+    },
+    {
+      "epoch": 0.6944604652440196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.509645431761062e-06,
+      "loss": 0.8453,
+      "step": 25286
+    },
+    {
+      "epoch": 0.6944879294718629,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5089019968365155e-06,
+      "loss": 0.7701,
+      "step": 25287
+    },
+    {
+      "epoch": 0.6945153936997062,
+      "grad_norm": 0.0,
+      "learning_rate": 4.508158605359664e-06,
+      "loss": 0.7718,
+      "step": 25288
+    },
+    {
+      "epoch": 0.6945428579275493,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5074152573363915e-06,
+      "loss": 0.8264,
+      "step": 25289
+    },
+    {
+      "epoch": 0.6945703221553926,
+      "grad_norm": 0.0,
+      "learning_rate": 4.506671952772576e-06,
+      "loss": 0.8359,
+      "step": 25290
+    },
+    {
+      "epoch": 0.6945977863832359,
+      "grad_norm": 0.0,
+      "learning_rate": 4.505928691674097e-06,
+      "loss": 0.8574,
+      "step": 25291
+    },
+    {
+      "epoch": 0.694625250611079,
+      "grad_norm": 0.0,
+      "learning_rate": 4.505185474046844e-06,
+      "loss": 0.7915,
+      "step": 25292
+    },
+    {
+      "epoch": 0.6946527148389223,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5044422998966875e-06,
+      "loss": 0.7907,
+      "step": 25293
+    },
+    {
+      "epoch": 0.6946801790667655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5036991692295125e-06,
+      "loss": 0.8211,
+      "step": 25294
+    },
+    {
+      "epoch": 0.6947076432946088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.502956082051203e-06,
+      "loss": 0.7733,
+      "step": 25295
+    },
+    {
+      "epoch": 0.694735107522452,
+      "grad_norm": 0.0,
+      "learning_rate": 4.50221303836763e-06,
+      "loss": 0.8338,
+      "step": 25296
+    },
+    {
+      "epoch": 0.6947625717502952,
+      "grad_norm": 0.0,
+      "learning_rate": 4.501470038184678e-06,
+      "loss": 0.8388,
+      "step": 25297
+    },
+    {
+      "epoch": 0.6947900359781385,
+      "grad_norm": 0.0,
+      "learning_rate": 4.500727081508223e-06,
+      "loss": 0.8774,
+      "step": 25298
+    },
+    {
+      "epoch": 0.6948175002059818,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4999841683441515e-06,
+      "loss": 0.8311,
+      "step": 25299
+    },
+    {
+      "epoch": 0.6948449644338249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.499241298698331e-06,
+      "loss": 0.8069,
+      "step": 25300
+    },
+    {
+      "epoch": 0.6948724286616682,
+      "grad_norm": 0.0,
+      "learning_rate": 4.498498472576648e-06,
+      "loss": 0.8728,
+      "step": 25301
+    },
+    {
+      "epoch": 0.6948998928895114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4977556899849715e-06,
+      "loss": 0.9137,
+      "step": 25302
+    },
+    {
+      "epoch": 0.6949273571173546,
+      "grad_norm": 0.0,
+      "learning_rate": 4.497012950929185e-06,
+      "loss": 0.8547,
+      "step": 25303
+    },
+    {
+      "epoch": 0.6949548213451979,
+      "grad_norm": 0.0,
+      "learning_rate": 4.496270255415163e-06,
+      "loss": 0.96,
+      "step": 25304
+    },
+    {
+      "epoch": 0.6949822855730411,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4955276034487825e-06,
+      "loss": 0.8326,
+      "step": 25305
+    },
+    {
+      "epoch": 0.6950097498008844,
+      "grad_norm": 0.0,
+      "learning_rate": 4.49478499503592e-06,
+      "loss": 0.8121,
+      "step": 25306
+    },
+    {
+      "epoch": 0.6950372140287275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.494042430182455e-06,
+      "loss": 0.9159,
+      "step": 25307
+    },
+    {
+      "epoch": 0.6950646782565708,
+      "grad_norm": 0.0,
+      "learning_rate": 4.493299908894258e-06,
+      "loss": 0.8293,
+      "step": 25308
+    },
+    {
+      "epoch": 0.6950921424844141,
+      "grad_norm": 0.0,
+      "learning_rate": 4.492557431177202e-06,
+      "loss": 0.9164,
+      "step": 25309
+    },
+    {
+      "epoch": 0.6951196067122573,
+      "grad_norm": 0.0,
+      "learning_rate": 4.491814997037166e-06,
+      "loss": 0.8385,
+      "step": 25310
+    },
+    {
+      "epoch": 0.6951470709401005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.491072606480023e-06,
+      "loss": 0.7314,
+      "step": 25311
+    },
+    {
+      "epoch": 0.6951745351679438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4903302595116475e-06,
+      "loss": 0.8097,
+      "step": 25312
+    },
+    {
+      "epoch": 0.695201999395787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4895879561379166e-06,
+      "loss": 0.8046,
+      "step": 25313
+    },
+    {
+      "epoch": 0.6952294636236303,
+      "grad_norm": 0.0,
+      "learning_rate": 4.488845696364697e-06,
+      "loss": 0.7744,
+      "step": 25314
+    },
+    {
+      "epoch": 0.6952569278514734,
+      "grad_norm": 0.0,
+      "learning_rate": 4.488103480197864e-06,
+      "loss": 0.8876,
+      "step": 25315
+    },
+    {
+      "epoch": 0.6952843920793167,
+      "grad_norm": 0.0,
+      "learning_rate": 4.487361307643295e-06,
+      "loss": 0.8339,
+      "step": 25316
+    },
+    {
+      "epoch": 0.69531185630716,
+      "grad_norm": 0.0,
+      "learning_rate": 4.486619178706857e-06,
+      "loss": 0.8723,
+      "step": 25317
+    },
+    {
+      "epoch": 0.6953393205350031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.485877093394422e-06,
+      "loss": 0.8645,
+      "step": 25318
+    },
+    {
+      "epoch": 0.6953667847628464,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4851350517118675e-06,
+      "loss": 0.8439,
+      "step": 25319
+    },
+    {
+      "epoch": 0.6953942489906896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.484393053665058e-06,
+      "loss": 0.8512,
+      "step": 25320
+    },
+    {
+      "epoch": 0.6954217132185329,
+      "grad_norm": 0.0,
+      "learning_rate": 4.483651099259866e-06,
+      "loss": 0.8388,
+      "step": 25321
+    },
+    {
+      "epoch": 0.6954491774463761,
+      "grad_norm": 0.0,
+      "learning_rate": 4.482909188502163e-06,
+      "loss": 0.8942,
+      "step": 25322
+    },
+    {
+      "epoch": 0.6954766416742193,
+      "grad_norm": 0.0,
+      "learning_rate": 4.482167321397825e-06,
+      "loss": 0.7902,
+      "step": 25323
+    },
+    {
+      "epoch": 0.6955041059020626,
+      "grad_norm": 0.0,
+      "learning_rate": 4.481425497952716e-06,
+      "loss": 0.7509,
+      "step": 25324
+    },
+    {
+      "epoch": 0.6955315701299059,
+      "grad_norm": 0.0,
+      "learning_rate": 4.480683718172701e-06,
+      "loss": 0.866,
+      "step": 25325
+    },
+    {
+      "epoch": 0.695559034357749,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4799419820636555e-06,
+      "loss": 0.8928,
+      "step": 25326
+    },
+    {
+      "epoch": 0.6955864985855923,
+      "grad_norm": 0.0,
+      "learning_rate": 4.479200289631448e-06,
+      "loss": 0.8014,
+      "step": 25327
+    },
+    {
+      "epoch": 0.6956139628134355,
+      "grad_norm": 0.0,
+      "learning_rate": 4.478458640881945e-06,
+      "loss": 0.7365,
+      "step": 25328
+    },
+    {
+      "epoch": 0.6956414270412787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.477717035821018e-06,
+      "loss": 0.7283,
+      "step": 25329
+    },
+    {
+      "epoch": 0.695668891269122,
+      "grad_norm": 0.0,
+      "learning_rate": 4.476975474454534e-06,
+      "loss": 0.7993,
+      "step": 25330
+    },
+    {
+      "epoch": 0.6956963554969652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.476233956788358e-06,
+      "loss": 0.9392,
+      "step": 25331
+    },
+    {
+      "epoch": 0.6957238197248085,
+      "grad_norm": 0.0,
+      "learning_rate": 4.475492482828361e-06,
+      "loss": 0.818,
+      "step": 25332
+    },
+    {
+      "epoch": 0.6957512839526516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.474751052580404e-06,
+      "loss": 0.8583,
+      "step": 25333
+    },
+    {
+      "epoch": 0.6957787481804949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.474009666050358e-06,
+      "loss": 0.8496,
+      "step": 25334
+    },
+    {
+      "epoch": 0.6958062124083382,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4732683232440865e-06,
+      "loss": 0.8548,
+      "step": 25335
+    },
+    {
+      "epoch": 0.6958336766361813,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4725270241674615e-06,
+      "loss": 0.8759,
+      "step": 25336
+    },
+    {
+      "epoch": 0.6958611408640246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.47178576882634e-06,
+      "loss": 0.7841,
+      "step": 25337
+    },
+    {
+      "epoch": 0.6958886050918679,
+      "grad_norm": 0.0,
+      "learning_rate": 4.471044557226591e-06,
+      "loss": 0.8792,
+      "step": 25338
+    },
+    {
+      "epoch": 0.6959160693197111,
+      "grad_norm": 0.0,
+      "learning_rate": 4.470303389374079e-06,
+      "loss": 0.8561,
+      "step": 25339
+    },
+    {
+      "epoch": 0.6959435335475543,
+      "grad_norm": 0.0,
+      "learning_rate": 4.469562265274673e-06,
+      "loss": 0.8969,
+      "step": 25340
+    },
+    {
+      "epoch": 0.6959709977753975,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4688211849342274e-06,
+      "loss": 0.8255,
+      "step": 25341
+    },
+    {
+      "epoch": 0.6959984620032408,
+      "grad_norm": 0.0,
+      "learning_rate": 4.468080148358616e-06,
+      "loss": 0.7281,
+      "step": 25342
+    },
+    {
+      "epoch": 0.6960259262310841,
+      "grad_norm": 0.0,
+      "learning_rate": 4.467339155553694e-06,
+      "loss": 0.8433,
+      "step": 25343
+    },
+    {
+      "epoch": 0.6960533904589272,
+      "grad_norm": 0.0,
+      "learning_rate": 4.466598206525327e-06,
+      "loss": 0.8334,
+      "step": 25344
+    },
+    {
+      "epoch": 0.6960808546867705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4658573012793785e-06,
+      "loss": 0.7963,
+      "step": 25345
+    },
+    {
+      "epoch": 0.6961083189146137,
+      "grad_norm": 0.0,
+      "learning_rate": 4.465116439821711e-06,
+      "loss": 0.7959,
+      "step": 25346
+    },
+    {
+      "epoch": 0.6961357831424569,
+      "grad_norm": 0.0,
+      "learning_rate": 4.464375622158189e-06,
+      "loss": 0.8692,
+      "step": 25347
+    },
+    {
+      "epoch": 0.6961632473703002,
+      "grad_norm": 0.0,
+      "learning_rate": 4.463634848294673e-06,
+      "loss": 0.8784,
+      "step": 25348
+    },
+    {
+      "epoch": 0.6961907115981434,
+      "grad_norm": 0.0,
+      "learning_rate": 4.462894118237017e-06,
+      "loss": 0.8153,
+      "step": 25349
+    },
+    {
+      "epoch": 0.6962181758259867,
+      "grad_norm": 0.0,
+      "learning_rate": 4.462153431991088e-06,
+      "loss": 0.7695,
+      "step": 25350
+    },
+    {
+      "epoch": 0.6962456400538299,
+      "grad_norm": 0.0,
+      "learning_rate": 4.461412789562748e-06,
+      "loss": 0.8226,
+      "step": 25351
+    },
+    {
+      "epoch": 0.6962731042816731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.460672190957854e-06,
+      "loss": 0.9004,
+      "step": 25352
+    },
+    {
+      "epoch": 0.6963005685095164,
+      "grad_norm": 0.0,
+      "learning_rate": 4.459931636182267e-06,
+      "loss": 0.8177,
+      "step": 25353
+    },
+    {
+      "epoch": 0.6963280327373595,
+      "grad_norm": 0.0,
+      "learning_rate": 4.459191125241852e-06,
+      "loss": 0.7485,
+      "step": 25354
+    },
+    {
+      "epoch": 0.6963554969652028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.458450658142459e-06,
+      "loss": 0.8017,
+      "step": 25355
+    },
+    {
+      "epoch": 0.6963829611930461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.457710234889954e-06,
+      "loss": 0.8107,
+      "step": 25356
+    },
+    {
+      "epoch": 0.6964104254208893,
+      "grad_norm": 0.0,
+      "learning_rate": 4.456969855490189e-06,
+      "loss": 0.8239,
+      "step": 25357
+    },
+    {
+      "epoch": 0.6964378896487325,
+      "grad_norm": 0.0,
+      "learning_rate": 4.456229519949026e-06,
+      "loss": 0.8658,
+      "step": 25358
+    },
+    {
+      "epoch": 0.6964653538765757,
+      "grad_norm": 0.0,
+      "learning_rate": 4.455489228272323e-06,
+      "loss": 0.85,
+      "step": 25359
+    },
+    {
+      "epoch": 0.696492818104419,
+      "grad_norm": 0.0,
+      "learning_rate": 4.454748980465939e-06,
+      "loss": 0.8486,
+      "step": 25360
+    },
+    {
+      "epoch": 0.6965202823322623,
+      "grad_norm": 0.0,
+      "learning_rate": 4.454008776535727e-06,
+      "loss": 0.7943,
+      "step": 25361
+    },
+    {
+      "epoch": 0.6965477465601054,
+      "grad_norm": 0.0,
+      "learning_rate": 4.453268616487545e-06,
+      "loss": 0.8951,
+      "step": 25362
+    },
+    {
+      "epoch": 0.6965752107879487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.452528500327254e-06,
+      "loss": 0.7807,
+      "step": 25363
+    },
+    {
+      "epoch": 0.696602675015792,
+      "grad_norm": 0.0,
+      "learning_rate": 4.451788428060704e-06,
+      "loss": 0.8498,
+      "step": 25364
+    },
+    {
+      "epoch": 0.6966301392436351,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4510483996937515e-06,
+      "loss": 0.8189,
+      "step": 25365
+    },
+    {
+      "epoch": 0.6966576034714784,
+      "grad_norm": 0.0,
+      "learning_rate": 4.450308415232258e-06,
+      "loss": 0.8027,
+      "step": 25366
+    },
+    {
+      "epoch": 0.6966850676993216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.44956847468207e-06,
+      "loss": 0.8601,
+      "step": 25367
+    },
+    {
+      "epoch": 0.6967125319271649,
+      "grad_norm": 0.0,
+      "learning_rate": 4.448828578049047e-06,
+      "loss": 0.8108,
+      "step": 25368
+    },
+    {
+      "epoch": 0.6967399961550081,
+      "grad_norm": 0.0,
+      "learning_rate": 4.448088725339041e-06,
+      "loss": 0.8528,
+      "step": 25369
+    },
+    {
+      "epoch": 0.6967674603828513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.44734891655791e-06,
+      "loss": 0.7439,
+      "step": 25370
+    },
+    {
+      "epoch": 0.6967949246106946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4466091517115074e-06,
+      "loss": 0.9408,
+      "step": 25371
+    },
+    {
+      "epoch": 0.6968223888385378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.445869430805684e-06,
+      "loss": 0.7984,
+      "step": 25372
+    },
+    {
+      "epoch": 0.696849853066381,
+      "grad_norm": 0.0,
+      "learning_rate": 4.44512975384629e-06,
+      "loss": 0.865,
+      "step": 25373
+    },
+    {
+      "epoch": 0.6968773172942243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.444390120839182e-06,
+      "loss": 0.7767,
+      "step": 25374
+    },
+    {
+      "epoch": 0.6969047815220675,
+      "grad_norm": 0.0,
+      "learning_rate": 4.44365053179021e-06,
+      "loss": 0.8689,
+      "step": 25375
+    },
+    {
+      "epoch": 0.6969322457499108,
+      "grad_norm": 0.0,
+      "learning_rate": 4.442910986705228e-06,
+      "loss": 0.7865,
+      "step": 25376
+    },
+    {
+      "epoch": 0.696959709977754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4421714855900866e-06,
+      "loss": 0.8137,
+      "step": 25377
+    },
+    {
+      "epoch": 0.6969871742055972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.441432028450641e-06,
+      "loss": 0.7862,
+      "step": 25378
+    },
+    {
+      "epoch": 0.6970146384334405,
+      "grad_norm": 0.0,
+      "learning_rate": 4.440692615292735e-06,
+      "loss": 0.7475,
+      "step": 25379
+    },
+    {
+      "epoch": 0.6970421026612836,
+      "grad_norm": 0.0,
+      "learning_rate": 4.439953246122226e-06,
+      "loss": 0.9618,
+      "step": 25380
+    },
+    {
+      "epoch": 0.6970695668891269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.439213920944957e-06,
+      "loss": 0.7803,
+      "step": 25381
+    },
+    {
+      "epoch": 0.6970970311169702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.438474639766781e-06,
+      "loss": 0.9035,
+      "step": 25382
+    },
+    {
+      "epoch": 0.6971244953448134,
+      "grad_norm": 0.0,
+      "learning_rate": 4.437735402593553e-06,
+      "loss": 0.7889,
+      "step": 25383
+    },
+    {
+      "epoch": 0.6971519595726566,
+      "grad_norm": 0.0,
+      "learning_rate": 4.436996209431113e-06,
+      "loss": 0.8773,
+      "step": 25384
+    },
+    {
+      "epoch": 0.6971794238004998,
+      "grad_norm": 0.0,
+      "learning_rate": 4.436257060285313e-06,
+      "loss": 0.7014,
+      "step": 25385
+    },
+    {
+      "epoch": 0.6972068880283431,
+      "grad_norm": 0.0,
+      "learning_rate": 4.435517955162003e-06,
+      "loss": 0.7977,
+      "step": 25386
+    },
+    {
+      "epoch": 0.6972343522561864,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4347788940670335e-06,
+      "loss": 0.8015,
+      "step": 25387
+    },
+    {
+      "epoch": 0.6972618164840295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.434039877006247e-06,
+      "loss": 0.908,
+      "step": 25388
+    },
+    {
+      "epoch": 0.6972892807118728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.433300903985495e-06,
+      "loss": 0.8149,
+      "step": 25389
+    },
+    {
+      "epoch": 0.6973167449397161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.43256197501062e-06,
+      "loss": 0.7954,
+      "step": 25390
+    },
+    {
+      "epoch": 0.6973442091675592,
+      "grad_norm": 0.0,
+      "learning_rate": 4.431823090087471e-06,
+      "loss": 0.8529,
+      "step": 25391
+    },
+    {
+      "epoch": 0.6973716733954025,
+      "grad_norm": 0.0,
+      "learning_rate": 4.431084249221893e-06,
+      "loss": 0.9,
+      "step": 25392
+    },
+    {
+      "epoch": 0.6973991376232457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.430345452419735e-06,
+      "loss": 0.8151,
+      "step": 25393
+    },
+    {
+      "epoch": 0.697426601851089,
+      "grad_norm": 0.0,
+      "learning_rate": 4.429606699686842e-06,
+      "loss": 0.779,
+      "step": 25394
+    },
+    {
+      "epoch": 0.6974540660789322,
+      "grad_norm": 0.0,
+      "learning_rate": 4.428867991029061e-06,
+      "loss": 0.8145,
+      "step": 25395
+    },
+    {
+      "epoch": 0.6974815303067754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.428129326452235e-06,
+      "loss": 0.7938,
+      "step": 25396
+    },
+    {
+      "epoch": 0.6975089945346187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.427390705962205e-06,
+      "loss": 0.8876,
+      "step": 25397
+    },
+    {
+      "epoch": 0.6975364587624618,
+      "grad_norm": 0.0,
+      "learning_rate": 4.426652129564818e-06,
+      "loss": 0.8418,
+      "step": 25398
+    },
+    {
+      "epoch": 0.6975639229903051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.425913597265917e-06,
+      "loss": 0.8194,
+      "step": 25399
+    },
+    {
+      "epoch": 0.6975913872181484,
+      "grad_norm": 0.0,
+      "learning_rate": 4.425175109071348e-06,
+      "loss": 0.7291,
+      "step": 25400
+    },
+    {
+      "epoch": 0.6976188514459916,
+      "grad_norm": 0.0,
+      "learning_rate": 4.424436664986957e-06,
+      "loss": 0.8784,
+      "step": 25401
+    },
+    {
+      "epoch": 0.6976463156738348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.423698265018579e-06,
+      "loss": 0.8826,
+      "step": 25402
+    },
+    {
+      "epoch": 0.6976737799016781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4229599091720595e-06,
+      "loss": 0.86,
+      "step": 25403
+    },
+    {
+      "epoch": 0.6977012441295213,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4222215974532466e-06,
+      "loss": 0.8229,
+      "step": 25404
+    },
+    {
+      "epoch": 0.6977287083573646,
+      "grad_norm": 0.0,
+      "learning_rate": 4.421483329867972e-06,
+      "loss": 0.8246,
+      "step": 25405
+    },
+    {
+      "epoch": 0.6977561725852077,
+      "grad_norm": 0.0,
+      "learning_rate": 4.420745106422083e-06,
+      "loss": 0.8276,
+      "step": 25406
+    },
+    {
+      "epoch": 0.697783636813051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.420006927121423e-06,
+      "loss": 0.8312,
+      "step": 25407
+    },
+    {
+      "epoch": 0.6978111010408943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4192687919718265e-06,
+      "loss": 0.836,
+      "step": 25408
+    },
+    {
+      "epoch": 0.6978385652687374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.418530700979138e-06,
+      "loss": 0.7975,
+      "step": 25409
+    },
+    {
+      "epoch": 0.6978660294965807,
+      "grad_norm": 0.0,
+      "learning_rate": 4.417792654149196e-06,
+      "loss": 0.8972,
+      "step": 25410
+    },
+    {
+      "epoch": 0.6978934937244239,
+      "grad_norm": 0.0,
+      "learning_rate": 4.417054651487843e-06,
+      "loss": 0.7816,
+      "step": 25411
+    },
+    {
+      "epoch": 0.6979209579522672,
+      "grad_norm": 0.0,
+      "learning_rate": 4.416316693000914e-06,
+      "loss": 0.8649,
+      "step": 25412
+    },
+    {
+      "epoch": 0.6979484221801104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.415578778694254e-06,
+      "loss": 0.7839,
+      "step": 25413
+    },
+    {
+      "epoch": 0.6979758864079536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.414840908573693e-06,
+      "loss": 0.8506,
+      "step": 25414
+    },
+    {
+      "epoch": 0.6980033506357969,
+      "grad_norm": 0.0,
+      "learning_rate": 4.414103082645076e-06,
+      "loss": 0.9106,
+      "step": 25415
+    },
+    {
+      "epoch": 0.6980308148636402,
+      "grad_norm": 0.0,
+      "learning_rate": 4.413365300914237e-06,
+      "loss": 0.763,
+      "step": 25416
+    },
+    {
+      "epoch": 0.6980582790914833,
+      "grad_norm": 0.0,
+      "learning_rate": 4.412627563387018e-06,
+      "loss": 0.7703,
+      "step": 25417
+    },
+    {
+      "epoch": 0.6980857433193266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.411889870069253e-06,
+      "loss": 0.7656,
+      "step": 25418
+    },
+    {
+      "epoch": 0.6981132075471698,
+      "grad_norm": 0.0,
+      "learning_rate": 4.411152220966783e-06,
+      "loss": 0.8545,
+      "step": 25419
+    },
+    {
+      "epoch": 0.698140671775013,
+      "grad_norm": 0.0,
+      "learning_rate": 4.410414616085443e-06,
+      "loss": 0.9019,
+      "step": 25420
+    },
+    {
+      "epoch": 0.6981681360028563,
+      "grad_norm": 0.0,
+      "learning_rate": 4.409677055431063e-06,
+      "loss": 0.8075,
+      "step": 25421
+    },
+    {
+      "epoch": 0.6981956002306995,
+      "grad_norm": 0.0,
+      "learning_rate": 4.408939539009485e-06,
+      "loss": 0.8859,
+      "step": 25422
+    },
+    {
+      "epoch": 0.6982230644585428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.408202066826542e-06,
+      "loss": 0.8356,
+      "step": 25423
+    },
+    {
+      "epoch": 0.6982505286863859,
+      "grad_norm": 0.0,
+      "learning_rate": 4.40746463888807e-06,
+      "loss": 0.8406,
+      "step": 25424
+    },
+    {
+      "epoch": 0.6982779929142292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.406727255199908e-06,
+      "loss": 0.801,
+      "step": 25425
+    },
+    {
+      "epoch": 0.6983054571420725,
+      "grad_norm": 0.0,
+      "learning_rate": 4.405989915767883e-06,
+      "loss": 0.882,
+      "step": 25426
+    },
+    {
+      "epoch": 0.6983329213699156,
+      "grad_norm": 0.0,
+      "learning_rate": 4.405252620597833e-06,
+      "loss": 0.912,
+      "step": 25427
+    },
+    {
+      "epoch": 0.6983603855977589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.404515369695595e-06,
+      "loss": 0.7825,
+      "step": 25428
+    },
+    {
+      "epoch": 0.6983878498256022,
+      "grad_norm": 0.0,
+      "learning_rate": 4.403778163066996e-06,
+      "loss": 0.7066,
+      "step": 25429
+    },
+    {
+      "epoch": 0.6984153140534454,
+      "grad_norm": 0.0,
+      "learning_rate": 4.40304100071787e-06,
+      "loss": 0.8259,
+      "step": 25430
+    },
+    {
+      "epoch": 0.6984427782812886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.402303882654057e-06,
+      "loss": 0.8196,
+      "step": 25431
+    },
+    {
+      "epoch": 0.6984702425091318,
+      "grad_norm": 0.0,
+      "learning_rate": 4.40156680888138e-06,
+      "loss": 0.7677,
+      "step": 25432
+    },
+    {
+      "epoch": 0.6984977067369751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.400829779405675e-06,
+      "loss": 0.9242,
+      "step": 25433
+    },
+    {
+      "epoch": 0.6985251709648184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4000927942327735e-06,
+      "loss": 0.9499,
+      "step": 25434
+    },
+    {
+      "epoch": 0.6985526351926615,
+      "grad_norm": 0.0,
+      "learning_rate": 4.399355853368511e-06,
+      "loss": 0.8866,
+      "step": 25435
+    },
+    {
+      "epoch": 0.6985800994205048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.39861895681871e-06,
+      "loss": 0.8007,
+      "step": 25436
+    },
+    {
+      "epoch": 0.698607563648348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.397882104589211e-06,
+      "loss": 0.796,
+      "step": 25437
+    },
+    {
+      "epoch": 0.6986350278761913,
+      "grad_norm": 0.0,
+      "learning_rate": 4.397145296685833e-06,
+      "loss": 0.8555,
+      "step": 25438
+    },
+    {
+      "epoch": 0.6986624921040345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3964085331144124e-06,
+      "loss": 0.8021,
+      "step": 25439
+    },
+    {
+      "epoch": 0.6986899563318777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.395671813880779e-06,
+      "loss": 0.7543,
+      "step": 25440
+    },
+    {
+      "epoch": 0.698717420559721,
+      "grad_norm": 0.0,
+      "learning_rate": 4.39493513899076e-06,
+      "loss": 0.8465,
+      "step": 25441
+    },
+    {
+      "epoch": 0.6987448847875642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.394198508450191e-06,
+      "loss": 0.8543,
+      "step": 25442
+    },
+    {
+      "epoch": 0.6987723490154074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3934619222648885e-06,
+      "loss": 0.8219,
+      "step": 25443
+    },
+    {
+      "epoch": 0.6987998132432507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.392725380440692e-06,
+      "loss": 0.8394,
+      "step": 25444
+    },
+    {
+      "epoch": 0.6988272774710939,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391988882983421e-06,
+      "loss": 0.8155,
+      "step": 25445
+    },
+    {
+      "epoch": 0.6988547416989371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391252429898906e-06,
+      "loss": 0.9193,
+      "step": 25446
+    },
+    {
+      "epoch": 0.6988822059267804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3905160211929746e-06,
+      "loss": 0.8392,
+      "step": 25447
+    },
+    {
+      "epoch": 0.6989096701546236,
+      "grad_norm": 0.0,
+      "learning_rate": 4.389779656871457e-06,
+      "loss": 0.8554,
+      "step": 25448
+    },
+    {
+      "epoch": 0.6989371343824669,
+      "grad_norm": 0.0,
+      "learning_rate": 4.389043336940172e-06,
+      "loss": 0.9016,
+      "step": 25449
+    },
+    {
+      "epoch": 0.69896459861031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.38830706140495e-06,
+      "loss": 0.801,
+      "step": 25450
+    },
+    {
+      "epoch": 0.6989920628381533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.387570830271615e-06,
+      "loss": 0.7754,
+      "step": 25451
+    },
+    {
+      "epoch": 0.6990195270659966,
+      "grad_norm": 0.0,
+      "learning_rate": 4.386834643545999e-06,
+      "loss": 0.8396,
+      "step": 25452
+    },
+    {
+      "epoch": 0.6990469912938397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.386098501233916e-06,
+      "loss": 0.845,
+      "step": 25453
+    },
+    {
+      "epoch": 0.699074455521683,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3853624033412015e-06,
+      "loss": 0.8656,
+      "step": 25454
+    },
+    {
+      "epoch": 0.6991019197495263,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3846263498736705e-06,
+      "loss": 0.8693,
+      "step": 25455
+    },
+    {
+      "epoch": 0.6991293839773695,
+      "grad_norm": 0.0,
+      "learning_rate": 4.383890340837151e-06,
+      "loss": 0.8061,
+      "step": 25456
+    },
+    {
+      "epoch": 0.6991568482052127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.383154376237468e-06,
+      "loss": 0.8747,
+      "step": 25457
+    },
+    {
+      "epoch": 0.6991843124330559,
+      "grad_norm": 0.0,
+      "learning_rate": 4.382418456080441e-06,
+      "loss": 0.8603,
+      "step": 25458
+    },
+    {
+      "epoch": 0.6992117766608992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3816825803719e-06,
+      "loss": 0.778,
+      "step": 25459
+    },
+    {
+      "epoch": 0.6992392408887425,
+      "grad_norm": 0.0,
+      "learning_rate": 4.380946749117663e-06,
+      "loss": 0.7931,
+      "step": 25460
+    },
+    {
+      "epoch": 0.6992667051165856,
+      "grad_norm": 0.0,
+      "learning_rate": 4.380210962323547e-06,
+      "loss": 0.7939,
+      "step": 25461
+    },
+    {
+      "epoch": 0.6992941693444289,
+      "grad_norm": 0.0,
+      "learning_rate": 4.379475219995379e-06,
+      "loss": 0.7938,
+      "step": 25462
+    },
+    {
+      "epoch": 0.6993216335722721,
+      "grad_norm": 0.0,
+      "learning_rate": 4.378739522138981e-06,
+      "loss": 0.907,
+      "step": 25463
+    },
+    {
+      "epoch": 0.6993490978001153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.378003868760174e-06,
+      "loss": 0.8009,
+      "step": 25464
+    },
+    {
+      "epoch": 0.6993765620279586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.377268259864777e-06,
+      "loss": 0.8542,
+      "step": 25465
+    },
+    {
+      "epoch": 0.6994040262558018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.376532695458616e-06,
+      "loss": 0.8035,
+      "step": 25466
+    },
+    {
+      "epoch": 0.6994314904836451,
+      "grad_norm": 0.0,
+      "learning_rate": 4.375797175547503e-06,
+      "loss": 0.7996,
+      "step": 25467
+    },
+    {
+      "epoch": 0.6994589547114883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3750617001372645e-06,
+      "loss": 0.7854,
+      "step": 25468
+    },
+    {
+      "epoch": 0.6994864189393315,
+      "grad_norm": 0.0,
+      "learning_rate": 4.374326269233713e-06,
+      "loss": 0.7629,
+      "step": 25469
+    },
+    {
+      "epoch": 0.6995138831671748,
+      "grad_norm": 0.0,
+      "learning_rate": 4.373590882842671e-06,
+      "loss": 0.8458,
+      "step": 25470
+    },
+    {
+      "epoch": 0.6995413473950179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.372855540969958e-06,
+      "loss": 0.8876,
+      "step": 25471
+    },
+    {
+      "epoch": 0.6995688116228612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.372120243621395e-06,
+      "loss": 0.8214,
+      "step": 25472
+    },
+    {
+      "epoch": 0.6995962758507045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.371384990802794e-06,
+      "loss": 0.9131,
+      "step": 25473
+    },
+    {
+      "epoch": 0.6996237400785477,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3706497825199746e-06,
+      "loss": 0.8994,
+      "step": 25474
+    },
+    {
+      "epoch": 0.6996512043063909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.369914618778754e-06,
+      "loss": 0.8901,
+      "step": 25475
+    },
+    {
+      "epoch": 0.6996786685342341,
+      "grad_norm": 0.0,
+      "learning_rate": 4.369179499584955e-06,
+      "loss": 0.8885,
+      "step": 25476
+    },
+    {
+      "epoch": 0.6997061327620774,
+      "grad_norm": 0.0,
+      "learning_rate": 4.368444424944384e-06,
+      "loss": 0.8776,
+      "step": 25477
+    },
+    {
+      "epoch": 0.6997335969899207,
+      "grad_norm": 0.0,
+      "learning_rate": 4.367709394862865e-06,
+      "loss": 0.7616,
+      "step": 25478
+    },
+    {
+      "epoch": 0.6997610612177638,
+      "grad_norm": 0.0,
+      "learning_rate": 4.36697440934621e-06,
+      "loss": 0.8877,
+      "step": 25479
+    },
+    {
+      "epoch": 0.6997885254456071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.366239468400233e-06,
+      "loss": 0.8501,
+      "step": 25480
+    },
+    {
+      "epoch": 0.6998159896734504,
+      "grad_norm": 0.0,
+      "learning_rate": 4.365504572030753e-06,
+      "loss": 0.7939,
+      "step": 25481
+    },
+    {
+      "epoch": 0.6998434539012935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.364769720243583e-06,
+      "loss": 0.8502,
+      "step": 25482
+    },
+    {
+      "epoch": 0.6998709181291368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.364034913044541e-06,
+      "loss": 0.9095,
+      "step": 25483
+    },
+    {
+      "epoch": 0.69989838235698,
+      "grad_norm": 0.0,
+      "learning_rate": 4.363300150439439e-06,
+      "loss": 0.7888,
+      "step": 25484
+    },
+    {
+      "epoch": 0.6999258465848233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.362565432434086e-06,
+      "loss": 0.739,
+      "step": 25485
+    },
+    {
+      "epoch": 0.6999533108126665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.361830759034299e-06,
+      "loss": 0.7967,
+      "step": 25486
+    },
+    {
+      "epoch": 0.6999807750405097,
+      "grad_norm": 0.0,
+      "learning_rate": 4.36109613024589e-06,
+      "loss": 0.8756,
+      "step": 25487
+    },
+    {
+      "epoch": 0.700008239268353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.360361546074674e-06,
+      "loss": 0.7714,
+      "step": 25488
+    },
+    {
+      "epoch": 0.7000357034961963,
+      "grad_norm": 0.0,
+      "learning_rate": 4.359627006526461e-06,
+      "loss": 0.8745,
+      "step": 25489
+    },
+    {
+      "epoch": 0.7000631677240394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.358892511607068e-06,
+      "loss": 0.9783,
+      "step": 25490
+    },
+    {
+      "epoch": 0.7000906319518827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.358158061322299e-06,
+      "loss": 0.828,
+      "step": 25491
+    },
+    {
+      "epoch": 0.7001180961797259,
+      "grad_norm": 0.0,
+      "learning_rate": 4.357423655677973e-06,
+      "loss": 0.8382,
+      "step": 25492
+    },
+    {
+      "epoch": 0.7001455604075691,
+      "grad_norm": 0.0,
+      "learning_rate": 4.356689294679892e-06,
+      "loss": 0.8643,
+      "step": 25493
+    },
+    {
+      "epoch": 0.7001730246354124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.355954978333872e-06,
+      "loss": 0.7731,
+      "step": 25494
+    },
+    {
+      "epoch": 0.7002004888632556,
+      "grad_norm": 0.0,
+      "learning_rate": 4.355220706645722e-06,
+      "loss": 0.831,
+      "step": 25495
+    },
+    {
+      "epoch": 0.7002279530910989,
+      "grad_norm": 0.0,
+      "learning_rate": 4.354486479621257e-06,
+      "loss": 0.8012,
+      "step": 25496
+    },
+    {
+      "epoch": 0.700255417318942,
+      "grad_norm": 0.0,
+      "learning_rate": 4.353752297266277e-06,
+      "loss": 0.8208,
+      "step": 25497
+    },
+    {
+      "epoch": 0.7002828815467853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.353018159586597e-06,
+      "loss": 0.7884,
+      "step": 25498
+    },
+    {
+      "epoch": 0.7003103457746286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.352284066588028e-06,
+      "loss": 0.8153,
+      "step": 25499
+    },
+    {
+      "epoch": 0.7003378100024718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.351550018276371e-06,
+      "loss": 0.9103,
+      "step": 25500
+    },
+    {
+      "epoch": 0.700365274230315,
+      "grad_norm": 0.0,
+      "learning_rate": 4.35081601465744e-06,
+      "loss": 0.8669,
+      "step": 25501
+    },
+    {
+      "epoch": 0.7003927384581583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.350082055737043e-06,
+      "loss": 0.8536,
+      "step": 25502
+    },
+    {
+      "epoch": 0.7004202026860015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.349348141520981e-06,
+      "loss": 1.0078,
+      "step": 25503
+    },
+    {
+      "epoch": 0.7004476669138447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.348614272015067e-06,
+      "loss": 0.8034,
+      "step": 25504
+    },
+    {
+      "epoch": 0.7004751311416879,
+      "grad_norm": 0.0,
+      "learning_rate": 4.347880447225106e-06,
+      "loss": 0.8541,
+      "step": 25505
+    },
+    {
+      "epoch": 0.7005025953695312,
+      "grad_norm": 0.0,
+      "learning_rate": 4.347146667156904e-06,
+      "loss": 0.8569,
+      "step": 25506
+    },
+    {
+      "epoch": 0.7005300595973745,
+      "grad_norm": 0.0,
+      "learning_rate": 4.34641293181627e-06,
+      "loss": 0.7267,
+      "step": 25507
+    },
+    {
+      "epoch": 0.7005575238252176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.345679241209007e-06,
+      "loss": 0.831,
+      "step": 25508
+    },
+    {
+      "epoch": 0.7005849880530609,
+      "grad_norm": 0.0,
+      "learning_rate": 4.344945595340918e-06,
+      "loss": 0.7333,
+      "step": 25509
+    },
+    {
+      "epoch": 0.7006124522809041,
+      "grad_norm": 0.0,
+      "learning_rate": 4.344211994217809e-06,
+      "loss": 0.9208,
+      "step": 25510
+    },
+    {
+      "epoch": 0.7006399165087474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.343478437845485e-06,
+      "loss": 0.8922,
+      "step": 25511
+    },
+    {
+      "epoch": 0.7006673807365906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.342744926229751e-06,
+      "loss": 0.7842,
+      "step": 25512
+    },
+    {
+      "epoch": 0.7006948449644338,
+      "grad_norm": 0.0,
+      "learning_rate": 4.342011459376415e-06,
+      "loss": 0.8079,
+      "step": 25513
+    },
+    {
+      "epoch": 0.7007223091922771,
+      "grad_norm": 0.0,
+      "learning_rate": 4.341278037291272e-06,
+      "loss": 0.8529,
+      "step": 25514
+    },
+    {
+      "epoch": 0.7007497734201203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3405446599801295e-06,
+      "loss": 0.8565,
+      "step": 25515
+    },
+    {
+      "epoch": 0.7007772376479635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.339811327448793e-06,
+      "loss": 0.8735,
+      "step": 25516
+    },
+    {
+      "epoch": 0.7008047018758068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.339078039703059e-06,
+      "loss": 0.735,
+      "step": 25517
+    },
+    {
+      "epoch": 0.70083216610365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.338344796748731e-06,
+      "loss": 0.8351,
+      "step": 25518
+    },
+    {
+      "epoch": 0.7008596303314932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.337611598591617e-06,
+      "loss": 0.8294,
+      "step": 25519
+    },
+    {
+      "epoch": 0.7008870945593365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.336878445237509e-06,
+      "loss": 0.8045,
+      "step": 25520
+    },
+    {
+      "epoch": 0.7009145587871797,
+      "grad_norm": 0.0,
+      "learning_rate": 4.336145336692212e-06,
+      "loss": 0.844,
+      "step": 25521
+    },
+    {
+      "epoch": 0.700942023015023,
+      "grad_norm": 0.0,
+      "learning_rate": 4.335412272961528e-06,
+      "loss": 0.8884,
+      "step": 25522
+    },
+    {
+      "epoch": 0.7009694872428661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3346792540512596e-06,
+      "loss": 0.8481,
+      "step": 25523
+    },
+    {
+      "epoch": 0.7009969514707094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.333946279967201e-06,
+      "loss": 0.9084,
+      "step": 25524
+    },
+    {
+      "epoch": 0.7010244156985527,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3332133507151565e-06,
+      "loss": 0.7717,
+      "step": 25525
+    },
+    {
+      "epoch": 0.7010518799263958,
+      "grad_norm": 0.0,
+      "learning_rate": 4.332480466300919e-06,
+      "loss": 0.9102,
+      "step": 25526
+    },
+    {
+      "epoch": 0.7010793441542391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3317476267302925e-06,
+      "loss": 0.9849,
+      "step": 25527
+    },
+    {
+      "epoch": 0.7011068083820824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3310148320090736e-06,
+      "loss": 0.8703,
+      "step": 25528
+    },
+    {
+      "epoch": 0.7011342726099256,
+      "grad_norm": 0.0,
+      "learning_rate": 4.330282082143062e-06,
+      "loss": 0.9108,
+      "step": 25529
+    },
+    {
+      "epoch": 0.7011617368377688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.329549377138055e-06,
+      "loss": 0.8114,
+      "step": 25530
+    },
+    {
+      "epoch": 0.701189201065612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.328816716999853e-06,
+      "loss": 0.8628,
+      "step": 25531
+    },
+    {
+      "epoch": 0.7012166652934553,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3280841017342504e-06,
+      "loss": 0.819,
+      "step": 25532
+    },
+    {
+      "epoch": 0.7012441295212986,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3273515313470395e-06,
+      "loss": 0.7305,
+      "step": 25533
+    },
+    {
+      "epoch": 0.7012715937491417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3266190058440205e-06,
+      "loss": 0.7943,
+      "step": 25534
+    },
+    {
+      "epoch": 0.701299057976985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.32588652523099e-06,
+      "loss": 0.894,
+      "step": 25535
+    },
+    {
+      "epoch": 0.7013265222048282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.325154089513742e-06,
+      "loss": 0.7556,
+      "step": 25536
+    },
+    {
+      "epoch": 0.7013539864326714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.324421698698078e-06,
+      "loss": 0.8319,
+      "step": 25537
+    },
+    {
+      "epoch": 0.7013814506605147,
+      "grad_norm": 0.0,
+      "learning_rate": 4.323689352789785e-06,
+      "loss": 0.9076,
+      "step": 25538
+    },
+    {
+      "epoch": 0.7014089148883579,
+      "grad_norm": 0.0,
+      "learning_rate": 4.322957051794659e-06,
+      "loss": 0.9227,
+      "step": 25539
+    },
+    {
+      "epoch": 0.7014363791162012,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3222247957185e-06,
+      "loss": 0.7334,
+      "step": 25540
+    },
+    {
+      "epoch": 0.7014638433440444,
+      "grad_norm": 0.0,
+      "learning_rate": 4.321492584567096e-06,
+      "loss": 0.798,
+      "step": 25541
+    },
+    {
+      "epoch": 0.7014913075718876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3207604183462395e-06,
+      "loss": 0.8532,
+      "step": 25542
+    },
+    {
+      "epoch": 0.7015187717997309,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3200282970617315e-06,
+      "loss": 0.9457,
+      "step": 25543
+    },
+    {
+      "epoch": 0.701546236027574,
+      "grad_norm": 0.0,
+      "learning_rate": 4.319296220719356e-06,
+      "loss": 0.9265,
+      "step": 25544
+    },
+    {
+      "epoch": 0.7015737002554173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.31856418932491e-06,
+      "loss": 0.9384,
+      "step": 25545
+    },
+    {
+      "epoch": 0.7016011644832606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.317832202884183e-06,
+      "loss": 0.8865,
+      "step": 25546
+    },
+    {
+      "epoch": 0.7016286287111038,
+      "grad_norm": 0.0,
+      "learning_rate": 4.317100261402975e-06,
+      "loss": 0.7957,
+      "step": 25547
+    },
+    {
+      "epoch": 0.701656092938947,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3163683648870645e-06,
+      "loss": 0.8399,
+      "step": 25548
+    },
+    {
+      "epoch": 0.7016835571667902,
+      "grad_norm": 0.0,
+      "learning_rate": 4.315636513342255e-06,
+      "loss": 0.7827,
+      "step": 25549
+    },
+    {
+      "epoch": 0.7017110213946335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.314904706774327e-06,
+      "loss": 0.8684,
+      "step": 25550
+    },
+    {
+      "epoch": 0.7017384856224768,
+      "grad_norm": 0.0,
+      "learning_rate": 4.314172945189074e-06,
+      "loss": 0.8882,
+      "step": 25551
+    },
+    {
+      "epoch": 0.7017659498503199,
+      "grad_norm": 0.0,
+      "learning_rate": 4.313441228592289e-06,
+      "loss": 0.8613,
+      "step": 25552
+    },
+    {
+      "epoch": 0.7017934140781632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3127095569897594e-06,
+      "loss": 0.7583,
+      "step": 25553
+    },
+    {
+      "epoch": 0.7018208783060065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.311977930387274e-06,
+      "loss": 0.8509,
+      "step": 25554
+    },
+    {
+      "epoch": 0.7018483425338496,
+      "grad_norm": 0.0,
+      "learning_rate": 4.311246348790627e-06,
+      "loss": 0.8778,
+      "step": 25555
+    },
+    {
+      "epoch": 0.7018758067616929,
+      "grad_norm": 0.0,
+      "learning_rate": 4.310514812205602e-06,
+      "loss": 0.7829,
+      "step": 25556
+    },
+    {
+      "epoch": 0.7019032709895361,
+      "grad_norm": 0.0,
+      "learning_rate": 4.309783320637985e-06,
+      "loss": 0.8451,
+      "step": 25557
+    },
+    {
+      "epoch": 0.7019307352173794,
+      "grad_norm": 0.0,
+      "learning_rate": 4.309051874093565e-06,
+      "loss": 0.6754,
+      "step": 25558
+    },
+    {
+      "epoch": 0.7019581994452226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.308320472578131e-06,
+      "loss": 0.8894,
+      "step": 25559
+    },
+    {
+      "epoch": 0.7019856636730658,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3075891160974714e-06,
+      "loss": 0.8961,
+      "step": 25560
+    },
+    {
+      "epoch": 0.7020131279009091,
+      "grad_norm": 0.0,
+      "learning_rate": 4.306857804657375e-06,
+      "loss": 0.8357,
+      "step": 25561
+    },
+    {
+      "epoch": 0.7020405921287522,
+      "grad_norm": 0.0,
+      "learning_rate": 4.306126538263621e-06,
+      "loss": 0.8034,
+      "step": 25562
+    },
+    {
+      "epoch": 0.7020680563565955,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3053953169219994e-06,
+      "loss": 0.8198,
+      "step": 25563
+    },
+    {
+      "epoch": 0.7020955205844388,
+      "grad_norm": 0.0,
+      "learning_rate": 4.304664140638301e-06,
+      "loss": 0.8066,
+      "step": 25564
+    },
+    {
+      "epoch": 0.702122984812282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.303933009418301e-06,
+      "loss": 0.8545,
+      "step": 25565
+    },
+    {
+      "epoch": 0.7021504490401252,
+      "grad_norm": 0.0,
+      "learning_rate": 4.303201923267792e-06,
+      "loss": 0.8325,
+      "step": 25566
+    },
+    {
+      "epoch": 0.7021779132679685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.302470882192554e-06,
+      "loss": 0.9371,
+      "step": 25567
+    },
+    {
+      "epoch": 0.7022053774958117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.301739886198374e-06,
+      "loss": 0.79,
+      "step": 25568
+    },
+    {
+      "epoch": 0.702232841723655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3010089352910345e-06,
+      "loss": 0.9045,
+      "step": 25569
+    },
+    {
+      "epoch": 0.7022603059514981,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3002780294763205e-06,
+      "loss": 0.8189,
+      "step": 25570
+    },
+    {
+      "epoch": 0.7022877701793414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2995471687600165e-06,
+      "loss": 0.776,
+      "step": 25571
+    },
+    {
+      "epoch": 0.7023152344071847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.298816353147901e-06,
+      "loss": 0.879,
+      "step": 25572
+    },
+    {
+      "epoch": 0.7023426986350279,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2980855826457635e-06,
+      "loss": 0.8423,
+      "step": 25573
+    },
+    {
+      "epoch": 0.7023701628628711,
+      "grad_norm": 0.0,
+      "learning_rate": 4.297354857259377e-06,
+      "loss": 0.9484,
+      "step": 25574
+    },
+    {
+      "epoch": 0.7023976270907143,
+      "grad_norm": 0.0,
+      "learning_rate": 4.296624176994527e-06,
+      "loss": 0.8058,
+      "step": 25575
+    },
+    {
+      "epoch": 0.7024250913185576,
+      "grad_norm": 0.0,
+      "learning_rate": 4.295893541856997e-06,
+      "loss": 0.9079,
+      "step": 25576
+    },
+    {
+      "epoch": 0.7024525555464008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2951629518525666e-06,
+      "loss": 0.896,
+      "step": 25577
+    },
+    {
+      "epoch": 0.702480019774244,
+      "grad_norm": 0.0,
+      "learning_rate": 4.29443240698702e-06,
+      "loss": 0.8185,
+      "step": 25578
+    },
+    {
+      "epoch": 0.7025074840020873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.293701907266131e-06,
+      "loss": 0.8285,
+      "step": 25579
+    },
+    {
+      "epoch": 0.7025349482299306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.292971452695686e-06,
+      "loss": 0.7619,
+      "step": 25580
+    },
+    {
+      "epoch": 0.7025624124577737,
+      "grad_norm": 0.0,
+      "learning_rate": 4.292241043281459e-06,
+      "loss": 0.8383,
+      "step": 25581
+    },
+    {
+      "epoch": 0.702589876685617,
+      "grad_norm": 0.0,
+      "learning_rate": 4.291510679029232e-06,
+      "loss": 0.8212,
+      "step": 25582
+    },
+    {
+      "epoch": 0.7026173409134602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.290780359944783e-06,
+      "loss": 0.8369,
+      "step": 25583
+    },
+    {
+      "epoch": 0.7026448051413035,
+      "grad_norm": 0.0,
+      "learning_rate": 4.290050086033894e-06,
+      "loss": 0.7118,
+      "step": 25584
+    },
+    {
+      "epoch": 0.7026722693691467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.289319857302338e-06,
+      "loss": 0.8382,
+      "step": 25585
+    },
+    {
+      "epoch": 0.7026997335969899,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2885896737558965e-06,
+      "loss": 0.9283,
+      "step": 25586
+    },
+    {
+      "epoch": 0.7027271978248332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.287859535400344e-06,
+      "loss": 0.8747,
+      "step": 25587
+    },
+    {
+      "epoch": 0.7027546620526763,
+      "grad_norm": 0.0,
+      "learning_rate": 4.287129442241464e-06,
+      "loss": 0.8261,
+      "step": 25588
+    },
+    {
+      "epoch": 0.7027821262805196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.286399394285024e-06,
+      "loss": 0.7961,
+      "step": 25589
+    },
+    {
+      "epoch": 0.7028095905083629,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2856693915368104e-06,
+      "loss": 0.8285,
+      "step": 25590
+    },
+    {
+      "epoch": 0.7028370547362061,
+      "grad_norm": 0.0,
+      "learning_rate": 4.284939434002591e-06,
+      "loss": 0.9196,
+      "step": 25591
+    },
+    {
+      "epoch": 0.7028645189640493,
+      "grad_norm": 0.0,
+      "learning_rate": 4.284209521688143e-06,
+      "loss": 0.8549,
+      "step": 25592
+    },
+    {
+      "epoch": 0.7028919831918926,
+      "grad_norm": 0.0,
+      "learning_rate": 4.283479654599244e-06,
+      "loss": 0.9148,
+      "step": 25593
+    },
+    {
+      "epoch": 0.7029194474197358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.282749832741668e-06,
+      "loss": 0.7972,
+      "step": 25594
+    },
+    {
+      "epoch": 0.702946911647579,
+      "grad_norm": 0.0,
+      "learning_rate": 4.282020056121194e-06,
+      "loss": 0.8435,
+      "step": 25595
+    },
+    {
+      "epoch": 0.7029743758754222,
+      "grad_norm": 0.0,
+      "learning_rate": 4.281290324743589e-06,
+      "loss": 0.8818,
+      "step": 25596
+    },
+    {
+      "epoch": 0.7030018401032655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.280560638614634e-06,
+      "loss": 0.891,
+      "step": 25597
+    },
+    {
+      "epoch": 0.7030293043311088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.279830997740093e-06,
+      "loss": 0.8243,
+      "step": 25598
+    },
+    {
+      "epoch": 0.7030567685589519,
+      "grad_norm": 0.0,
+      "learning_rate": 4.279101402125747e-06,
+      "loss": 0.7398,
+      "step": 25599
+    },
+    {
+      "epoch": 0.7030842327867952,
+      "grad_norm": 0.0,
+      "learning_rate": 4.278371851777366e-06,
+      "loss": 0.7403,
+      "step": 25600
+    },
+    {
+      "epoch": 0.7031116970146384,
+      "grad_norm": 0.0,
+      "learning_rate": 4.277642346700724e-06,
+      "loss": 0.8928,
+      "step": 25601
+    },
+    {
+      "epoch": 0.7031391612424817,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2769128869015954e-06,
+      "loss": 0.7882,
+      "step": 25602
+    },
+    {
+      "epoch": 0.7031666254703249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.276183472385744e-06,
+      "loss": 0.7692,
+      "step": 25603
+    },
+    {
+      "epoch": 0.7031940896981681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.275454103158952e-06,
+      "loss": 0.817,
+      "step": 25604
+    },
+    {
+      "epoch": 0.7032215539260114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2747247792269795e-06,
+      "loss": 0.7661,
+      "step": 25605
+    },
+    {
+      "epoch": 0.7032490181538547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.273995500595603e-06,
+      "loss": 0.8641,
+      "step": 25606
+    },
+    {
+      "epoch": 0.7032764823816978,
+      "grad_norm": 0.0,
+      "learning_rate": 4.273266267270591e-06,
+      "loss": 0.8301,
+      "step": 25607
+    },
+    {
+      "epoch": 0.7033039466095411,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2725370792577205e-06,
+      "loss": 0.8437,
+      "step": 25608
+    },
+    {
+      "epoch": 0.7033314108373843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.27180793656275e-06,
+      "loss": 0.7954,
+      "step": 25609
+    },
+    {
+      "epoch": 0.7033588750652275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2710788391914545e-06,
+      "loss": 0.807,
+      "step": 25610
+    },
+    {
+      "epoch": 0.7033863392930708,
+      "grad_norm": 0.0,
+      "learning_rate": 4.270349787149603e-06,
+      "loss": 0.8492,
+      "step": 25611
+    },
+    {
+      "epoch": 0.703413803520914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.269620780442965e-06,
+      "loss": 0.8607,
+      "step": 25612
+    },
+    {
+      "epoch": 0.7034412677487573,
+      "grad_norm": 0.0,
+      "learning_rate": 4.268891819077305e-06,
+      "loss": 0.8225,
+      "step": 25613
+    },
+    {
+      "epoch": 0.7034687319766004,
+      "grad_norm": 0.0,
+      "learning_rate": 4.268162903058397e-06,
+      "loss": 0.8998,
+      "step": 25614
+    },
+    {
+      "epoch": 0.7034961962044437,
+      "grad_norm": 0.0,
+      "learning_rate": 4.267434032392001e-06,
+      "loss": 0.8446,
+      "step": 25615
+    },
+    {
+      "epoch": 0.703523660432287,
+      "grad_norm": 0.0,
+      "learning_rate": 4.266705207083885e-06,
+      "loss": 0.8175,
+      "step": 25616
+    },
+    {
+      "epoch": 0.7035511246601301,
+      "grad_norm": 0.0,
+      "learning_rate": 4.265976427139821e-06,
+      "loss": 0.8885,
+      "step": 25617
+    },
+    {
+      "epoch": 0.7035785888879734,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2652476925655716e-06,
+      "loss": 0.8627,
+      "step": 25618
+    },
+    {
+      "epoch": 0.7036060531158167,
+      "grad_norm": 0.0,
+      "learning_rate": 4.264519003366907e-06,
+      "loss": 0.8266,
+      "step": 25619
+    },
+    {
+      "epoch": 0.7036335173436599,
+      "grad_norm": 0.0,
+      "learning_rate": 4.26379035954959e-06,
+      "loss": 0.7924,
+      "step": 25620
+    },
+    {
+      "epoch": 0.7036609815715031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.263061761119383e-06,
+      "loss": 0.8246,
+      "step": 25621
+    },
+    {
+      "epoch": 0.7036884457993463,
+      "grad_norm": 0.0,
+      "learning_rate": 4.262333208082052e-06,
+      "loss": 0.8469,
+      "step": 25622
+    },
+    {
+      "epoch": 0.7037159100271896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.261604700443365e-06,
+      "loss": 0.8141,
+      "step": 25623
+    },
+    {
+      "epoch": 0.7037433742550329,
+      "grad_norm": 0.0,
+      "learning_rate": 4.260876238209084e-06,
+      "loss": 0.8743,
+      "step": 25624
+    },
+    {
+      "epoch": 0.703770838482876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2601478213849754e-06,
+      "loss": 0.8637,
+      "step": 25625
+    },
+    {
+      "epoch": 0.7037983027107193,
+      "grad_norm": 0.0,
+      "learning_rate": 4.259419449976799e-06,
+      "loss": 0.7808,
+      "step": 25626
+    },
+    {
+      "epoch": 0.7038257669385625,
+      "grad_norm": 0.0,
+      "learning_rate": 4.258691123990318e-06,
+      "loss": 0.7529,
+      "step": 25627
+    },
+    {
+      "epoch": 0.7038532311664057,
+      "grad_norm": 0.0,
+      "learning_rate": 4.257962843431301e-06,
+      "loss": 0.7704,
+      "step": 25628
+    },
+    {
+      "epoch": 0.703880695394249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2572346083055034e-06,
+      "loss": 0.8575,
+      "step": 25629
+    },
+    {
+      "epoch": 0.7039081596220922,
+      "grad_norm": 0.0,
+      "learning_rate": 4.25650641861869e-06,
+      "loss": 0.8153,
+      "step": 25630
+    },
+    {
+      "epoch": 0.7039356238499355,
+      "grad_norm": 0.0,
+      "learning_rate": 4.255778274376625e-06,
+      "loss": 0.7922,
+      "step": 25631
+    },
+    {
+      "epoch": 0.7039630880777787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.255050175585065e-06,
+      "loss": 0.8582,
+      "step": 25632
+    },
+    {
+      "epoch": 0.7039905523056219,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2543221222497736e-06,
+      "loss": 0.8924,
+      "step": 25633
+    },
+    {
+      "epoch": 0.7040180165334652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.25359411437651e-06,
+      "loss": 0.8832,
+      "step": 25634
+    },
+    {
+      "epoch": 0.7040454807613084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.252866151971037e-06,
+      "loss": 0.8135,
+      "step": 25635
+    },
+    {
+      "epoch": 0.7040729449891516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.252138235039116e-06,
+      "loss": 0.8233,
+      "step": 25636
+    },
+    {
+      "epoch": 0.7041004092169949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2514103635865035e-06,
+      "loss": 0.857,
+      "step": 25637
+    },
+    {
+      "epoch": 0.7041278734448381,
+      "grad_norm": 0.0,
+      "learning_rate": 4.250682537618956e-06,
+      "loss": 0.8652,
+      "step": 25638
+    },
+    {
+      "epoch": 0.7041553376726813,
+      "grad_norm": 0.0,
+      "learning_rate": 4.249954757142235e-06,
+      "loss": 0.7808,
+      "step": 25639
+    },
+    {
+      "epoch": 0.7041828019005245,
+      "grad_norm": 0.0,
+      "learning_rate": 4.249227022162099e-06,
+      "loss": 0.7159,
+      "step": 25640
+    },
+    {
+      "epoch": 0.7042102661283678,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2484993326843075e-06,
+      "loss": 0.7676,
+      "step": 25641
+    },
+    {
+      "epoch": 0.7042377303562111,
+      "grad_norm": 0.0,
+      "learning_rate": 4.247771688714617e-06,
+      "loss": 0.8527,
+      "step": 25642
+    },
+    {
+      "epoch": 0.7042651945840542,
+      "grad_norm": 0.0,
+      "learning_rate": 4.247044090258789e-06,
+      "loss": 0.7367,
+      "step": 25643
+    },
+    {
+      "epoch": 0.7042926588118975,
+      "grad_norm": 0.0,
+      "learning_rate": 4.246316537322574e-06,
+      "loss": 0.9397,
+      "step": 25644
+    },
+    {
+      "epoch": 0.7043201230397408,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2455890299117306e-06,
+      "loss": 0.8138,
+      "step": 25645
+    },
+    {
+      "epoch": 0.704347587267584,
+      "grad_norm": 0.0,
+      "learning_rate": 4.244861568032015e-06,
+      "loss": 0.8536,
+      "step": 25646
+    },
+    {
+      "epoch": 0.7043750514954272,
+      "grad_norm": 0.0,
+      "learning_rate": 4.244134151689183e-06,
+      "loss": 0.8783,
+      "step": 25647
+    },
+    {
+      "epoch": 0.7044025157232704,
+      "grad_norm": 0.0,
+      "learning_rate": 4.243406780888992e-06,
+      "loss": 0.8922,
+      "step": 25648
+    },
+    {
+      "epoch": 0.7044299799511137,
+      "grad_norm": 0.0,
+      "learning_rate": 4.242679455637201e-06,
+      "loss": 0.8944,
+      "step": 25649
+    },
+    {
+      "epoch": 0.704457444178957,
+      "grad_norm": 0.0,
+      "learning_rate": 4.241952175939555e-06,
+      "loss": 0.8549,
+      "step": 25650
+    },
+    {
+      "epoch": 0.7044849084068001,
+      "grad_norm": 0.0,
+      "learning_rate": 4.241224941801816e-06,
+      "loss": 0.7592,
+      "step": 25651
+    },
+    {
+      "epoch": 0.7045123726346434,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2404977532297375e-06,
+      "loss": 0.8352,
+      "step": 25652
+    },
+    {
+      "epoch": 0.7045398368624866,
+      "grad_norm": 0.0,
+      "learning_rate": 4.239770610229069e-06,
+      "loss": 0.8802,
+      "step": 25653
+    },
+    {
+      "epoch": 0.7045673010903298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.239043512805567e-06,
+      "loss": 0.7681,
+      "step": 25654
+    },
+    {
+      "epoch": 0.7045947653181731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.238316460964987e-06,
+      "loss": 0.8241,
+      "step": 25655
+    },
+    {
+      "epoch": 0.7046222295460163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.237589454713077e-06,
+      "loss": 0.7656,
+      "step": 25656
+    },
+    {
+      "epoch": 0.7046496937738596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2368624940555905e-06,
+      "loss": 0.8125,
+      "step": 25657
+    },
+    {
+      "epoch": 0.7046771580017028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.236135578998281e-06,
+      "loss": 0.8073,
+      "step": 25658
+    },
+    {
+      "epoch": 0.704704622229546,
+      "grad_norm": 0.0,
+      "learning_rate": 4.235408709546901e-06,
+      "loss": 0.8765,
+      "step": 25659
+    },
+    {
+      "epoch": 0.7047320864573893,
+      "grad_norm": 0.0,
+      "learning_rate": 4.234681885707198e-06,
+      "loss": 0.8673,
+      "step": 25660
+    },
+    {
+      "epoch": 0.7047595506852324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.233955107484929e-06,
+      "loss": 0.9223,
+      "step": 25661
+    },
+    {
+      "epoch": 0.7047870149130757,
+      "grad_norm": 0.0,
+      "learning_rate": 4.233228374885837e-06,
+      "loss": 0.8283,
+      "step": 25662
+    },
+    {
+      "epoch": 0.704814479140919,
+      "grad_norm": 0.0,
+      "learning_rate": 4.232501687915677e-06,
+      "loss": 0.7194,
+      "step": 25663
+    },
+    {
+      "epoch": 0.7048419433687622,
+      "grad_norm": 0.0,
+      "learning_rate": 4.231775046580197e-06,
+      "loss": 0.8663,
+      "step": 25664
+    },
+    {
+      "epoch": 0.7048694075966054,
+      "grad_norm": 0.0,
+      "learning_rate": 4.231048450885146e-06,
+      "loss": 0.8982,
+      "step": 25665
+    },
+    {
+      "epoch": 0.7048968718244486,
+      "grad_norm": 0.0,
+      "learning_rate": 4.230321900836275e-06,
+      "loss": 0.8619,
+      "step": 25666
+    },
+    {
+      "epoch": 0.7049243360522919,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2295953964393365e-06,
+      "loss": 0.7749,
+      "step": 25667
+    },
+    {
+      "epoch": 0.7049518002801352,
+      "grad_norm": 0.0,
+      "learning_rate": 4.228868937700073e-06,
+      "loss": 0.9154,
+      "step": 25668
+    },
+    {
+      "epoch": 0.7049792645079783,
+      "grad_norm": 0.0,
+      "learning_rate": 4.228142524624233e-06,
+      "loss": 0.8912,
+      "step": 25669
+    },
+    {
+      "epoch": 0.7050067287358216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.227416157217563e-06,
+      "loss": 0.8703,
+      "step": 25670
+    },
+    {
+      "epoch": 0.7050341929636649,
+      "grad_norm": 0.0,
+      "learning_rate": 4.226689835485812e-06,
+      "loss": 0.8358,
+      "step": 25671
+    },
+    {
+      "epoch": 0.705061657191508,
+      "grad_norm": 0.0,
+      "learning_rate": 4.225963559434729e-06,
+      "loss": 0.7521,
+      "step": 25672
+    },
+    {
+      "epoch": 0.7050891214193513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.225237329070062e-06,
+      "loss": 0.7979,
+      "step": 25673
+    },
+    {
+      "epoch": 0.7051165856471945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.22451114439755e-06,
+      "loss": 0.8362,
+      "step": 25674
+    },
+    {
+      "epoch": 0.7051440498750378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.223785005422944e-06,
+      "loss": 0.8609,
+      "step": 25675
+    },
+    {
+      "epoch": 0.705171514102881,
+      "grad_norm": 0.0,
+      "learning_rate": 4.223058912151992e-06,
+      "loss": 0.8866,
+      "step": 25676
+    },
+    {
+      "epoch": 0.7051989783307242,
+      "grad_norm": 0.0,
+      "learning_rate": 4.222332864590432e-06,
+      "loss": 0.818,
+      "step": 25677
+    },
+    {
+      "epoch": 0.7052264425585675,
+      "grad_norm": 0.0,
+      "learning_rate": 4.221606862744011e-06,
+      "loss": 0.8857,
+      "step": 25678
+    },
+    {
+      "epoch": 0.7052539067864106,
+      "grad_norm": 0.0,
+      "learning_rate": 4.220880906618479e-06,
+      "loss": 0.774,
+      "step": 25679
+    },
+    {
+      "epoch": 0.7052813710142539,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2201549962195735e-06,
+      "loss": 0.8064,
+      "step": 25680
+    },
+    {
+      "epoch": 0.7053088352420972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.21942913155304e-06,
+      "loss": 0.8788,
+      "step": 25681
+    },
+    {
+      "epoch": 0.7053362994699404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2187033126246225e-06,
+      "loss": 0.8469,
+      "step": 25682
+    },
+    {
+      "epoch": 0.7053637636977836,
+      "grad_norm": 0.0,
+      "learning_rate": 4.217977539440067e-06,
+      "loss": 0.8385,
+      "step": 25683
+    },
+    {
+      "epoch": 0.7053912279256269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.21725181200511e-06,
+      "loss": 0.7958,
+      "step": 25684
+    },
+    {
+      "epoch": 0.7054186921534701,
+      "grad_norm": 0.0,
+      "learning_rate": 4.216526130325499e-06,
+      "loss": 0.9257,
+      "step": 25685
+    },
+    {
+      "epoch": 0.7054461563813134,
+      "grad_norm": 0.0,
+      "learning_rate": 4.215800494406971e-06,
+      "loss": 0.9238,
+      "step": 25686
+    },
+    {
+      "epoch": 0.7054736206091565,
+      "grad_norm": 0.0,
+      "learning_rate": 4.21507490425527e-06,
+      "loss": 0.8146,
+      "step": 25687
+    },
+    {
+      "epoch": 0.7055010848369998,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2143493598761375e-06,
+      "loss": 0.7544,
+      "step": 25688
+    },
+    {
+      "epoch": 0.7055285490648431,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2136238612753135e-06,
+      "loss": 0.8089,
+      "step": 25689
+    },
+    {
+      "epoch": 0.7055560132926862,
+      "grad_norm": 0.0,
+      "learning_rate": 4.212898408458542e-06,
+      "loss": 0.816,
+      "step": 25690
+    },
+    {
+      "epoch": 0.7055834775205295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.212173001431557e-06,
+      "loss": 0.704,
+      "step": 25691
+    },
+    {
+      "epoch": 0.7056109417483728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.211447640200106e-06,
+      "loss": 0.779,
+      "step": 25692
+    },
+    {
+      "epoch": 0.705638405976216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.210722324769918e-06,
+      "loss": 0.8346,
+      "step": 25693
+    },
+    {
+      "epoch": 0.7056658702040592,
+      "grad_norm": 0.0,
+      "learning_rate": 4.209997055146739e-06,
+      "loss": 0.8014,
+      "step": 25694
+    },
+    {
+      "epoch": 0.7056933344319024,
+      "grad_norm": 0.0,
+      "learning_rate": 4.209271831336306e-06,
+      "loss": 0.7645,
+      "step": 25695
+    },
+    {
+      "epoch": 0.7057207986597457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.20854665334436e-06,
+      "loss": 0.8165,
+      "step": 25696
+    },
+    {
+      "epoch": 0.705748262887589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.207821521176633e-06,
+      "loss": 0.8683,
+      "step": 25697
+    },
+    {
+      "epoch": 0.7057757271154321,
+      "grad_norm": 0.0,
+      "learning_rate": 4.207096434838865e-06,
+      "loss": 0.9014,
+      "step": 25698
+    },
+    {
+      "epoch": 0.7058031913432754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.206371394336796e-06,
+      "loss": 0.8386,
+      "step": 25699
+    },
+    {
+      "epoch": 0.7058306555711186,
+      "grad_norm": 0.0,
+      "learning_rate": 4.205646399676163e-06,
+      "loss": 0.8477,
+      "step": 25700
+    },
+    {
+      "epoch": 0.7058581197989618,
+      "grad_norm": 0.0,
+      "learning_rate": 4.204921450862697e-06,
+      "loss": 0.8706,
+      "step": 25701
+    },
+    {
+      "epoch": 0.7058855840268051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.204196547902141e-06,
+      "loss": 0.8334,
+      "step": 25702
+    },
+    {
+      "epoch": 0.7059130482546483,
+      "grad_norm": 0.0,
+      "learning_rate": 4.203471690800223e-06,
+      "loss": 0.7406,
+      "step": 25703
+    },
+    {
+      "epoch": 0.7059405124824916,
+      "grad_norm": 0.0,
+      "learning_rate": 4.202746879562682e-06,
+      "loss": 0.7924,
+      "step": 25704
+    },
+    {
+      "epoch": 0.7059679767103348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.202022114195255e-06,
+      "loss": 0.8368,
+      "step": 25705
+    },
+    {
+      "epoch": 0.705995440938178,
+      "grad_norm": 0.0,
+      "learning_rate": 4.201297394703673e-06,
+      "loss": 0.8672,
+      "step": 25706
+    },
+    {
+      "epoch": 0.7060229051660213,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2005727210936764e-06,
+      "loss": 0.8275,
+      "step": 25707
+    },
+    {
+      "epoch": 0.7060503693938645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.199848093370992e-06,
+      "loss": 0.9337,
+      "step": 25708
+    },
+    {
+      "epoch": 0.7060778336217077,
+      "grad_norm": 0.0,
+      "learning_rate": 4.199123511541359e-06,
+      "loss": 0.8511,
+      "step": 25709
+    },
+    {
+      "epoch": 0.706105297849551,
+      "grad_norm": 0.0,
+      "learning_rate": 4.198398975610505e-06,
+      "loss": 0.716,
+      "step": 25710
+    },
+    {
+      "epoch": 0.7061327620773942,
+      "grad_norm": 0.0,
+      "learning_rate": 4.197674485584164e-06,
+      "loss": 0.8862,
+      "step": 25711
+    },
+    {
+      "epoch": 0.7061602263052374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.196950041468072e-06,
+      "loss": 0.8354,
+      "step": 25712
+    },
+    {
+      "epoch": 0.7061876905330806,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1962256432679586e-06,
+      "loss": 0.7933,
+      "step": 25713
+    },
+    {
+      "epoch": 0.7062151547609239,
+      "grad_norm": 0.0,
+      "learning_rate": 4.195501290989559e-06,
+      "loss": 0.8385,
+      "step": 25714
+    },
+    {
+      "epoch": 0.7062426189887672,
+      "grad_norm": 0.0,
+      "learning_rate": 4.194776984638599e-06,
+      "loss": 0.7342,
+      "step": 25715
+    },
+    {
+      "epoch": 0.7062700832166103,
+      "grad_norm": 0.0,
+      "learning_rate": 4.194052724220815e-06,
+      "loss": 0.7423,
+      "step": 25716
+    },
+    {
+      "epoch": 0.7062975474444536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1933285097419315e-06,
+      "loss": 0.9562,
+      "step": 25717
+    },
+    {
+      "epoch": 0.7063250116722969,
+      "grad_norm": 0.0,
+      "learning_rate": 4.192604341207682e-06,
+      "loss": 0.8612,
+      "step": 25718
+    },
+    {
+      "epoch": 0.70635247590014,
+      "grad_norm": 0.0,
+      "learning_rate": 4.191880218623797e-06,
+      "loss": 0.8174,
+      "step": 25719
+    },
+    {
+      "epoch": 0.7063799401279833,
+      "grad_norm": 0.0,
+      "learning_rate": 4.191156141996009e-06,
+      "loss": 0.8128,
+      "step": 25720
+    },
+    {
+      "epoch": 0.7064074043558265,
+      "grad_norm": 0.0,
+      "learning_rate": 4.19043211133004e-06,
+      "loss": 0.7639,
+      "step": 25721
+    },
+    {
+      "epoch": 0.7064348685836698,
+      "grad_norm": 0.0,
+      "learning_rate": 4.189708126631623e-06,
+      "loss": 0.823,
+      "step": 25722
+    },
+    {
+      "epoch": 0.706462332811513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.188984187906485e-06,
+      "loss": 0.8428,
+      "step": 25723
+    },
+    {
+      "epoch": 0.7064897970393562,
+      "grad_norm": 0.0,
+      "learning_rate": 4.18826029516036e-06,
+      "loss": 0.823,
+      "step": 25724
+    },
+    {
+      "epoch": 0.7065172612671995,
+      "grad_norm": 0.0,
+      "learning_rate": 4.187536448398966e-06,
+      "loss": 0.8589,
+      "step": 25725
+    },
+    {
+      "epoch": 0.7065447254950427,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1868126476280395e-06,
+      "loss": 0.8234,
+      "step": 25726
+    },
+    {
+      "epoch": 0.7065721897228859,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1860888928533e-06,
+      "loss": 0.8184,
+      "step": 25727
+    },
+    {
+      "epoch": 0.7065996539507292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1853651840804754e-06,
+      "loss": 0.8552,
+      "step": 25728
+    },
+    {
+      "epoch": 0.7066271181785724,
+      "grad_norm": 0.0,
+      "learning_rate": 4.184641521315296e-06,
+      "loss": 0.8361,
+      "step": 25729
+    },
+    {
+      "epoch": 0.7066545824064157,
+      "grad_norm": 0.0,
+      "learning_rate": 4.183917904563484e-06,
+      "loss": 0.8418,
+      "step": 25730
+    },
+    {
+      "epoch": 0.7066820466342589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1831943338307705e-06,
+      "loss": 0.8107,
+      "step": 25731
+    },
+    {
+      "epoch": 0.7067095108621021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.182470809122873e-06,
+      "loss": 0.8072,
+      "step": 25732
+    },
+    {
+      "epoch": 0.7067369750899454,
+      "grad_norm": 0.0,
+      "learning_rate": 4.181747330445524e-06,
+      "loss": 0.8439,
+      "step": 25733
+    },
+    {
+      "epoch": 0.7067644393177885,
+      "grad_norm": 0.0,
+      "learning_rate": 4.181023897804439e-06,
+      "loss": 0.9329,
+      "step": 25734
+    },
+    {
+      "epoch": 0.7067919035456318,
+      "grad_norm": 0.0,
+      "learning_rate": 4.180300511205348e-06,
+      "loss": 0.8182,
+      "step": 25735
+    },
+    {
+      "epoch": 0.7068193677734751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.179577170653974e-06,
+      "loss": 0.8379,
+      "step": 25736
+    },
+    {
+      "epoch": 0.7068468320013183,
+      "grad_norm": 0.0,
+      "learning_rate": 4.17885387615604e-06,
+      "loss": 0.8297,
+      "step": 25737
+    },
+    {
+      "epoch": 0.7068742962291615,
+      "grad_norm": 0.0,
+      "learning_rate": 4.178130627717273e-06,
+      "loss": 0.7357,
+      "step": 25738
+    },
+    {
+      "epoch": 0.7069017604570047,
+      "grad_norm": 0.0,
+      "learning_rate": 4.177407425343387e-06,
+      "loss": 0.8086,
+      "step": 25739
+    },
+    {
+      "epoch": 0.706929224684848,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1766842690401135e-06,
+      "loss": 0.8837,
+      "step": 25740
+    },
+    {
+      "epoch": 0.7069566889126913,
+      "grad_norm": 0.0,
+      "learning_rate": 4.175961158813165e-06,
+      "loss": 0.8643,
+      "step": 25741
+    },
+    {
+      "epoch": 0.7069841531405344,
+      "grad_norm": 0.0,
+      "learning_rate": 4.175238094668268e-06,
+      "loss": 0.8372,
+      "step": 25742
+    },
+    {
+      "epoch": 0.7070116173683777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.174515076611144e-06,
+      "loss": 0.8463,
+      "step": 25743
+    },
+    {
+      "epoch": 0.707039081596221,
+      "grad_norm": 0.0,
+      "learning_rate": 4.173792104647517e-06,
+      "loss": 0.8118,
+      "step": 25744
+    },
+    {
+      "epoch": 0.7070665458240641,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1730691787831e-06,
+      "loss": 0.8715,
+      "step": 25745
+    },
+    {
+      "epoch": 0.7070940100519074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.172346299023616e-06,
+      "loss": 0.8945,
+      "step": 25746
+    },
+    {
+      "epoch": 0.7071214742797506,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1716234653747866e-06,
+      "loss": 0.8207,
+      "step": 25747
+    },
+    {
+      "epoch": 0.7071489385075939,
+      "grad_norm": 0.0,
+      "learning_rate": 4.170900677842334e-06,
+      "loss": 0.8723,
+      "step": 25748
+    },
+    {
+      "epoch": 0.7071764027354371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1701779364319715e-06,
+      "loss": 0.8778,
+      "step": 25749
+    },
+    {
+      "epoch": 0.7072038669632803,
+      "grad_norm": 0.0,
+      "learning_rate": 4.169455241149417e-06,
+      "loss": 0.809,
+      "step": 25750
+    },
+    {
+      "epoch": 0.7072313311911236,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1687325920003905e-06,
+      "loss": 0.8378,
+      "step": 25751
+    },
+    {
+      "epoch": 0.7072587954189667,
+      "grad_norm": 0.0,
+      "learning_rate": 4.168009988990613e-06,
+      "loss": 0.8971,
+      "step": 25752
+    },
+    {
+      "epoch": 0.70728625964681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.167287432125796e-06,
+      "loss": 0.755,
+      "step": 25753
+    },
+    {
+      "epoch": 0.7073137238746533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1665649214116624e-06,
+      "loss": 0.8928,
+      "step": 25754
+    },
+    {
+      "epoch": 0.7073411881024965,
+      "grad_norm": 0.0,
+      "learning_rate": 4.165842456853929e-06,
+      "loss": 0.836,
+      "step": 25755
+    },
+    {
+      "epoch": 0.7073686523303397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.165120038458308e-06,
+      "loss": 0.7994,
+      "step": 25756
+    },
+    {
+      "epoch": 0.707396116558183,
+      "grad_norm": 0.0,
+      "learning_rate": 4.164397666230519e-06,
+      "loss": 0.8774,
+      "step": 25757
+    },
+    {
+      "epoch": 0.7074235807860262,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1636753401762756e-06,
+      "loss": 0.7609,
+      "step": 25758
+    },
+    {
+      "epoch": 0.7074510450138695,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1629530603012915e-06,
+      "loss": 0.7364,
+      "step": 25759
+    },
+    {
+      "epoch": 0.7074785092417126,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1622308266112846e-06,
+      "loss": 0.8176,
+      "step": 25760
+    },
+    {
+      "epoch": 0.7075059734695559,
+      "grad_norm": 0.0,
+      "learning_rate": 4.161508639111972e-06,
+      "loss": 0.8437,
+      "step": 25761
+    },
+    {
+      "epoch": 0.7075334376973992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.160786497809061e-06,
+      "loss": 0.9157,
+      "step": 25762
+    },
+    {
+      "epoch": 0.7075609019252423,
+      "grad_norm": 0.0,
+      "learning_rate": 4.160064402708268e-06,
+      "loss": 0.8585,
+      "step": 25763
+    },
+    {
+      "epoch": 0.7075883661530856,
+      "grad_norm": 0.0,
+      "learning_rate": 4.159342353815313e-06,
+      "loss": 0.9004,
+      "step": 25764
+    },
+    {
+      "epoch": 0.7076158303809288,
+      "grad_norm": 0.0,
+      "learning_rate": 4.158620351135899e-06,
+      "loss": 0.8811,
+      "step": 25765
+    },
+    {
+      "epoch": 0.7076432946087721,
+      "grad_norm": 0.0,
+      "learning_rate": 4.157898394675745e-06,
+      "loss": 0.8167,
+      "step": 25766
+    },
+    {
+      "epoch": 0.7076707588366153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1571764844405635e-06,
+      "loss": 0.9408,
+      "step": 25767
+    },
+    {
+      "epoch": 0.7076982230644585,
+      "grad_norm": 0.0,
+      "learning_rate": 4.156454620436064e-06,
+      "loss": 0.7536,
+      "step": 25768
+    },
+    {
+      "epoch": 0.7077256872923018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.155732802667957e-06,
+      "loss": 0.7534,
+      "step": 25769
+    },
+    {
+      "epoch": 0.7077531515201451,
+      "grad_norm": 0.0,
+      "learning_rate": 4.155011031141957e-06,
+      "loss": 0.8885,
+      "step": 25770
+    },
+    {
+      "epoch": 0.7077806157479882,
+      "grad_norm": 0.0,
+      "learning_rate": 4.154289305863773e-06,
+      "loss": 0.8676,
+      "step": 25771
+    },
+    {
+      "epoch": 0.7078080799758315,
+      "grad_norm": 0.0,
+      "learning_rate": 4.15356762683912e-06,
+      "loss": 0.8966,
+      "step": 25772
+    },
+    {
+      "epoch": 0.7078355442036747,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1528459940737055e-06,
+      "loss": 0.8314,
+      "step": 25773
+    },
+    {
+      "epoch": 0.707863008431518,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1521244075732335e-06,
+      "loss": 0.8426,
+      "step": 25774
+    },
+    {
+      "epoch": 0.7078904726593612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1514028673434194e-06,
+      "loss": 0.7512,
+      "step": 25775
+    },
+    {
+      "epoch": 0.7079179368872044,
+      "grad_norm": 0.0,
+      "learning_rate": 4.150681373389971e-06,
+      "loss": 0.8239,
+      "step": 25776
+    },
+    {
+      "epoch": 0.7079454011150477,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1499599257185985e-06,
+      "loss": 0.7784,
+      "step": 25777
+    },
+    {
+      "epoch": 0.7079728653428908,
+      "grad_norm": 0.0,
+      "learning_rate": 4.149238524335008e-06,
+      "loss": 0.823,
+      "step": 25778
+    },
+    {
+      "epoch": 0.7080003295707341,
+      "grad_norm": 0.0,
+      "learning_rate": 4.148517169244912e-06,
+      "loss": 0.8383,
+      "step": 25779
+    },
+    {
+      "epoch": 0.7080277937985774,
+      "grad_norm": 0.0,
+      "learning_rate": 4.147795860454016e-06,
+      "loss": 0.8201,
+      "step": 25780
+    },
+    {
+      "epoch": 0.7080552580264206,
+      "grad_norm": 0.0,
+      "learning_rate": 4.147074597968022e-06,
+      "loss": 0.9486,
+      "step": 25781
+    },
+    {
+      "epoch": 0.7080827222542638,
+      "grad_norm": 0.0,
+      "learning_rate": 4.146353381792641e-06,
+      "loss": 0.8433,
+      "step": 25782
+    },
+    {
+      "epoch": 0.7081101864821071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.145632211933579e-06,
+      "loss": 0.8668,
+      "step": 25783
+    },
+    {
+      "epoch": 0.7081376507099503,
+      "grad_norm": 0.0,
+      "learning_rate": 4.144911088396543e-06,
+      "loss": 0.8655,
+      "step": 25784
+    },
+    {
+      "epoch": 0.7081651149377935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.144190011187241e-06,
+      "loss": 0.8046,
+      "step": 25785
+    },
+    {
+      "epoch": 0.7081925791656367,
+      "grad_norm": 0.0,
+      "learning_rate": 4.143468980311375e-06,
+      "loss": 0.8722,
+      "step": 25786
+    },
+    {
+      "epoch": 0.70822004339348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.142747995774649e-06,
+      "loss": 0.8583,
+      "step": 25787
+    },
+    {
+      "epoch": 0.7082475076213233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.142027057582774e-06,
+      "loss": 0.791,
+      "step": 25788
+    },
+    {
+      "epoch": 0.7082749718491664,
+      "grad_norm": 0.0,
+      "learning_rate": 4.141306165741447e-06,
+      "loss": 0.8885,
+      "step": 25789
+    },
+    {
+      "epoch": 0.7083024360770097,
+      "grad_norm": 0.0,
+      "learning_rate": 4.140585320256374e-06,
+      "loss": 0.7941,
+      "step": 25790
+    },
+    {
+      "epoch": 0.7083299003048529,
+      "grad_norm": 0.0,
+      "learning_rate": 4.139864521133263e-06,
+      "loss": 0.8008,
+      "step": 25791
+    },
+    {
+      "epoch": 0.7083573645326962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.139143768377812e-06,
+      "loss": 0.8571,
+      "step": 25792
+    },
+    {
+      "epoch": 0.7083848287605394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.138423061995724e-06,
+      "loss": 0.8507,
+      "step": 25793
+    },
+    {
+      "epoch": 0.7084122929883826,
+      "grad_norm": 0.0,
+      "learning_rate": 4.137702401992704e-06,
+      "loss": 0.7683,
+      "step": 25794
+    },
+    {
+      "epoch": 0.7084397572162259,
+      "grad_norm": 0.0,
+      "learning_rate": 4.136981788374454e-06,
+      "loss": 0.7976,
+      "step": 25795
+    },
+    {
+      "epoch": 0.7084672214440692,
+      "grad_norm": 0.0,
+      "learning_rate": 4.136261221146678e-06,
+      "loss": 0.8675,
+      "step": 25796
+    },
+    {
+      "epoch": 0.7084946856719123,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1355407003150745e-06,
+      "loss": 0.9057,
+      "step": 25797
+    },
+    {
+      "epoch": 0.7085221498997556,
+      "grad_norm": 0.0,
+      "learning_rate": 4.134820225885342e-06,
+      "loss": 0.787,
+      "step": 25798
+    },
+    {
+      "epoch": 0.7085496141275988,
+      "grad_norm": 0.0,
+      "learning_rate": 4.134099797863182e-06,
+      "loss": 0.8457,
+      "step": 25799
+    },
+    {
+      "epoch": 0.708577078355442,
+      "grad_norm": 0.0,
+      "learning_rate": 4.133379416254299e-06,
+      "loss": 0.889,
+      "step": 25800
+    },
+    {
+      "epoch": 0.7086045425832853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.132659081064389e-06,
+      "loss": 0.877,
+      "step": 25801
+    },
+    {
+      "epoch": 0.7086320068111285,
+      "grad_norm": 0.0,
+      "learning_rate": 4.131938792299154e-06,
+      "loss": 0.7743,
+      "step": 25802
+    },
+    {
+      "epoch": 0.7086594710389718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.131218549964295e-06,
+      "loss": 0.7932,
+      "step": 25803
+    },
+    {
+      "epoch": 0.7086869352668149,
+      "grad_norm": 0.0,
+      "learning_rate": 4.13049835406551e-06,
+      "loss": 0.9387,
+      "step": 25804
+    },
+    {
+      "epoch": 0.7087143994946582,
+      "grad_norm": 0.0,
+      "learning_rate": 4.12977820460849e-06,
+      "loss": 0.7691,
+      "step": 25805
+    },
+    {
+      "epoch": 0.7087418637225015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.12905810159894e-06,
+      "loss": 0.8437,
+      "step": 25806
+    },
+    {
+      "epoch": 0.7087693279503446,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1283380450425555e-06,
+      "loss": 0.8169,
+      "step": 25807
+    },
+    {
+      "epoch": 0.7087967921781879,
+      "grad_norm": 0.0,
+      "learning_rate": 4.127618034945039e-06,
+      "loss": 0.7019,
+      "step": 25808
+    },
+    {
+      "epoch": 0.7088242564060312,
+      "grad_norm": 0.0,
+      "learning_rate": 4.12689807131208e-06,
+      "loss": 0.8431,
+      "step": 25809
+    },
+    {
+      "epoch": 0.7088517206338744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.126178154149378e-06,
+      "loss": 0.8019,
+      "step": 25810
+    },
+    {
+      "epoch": 0.7088791848617176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.12545828346263e-06,
+      "loss": 0.8142,
+      "step": 25811
+    },
+    {
+      "epoch": 0.7089066490895608,
+      "grad_norm": 0.0,
+      "learning_rate": 4.124738459257536e-06,
+      "loss": 0.8336,
+      "step": 25812
+    },
+    {
+      "epoch": 0.7089341133174041,
+      "grad_norm": 0.0,
+      "learning_rate": 4.124018681539782e-06,
+      "loss": 0.75,
+      "step": 25813
+    },
+    {
+      "epoch": 0.7089615775452474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.123298950315073e-06,
+      "loss": 0.8773,
+      "step": 25814
+    },
+    {
+      "epoch": 0.7089890417730905,
+      "grad_norm": 0.0,
+      "learning_rate": 4.122579265589095e-06,
+      "loss": 0.942,
+      "step": 25815
+    },
+    {
+      "epoch": 0.7090165060009338,
+      "grad_norm": 0.0,
+      "learning_rate": 4.121859627367547e-06,
+      "loss": 0.7299,
+      "step": 25816
+    },
+    {
+      "epoch": 0.709043970228777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.121140035656123e-06,
+      "loss": 0.8224,
+      "step": 25817
+    },
+    {
+      "epoch": 0.7090714344566202,
+      "grad_norm": 0.0,
+      "learning_rate": 4.120420490460515e-06,
+      "loss": 0.8664,
+      "step": 25818
+    },
+    {
+      "epoch": 0.7090988986844635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.119700991786421e-06,
+      "loss": 0.798,
+      "step": 25819
+    },
+    {
+      "epoch": 0.7091263629123067,
+      "grad_norm": 0.0,
+      "learning_rate": 4.118981539639529e-06,
+      "loss": 0.8509,
+      "step": 25820
+    },
+    {
+      "epoch": 0.70915382714015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.118262134025536e-06,
+      "loss": 0.8042,
+      "step": 25821
+    },
+    {
+      "epoch": 0.7091812913679932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.117542774950127e-06,
+      "loss": 0.7747,
+      "step": 25822
+    },
+    {
+      "epoch": 0.7092087555958364,
+      "grad_norm": 0.0,
+      "learning_rate": 4.116823462418998e-06,
+      "loss": 0.768,
+      "step": 25823
+    },
+    {
+      "epoch": 0.7092362198236797,
+      "grad_norm": 0.0,
+      "learning_rate": 4.116104196437842e-06,
+      "loss": 0.6877,
+      "step": 25824
+    },
+    {
+      "epoch": 0.7092636840515228,
+      "grad_norm": 0.0,
+      "learning_rate": 4.115384977012348e-06,
+      "loss": 0.8669,
+      "step": 25825
+    },
+    {
+      "epoch": 0.7092911482793661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.114665804148211e-06,
+      "loss": 0.7872,
+      "step": 25826
+    },
+    {
+      "epoch": 0.7093186125072094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.113946677851115e-06,
+      "loss": 0.8009,
+      "step": 25827
+    },
+    {
+      "epoch": 0.7093460767350526,
+      "grad_norm": 0.0,
+      "learning_rate": 4.113227598126755e-06,
+      "loss": 0.8245,
+      "step": 25828
+    },
+    {
+      "epoch": 0.7093735409628958,
+      "grad_norm": 0.0,
+      "learning_rate": 4.112508564980817e-06,
+      "loss": 0.9012,
+      "step": 25829
+    },
+    {
+      "epoch": 0.709401005190739,
+      "grad_norm": 0.0,
+      "learning_rate": 4.111789578418991e-06,
+      "loss": 0.8184,
+      "step": 25830
+    },
+    {
+      "epoch": 0.7094284694185823,
+      "grad_norm": 0.0,
+      "learning_rate": 4.111070638446966e-06,
+      "loss": 0.7993,
+      "step": 25831
+    },
+    {
+      "epoch": 0.7094559336464256,
+      "grad_norm": 0.0,
+      "learning_rate": 4.110351745070435e-06,
+      "loss": 0.8029,
+      "step": 25832
+    },
+    {
+      "epoch": 0.7094833978742687,
+      "grad_norm": 0.0,
+      "learning_rate": 4.109632898295079e-06,
+      "loss": 0.8896,
+      "step": 25833
+    },
+    {
+      "epoch": 0.709510862102112,
+      "grad_norm": 0.0,
+      "learning_rate": 4.108914098126591e-06,
+      "loss": 0.7744,
+      "step": 25834
+    },
+    {
+      "epoch": 0.7095383263299553,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1081953445706545e-06,
+      "loss": 0.7948,
+      "step": 25835
+    },
+    {
+      "epoch": 0.7095657905577984,
+      "grad_norm": 0.0,
+      "learning_rate": 4.107476637632963e-06,
+      "loss": 0.7602,
+      "step": 25836
+    },
+    {
+      "epoch": 0.7095932547856417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.106757977319195e-06,
+      "loss": 0.8699,
+      "step": 25837
+    },
+    {
+      "epoch": 0.7096207190134849,
+      "grad_norm": 0.0,
+      "learning_rate": 4.106039363635045e-06,
+      "loss": 0.7752,
+      "step": 25838
+    },
+    {
+      "epoch": 0.7096481832413282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.10532079658619e-06,
+      "loss": 0.7736,
+      "step": 25839
+    },
+    {
+      "epoch": 0.7096756474691714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1046022761783215e-06,
+      "loss": 0.8347,
+      "step": 25840
+    },
+    {
+      "epoch": 0.7097031116970146,
+      "grad_norm": 0.0,
+      "learning_rate": 4.103883802417122e-06,
+      "loss": 0.8647,
+      "step": 25841
+    },
+    {
+      "epoch": 0.7097305759248579,
+      "grad_norm": 0.0,
+      "learning_rate": 4.103165375308279e-06,
+      "loss": 0.7641,
+      "step": 25842
+    },
+    {
+      "epoch": 0.709758040152701,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1024469948574785e-06,
+      "loss": 0.8554,
+      "step": 25843
+    },
+    {
+      "epoch": 0.7097855043805443,
+      "grad_norm": 0.0,
+      "learning_rate": 4.101728661070399e-06,
+      "loss": 0.8667,
+      "step": 25844
+    },
+    {
+      "epoch": 0.7098129686083876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.10101037395273e-06,
+      "loss": 0.7743,
+      "step": 25845
+    },
+    {
+      "epoch": 0.7098404328362308,
+      "grad_norm": 0.0,
+      "learning_rate": 4.100292133510149e-06,
+      "loss": 0.8488,
+      "step": 25846
+    },
+    {
+      "epoch": 0.709867897064074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.099573939748341e-06,
+      "loss": 0.8575,
+      "step": 25847
+    },
+    {
+      "epoch": 0.7098953612919173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.098855792672991e-06,
+      "loss": 0.9439,
+      "step": 25848
+    },
+    {
+      "epoch": 0.7099228255197605,
+      "grad_norm": 0.0,
+      "learning_rate": 4.09813769228978e-06,
+      "loss": 0.8816,
+      "step": 25849
+    },
+    {
+      "epoch": 0.7099502897476038,
+      "grad_norm": 0.0,
+      "learning_rate": 4.097419638604392e-06,
+      "loss": 0.851,
+      "step": 25850
+    },
+    {
+      "epoch": 0.7099777539754469,
+      "grad_norm": 0.0,
+      "learning_rate": 4.096701631622503e-06,
+      "loss": 0.82,
+      "step": 25851
+    },
+    {
+      "epoch": 0.7100052182032902,
+      "grad_norm": 0.0,
+      "learning_rate": 4.095983671349802e-06,
+      "loss": 0.7796,
+      "step": 25852
+    },
+    {
+      "epoch": 0.7100326824311335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0952657577919595e-06,
+      "loss": 0.8689,
+      "step": 25853
+    },
+    {
+      "epoch": 0.7100601466589767,
+      "grad_norm": 0.0,
+      "learning_rate": 4.094547890954663e-06,
+      "loss": 0.7303,
+      "step": 25854
+    },
+    {
+      "epoch": 0.7100876108868199,
+      "grad_norm": 0.0,
+      "learning_rate": 4.093830070843592e-06,
+      "loss": 0.7885,
+      "step": 25855
+    },
+    {
+      "epoch": 0.7101150751146631,
+      "grad_norm": 0.0,
+      "learning_rate": 4.093112297464428e-06,
+      "loss": 0.833,
+      "step": 25856
+    },
+    {
+      "epoch": 0.7101425393425064,
+      "grad_norm": 0.0,
+      "learning_rate": 4.092394570822843e-06,
+      "loss": 0.641,
+      "step": 25857
+    },
+    {
+      "epoch": 0.7101700035703497,
+      "grad_norm": 0.0,
+      "learning_rate": 4.091676890924522e-06,
+      "loss": 0.959,
+      "step": 25858
+    },
+    {
+      "epoch": 0.7101974677981928,
+      "grad_norm": 0.0,
+      "learning_rate": 4.090959257775141e-06,
+      "loss": 0.8456,
+      "step": 25859
+    },
+    {
+      "epoch": 0.7102249320260361,
+      "grad_norm": 0.0,
+      "learning_rate": 4.090241671380384e-06,
+      "loss": 0.883,
+      "step": 25860
+    },
+    {
+      "epoch": 0.7102523962538794,
+      "grad_norm": 0.0,
+      "learning_rate": 4.08952413174592e-06,
+      "loss": 0.892,
+      "step": 25861
+    },
+    {
+      "epoch": 0.7102798604817225,
+      "grad_norm": 0.0,
+      "learning_rate": 4.088806638877434e-06,
+      "loss": 0.8439,
+      "step": 25862
+    },
+    {
+      "epoch": 0.7103073247095658,
+      "grad_norm": 0.0,
+      "learning_rate": 4.088089192780596e-06,
+      "loss": 0.7938,
+      "step": 25863
+    },
+    {
+      "epoch": 0.710334788937409,
+      "grad_norm": 0.0,
+      "learning_rate": 4.087371793461087e-06,
+      "loss": 0.8607,
+      "step": 25864
+    },
+    {
+      "epoch": 0.7103622531652523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.086654440924582e-06,
+      "loss": 0.7854,
+      "step": 25865
+    },
+    {
+      "epoch": 0.7103897173930955,
+      "grad_norm": 0.0,
+      "learning_rate": 4.085937135176759e-06,
+      "loss": 0.831,
+      "step": 25866
+    },
+    {
+      "epoch": 0.7104171816209387,
+      "grad_norm": 0.0,
+      "learning_rate": 4.085219876223294e-06,
+      "loss": 0.8491,
+      "step": 25867
+    },
+    {
+      "epoch": 0.710444645848782,
+      "grad_norm": 0.0,
+      "learning_rate": 4.084502664069858e-06,
+      "loss": 0.7768,
+      "step": 25868
+    },
+    {
+      "epoch": 0.7104721100766251,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0837854987221315e-06,
+      "loss": 0.879,
+      "step": 25869
+    },
+    {
+      "epoch": 0.7104995743044684,
+      "grad_norm": 0.0,
+      "learning_rate": 4.083068380185783e-06,
+      "loss": 0.9391,
+      "step": 25870
+    },
+    {
+      "epoch": 0.7105270385323117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.082351308466489e-06,
+      "loss": 0.8435,
+      "step": 25871
+    },
+    {
+      "epoch": 0.7105545027601549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.081634283569923e-06,
+      "loss": 0.7789,
+      "step": 25872
+    },
+    {
+      "epoch": 0.7105819669879981,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0809173055017634e-06,
+      "loss": 0.81,
+      "step": 25873
+    },
+    {
+      "epoch": 0.7106094312158414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.080200374267675e-06,
+      "loss": 0.7812,
+      "step": 25874
+    },
+    {
+      "epoch": 0.7106368954436846,
+      "grad_norm": 0.0,
+      "learning_rate": 4.079483489873335e-06,
+      "loss": 0.9175,
+      "step": 25875
+    },
+    {
+      "epoch": 0.7106643596715279,
+      "grad_norm": 0.0,
+      "learning_rate": 4.078766652324418e-06,
+      "loss": 0.8761,
+      "step": 25876
+    },
+    {
+      "epoch": 0.710691823899371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.078049861626589e-06,
+      "loss": 0.8281,
+      "step": 25877
+    },
+    {
+      "epoch": 0.7107192881272143,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0773331177855245e-06,
+      "loss": 0.8196,
+      "step": 25878
+    },
+    {
+      "epoch": 0.7107467523550576,
+      "grad_norm": 0.0,
+      "learning_rate": 4.076616420806898e-06,
+      "loss": 0.7802,
+      "step": 25879
+    },
+    {
+      "epoch": 0.7107742165829007,
+      "grad_norm": 0.0,
+      "learning_rate": 4.075899770696373e-06,
+      "loss": 0.776,
+      "step": 25880
+    },
+    {
+      "epoch": 0.710801680810744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.075183167459625e-06,
+      "loss": 0.8342,
+      "step": 25881
+    },
+    {
+      "epoch": 0.7108291450385873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.07446661110232e-06,
+      "loss": 0.8387,
+      "step": 25882
+    },
+    {
+      "epoch": 0.7108566092664305,
+      "grad_norm": 0.0,
+      "learning_rate": 4.073750101630133e-06,
+      "loss": 0.8131,
+      "step": 25883
+    },
+    {
+      "epoch": 0.7108840734942737,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0730336390487336e-06,
+      "loss": 0.7821,
+      "step": 25884
+    },
+    {
+      "epoch": 0.7109115377221169,
+      "grad_norm": 0.0,
+      "learning_rate": 4.072317223363787e-06,
+      "loss": 0.7892,
+      "step": 25885
+    },
+    {
+      "epoch": 0.7109390019499602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0716008545809605e-06,
+      "loss": 0.935,
+      "step": 25886
+    },
+    {
+      "epoch": 0.7109664661778035,
+      "grad_norm": 0.0,
+      "learning_rate": 4.070884532705923e-06,
+      "loss": 0.8708,
+      "step": 25887
+    },
+    {
+      "epoch": 0.7109939304056466,
+      "grad_norm": 0.0,
+      "learning_rate": 4.070168257744345e-06,
+      "loss": 0.8668,
+      "step": 25888
+    },
+    {
+      "epoch": 0.7110213946334899,
+      "grad_norm": 0.0,
+      "learning_rate": 4.069452029701892e-06,
+      "loss": 0.8629,
+      "step": 25889
+    },
+    {
+      "epoch": 0.7110488588613331,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0687358485842335e-06,
+      "loss": 0.7744,
+      "step": 25890
+    },
+    {
+      "epoch": 0.7110763230891763,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0680197143970365e-06,
+      "loss": 0.8986,
+      "step": 25891
+    },
+    {
+      "epoch": 0.7111037873170196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.067303627145962e-06,
+      "loss": 0.8722,
+      "step": 25892
+    },
+    {
+      "epoch": 0.7111312515448628,
+      "grad_norm": 0.0,
+      "learning_rate": 4.066587586836684e-06,
+      "loss": 0.8192,
+      "step": 25893
+    },
+    {
+      "epoch": 0.7111587157727061,
+      "grad_norm": 0.0,
+      "learning_rate": 4.065871593474859e-06,
+      "loss": 0.8393,
+      "step": 25894
+    },
+    {
+      "epoch": 0.7111861800005493,
+      "grad_norm": 0.0,
+      "learning_rate": 4.065155647066156e-06,
+      "loss": 0.8994,
+      "step": 25895
+    },
+    {
+      "epoch": 0.7112136442283925,
+      "grad_norm": 0.0,
+      "learning_rate": 4.064439747616241e-06,
+      "loss": 0.8837,
+      "step": 25896
+    },
+    {
+      "epoch": 0.7112411084562358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.063723895130781e-06,
+      "loss": 0.828,
+      "step": 25897
+    },
+    {
+      "epoch": 0.711268572684079,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0630080896154346e-06,
+      "loss": 0.8693,
+      "step": 25898
+    },
+    {
+      "epoch": 0.7112960369119222,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0622923310758665e-06,
+      "loss": 0.8738,
+      "step": 25899
+    },
+    {
+      "epoch": 0.7113235011397655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0615766195177466e-06,
+      "loss": 0.8558,
+      "step": 25900
+    },
+    {
+      "epoch": 0.7113509653676087,
+      "grad_norm": 0.0,
+      "learning_rate": 4.060860954946728e-06,
+      "loss": 0.8714,
+      "step": 25901
+    },
+    {
+      "epoch": 0.7113784295954519,
+      "grad_norm": 0.0,
+      "learning_rate": 4.060145337368479e-06,
+      "loss": 0.808,
+      "step": 25902
+    },
+    {
+      "epoch": 0.7114058938232951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.059429766788664e-06,
+      "loss": 0.7773,
+      "step": 25903
+    },
+    {
+      "epoch": 0.7114333580511384,
+      "grad_norm": 0.0,
+      "learning_rate": 4.058714243212938e-06,
+      "loss": 0.7386,
+      "step": 25904
+    },
+    {
+      "epoch": 0.7114608222789817,
+      "grad_norm": 0.0,
+      "learning_rate": 4.057998766646967e-06,
+      "loss": 0.9011,
+      "step": 25905
+    },
+    {
+      "epoch": 0.7114882865068248,
+      "grad_norm": 0.0,
+      "learning_rate": 4.057283337096412e-06,
+      "loss": 0.7484,
+      "step": 25906
+    },
+    {
+      "epoch": 0.7115157507346681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.056567954566931e-06,
+      "loss": 0.808,
+      "step": 25907
+    },
+    {
+      "epoch": 0.7115432149625114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.055852619064192e-06,
+      "loss": 0.6907,
+      "step": 25908
+    },
+    {
+      "epoch": 0.7115706791903545,
+      "grad_norm": 0.0,
+      "learning_rate": 4.05513733059385e-06,
+      "loss": 0.9084,
+      "step": 25909
+    },
+    {
+      "epoch": 0.7115981434181978,
+      "grad_norm": 0.0,
+      "learning_rate": 4.054422089161559e-06,
+      "loss": 0.8191,
+      "step": 25910
+    },
+    {
+      "epoch": 0.711625607646041,
+      "grad_norm": 0.0,
+      "learning_rate": 4.053706894772985e-06,
+      "loss": 0.8577,
+      "step": 25911
+    },
+    {
+      "epoch": 0.7116530718738843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.052991747433786e-06,
+      "loss": 0.9981,
+      "step": 25912
+    },
+    {
+      "epoch": 0.7116805361017275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.052276647149618e-06,
+      "loss": 0.8649,
+      "step": 25913
+    },
+    {
+      "epoch": 0.7117080003295707,
+      "grad_norm": 0.0,
+      "learning_rate": 4.051561593926142e-06,
+      "loss": 0.9175,
+      "step": 25914
+    },
+    {
+      "epoch": 0.711735464557414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.050846587769019e-06,
+      "loss": 0.7717,
+      "step": 25915
+    },
+    {
+      "epoch": 0.7117629287852572,
+      "grad_norm": 0.0,
+      "learning_rate": 4.050131628683897e-06,
+      "loss": 0.7907,
+      "step": 25916
+    },
+    {
+      "epoch": 0.7117903930131004,
+      "grad_norm": 0.0,
+      "learning_rate": 4.049416716676443e-06,
+      "loss": 0.7825,
+      "step": 25917
+    },
+    {
+      "epoch": 0.7118178572409437,
+      "grad_norm": 0.0,
+      "learning_rate": 4.048701851752306e-06,
+      "loss": 0.8483,
+      "step": 25918
+    },
+    {
+      "epoch": 0.7118453214687869,
+      "grad_norm": 0.0,
+      "learning_rate": 4.047987033917144e-06,
+      "loss": 0.8408,
+      "step": 25919
+    },
+    {
+      "epoch": 0.7118727856966302,
+      "grad_norm": 0.0,
+      "learning_rate": 4.047272263176615e-06,
+      "loss": 0.8658,
+      "step": 25920
+    },
+    {
+      "epoch": 0.7119002499244734,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0465575395363774e-06,
+      "loss": 0.8874,
+      "step": 25921
+    },
+    {
+      "epoch": 0.7119277141523166,
+      "grad_norm": 0.0,
+      "learning_rate": 4.045842863002078e-06,
+      "loss": 0.764,
+      "step": 25922
+    },
+    {
+      "epoch": 0.7119551783801599,
+      "grad_norm": 0.0,
+      "learning_rate": 4.045128233579379e-06,
+      "loss": 0.8527,
+      "step": 25923
+    },
+    {
+      "epoch": 0.711982642608003,
+      "grad_norm": 0.0,
+      "learning_rate": 4.044413651273933e-06,
+      "loss": 0.9039,
+      "step": 25924
+    },
+    {
+      "epoch": 0.7120101068358463,
+      "grad_norm": 0.0,
+      "learning_rate": 4.043699116091391e-06,
+      "loss": 0.883,
+      "step": 25925
+    },
+    {
+      "epoch": 0.7120375710636896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.042984628037411e-06,
+      "loss": 0.8982,
+      "step": 25926
+    },
+    {
+      "epoch": 0.7120650352915328,
+      "grad_norm": 0.0,
+      "learning_rate": 4.042270187117642e-06,
+      "loss": 0.8571,
+      "step": 25927
+    },
+    {
+      "epoch": 0.712092499519376,
+      "grad_norm": 0.0,
+      "learning_rate": 4.041555793337738e-06,
+      "loss": 0.8308,
+      "step": 25928
+    },
+    {
+      "epoch": 0.7121199637472192,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0408414467033526e-06,
+      "loss": 0.7987,
+      "step": 25929
+    },
+    {
+      "epoch": 0.7121474279750625,
+      "grad_norm": 0.0,
+      "learning_rate": 4.040127147220139e-06,
+      "loss": 0.8067,
+      "step": 25930
+    },
+    {
+      "epoch": 0.7121748922029058,
+      "grad_norm": 0.0,
+      "learning_rate": 4.039412894893747e-06,
+      "loss": 0.7605,
+      "step": 25931
+    },
+    {
+      "epoch": 0.7122023564307489,
+      "grad_norm": 0.0,
+      "learning_rate": 4.038698689729834e-06,
+      "loss": 0.7585,
+      "step": 25932
+    },
+    {
+      "epoch": 0.7122298206585922,
+      "grad_norm": 0.0,
+      "learning_rate": 4.037984531734045e-06,
+      "loss": 0.7755,
+      "step": 25933
+    },
+    {
+      "epoch": 0.7122572848864355,
+      "grad_norm": 0.0,
+      "learning_rate": 4.037270420912029e-06,
+      "loss": 0.8762,
+      "step": 25934
+    },
+    {
+      "epoch": 0.7122847491142786,
+      "grad_norm": 0.0,
+      "learning_rate": 4.036556357269439e-06,
+      "loss": 0.8742,
+      "step": 25935
+    },
+    {
+      "epoch": 0.7123122133421219,
+      "grad_norm": 0.0,
+      "learning_rate": 4.035842340811925e-06,
+      "loss": 0.9073,
+      "step": 25936
+    },
+    {
+      "epoch": 0.7123396775699651,
+      "grad_norm": 0.0,
+      "learning_rate": 4.035128371545136e-06,
+      "loss": 0.8119,
+      "step": 25937
+    },
+    {
+      "epoch": 0.7123671417978084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.034414449474727e-06,
+      "loss": 0.8513,
+      "step": 25938
+    },
+    {
+      "epoch": 0.7123946060256516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.033700574606337e-06,
+      "loss": 0.7846,
+      "step": 25939
+    },
+    {
+      "epoch": 0.7124220702534948,
+      "grad_norm": 0.0,
+      "learning_rate": 4.032986746945623e-06,
+      "loss": 0.7394,
+      "step": 25940
+    },
+    {
+      "epoch": 0.7124495344813381,
+      "grad_norm": 0.0,
+      "learning_rate": 4.032272966498226e-06,
+      "loss": 0.7371,
+      "step": 25941
+    },
+    {
+      "epoch": 0.7124769987091812,
+      "grad_norm": 0.0,
+      "learning_rate": 4.031559233269796e-06,
+      "loss": 0.7555,
+      "step": 25942
+    },
+    {
+      "epoch": 0.7125044629370245,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0308455472659825e-06,
+      "loss": 0.8461,
+      "step": 25943
+    },
+    {
+      "epoch": 0.7125319271648678,
+      "grad_norm": 0.0,
+      "learning_rate": 4.030131908492434e-06,
+      "loss": 0.8743,
+      "step": 25944
+    },
+    {
+      "epoch": 0.712559391392711,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0294183169547915e-06,
+      "loss": 0.7957,
+      "step": 25945
+    },
+    {
+      "epoch": 0.7125868556205542,
+      "grad_norm": 0.0,
+      "learning_rate": 4.028704772658704e-06,
+      "loss": 0.7866,
+      "step": 25946
+    },
+    {
+      "epoch": 0.7126143198483975,
+      "grad_norm": 0.0,
+      "learning_rate": 4.027991275609817e-06,
+      "loss": 0.818,
+      "step": 25947
+    },
+    {
+      "epoch": 0.7126417840762407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.02727782581378e-06,
+      "loss": 0.8394,
+      "step": 25948
+    },
+    {
+      "epoch": 0.712669248304084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.02656442327623e-06,
+      "loss": 0.8209,
+      "step": 25949
+    },
+    {
+      "epoch": 0.7126967125319271,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025851068002822e-06,
+      "loss": 0.7646,
+      "step": 25950
+    },
+    {
+      "epoch": 0.7127241767597704,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025137759999189e-06,
+      "loss": 0.9112,
+      "step": 25951
+    },
+    {
+      "epoch": 0.7127516409876137,
+      "grad_norm": 0.0,
+      "learning_rate": 4.024424499270981e-06,
+      "loss": 0.9179,
+      "step": 25952
+    },
+    {
+      "epoch": 0.7127791052154568,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0237112858238415e-06,
+      "loss": 0.8286,
+      "step": 25953
+    },
+    {
+      "epoch": 0.7128065694433001,
+      "grad_norm": 0.0,
+      "learning_rate": 4.022998119663413e-06,
+      "loss": 0.8164,
+      "step": 25954
+    },
+    {
+      "epoch": 0.7128340336711433,
+      "grad_norm": 0.0,
+      "learning_rate": 4.022285000795339e-06,
+      "loss": 0.8206,
+      "step": 25955
+    },
+    {
+      "epoch": 0.7128614978989866,
+      "grad_norm": 0.0,
+      "learning_rate": 4.021571929225264e-06,
+      "loss": 0.8007,
+      "step": 25956
+    },
+    {
+      "epoch": 0.7128889621268298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0208589049588285e-06,
+      "loss": 0.8188,
+      "step": 25957
+    },
+    {
+      "epoch": 0.712916426354673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.02014592800167e-06,
+      "loss": 0.8513,
+      "step": 25958
+    },
+    {
+      "epoch": 0.7129438905825163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.019432998359433e-06,
+      "loss": 0.8599,
+      "step": 25959
+    },
+    {
+      "epoch": 0.7129713548103596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.018720116037759e-06,
+      "loss": 0.9413,
+      "step": 25960
+    },
+    {
+      "epoch": 0.7129988190382027,
+      "grad_norm": 0.0,
+      "learning_rate": 4.018007281042288e-06,
+      "loss": 0.836,
+      "step": 25961
+    },
+    {
+      "epoch": 0.713026283266046,
+      "grad_norm": 0.0,
+      "learning_rate": 4.017294493378664e-06,
+      "loss": 0.7794,
+      "step": 25962
+    },
+    {
+      "epoch": 0.7130537474938892,
+      "grad_norm": 0.0,
+      "learning_rate": 4.016581753052521e-06,
+      "loss": 0.8077,
+      "step": 25963
+    },
+    {
+      "epoch": 0.7130812117217324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.015869060069504e-06,
+      "loss": 0.8247,
+      "step": 25964
+    },
+    {
+      "epoch": 0.7131086759495757,
+      "grad_norm": 0.0,
+      "learning_rate": 4.015156414435245e-06,
+      "loss": 0.8992,
+      "step": 25965
+    },
+    {
+      "epoch": 0.7131361401774189,
+      "grad_norm": 0.0,
+      "learning_rate": 4.014443816155387e-06,
+      "loss": 0.884,
+      "step": 25966
+    },
+    {
+      "epoch": 0.7131636044052622,
+      "grad_norm": 0.0,
+      "learning_rate": 4.013731265235568e-06,
+      "loss": 0.7575,
+      "step": 25967
+    },
+    {
+      "epoch": 0.7131910686331053,
+      "grad_norm": 0.0,
+      "learning_rate": 4.01301876168143e-06,
+      "loss": 0.8103,
+      "step": 25968
+    },
+    {
+      "epoch": 0.7132185328609486,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0123063054986026e-06,
+      "loss": 0.6815,
+      "step": 25969
+    },
+    {
+      "epoch": 0.7132459970887919,
+      "grad_norm": 0.0,
+      "learning_rate": 4.011593896692727e-06,
+      "loss": 0.9429,
+      "step": 25970
+    },
+    {
+      "epoch": 0.713273461316635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.010881535269441e-06,
+      "loss": 0.8461,
+      "step": 25971
+    },
+    {
+      "epoch": 0.7133009255444783,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0101692212343835e-06,
+      "loss": 0.8539,
+      "step": 25972
+    },
+    {
+      "epoch": 0.7133283897723216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.009456954593184e-06,
+      "loss": 0.8401,
+      "step": 25973
+    },
+    {
+      "epoch": 0.7133558540001648,
+      "grad_norm": 0.0,
+      "learning_rate": 4.008744735351484e-06,
+      "loss": 0.8575,
+      "step": 25974
+    },
+    {
+      "epoch": 0.713383318228008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.008032563514914e-06,
+      "loss": 0.7927,
+      "step": 25975
+    },
+    {
+      "epoch": 0.7134107824558512,
+      "grad_norm": 0.0,
+      "learning_rate": 4.007320439089112e-06,
+      "loss": 0.8031,
+      "step": 25976
+    },
+    {
+      "epoch": 0.7134382466836945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.00660836207971e-06,
+      "loss": 0.7661,
+      "step": 25977
+    },
+    {
+      "epoch": 0.7134657109115378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.005896332492346e-06,
+      "loss": 0.7724,
+      "step": 25978
+    },
+    {
+      "epoch": 0.7134931751393809,
+      "grad_norm": 0.0,
+      "learning_rate": 4.005184350332655e-06,
+      "loss": 0.8654,
+      "step": 25979
+    },
+    {
+      "epoch": 0.7135206393672242,
+      "grad_norm": 0.0,
+      "learning_rate": 4.004472415606265e-06,
+      "loss": 0.8094,
+      "step": 25980
+    },
+    {
+      "epoch": 0.7135481035950674,
+      "grad_norm": 0.0,
+      "learning_rate": 4.003760528318814e-06,
+      "loss": 0.929,
+      "step": 25981
+    },
+    {
+      "epoch": 0.7135755678229107,
+      "grad_norm": 0.0,
+      "learning_rate": 4.003048688475929e-06,
+      "loss": 0.7967,
+      "step": 25982
+    },
+    {
+      "epoch": 0.7136030320507539,
+      "grad_norm": 0.0,
+      "learning_rate": 4.002336896083246e-06,
+      "loss": 0.8812,
+      "step": 25983
+    },
+    {
+      "epoch": 0.7136304962785971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.001625151146396e-06,
+      "loss": 0.7786,
+      "step": 25984
+    },
+    {
+      "epoch": 0.7136579605064404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.000913453671017e-06,
+      "loss": 0.8615,
+      "step": 25985
+    },
+    {
+      "epoch": 0.7136854247342836,
+      "grad_norm": 0.0,
+      "learning_rate": 4.00020180366273e-06,
+      "loss": 0.8055,
+      "step": 25986
+    },
+    {
+      "epoch": 0.7137128889621268,
+      "grad_norm": 0.0,
+      "learning_rate": 3.99949020112717e-06,
+      "loss": 0.8446,
+      "step": 25987
+    },
+    {
+      "epoch": 0.7137403531899701,
+      "grad_norm": 0.0,
+      "learning_rate": 3.998778646069972e-06,
+      "loss": 0.7439,
+      "step": 25988
+    },
+    {
+      "epoch": 0.7137678174178133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.998067138496758e-06,
+      "loss": 0.7506,
+      "step": 25989
+    },
+    {
+      "epoch": 0.7137952816456565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.997355678413162e-06,
+      "loss": 0.869,
+      "step": 25990
+    },
+    {
+      "epoch": 0.7138227458734998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.996644265824816e-06,
+      "loss": 0.7948,
+      "step": 25991
+    },
+    {
+      "epoch": 0.713850210101343,
+      "grad_norm": 0.0,
+      "learning_rate": 3.995932900737343e-06,
+      "loss": 0.8456,
+      "step": 25992
+    },
+    {
+      "epoch": 0.7138776743291863,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9952215831563745e-06,
+      "loss": 0.8811,
+      "step": 25993
+    },
+    {
+      "epoch": 0.7139051385570294,
+      "grad_norm": 0.0,
+      "learning_rate": 3.994510313087538e-06,
+      "loss": 0.8019,
+      "step": 25994
+    },
+    {
+      "epoch": 0.7139326027848727,
+      "grad_norm": 0.0,
+      "learning_rate": 3.993799090536465e-06,
+      "loss": 0.739,
+      "step": 25995
+    },
+    {
+      "epoch": 0.713960067012716,
+      "grad_norm": 0.0,
+      "learning_rate": 3.993087915508781e-06,
+      "loss": 0.8255,
+      "step": 25996
+    },
+    {
+      "epoch": 0.7139875312405591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.992376788010113e-06,
+      "loss": 0.7333,
+      "step": 25997
+    },
+    {
+      "epoch": 0.7140149954684024,
+      "grad_norm": 0.0,
+      "learning_rate": 3.991665708046083e-06,
+      "loss": 0.8895,
+      "step": 25998
+    },
+    {
+      "epoch": 0.7140424596962457,
+      "grad_norm": 0.0,
+      "learning_rate": 3.990954675622323e-06,
+      "loss": 0.8851,
+      "step": 25999
+    },
+    {
+      "epoch": 0.7140699239240889,
+      "grad_norm": 0.0,
+      "learning_rate": 3.990243690744455e-06,
+      "loss": 0.825,
+      "step": 26000
+    },
+    {
+      "epoch": 0.7140973881519321,
+      "grad_norm": 0.0,
+      "learning_rate": 3.989532753418108e-06,
+      "loss": 0.7412,
+      "step": 26001
+    },
+    {
+      "epoch": 0.7141248523797753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.988821863648906e-06,
+      "loss": 0.726,
+      "step": 26002
+    },
+    {
+      "epoch": 0.7141523166076186,
+      "grad_norm": 0.0,
+      "learning_rate": 3.988111021442478e-06,
+      "loss": 0.795,
+      "step": 26003
+    },
+    {
+      "epoch": 0.7141797808354619,
+      "grad_norm": 0.0,
+      "learning_rate": 3.987400226804441e-06,
+      "loss": 0.8048,
+      "step": 26004
+    },
+    {
+      "epoch": 0.714207245063305,
+      "grad_norm": 0.0,
+      "learning_rate": 3.986689479740425e-06,
+      "loss": 0.8418,
+      "step": 26005
+    },
+    {
+      "epoch": 0.7142347092911483,
+      "grad_norm": 0.0,
+      "learning_rate": 3.985978780256048e-06,
+      "loss": 0.873,
+      "step": 26006
+    },
+    {
+      "epoch": 0.7142621735189915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.985268128356937e-06,
+      "loss": 0.8191,
+      "step": 26007
+    },
+    {
+      "epoch": 0.7142896377468347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.984557524048714e-06,
+      "loss": 0.7294,
+      "step": 26008
+    },
+    {
+      "epoch": 0.714317101974678,
+      "grad_norm": 0.0,
+      "learning_rate": 3.983846967337005e-06,
+      "loss": 0.8366,
+      "step": 26009
+    },
+    {
+      "epoch": 0.7143445662025212,
+      "grad_norm": 0.0,
+      "learning_rate": 3.983136458227427e-06,
+      "loss": 0.9136,
+      "step": 26010
+    },
+    {
+      "epoch": 0.7143720304303645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.982425996725603e-06,
+      "loss": 0.895,
+      "step": 26011
+    },
+    {
+      "epoch": 0.7143994946582077,
+      "grad_norm": 0.0,
+      "learning_rate": 3.98171558283716e-06,
+      "loss": 0.929,
+      "step": 26012
+    },
+    {
+      "epoch": 0.7144269588860509,
+      "grad_norm": 0.0,
+      "learning_rate": 3.98100521656771e-06,
+      "loss": 0.8497,
+      "step": 26013
+    },
+    {
+      "epoch": 0.7144544231138942,
+      "grad_norm": 0.0,
+      "learning_rate": 3.980294897922878e-06,
+      "loss": 0.861,
+      "step": 26014
+    },
+    {
+      "epoch": 0.7144818873417373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9795846269082894e-06,
+      "loss": 0.7563,
+      "step": 26015
+    },
+    {
+      "epoch": 0.7145093515695806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.978874403529555e-06,
+      "loss": 0.8643,
+      "step": 26016
+    },
+    {
+      "epoch": 0.7145368157974239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9781642277923e-06,
+      "loss": 0.8106,
+      "step": 26017
+    },
+    {
+      "epoch": 0.7145642800252671,
+      "grad_norm": 0.0,
+      "learning_rate": 3.97745409970214e-06,
+      "loss": 0.8832,
+      "step": 26018
+    },
+    {
+      "epoch": 0.7145917442531103,
+      "grad_norm": 0.0,
+      "learning_rate": 3.976744019264698e-06,
+      "loss": 0.8512,
+      "step": 26019
+    },
+    {
+      "epoch": 0.7146192084809535,
+      "grad_norm": 0.0,
+      "learning_rate": 3.976033986485592e-06,
+      "loss": 0.865,
+      "step": 26020
+    },
+    {
+      "epoch": 0.7146466727087968,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9753240013704405e-06,
+      "loss": 0.8048,
+      "step": 26021
+    },
+    {
+      "epoch": 0.7146741369366401,
+      "grad_norm": 0.0,
+      "learning_rate": 3.974614063924856e-06,
+      "loss": 0.8406,
+      "step": 26022
+    },
+    {
+      "epoch": 0.7147016011644832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.973904174154457e-06,
+      "loss": 0.8081,
+      "step": 26023
+    },
+    {
+      "epoch": 0.7147290653923265,
+      "grad_norm": 0.0,
+      "learning_rate": 3.973194332064864e-06,
+      "loss": 0.7979,
+      "step": 26024
+    },
+    {
+      "epoch": 0.7147565296201698,
+      "grad_norm": 0.0,
+      "learning_rate": 3.972484537661691e-06,
+      "loss": 0.684,
+      "step": 26025
+    },
+    {
+      "epoch": 0.7147839938480129,
+      "grad_norm": 0.0,
+      "learning_rate": 3.971774790950556e-06,
+      "loss": 0.8848,
+      "step": 26026
+    },
+    {
+      "epoch": 0.7148114580758562,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9710650919370764e-06,
+      "loss": 0.9052,
+      "step": 26027
+    },
+    {
+      "epoch": 0.7148389223036994,
+      "grad_norm": 0.0,
+      "learning_rate": 3.970355440626863e-06,
+      "loss": 0.8945,
+      "step": 26028
+    },
+    {
+      "epoch": 0.7148663865315427,
+      "grad_norm": 0.0,
+      "learning_rate": 3.969645837025535e-06,
+      "loss": 0.8368,
+      "step": 26029
+    },
+    {
+      "epoch": 0.7148938507593859,
+      "grad_norm": 0.0,
+      "learning_rate": 3.968936281138703e-06,
+      "loss": 0.8132,
+      "step": 26030
+    },
+    {
+      "epoch": 0.7149213149872291,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9682267729719835e-06,
+      "loss": 0.7733,
+      "step": 26031
+    },
+    {
+      "epoch": 0.7149487792150724,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9675173125309895e-06,
+      "loss": 0.8951,
+      "step": 26032
+    },
+    {
+      "epoch": 0.7149762434429155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.966807899821339e-06,
+      "loss": 0.7866,
+      "step": 26033
+    },
+    {
+      "epoch": 0.7150037076707588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.966098534848638e-06,
+      "loss": 0.7829,
+      "step": 26034
+    },
+    {
+      "epoch": 0.7150311718986021,
+      "grad_norm": 0.0,
+      "learning_rate": 3.965389217618501e-06,
+      "loss": 0.7813,
+      "step": 26035
+    },
+    {
+      "epoch": 0.7150586361264453,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9646799481365485e-06,
+      "loss": 0.9133,
+      "step": 26036
+    },
+    {
+      "epoch": 0.7150861003542885,
+      "grad_norm": 0.0,
+      "learning_rate": 3.963970726408381e-06,
+      "loss": 0.8503,
+      "step": 26037
+    },
+    {
+      "epoch": 0.7151135645821318,
+      "grad_norm": 0.0,
+      "learning_rate": 3.963261552439615e-06,
+      "loss": 0.8468,
+      "step": 26038
+    },
+    {
+      "epoch": 0.715141028809975,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9625524262358665e-06,
+      "loss": 0.826,
+      "step": 26039
+    },
+    {
+      "epoch": 0.7151684930378183,
+      "grad_norm": 0.0,
+      "learning_rate": 3.961843347802739e-06,
+      "loss": 0.8041,
+      "step": 26040
+    },
+    {
+      "epoch": 0.7151959572656614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.961134317145844e-06,
+      "loss": 0.7836,
+      "step": 26041
+    },
+    {
+      "epoch": 0.7152234214935047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.960425334270795e-06,
+      "loss": 0.945,
+      "step": 26042
+    },
+    {
+      "epoch": 0.715250885721348,
+      "grad_norm": 0.0,
+      "learning_rate": 3.959716399183199e-06,
+      "loss": 0.7524,
+      "step": 26043
+    },
+    {
+      "epoch": 0.7152783499491912,
+      "grad_norm": 0.0,
+      "learning_rate": 3.959007511888671e-06,
+      "loss": 0.7953,
+      "step": 26044
+    },
+    {
+      "epoch": 0.7153058141770344,
+      "grad_norm": 0.0,
+      "learning_rate": 3.958298672392815e-06,
+      "loss": 0.7778,
+      "step": 26045
+    },
+    {
+      "epoch": 0.7153332784048776,
+      "grad_norm": 0.0,
+      "learning_rate": 3.957589880701237e-06,
+      "loss": 0.8468,
+      "step": 26046
+    },
+    {
+      "epoch": 0.7153607426327209,
+      "grad_norm": 0.0,
+      "learning_rate": 3.956881136819548e-06,
+      "loss": 0.7975,
+      "step": 26047
+    },
+    {
+      "epoch": 0.7153882068605641,
+      "grad_norm": 0.0,
+      "learning_rate": 3.956172440753357e-06,
+      "loss": 0.841,
+      "step": 26048
+    },
+    {
+      "epoch": 0.7154156710884073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.955463792508271e-06,
+      "loss": 0.8528,
+      "step": 26049
+    },
+    {
+      "epoch": 0.7154431353162506,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9547551920899e-06,
+      "loss": 0.8171,
+      "step": 26050
+    },
+    {
+      "epoch": 0.7154705995440939,
+      "grad_norm": 0.0,
+      "learning_rate": 3.954046639503843e-06,
+      "loss": 0.8815,
+      "step": 26051
+    },
+    {
+      "epoch": 0.715498063771937,
+      "grad_norm": 0.0,
+      "learning_rate": 3.953338134755713e-06,
+      "loss": 0.8055,
+      "step": 26052
+    },
+    {
+      "epoch": 0.7155255279997803,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9526296778511165e-06,
+      "loss": 0.8764,
+      "step": 26053
+    },
+    {
+      "epoch": 0.7155529922276235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9519212687956535e-06,
+      "loss": 0.8119,
+      "step": 26054
+    },
+    {
+      "epoch": 0.7155804564554668,
+      "grad_norm": 0.0,
+      "learning_rate": 3.951212907594931e-06,
+      "loss": 0.8412,
+      "step": 26055
+    },
+    {
+      "epoch": 0.71560792068331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.95050459425456e-06,
+      "loss": 0.9086,
+      "step": 26056
+    },
+    {
+      "epoch": 0.7156353849111532,
+      "grad_norm": 0.0,
+      "learning_rate": 3.949796328780135e-06,
+      "loss": 1.0137,
+      "step": 26057
+    },
+    {
+      "epoch": 0.7156628491389965,
+      "grad_norm": 0.0,
+      "learning_rate": 3.949088111177267e-06,
+      "loss": 0.7802,
+      "step": 26058
+    },
+    {
+      "epoch": 0.7156903133668396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.948379941451557e-06,
+      "loss": 0.889,
+      "step": 26059
+    },
+    {
+      "epoch": 0.7157177775946829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.947671819608612e-06,
+      "loss": 0.8578,
+      "step": 26060
+    },
+    {
+      "epoch": 0.7157452418225262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.946963745654029e-06,
+      "loss": 0.8143,
+      "step": 26061
+    },
+    {
+      "epoch": 0.7157727060503694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.946255719593416e-06,
+      "loss": 0.8314,
+      "step": 26062
+    },
+    {
+      "epoch": 0.7158001702782126,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9455477414323705e-06,
+      "loss": 0.7782,
+      "step": 26063
+    },
+    {
+      "epoch": 0.7158276345060559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.944839811176497e-06,
+      "loss": 0.7875,
+      "step": 26064
+    },
+    {
+      "epoch": 0.7158550987338991,
+      "grad_norm": 0.0,
+      "learning_rate": 3.944131928831395e-06,
+      "loss": 0.8788,
+      "step": 26065
+    },
+    {
+      "epoch": 0.7158825629617424,
+      "grad_norm": 0.0,
+      "learning_rate": 3.943424094402667e-06,
+      "loss": 0.8123,
+      "step": 26066
+    },
+    {
+      "epoch": 0.7159100271895855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.942716307895914e-06,
+      "loss": 0.9254,
+      "step": 26067
+    },
+    {
+      "epoch": 0.7159374914174288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.94200856931674e-06,
+      "loss": 0.891,
+      "step": 26068
+    },
+    {
+      "epoch": 0.7159649556452721,
+      "grad_norm": 0.0,
+      "learning_rate": 3.941300878670741e-06,
+      "loss": 0.8526,
+      "step": 26069
+    },
+    {
+      "epoch": 0.7159924198731152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.940593235963512e-06,
+      "loss": 0.8037,
+      "step": 26070
+    },
+    {
+      "epoch": 0.7160198841009585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.939885641200657e-06,
+      "loss": 0.9012,
+      "step": 26071
+    },
+    {
+      "epoch": 0.7160473483288017,
+      "grad_norm": 0.0,
+      "learning_rate": 3.939178094387774e-06,
+      "loss": 0.7658,
+      "step": 26072
+    },
+    {
+      "epoch": 0.716074812556645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.938470595530463e-06,
+      "loss": 0.7784,
+      "step": 26073
+    },
+    {
+      "epoch": 0.7161022767844882,
+      "grad_norm": 0.0,
+      "learning_rate": 3.937763144634323e-06,
+      "loss": 0.8178,
+      "step": 26074
+    },
+    {
+      "epoch": 0.7161297410123314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.937055741704945e-06,
+      "loss": 0.8058,
+      "step": 26075
+    },
+    {
+      "epoch": 0.7161572052401747,
+      "grad_norm": 0.0,
+      "learning_rate": 3.936348386747932e-06,
+      "loss": 0.8388,
+      "step": 26076
+    },
+    {
+      "epoch": 0.716184669468018,
+      "grad_norm": 0.0,
+      "learning_rate": 3.935641079768881e-06,
+      "loss": 0.8347,
+      "step": 26077
+    },
+    {
+      "epoch": 0.7162121336958611,
+      "grad_norm": 0.0,
+      "learning_rate": 3.934933820773385e-06,
+      "loss": 0.9095,
+      "step": 26078
+    },
+    {
+      "epoch": 0.7162395979237044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9342266097670415e-06,
+      "loss": 0.8682,
+      "step": 26079
+    },
+    {
+      "epoch": 0.7162670621515476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.93351944675545e-06,
+      "loss": 0.86,
+      "step": 26080
+    },
+    {
+      "epoch": 0.7162945263793908,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9328123317442e-06,
+      "loss": 0.871,
+      "step": 26081
+    },
+    {
+      "epoch": 0.7163219906072341,
+      "grad_norm": 0.0,
+      "learning_rate": 3.932105264738888e-06,
+      "loss": 0.8782,
+      "step": 26082
+    },
+    {
+      "epoch": 0.7163494548350773,
+      "grad_norm": 0.0,
+      "learning_rate": 3.931398245745111e-06,
+      "loss": 0.8523,
+      "step": 26083
+    },
+    {
+      "epoch": 0.7163769190629206,
+      "grad_norm": 0.0,
+      "learning_rate": 3.930691274768464e-06,
+      "loss": 0.9652,
+      "step": 26084
+    },
+    {
+      "epoch": 0.7164043832907638,
+      "grad_norm": 0.0,
+      "learning_rate": 3.929984351814535e-06,
+      "loss": 0.826,
+      "step": 26085
+    },
+    {
+      "epoch": 0.716431847518607,
+      "grad_norm": 0.0,
+      "learning_rate": 3.929277476888925e-06,
+      "loss": 0.8235,
+      "step": 26086
+    },
+    {
+      "epoch": 0.7164593117464503,
+      "grad_norm": 0.0,
+      "learning_rate": 3.92857064999722e-06,
+      "loss": 0.9266,
+      "step": 26087
+    },
+    {
+      "epoch": 0.7164867759742934,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9278638711450155e-06,
+      "loss": 0.912,
+      "step": 26088
+    },
+    {
+      "epoch": 0.7165142402021367,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9271571403379045e-06,
+      "loss": 0.811,
+      "step": 26089
+    },
+    {
+      "epoch": 0.71654170442998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9264504575814786e-06,
+      "loss": 0.8848,
+      "step": 26090
+    },
+    {
+      "epoch": 0.7165691686578232,
+      "grad_norm": 0.0,
+      "learning_rate": 3.92574382288133e-06,
+      "loss": 0.8238,
+      "step": 26091
+    },
+    {
+      "epoch": 0.7165966328856664,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9250372362430515e-06,
+      "loss": 0.9362,
+      "step": 26092
+    },
+    {
+      "epoch": 0.7166240971135096,
+      "grad_norm": 0.0,
+      "learning_rate": 3.924330697672231e-06,
+      "loss": 0.9037,
+      "step": 26093
+    },
+    {
+      "epoch": 0.7166515613413529,
+      "grad_norm": 0.0,
+      "learning_rate": 3.923624207174458e-06,
+      "loss": 0.7672,
+      "step": 26094
+    },
+    {
+      "epoch": 0.7166790255691962,
+      "grad_norm": 0.0,
+      "learning_rate": 3.922917764755323e-06,
+      "loss": 0.9175,
+      "step": 26095
+    },
+    {
+      "epoch": 0.7167064897970393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.922211370420417e-06,
+      "loss": 0.8397,
+      "step": 26096
+    },
+    {
+      "epoch": 0.7167339540248826,
+      "grad_norm": 0.0,
+      "learning_rate": 3.921505024175328e-06,
+      "loss": 0.7881,
+      "step": 26097
+    },
+    {
+      "epoch": 0.7167614182527259,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9207987260256495e-06,
+      "loss": 0.7647,
+      "step": 26098
+    },
+    {
+      "epoch": 0.716788882480569,
+      "grad_norm": 0.0,
+      "learning_rate": 3.920092475976964e-06,
+      "loss": 0.8029,
+      "step": 26099
+    },
+    {
+      "epoch": 0.7168163467084123,
+      "grad_norm": 0.0,
+      "learning_rate": 3.919386274034864e-06,
+      "loss": 0.7511,
+      "step": 26100
+    },
+    {
+      "epoch": 0.7168438109362555,
+      "grad_norm": 0.0,
+      "learning_rate": 3.918680120204933e-06,
+      "loss": 0.8495,
+      "step": 26101
+    },
+    {
+      "epoch": 0.7168712751640988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.91797401449276e-06,
+      "loss": 0.8623,
+      "step": 26102
+    },
+    {
+      "epoch": 0.716898739391942,
+      "grad_norm": 0.0,
+      "learning_rate": 3.917267956903933e-06,
+      "loss": 0.7428,
+      "step": 26103
+    },
+    {
+      "epoch": 0.7169262036197852,
+      "grad_norm": 0.0,
+      "learning_rate": 3.916561947444041e-06,
+      "loss": 0.7325,
+      "step": 26104
+    },
+    {
+      "epoch": 0.7169536678476285,
+      "grad_norm": 0.0,
+      "learning_rate": 3.915855986118664e-06,
+      "loss": 0.8757,
+      "step": 26105
+    },
+    {
+      "epoch": 0.7169811320754716,
+      "grad_norm": 0.0,
+      "learning_rate": 3.915150072933392e-06,
+      "loss": 0.8176,
+      "step": 26106
+    },
+    {
+      "epoch": 0.7170085963033149,
+      "grad_norm": 0.0,
+      "learning_rate": 3.914444207893808e-06,
+      "loss": 0.801,
+      "step": 26107
+    },
+    {
+      "epoch": 0.7170360605311582,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9137383910055035e-06,
+      "loss": 0.8318,
+      "step": 26108
+    },
+    {
+      "epoch": 0.7170635247590014,
+      "grad_norm": 0.0,
+      "learning_rate": 3.913032622274059e-06,
+      "loss": 0.8382,
+      "step": 26109
+    },
+    {
+      "epoch": 0.7170909889868446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.912326901705054e-06,
+      "loss": 0.8197,
+      "step": 26110
+    },
+    {
+      "epoch": 0.7171184532146879,
+      "grad_norm": 0.0,
+      "learning_rate": 3.911621229304077e-06,
+      "loss": 0.7238,
+      "step": 26111
+    },
+    {
+      "epoch": 0.7171459174425311,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9109156050767106e-06,
+      "loss": 0.8057,
+      "step": 26112
+    },
+    {
+      "epoch": 0.7171733816703744,
+      "grad_norm": 0.0,
+      "learning_rate": 3.910210029028538e-06,
+      "loss": 0.8284,
+      "step": 26113
+    },
+    {
+      "epoch": 0.7172008458982175,
+      "grad_norm": 0.0,
+      "learning_rate": 3.909504501165145e-06,
+      "loss": 0.8148,
+      "step": 26114
+    },
+    {
+      "epoch": 0.7172283101260608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.908799021492112e-06,
+      "loss": 0.8099,
+      "step": 26115
+    },
+    {
+      "epoch": 0.7172557743539041,
+      "grad_norm": 0.0,
+      "learning_rate": 3.908093590015019e-06,
+      "loss": 0.8238,
+      "step": 26116
+    },
+    {
+      "epoch": 0.7172832385817473,
+      "grad_norm": 0.0,
+      "learning_rate": 3.907388206739453e-06,
+      "loss": 0.7293,
+      "step": 26117
+    },
+    {
+      "epoch": 0.7173107028095905,
+      "grad_norm": 0.0,
+      "learning_rate": 3.906682871670988e-06,
+      "loss": 0.8786,
+      "step": 26118
+    },
+    {
+      "epoch": 0.7173381670374337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.905977584815208e-06,
+      "loss": 0.8021,
+      "step": 26119
+    },
+    {
+      "epoch": 0.717365631265277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.905272346177694e-06,
+      "loss": 0.7654,
+      "step": 26120
+    },
+    {
+      "epoch": 0.7173930954931202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.90456715576403e-06,
+      "loss": 0.8602,
+      "step": 26121
+    },
+    {
+      "epoch": 0.7174205597209634,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9038620135797874e-06,
+      "loss": 0.8116,
+      "step": 26122
+    },
+    {
+      "epoch": 0.7174480239488067,
+      "grad_norm": 0.0,
+      "learning_rate": 3.903156919630551e-06,
+      "loss": 0.7875,
+      "step": 26123
+    },
+    {
+      "epoch": 0.71747548817665,
+      "grad_norm": 0.0,
+      "learning_rate": 3.902451873921901e-06,
+      "loss": 0.7972,
+      "step": 26124
+    },
+    {
+      "epoch": 0.7175029524044931,
+      "grad_norm": 0.0,
+      "learning_rate": 3.901746876459412e-06,
+      "loss": 0.8636,
+      "step": 26125
+    },
+    {
+      "epoch": 0.7175304166323364,
+      "grad_norm": 0.0,
+      "learning_rate": 3.901041927248662e-06,
+      "loss": 0.8277,
+      "step": 26126
+    },
+    {
+      "epoch": 0.7175578808601796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.900337026295236e-06,
+      "loss": 0.6548,
+      "step": 26127
+    },
+    {
+      "epoch": 0.7175853450880229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8996321736047015e-06,
+      "loss": 0.8385,
+      "step": 26128
+    },
+    {
+      "epoch": 0.7176128093158661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.89892736918264e-06,
+      "loss": 0.8718,
+      "step": 26129
+    },
+    {
+      "epoch": 0.7176402735437093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.89822261303463e-06,
+      "loss": 0.8473,
+      "step": 26130
+    },
+    {
+      "epoch": 0.7176677377715526,
+      "grad_norm": 0.0,
+      "learning_rate": 3.897517905166246e-06,
+      "loss": 0.843,
+      "step": 26131
+    },
+    {
+      "epoch": 0.7176952019993957,
+      "grad_norm": 0.0,
+      "learning_rate": 3.896813245583066e-06,
+      "loss": 0.7773,
+      "step": 26132
+    },
+    {
+      "epoch": 0.717722666227239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.896108634290665e-06,
+      "loss": 0.8441,
+      "step": 26133
+    },
+    {
+      "epoch": 0.7177501304550823,
+      "grad_norm": 0.0,
+      "learning_rate": 3.895404071294614e-06,
+      "loss": 0.8153,
+      "step": 26134
+    },
+    {
+      "epoch": 0.7177775946829255,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8946995566004894e-06,
+      "loss": 0.8981,
+      "step": 26135
+    },
+    {
+      "epoch": 0.7178050589107687,
+      "grad_norm": 0.0,
+      "learning_rate": 3.893995090213868e-06,
+      "loss": 0.7921,
+      "step": 26136
+    },
+    {
+      "epoch": 0.717832523138612,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8932906721403225e-06,
+      "loss": 0.907,
+      "step": 26137
+    },
+    {
+      "epoch": 0.7178599873664552,
+      "grad_norm": 0.0,
+      "learning_rate": 3.892586302385426e-06,
+      "loss": 0.7518,
+      "step": 26138
+    },
+    {
+      "epoch": 0.7178874515942985,
+      "grad_norm": 0.0,
+      "learning_rate": 3.891881980954757e-06,
+      "loss": 0.8797,
+      "step": 26139
+    },
+    {
+      "epoch": 0.7179149158221416,
+      "grad_norm": 0.0,
+      "learning_rate": 3.891177707853879e-06,
+      "loss": 0.8271,
+      "step": 26140
+    },
+    {
+      "epoch": 0.7179423800499849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.890473483088374e-06,
+      "loss": 0.7905,
+      "step": 26141
+    },
+    {
+      "epoch": 0.7179698442778282,
+      "grad_norm": 0.0,
+      "learning_rate": 3.889769306663805e-06,
+      "loss": 0.8883,
+      "step": 26142
+    },
+    {
+      "epoch": 0.7179973085056713,
+      "grad_norm": 0.0,
+      "learning_rate": 3.889065178585749e-06,
+      "loss": 0.803,
+      "step": 26143
+    },
+    {
+      "epoch": 0.7180247727335146,
+      "grad_norm": 0.0,
+      "learning_rate": 3.888361098859777e-06,
+      "loss": 0.772,
+      "step": 26144
+    },
+    {
+      "epoch": 0.7180522369613578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8876570674914615e-06,
+      "loss": 0.8308,
+      "step": 26145
+    },
+    {
+      "epoch": 0.7180797011892011,
+      "grad_norm": 0.0,
+      "learning_rate": 3.886953084486368e-06,
+      "loss": 0.8944,
+      "step": 26146
+    },
+    {
+      "epoch": 0.7181071654170443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.886249149850068e-06,
+      "loss": 0.8286,
+      "step": 26147
+    },
+    {
+      "epoch": 0.7181346296448875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.885545263588139e-06,
+      "loss": 0.7499,
+      "step": 26148
+    },
+    {
+      "epoch": 0.7181620938727308,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8848414257061395e-06,
+      "loss": 0.8389,
+      "step": 26149
+    },
+    {
+      "epoch": 0.718189558100574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8841376362096436e-06,
+      "loss": 0.875,
+      "step": 26150
+    },
+    {
+      "epoch": 0.7182170223284172,
+      "grad_norm": 0.0,
+      "learning_rate": 3.883433895104224e-06,
+      "loss": 0.7092,
+      "step": 26151
+    },
+    {
+      "epoch": 0.7182444865562605,
+      "grad_norm": 0.0,
+      "learning_rate": 3.882730202395441e-06,
+      "loss": 0.8083,
+      "step": 26152
+    },
+    {
+      "epoch": 0.7182719507841037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.882026558088866e-06,
+      "loss": 0.7737,
+      "step": 26153
+    },
+    {
+      "epoch": 0.7182994150119469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.881322962190067e-06,
+      "loss": 0.7815,
+      "step": 26154
+    },
+    {
+      "epoch": 0.7183268792397902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.88061941470461e-06,
+      "loss": 0.8248,
+      "step": 26155
+    },
+    {
+      "epoch": 0.7183543434676334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8799159156380675e-06,
+      "loss": 0.8113,
+      "step": 26156
+    },
+    {
+      "epoch": 0.7183818076954767,
+      "grad_norm": 0.0,
+      "learning_rate": 3.879212464995999e-06,
+      "loss": 0.841,
+      "step": 26157
+    },
+    {
+      "epoch": 0.7184092719233198,
+      "grad_norm": 0.0,
+      "learning_rate": 3.878509062783972e-06,
+      "loss": 0.833,
+      "step": 26158
+    },
+    {
+      "epoch": 0.7184367361511631,
+      "grad_norm": 0.0,
+      "learning_rate": 3.877805709007549e-06,
+      "loss": 0.8283,
+      "step": 26159
+    },
+    {
+      "epoch": 0.7184642003790064,
+      "grad_norm": 0.0,
+      "learning_rate": 3.877102403672301e-06,
+      "loss": 0.9351,
+      "step": 26160
+    },
+    {
+      "epoch": 0.7184916646068495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.87639914678379e-06,
+      "loss": 0.8012,
+      "step": 26161
+    },
+    {
+      "epoch": 0.7185191288346928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.875695938347581e-06,
+      "loss": 0.8108,
+      "step": 26162
+    },
+    {
+      "epoch": 0.7185465930625361,
+      "grad_norm": 0.0,
+      "learning_rate": 3.874992778369242e-06,
+      "loss": 0.9495,
+      "step": 26163
+    },
+    {
+      "epoch": 0.7185740572903793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.874289666854328e-06,
+      "loss": 0.8074,
+      "step": 26164
+    },
+    {
+      "epoch": 0.7186015215182225,
+      "grad_norm": 0.0,
+      "learning_rate": 3.873586603808413e-06,
+      "loss": 0.7504,
+      "step": 26165
+    },
+    {
+      "epoch": 0.7186289857460657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8728835892370476e-06,
+      "loss": 0.8466,
+      "step": 26166
+    },
+    {
+      "epoch": 0.718656449973909,
+      "grad_norm": 0.0,
+      "learning_rate": 3.872180623145803e-06,
+      "loss": 0.8311,
+      "step": 26167
+    },
+    {
+      "epoch": 0.7186839142017523,
+      "grad_norm": 0.0,
+      "learning_rate": 3.871477705540241e-06,
+      "loss": 0.8541,
+      "step": 26168
+    },
+    {
+      "epoch": 0.7187113784295954,
+      "grad_norm": 0.0,
+      "learning_rate": 3.870774836425918e-06,
+      "loss": 0.9257,
+      "step": 26169
+    },
+    {
+      "epoch": 0.7187388426574387,
+      "grad_norm": 0.0,
+      "learning_rate": 3.870072015808398e-06,
+      "loss": 0.8793,
+      "step": 26170
+    },
+    {
+      "epoch": 0.7187663068852819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.869369243693244e-06,
+      "loss": 0.9066,
+      "step": 26171
+    },
+    {
+      "epoch": 0.7187937711131251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.868666520086018e-06,
+      "loss": 0.9072,
+      "step": 26172
+    },
+    {
+      "epoch": 0.7188212353409684,
+      "grad_norm": 0.0,
+      "learning_rate": 3.867963844992275e-06,
+      "loss": 0.8874,
+      "step": 26173
+    },
+    {
+      "epoch": 0.7188486995688116,
+      "grad_norm": 0.0,
+      "learning_rate": 3.86726121841758e-06,
+      "loss": 0.9575,
+      "step": 26174
+    },
+    {
+      "epoch": 0.7188761637966549,
+      "grad_norm": 0.0,
+      "learning_rate": 3.866558640367487e-06,
+      "loss": 0.8216,
+      "step": 26175
+    },
+    {
+      "epoch": 0.7189036280244981,
+      "grad_norm": 0.0,
+      "learning_rate": 3.865856110847558e-06,
+      "loss": 0.8218,
+      "step": 26176
+    },
+    {
+      "epoch": 0.7189310922523413,
+      "grad_norm": 0.0,
+      "learning_rate": 3.865153629863351e-06,
+      "loss": 0.8937,
+      "step": 26177
+    },
+    {
+      "epoch": 0.7189585564801846,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8644511974204256e-06,
+      "loss": 0.88,
+      "step": 26178
+    },
+    {
+      "epoch": 0.7189860207080278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.863748813524339e-06,
+      "loss": 0.9421,
+      "step": 26179
+    },
+    {
+      "epoch": 0.719013484935871,
+      "grad_norm": 0.0,
+      "learning_rate": 3.863046478180651e-06,
+      "loss": 0.8534,
+      "step": 26180
+    },
+    {
+      "epoch": 0.7190409491637143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.862344191394917e-06,
+      "loss": 0.9373,
+      "step": 26181
+    },
+    {
+      "epoch": 0.7190684133915575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.86164195317269e-06,
+      "loss": 0.8842,
+      "step": 26182
+    },
+    {
+      "epoch": 0.7190958776194007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.86093976351953e-06,
+      "loss": 0.8728,
+      "step": 26183
+    },
+    {
+      "epoch": 0.7191233418472439,
+      "grad_norm": 0.0,
+      "learning_rate": 3.860237622440992e-06,
+      "loss": 0.8714,
+      "step": 26184
+    },
+    {
+      "epoch": 0.7191508060750872,
+      "grad_norm": 0.0,
+      "learning_rate": 3.859535529942633e-06,
+      "loss": 0.905,
+      "step": 26185
+    },
+    {
+      "epoch": 0.7191782703029305,
+      "grad_norm": 0.0,
+      "learning_rate": 3.85883348603001e-06,
+      "loss": 0.8323,
+      "step": 26186
+    },
+    {
+      "epoch": 0.7192057345307736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.858131490708673e-06,
+      "loss": 0.844,
+      "step": 26187
+    },
+    {
+      "epoch": 0.7192331987586169,
+      "grad_norm": 0.0,
+      "learning_rate": 3.857429543984178e-06,
+      "loss": 0.8515,
+      "step": 26188
+    },
+    {
+      "epoch": 0.7192606629864602,
+      "grad_norm": 0.0,
+      "learning_rate": 3.856727645862084e-06,
+      "loss": 0.8784,
+      "step": 26189
+    },
+    {
+      "epoch": 0.7192881272143034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8560257963479366e-06,
+      "loss": 0.8507,
+      "step": 26190
+    },
+    {
+      "epoch": 0.7193155914421466,
+      "grad_norm": 0.0,
+      "learning_rate": 3.855323995447292e-06,
+      "loss": 0.7556,
+      "step": 26191
+    },
+    {
+      "epoch": 0.7193430556699898,
+      "grad_norm": 0.0,
+      "learning_rate": 3.854622243165709e-06,
+      "loss": 0.7606,
+      "step": 26192
+    },
+    {
+      "epoch": 0.7193705198978331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.853920539508731e-06,
+      "loss": 0.8981,
+      "step": 26193
+    },
+    {
+      "epoch": 0.7193979841256763,
+      "grad_norm": 0.0,
+      "learning_rate": 3.853218884481915e-06,
+      "loss": 0.9037,
+      "step": 26194
+    },
+    {
+      "epoch": 0.7194254483535195,
+      "grad_norm": 0.0,
+      "learning_rate": 3.852517278090812e-06,
+      "loss": 0.8596,
+      "step": 26195
+    },
+    {
+      "epoch": 0.7194529125813628,
+      "grad_norm": 0.0,
+      "learning_rate": 3.851815720340978e-06,
+      "loss": 0.8162,
+      "step": 26196
+    },
+    {
+      "epoch": 0.719480376809206,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8511142112379555e-06,
+      "loss": 0.8034,
+      "step": 26197
+    },
+    {
+      "epoch": 0.7195078410370492,
+      "grad_norm": 0.0,
+      "learning_rate": 3.850412750787302e-06,
+      "loss": 0.8664,
+      "step": 26198
+    },
+    {
+      "epoch": 0.7195353052648925,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8497113389945615e-06,
+      "loss": 0.7483,
+      "step": 26199
+    },
+    {
+      "epoch": 0.7195627694927357,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8490099758652876e-06,
+      "loss": 0.7856,
+      "step": 26200
+    },
+    {
+      "epoch": 0.719590233720579,
+      "grad_norm": 0.0,
+      "learning_rate": 3.84830866140503e-06,
+      "loss": 0.8849,
+      "step": 26201
+    },
+    {
+      "epoch": 0.7196176979484222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.847607395619337e-06,
+      "loss": 0.7628,
+      "step": 26202
+    },
+    {
+      "epoch": 0.7196451621762654,
+      "grad_norm": 0.0,
+      "learning_rate": 3.846906178513757e-06,
+      "loss": 0.8027,
+      "step": 26203
+    },
+    {
+      "epoch": 0.7196726264041087,
+      "grad_norm": 0.0,
+      "learning_rate": 3.846205010093843e-06,
+      "loss": 0.8124,
+      "step": 26204
+    },
+    {
+      "epoch": 0.7197000906319518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.845503890365138e-06,
+      "loss": 0.8744,
+      "step": 26205
+    },
+    {
+      "epoch": 0.7197275548597951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.844802819333187e-06,
+      "loss": 0.9647,
+      "step": 26206
+    },
+    {
+      "epoch": 0.7197550190876384,
+      "grad_norm": 0.0,
+      "learning_rate": 3.844101797003541e-06,
+      "loss": 0.8251,
+      "step": 26207
+    },
+    {
+      "epoch": 0.7197824833154816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8434008233817455e-06,
+      "loss": 0.8382,
+      "step": 26208
+    },
+    {
+      "epoch": 0.7198099475433248,
+      "grad_norm": 0.0,
+      "learning_rate": 3.842699898473347e-06,
+      "loss": 0.7128,
+      "step": 26209
+    },
+    {
+      "epoch": 0.719837411771168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8419990222838965e-06,
+      "loss": 0.8545,
+      "step": 26210
+    },
+    {
+      "epoch": 0.7198648759990113,
+      "grad_norm": 0.0,
+      "learning_rate": 3.841298194818931e-06,
+      "loss": 0.8176,
+      "step": 26211
+    },
+    {
+      "epoch": 0.7198923402268546,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8405974160839995e-06,
+      "loss": 0.8525,
+      "step": 26212
+    },
+    {
+      "epoch": 0.7199198044546977,
+      "grad_norm": 0.0,
+      "learning_rate": 3.839896686084652e-06,
+      "loss": 0.8455,
+      "step": 26213
+    },
+    {
+      "epoch": 0.719947268682541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8391960048264246e-06,
+      "loss": 0.8656,
+      "step": 26214
+    },
+    {
+      "epoch": 0.7199747329103843,
+      "grad_norm": 0.0,
+      "learning_rate": 3.838495372314864e-06,
+      "loss": 0.8886,
+      "step": 26215
+    },
+    {
+      "epoch": 0.7200021971382274,
+      "grad_norm": 0.0,
+      "learning_rate": 3.837794788555518e-06,
+      "loss": 0.8102,
+      "step": 26216
+    },
+    {
+      "epoch": 0.7200296613660707,
+      "grad_norm": 0.0,
+      "learning_rate": 3.837094253553923e-06,
+      "loss": 0.7785,
+      "step": 26217
+    },
+    {
+      "epoch": 0.7200571255939139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.836393767315626e-06,
+      "loss": 0.7657,
+      "step": 26218
+    },
+    {
+      "epoch": 0.7200845898217572,
+      "grad_norm": 0.0,
+      "learning_rate": 3.835693329846169e-06,
+      "loss": 0.8362,
+      "step": 26219
+    },
+    {
+      "epoch": 0.7201120540496004,
+      "grad_norm": 0.0,
+      "learning_rate": 3.834992941151097e-06,
+      "loss": 0.8442,
+      "step": 26220
+    },
+    {
+      "epoch": 0.7201395182774436,
+      "grad_norm": 0.0,
+      "learning_rate": 3.834292601235947e-06,
+      "loss": 0.8844,
+      "step": 26221
+    },
+    {
+      "epoch": 0.7201669825052869,
+      "grad_norm": 0.0,
+      "learning_rate": 3.833592310106266e-06,
+      "loss": 0.8973,
+      "step": 26222
+    },
+    {
+      "epoch": 0.72019444673313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.832892067767587e-06,
+      "loss": 0.8359,
+      "step": 26223
+    },
+    {
+      "epoch": 0.7202219109609733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.832191874225454e-06,
+      "loss": 0.7648,
+      "step": 26224
+    },
+    {
+      "epoch": 0.7202493751888166,
+      "grad_norm": 0.0,
+      "learning_rate": 3.83149172948541e-06,
+      "loss": 0.7207,
+      "step": 26225
+    },
+    {
+      "epoch": 0.7202768394166598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.830791633552991e-06,
+      "loss": 0.8889,
+      "step": 26226
+    },
+    {
+      "epoch": 0.720304303644503,
+      "grad_norm": 0.0,
+      "learning_rate": 3.830091586433742e-06,
+      "loss": 0.8751,
+      "step": 26227
+    },
+    {
+      "epoch": 0.7203317678723463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.829391588133196e-06,
+      "loss": 0.8444,
+      "step": 26228
+    },
+    {
+      "epoch": 0.7203592321001895,
+      "grad_norm": 0.0,
+      "learning_rate": 3.828691638656896e-06,
+      "loss": 0.8533,
+      "step": 26229
+    },
+    {
+      "epoch": 0.7203866963280328,
+      "grad_norm": 0.0,
+      "learning_rate": 3.827991738010376e-06,
+      "loss": 0.8222,
+      "step": 26230
+    },
+    {
+      "epoch": 0.7204141605558759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.827291886199176e-06,
+      "loss": 0.7623,
+      "step": 26231
+    },
+    {
+      "epoch": 0.7204416247837192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.826592083228834e-06,
+      "loss": 0.825,
+      "step": 26232
+    },
+    {
+      "epoch": 0.7204690890115625,
+      "grad_norm": 0.0,
+      "learning_rate": 3.82589232910489e-06,
+      "loss": 0.8146,
+      "step": 26233
+    },
+    {
+      "epoch": 0.7204965532394056,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8251926238328725e-06,
+      "loss": 0.8666,
+      "step": 26234
+    },
+    {
+      "epoch": 0.7205240174672489,
+      "grad_norm": 0.0,
+      "learning_rate": 3.824492967418325e-06,
+      "loss": 0.7527,
+      "step": 26235
+    },
+    {
+      "epoch": 0.7205514816950921,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8237933598667805e-06,
+      "loss": 0.8522,
+      "step": 26236
+    },
+    {
+      "epoch": 0.7205789459229354,
+      "grad_norm": 0.0,
+      "learning_rate": 3.82309380118378e-06,
+      "loss": 0.8679,
+      "step": 26237
+    },
+    {
+      "epoch": 0.7206064101507786,
+      "grad_norm": 0.0,
+      "learning_rate": 3.822394291374848e-06,
+      "loss": 0.8743,
+      "step": 26238
+    },
+    {
+      "epoch": 0.7206338743786218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.82169483044553e-06,
+      "loss": 0.8475,
+      "step": 26239
+    },
+    {
+      "epoch": 0.7206613386064651,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8209954184013535e-06,
+      "loss": 0.8878,
+      "step": 26240
+    },
+    {
+      "epoch": 0.7206888028343084,
+      "grad_norm": 0.0,
+      "learning_rate": 3.820296055247853e-06,
+      "loss": 0.8963,
+      "step": 26241
+    },
+    {
+      "epoch": 0.7207162670621515,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8195967409905655e-06,
+      "loss": 0.8174,
+      "step": 26242
+    },
+    {
+      "epoch": 0.7207437312899948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.818897475635022e-06,
+      "loss": 0.8124,
+      "step": 26243
+    },
+    {
+      "epoch": 0.720771195517838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.818198259186759e-06,
+      "loss": 0.8586,
+      "step": 26244
+    },
+    {
+      "epoch": 0.7207986597456812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.817499091651305e-06,
+      "loss": 0.7928,
+      "step": 26245
+    },
+    {
+      "epoch": 0.7208261239735245,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8167999730341924e-06,
+      "loss": 0.8806,
+      "step": 26246
+    },
+    {
+      "epoch": 0.7208535882013677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.816100903340951e-06,
+      "loss": 0.7824,
+      "step": 26247
+    },
+    {
+      "epoch": 0.720881052429211,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8154018825771155e-06,
+      "loss": 0.748,
+      "step": 26248
+    },
+    {
+      "epoch": 0.7209085166570541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.814702910748217e-06,
+      "loss": 0.8559,
+      "step": 26249
+    },
+    {
+      "epoch": 0.7209359808848974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8140039878597833e-06,
+      "loss": 0.8384,
+      "step": 26250
+    },
+    {
+      "epoch": 0.7209634451127407,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8133051139173517e-06,
+      "loss": 0.7977,
+      "step": 26251
+    },
+    {
+      "epoch": 0.7209909093405839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8126062889264415e-06,
+      "loss": 0.766,
+      "step": 26252
+    },
+    {
+      "epoch": 0.7210183735684271,
+      "grad_norm": 0.0,
+      "learning_rate": 3.811907512892591e-06,
+      "loss": 0.7339,
+      "step": 26253
+    },
+    {
+      "epoch": 0.7210458377962704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.811208785821322e-06,
+      "loss": 0.8085,
+      "step": 26254
+    },
+    {
+      "epoch": 0.7210733020241136,
+      "grad_norm": 0.0,
+      "learning_rate": 3.810510107718166e-06,
+      "loss": 0.8453,
+      "step": 26255
+    },
+    {
+      "epoch": 0.7211007662519568,
+      "grad_norm": 0.0,
+      "learning_rate": 3.809811478588652e-06,
+      "loss": 0.8316,
+      "step": 26256
+    },
+    {
+      "epoch": 0.7211282304798,
+      "grad_norm": 0.0,
+      "learning_rate": 3.809112898438311e-06,
+      "loss": 0.829,
+      "step": 26257
+    },
+    {
+      "epoch": 0.7211556947076433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8084143672726637e-06,
+      "loss": 0.6954,
+      "step": 26258
+    },
+    {
+      "epoch": 0.7211831589354866,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8077158850972397e-06,
+      "loss": 0.8546,
+      "step": 26259
+    },
+    {
+      "epoch": 0.7212106231633297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8070174519175696e-06,
+      "loss": 0.8822,
+      "step": 26260
+    },
+    {
+      "epoch": 0.721238087391173,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8063190677391727e-06,
+      "loss": 0.8591,
+      "step": 26261
+    },
+    {
+      "epoch": 0.7212655516190162,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8056207325675797e-06,
+      "loss": 0.8213,
+      "step": 26262
+    },
+    {
+      "epoch": 0.7212930158468595,
+      "grad_norm": 0.0,
+      "learning_rate": 3.804922446408318e-06,
+      "loss": 0.8085,
+      "step": 26263
+    },
+    {
+      "epoch": 0.7213204800747027,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8042242092669056e-06,
+      "loss": 0.8606,
+      "step": 26264
+    },
+    {
+      "epoch": 0.7213479443025459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.803526021148871e-06,
+      "loss": 0.8546,
+      "step": 26265
+    },
+    {
+      "epoch": 0.7213754085303892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80282788205974e-06,
+      "loss": 0.7759,
+      "step": 26266
+    },
+    {
+      "epoch": 0.7214028727582324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.802129792005035e-06,
+      "loss": 0.8285,
+      "step": 26267
+    },
+    {
+      "epoch": 0.7214303369860756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8014317509902822e-06,
+      "loss": 0.7828,
+      "step": 26268
+    },
+    {
+      "epoch": 0.7214578012139189,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8007337590210035e-06,
+      "loss": 0.8282,
+      "step": 26269
+    },
+    {
+      "epoch": 0.7214852654417621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8000358161027175e-06,
+      "loss": 0.8699,
+      "step": 26270
+    },
+    {
+      "epoch": 0.7215127296696053,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7993379222409486e-06,
+      "loss": 0.777,
+      "step": 26271
+    },
+    {
+      "epoch": 0.7215401938974486,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7986400774412215e-06,
+      "loss": 0.8177,
+      "step": 26272
+    },
+    {
+      "epoch": 0.7215676581252918,
+      "grad_norm": 0.0,
+      "learning_rate": 3.797942281709055e-06,
+      "loss": 0.8477,
+      "step": 26273
+    },
+    {
+      "epoch": 0.721595122353135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.797244535049972e-06,
+      "loss": 0.6984,
+      "step": 26274
+    },
+    {
+      "epoch": 0.7216225865809783,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7965468374694968e-06,
+      "loss": 0.7838,
+      "step": 26275
+    },
+    {
+      "epoch": 0.7216500508088215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.795849188973142e-06,
+      "loss": 0.7989,
+      "step": 26276
+    },
+    {
+      "epoch": 0.7216775150366648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.795151589566436e-06,
+      "loss": 0.7214,
+      "step": 26277
+    },
+    {
+      "epoch": 0.7217049792645079,
+      "grad_norm": 0.0,
+      "learning_rate": 3.79445403925489e-06,
+      "loss": 0.8825,
+      "step": 26278
+    },
+    {
+      "epoch": 0.7217324434923512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7937565380440268e-06,
+      "loss": 0.8324,
+      "step": 26279
+    },
+    {
+      "epoch": 0.7217599077201945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7930590859393666e-06,
+      "loss": 0.8367,
+      "step": 26280
+    },
+    {
+      "epoch": 0.7217873719480377,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7923616829464314e-06,
+      "loss": 0.891,
+      "step": 26281
+    },
+    {
+      "epoch": 0.7218148361758809,
+      "grad_norm": 0.0,
+      "learning_rate": 3.791664329070731e-06,
+      "loss": 0.8916,
+      "step": 26282
+    },
+    {
+      "epoch": 0.7218423004037241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7909670243177877e-06,
+      "loss": 0.7459,
+      "step": 26283
+    },
+    {
+      "epoch": 0.7218697646315674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.790269768693121e-06,
+      "loss": 0.832,
+      "step": 26284
+    },
+    {
+      "epoch": 0.7218972288594107,
+      "grad_norm": 0.0,
+      "learning_rate": 3.789572562202243e-06,
+      "loss": 0.8239,
+      "step": 26285
+    },
+    {
+      "epoch": 0.7219246930872538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7888754048506747e-06,
+      "loss": 0.8452,
+      "step": 26286
+    },
+    {
+      "epoch": 0.7219521573150971,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7881782966439285e-06,
+      "loss": 0.7822,
+      "step": 26287
+    },
+    {
+      "epoch": 0.7219796215429404,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7874812375875204e-06,
+      "loss": 0.8764,
+      "step": 26288
+    },
+    {
+      "epoch": 0.7220070857707835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.786784227686967e-06,
+      "loss": 0.7851,
+      "step": 26289
+    },
+    {
+      "epoch": 0.7220345499986268,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7860872669477844e-06,
+      "loss": 0.9456,
+      "step": 26290
+    },
+    {
+      "epoch": 0.72206201422647,
+      "grad_norm": 0.0,
+      "learning_rate": 3.785390355375487e-06,
+      "loss": 0.8371,
+      "step": 26291
+    },
+    {
+      "epoch": 0.7220894784543133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7846934929755906e-06,
+      "loss": 0.7922,
+      "step": 26292
+    },
+    {
+      "epoch": 0.7221169426821565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7839966797536078e-06,
+      "loss": 0.87,
+      "step": 26293
+    },
+    {
+      "epoch": 0.7221444069099997,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7832999157150475e-06,
+      "loss": 0.7805,
+      "step": 26294
+    },
+    {
+      "epoch": 0.722171871137843,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7826032008654256e-06,
+      "loss": 0.7815,
+      "step": 26295
+    },
+    {
+      "epoch": 0.7221993353656861,
+      "grad_norm": 0.0,
+      "learning_rate": 3.781906535210256e-06,
+      "loss": 0.8013,
+      "step": 26296
+    },
+    {
+      "epoch": 0.7222267995935294,
+      "grad_norm": 0.0,
+      "learning_rate": 3.781209918755051e-06,
+      "loss": 0.9178,
+      "step": 26297
+    },
+    {
+      "epoch": 0.7222542638213727,
+      "grad_norm": 0.0,
+      "learning_rate": 3.780513351505325e-06,
+      "loss": 0.7976,
+      "step": 26298
+    },
+    {
+      "epoch": 0.7222817280492159,
+      "grad_norm": 0.0,
+      "learning_rate": 3.779816833466583e-06,
+      "loss": 0.8941,
+      "step": 26299
+    },
+    {
+      "epoch": 0.7223091922770591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7791203646443398e-06,
+      "loss": 0.886,
+      "step": 26300
+    },
+    {
+      "epoch": 0.7223366565049024,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7784239450441097e-06,
+      "loss": 0.7752,
+      "step": 26301
+    },
+    {
+      "epoch": 0.7223641207327456,
+      "grad_norm": 0.0,
+      "learning_rate": 3.777727574671396e-06,
+      "loss": 0.8701,
+      "step": 26302
+    },
+    {
+      "epoch": 0.7223915849605889,
+      "grad_norm": 0.0,
+      "learning_rate": 3.777031253531712e-06,
+      "loss": 0.767,
+      "step": 26303
+    },
+    {
+      "epoch": 0.722419049188432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7763349816305706e-06,
+      "loss": 0.8186,
+      "step": 26304
+    },
+    {
+      "epoch": 0.7224465134162753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7756387589734735e-06,
+      "loss": 0.8485,
+      "step": 26305
+    },
+    {
+      "epoch": 0.7224739776441186,
+      "grad_norm": 0.0,
+      "learning_rate": 3.774942585565935e-06,
+      "loss": 0.7874,
+      "step": 26306
+    },
+    {
+      "epoch": 0.7225014418719617,
+      "grad_norm": 0.0,
+      "learning_rate": 3.774246461413461e-06,
+      "loss": 0.8448,
+      "step": 26307
+    },
+    {
+      "epoch": 0.722528906099805,
+      "grad_norm": 0.0,
+      "learning_rate": 3.773550386521564e-06,
+      "loss": 0.8057,
+      "step": 26308
+    },
+    {
+      "epoch": 0.7225563703276482,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7728543608957458e-06,
+      "loss": 0.8188,
+      "step": 26309
+    },
+    {
+      "epoch": 0.7225838345554915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7721583845415186e-06,
+      "loss": 0.8175,
+      "step": 26310
+    },
+    {
+      "epoch": 0.7226112987833347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.771462457464383e-06,
+      "loss": 0.8583,
+      "step": 26311
+    },
+    {
+      "epoch": 0.7226387630111779,
+      "grad_norm": 0.0,
+      "learning_rate": 3.77076657966985e-06,
+      "loss": 0.795,
+      "step": 26312
+    },
+    {
+      "epoch": 0.7226662272390212,
+      "grad_norm": 0.0,
+      "learning_rate": 3.770070751163425e-06,
+      "loss": 0.7938,
+      "step": 26313
+    },
+    {
+      "epoch": 0.7226936914668645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7693749719506134e-06,
+      "loss": 0.7526,
+      "step": 26314
+    },
+    {
+      "epoch": 0.7227211556947076,
+      "grad_norm": 0.0,
+      "learning_rate": 3.76867924203692e-06,
+      "loss": 0.8517,
+      "step": 26315
+    },
+    {
+      "epoch": 0.7227486199225509,
+      "grad_norm": 0.0,
+      "learning_rate": 3.767983561427854e-06,
+      "loss": 0.8297,
+      "step": 26316
+    },
+    {
+      "epoch": 0.7227760841503941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7672879301289154e-06,
+      "loss": 0.9174,
+      "step": 26317
+    },
+    {
+      "epoch": 0.7228035483782373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7665923481456057e-06,
+      "loss": 0.88,
+      "step": 26318
+    },
+    {
+      "epoch": 0.7228310126060806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.765896815483433e-06,
+      "loss": 0.8027,
+      "step": 26319
+    },
+    {
+      "epoch": 0.7228584768339238,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7652013321478984e-06,
+      "loss": 0.8612,
+      "step": 26320
+    },
+    {
+      "epoch": 0.7228859410617671,
+      "grad_norm": 0.0,
+      "learning_rate": 3.764505898144507e-06,
+      "loss": 0.8315,
+      "step": 26321
+    },
+    {
+      "epoch": 0.7229134052896102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7638105134787627e-06,
+      "loss": 0.7653,
+      "step": 26322
+    },
+    {
+      "epoch": 0.7229408695174535,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7631151781561626e-06,
+      "loss": 0.8534,
+      "step": 26323
+    },
+    {
+      "epoch": 0.7229683337452968,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7624198921822107e-06,
+      "loss": 0.7948,
+      "step": 26324
+    },
+    {
+      "epoch": 0.72299579797314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.761724655562413e-06,
+      "loss": 0.8579,
+      "step": 26325
+    },
+    {
+      "epoch": 0.7230232622009832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.761029468302262e-06,
+      "loss": 0.7708,
+      "step": 26326
+    },
+    {
+      "epoch": 0.7230507264288265,
+      "grad_norm": 0.0,
+      "learning_rate": 3.760334330407263e-06,
+      "loss": 0.8448,
+      "step": 26327
+    },
+    {
+      "epoch": 0.7230781906566697,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7596392418829187e-06,
+      "loss": 0.7296,
+      "step": 26328
+    },
+    {
+      "epoch": 0.723105654884513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7589442027347233e-06,
+      "loss": 0.796,
+      "step": 26329
+    },
+    {
+      "epoch": 0.7231331191123561,
+      "grad_norm": 0.0,
+      "learning_rate": 3.758249212968179e-06,
+      "loss": 0.8196,
+      "step": 26330
+    },
+    {
+      "epoch": 0.7231605833401994,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7575542725887836e-06,
+      "loss": 0.8512,
+      "step": 26331
+    },
+    {
+      "epoch": 0.7231880475680427,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7568593816020417e-06,
+      "loss": 0.7563,
+      "step": 26332
+    },
+    {
+      "epoch": 0.7232155117958858,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7561645400134427e-06,
+      "loss": 0.7018,
+      "step": 26333
+    },
+    {
+      "epoch": 0.7232429760237291,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7554697478284917e-06,
+      "loss": 0.7038,
+      "step": 26334
+    },
+    {
+      "epoch": 0.7232704402515723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.75477500505268e-06,
+      "loss": 0.972,
+      "step": 26335
+    },
+    {
+      "epoch": 0.7232979044794156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.754080311691507e-06,
+      "loss": 0.8117,
+      "step": 26336
+    },
+    {
+      "epoch": 0.7233253687072588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.753385667750471e-06,
+      "loss": 0.8421,
+      "step": 26337
+    },
+    {
+      "epoch": 0.723352832935102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7526910732350663e-06,
+      "loss": 0.7725,
+      "step": 26338
+    },
+    {
+      "epoch": 0.7233802971629453,
+      "grad_norm": 0.0,
+      "learning_rate": 3.751996528150791e-06,
+      "loss": 0.8693,
+      "step": 26339
+    },
+    {
+      "epoch": 0.7234077613907886,
+      "grad_norm": 0.0,
+      "learning_rate": 3.751302032503141e-06,
+      "loss": 0.9533,
+      "step": 26340
+    },
+    {
+      "epoch": 0.7234352256186317,
+      "grad_norm": 0.0,
+      "learning_rate": 3.750607586297612e-06,
+      "loss": 0.7518,
+      "step": 26341
+    },
+    {
+      "epoch": 0.723462689846475,
+      "grad_norm": 0.0,
+      "learning_rate": 3.749913189539691e-06,
+      "loss": 0.9291,
+      "step": 26342
+    },
+    {
+      "epoch": 0.7234901540743182,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7492188422348797e-06,
+      "loss": 0.7859,
+      "step": 26343
+    },
+    {
+      "epoch": 0.7235176183021614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.74852454438867e-06,
+      "loss": 0.8487,
+      "step": 26344
+    },
+    {
+      "epoch": 0.7235450825300047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.747830296006555e-06,
+      "loss": 0.9012,
+      "step": 26345
+    },
+    {
+      "epoch": 0.7235725467578479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7471360970940317e-06,
+      "loss": 0.7852,
+      "step": 26346
+    },
+    {
+      "epoch": 0.7236000109856912,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7464419476565874e-06,
+      "loss": 0.8631,
+      "step": 26347
+    },
+    {
+      "epoch": 0.7236274752135343,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7457478476997157e-06,
+      "loss": 0.8174,
+      "step": 26348
+    },
+    {
+      "epoch": 0.7236549394413776,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7450537972289136e-06,
+      "loss": 0.8222,
+      "step": 26349
+    },
+    {
+      "epoch": 0.7236824036692209,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7443597962496646e-06,
+      "loss": 0.9067,
+      "step": 26350
+    },
+    {
+      "epoch": 0.723709867897064,
+      "grad_norm": 0.0,
+      "learning_rate": 3.743665844767469e-06,
+      "loss": 0.7782,
+      "step": 26351
+    },
+    {
+      "epoch": 0.7237373321249073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7429719427878085e-06,
+      "loss": 0.863,
+      "step": 26352
+    },
+    {
+      "epoch": 0.7237647963527506,
+      "grad_norm": 0.0,
+      "learning_rate": 3.742278090316177e-06,
+      "loss": 0.8208,
+      "step": 26353
+    },
+    {
+      "epoch": 0.7237922605805938,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7415842873580664e-06,
+      "loss": 0.7806,
+      "step": 26354
+    },
+    {
+      "epoch": 0.723819724808437,
+      "grad_norm": 0.0,
+      "learning_rate": 3.740890533918965e-06,
+      "loss": 0.8299,
+      "step": 26355
+    },
+    {
+      "epoch": 0.7238471890362802,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7401968300043655e-06,
+      "loss": 0.9201,
+      "step": 26356
+    },
+    {
+      "epoch": 0.7238746532641235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.739503175619753e-06,
+      "loss": 0.8358,
+      "step": 26357
+    },
+    {
+      "epoch": 0.7239021174919668,
+      "grad_norm": 0.0,
+      "learning_rate": 3.738809570770614e-06,
+      "loss": 0.8139,
+      "step": 26358
+    },
+    {
+      "epoch": 0.7239295817198099,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7381160154624385e-06,
+      "loss": 0.9067,
+      "step": 26359
+    },
+    {
+      "epoch": 0.7239570459476532,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7374225097007146e-06,
+      "loss": 0.748,
+      "step": 26360
+    },
+    {
+      "epoch": 0.7239845101754964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7367290534909305e-06,
+      "loss": 0.8953,
+      "step": 26361
+    },
+    {
+      "epoch": 0.7240119744033396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7360356468385717e-06,
+      "loss": 0.7726,
+      "step": 26362
+    },
+    {
+      "epoch": 0.7240394386311829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7353422897491287e-06,
+      "loss": 0.8138,
+      "step": 26363
+    },
+    {
+      "epoch": 0.7240669028590261,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7346489822280808e-06,
+      "loss": 0.8506,
+      "step": 26364
+    },
+    {
+      "epoch": 0.7240943670868694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.73395572428092e-06,
+      "loss": 0.7801,
+      "step": 26365
+    },
+    {
+      "epoch": 0.7241218313147126,
+      "grad_norm": 0.0,
+      "learning_rate": 3.733262515913125e-06,
+      "loss": 0.8231,
+      "step": 26366
+    },
+    {
+      "epoch": 0.7241492955425558,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7325693571301858e-06,
+      "loss": 0.8828,
+      "step": 26367
+    },
+    {
+      "epoch": 0.7241767597703991,
+      "grad_norm": 0.0,
+      "learning_rate": 3.731876247937586e-06,
+      "loss": 0.8176,
+      "step": 26368
+    },
+    {
+      "epoch": 0.7242042239982422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7311831883408113e-06,
+      "loss": 0.913,
+      "step": 26369
+    },
+    {
+      "epoch": 0.7242316882260855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7304901783453407e-06,
+      "loss": 0.7461,
+      "step": 26370
+    },
+    {
+      "epoch": 0.7242591524539288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.72979721795666e-06,
+      "loss": 0.7751,
+      "step": 26371
+    },
+    {
+      "epoch": 0.724286616681772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7291043071802533e-06,
+      "loss": 0.8405,
+      "step": 26372
+    },
+    {
+      "epoch": 0.7243140809096152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7284114460216057e-06,
+      "loss": 0.8003,
+      "step": 26373
+    },
+    {
+      "epoch": 0.7243415451374584,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7277186344861926e-06,
+      "loss": 0.8989,
+      "step": 26374
+    },
+    {
+      "epoch": 0.7243690093653017,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7270258725795017e-06,
+      "loss": 0.8091,
+      "step": 26375
+    },
+    {
+      "epoch": 0.724396473593145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7263331603070094e-06,
+      "loss": 0.834,
+      "step": 26376
+    },
+    {
+      "epoch": 0.7244239378209881,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7256404976742e-06,
+      "loss": 0.7802,
+      "step": 26377
+    },
+    {
+      "epoch": 0.7244514020488314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7249478846865526e-06,
+      "loss": 0.7567,
+      "step": 26378
+    },
+    {
+      "epoch": 0.7244788662766747,
+      "grad_norm": 0.0,
+      "learning_rate": 3.724255321349549e-06,
+      "loss": 0.864,
+      "step": 26379
+    },
+    {
+      "epoch": 0.7245063305045178,
+      "grad_norm": 0.0,
+      "learning_rate": 3.723562807668671e-06,
+      "loss": 0.7588,
+      "step": 26380
+    },
+    {
+      "epoch": 0.7245337947323611,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7228703436493952e-06,
+      "loss": 0.8254,
+      "step": 26381
+    },
+    {
+      "epoch": 0.7245612589602043,
+      "grad_norm": 0.0,
+      "learning_rate": 3.722177929297198e-06,
+      "loss": 0.8012,
+      "step": 26382
+    },
+    {
+      "epoch": 0.7245887231880476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7214855646175607e-06,
+      "loss": 0.8055,
+      "step": 26383
+    },
+    {
+      "epoch": 0.7246161874158908,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7207932496159615e-06,
+      "loss": 0.9343,
+      "step": 26384
+    },
+    {
+      "epoch": 0.724643651643734,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7201009842978786e-06,
+      "loss": 0.8306,
+      "step": 26385
+    },
+    {
+      "epoch": 0.7246711158715773,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7194087686687884e-06,
+      "loss": 0.8978,
+      "step": 26386
+    },
+    {
+      "epoch": 0.7246985800994205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.718716602734174e-06,
+      "loss": 0.855,
+      "step": 26387
+    },
+    {
+      "epoch": 0.7247260443272637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.718024486499502e-06,
+      "loss": 0.7773,
+      "step": 26388
+    },
+    {
+      "epoch": 0.724753508555107,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7173324199702575e-06,
+      "loss": 0.9325,
+      "step": 26389
+    },
+    {
+      "epoch": 0.7247809727829502,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7166404031519097e-06,
+      "loss": 0.8388,
+      "step": 26390
+    },
+    {
+      "epoch": 0.7248084370107934,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7159484360499366e-06,
+      "loss": 0.8695,
+      "step": 26391
+    },
+    {
+      "epoch": 0.7248359012386367,
+      "grad_norm": 0.0,
+      "learning_rate": 3.715256518669814e-06,
+      "loss": 0.9314,
+      "step": 26392
+    },
+    {
+      "epoch": 0.7248633654664799,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7145646510170196e-06,
+      "loss": 0.9597,
+      "step": 26393
+    },
+    {
+      "epoch": 0.7248908296943232,
+      "grad_norm": 0.0,
+      "learning_rate": 3.713872833097021e-06,
+      "loss": 0.8743,
+      "step": 26394
+    },
+    {
+      "epoch": 0.7249182939221663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7131810649152943e-06,
+      "loss": 0.7835,
+      "step": 26395
+    },
+    {
+      "epoch": 0.7249457581500096,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7124893464773193e-06,
+      "loss": 0.8946,
+      "step": 26396
+    },
+    {
+      "epoch": 0.7249732223778529,
+      "grad_norm": 0.0,
+      "learning_rate": 3.71179767778856e-06,
+      "loss": 0.8265,
+      "step": 26397
+    },
+    {
+      "epoch": 0.725000686605696,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7111060588544933e-06,
+      "loss": 0.7889,
+      "step": 26398
+    },
+    {
+      "epoch": 0.7250281508335393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.710414489680596e-06,
+      "loss": 0.8351,
+      "step": 26399
+    },
+    {
+      "epoch": 0.7250556150613825,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7097229702723316e-06,
+      "loss": 0.8944,
+      "step": 26400
+    },
+    {
+      "epoch": 0.7250830792892258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.709031500635175e-06,
+      "loss": 0.9043,
+      "step": 26401
+    },
+    {
+      "epoch": 0.725110543517069,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7083400807745983e-06,
+      "loss": 0.7889,
+      "step": 26402
+    },
+    {
+      "epoch": 0.7251380077449122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7076487106960724e-06,
+      "loss": 0.9447,
+      "step": 26403
+    },
+    {
+      "epoch": 0.7251654719727555,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7069573904050703e-06,
+      "loss": 0.8123,
+      "step": 26404
+    },
+    {
+      "epoch": 0.7251929362005988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7062661199070592e-06,
+      "loss": 0.833,
+      "step": 26405
+    },
+    {
+      "epoch": 0.7252204004284419,
+      "grad_norm": 0.0,
+      "learning_rate": 3.705574899207506e-06,
+      "loss": 0.8079,
+      "step": 26406
+    },
+    {
+      "epoch": 0.7252478646562852,
+      "grad_norm": 0.0,
+      "learning_rate": 3.704883728311881e-06,
+      "loss": 0.924,
+      "step": 26407
+    },
+    {
+      "epoch": 0.7252753288841284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.704192607225655e-06,
+      "loss": 0.8424,
+      "step": 26408
+    },
+    {
+      "epoch": 0.7253027931119717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7035015359542956e-06,
+      "loss": 0.8975,
+      "step": 26409
+    },
+    {
+      "epoch": 0.7253302573398149,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7028105145032754e-06,
+      "loss": 0.8285,
+      "step": 26410
+    },
+    {
+      "epoch": 0.7253577215676581,
+      "grad_norm": 0.0,
+      "learning_rate": 3.702119542878053e-06,
+      "loss": 0.8517,
+      "step": 26411
+    },
+    {
+      "epoch": 0.7253851857955014,
+      "grad_norm": 0.0,
+      "learning_rate": 3.701428621084102e-06,
+      "loss": 0.7743,
+      "step": 26412
+    },
+    {
+      "epoch": 0.7254126500233445,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7007377491268902e-06,
+      "loss": 0.8241,
+      "step": 26413
+    },
+    {
+      "epoch": 0.7254401142511878,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7000469270118777e-06,
+      "loss": 0.8059,
+      "step": 26414
+    },
+    {
+      "epoch": 0.7254675784790311,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6993561547445344e-06,
+      "loss": 0.7627,
+      "step": 26415
+    },
+    {
+      "epoch": 0.7254950427068743,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6986654323303295e-06,
+      "loss": 0.7492,
+      "step": 26416
+    },
+    {
+      "epoch": 0.7255225069347175,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6979747597747205e-06,
+      "loss": 0.7707,
+      "step": 26417
+    },
+    {
+      "epoch": 0.7255499711625608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.697284137083177e-06,
+      "loss": 0.9403,
+      "step": 26418
+    },
+    {
+      "epoch": 0.725577435390404,
+      "grad_norm": 0.0,
+      "learning_rate": 3.696593564261163e-06,
+      "loss": 0.8016,
+      "step": 26419
+    },
+    {
+      "epoch": 0.7256048996182473,
+      "grad_norm": 0.0,
+      "learning_rate": 3.695903041314145e-06,
+      "loss": 0.8419,
+      "step": 26420
+    },
+    {
+      "epoch": 0.7256323638460904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6952125682475814e-06,
+      "loss": 0.7846,
+      "step": 26421
+    },
+    {
+      "epoch": 0.7256598280739337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.694522145066941e-06,
+      "loss": 0.7133,
+      "step": 26422
+    },
+    {
+      "epoch": 0.725687292301777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6938317717776815e-06,
+      "loss": 0.8051,
+      "step": 26423
+    },
+    {
+      "epoch": 0.7257147565296201,
+      "grad_norm": 0.0,
+      "learning_rate": 3.693141448385267e-06,
+      "loss": 0.8177,
+      "step": 26424
+    },
+    {
+      "epoch": 0.7257422207574634,
+      "grad_norm": 0.0,
+      "learning_rate": 3.692451174895161e-06,
+      "loss": 0.8789,
+      "step": 26425
+    },
+    {
+      "epoch": 0.7257696849853066,
+      "grad_norm": 0.0,
+      "learning_rate": 3.691760951312824e-06,
+      "loss": 0.778,
+      "step": 26426
+    },
+    {
+      "epoch": 0.7257971492131499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6910707776437182e-06,
+      "loss": 0.7856,
+      "step": 26427
+    },
+    {
+      "epoch": 0.7258246134409931,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6903806538933075e-06,
+      "loss": 0.8037,
+      "step": 26428
+    },
+    {
+      "epoch": 0.7258520776688363,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68969058006705e-06,
+      "loss": 0.7357,
+      "step": 26429
+    },
+    {
+      "epoch": 0.7258795418966796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6890005561704002e-06,
+      "loss": 0.7832,
+      "step": 26430
+    },
+    {
+      "epoch": 0.7259070061245229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.688310582208824e-06,
+      "loss": 0.824,
+      "step": 26431
+    },
+    {
+      "epoch": 0.725934470352366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.687620658187778e-06,
+      "loss": 0.8229,
+      "step": 26432
+    },
+    {
+      "epoch": 0.7259619345802093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6869307841127232e-06,
+      "loss": 0.7866,
+      "step": 26433
+    },
+    {
+      "epoch": 0.7259893988080525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6862409599891214e-06,
+      "loss": 0.8245,
+      "step": 26434
+    },
+    {
+      "epoch": 0.7260168630358957,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6855511858224234e-06,
+      "loss": 0.9413,
+      "step": 26435
+    },
+    {
+      "epoch": 0.726044327263739,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68486146161809e-06,
+      "loss": 0.8301,
+      "step": 26436
+    },
+    {
+      "epoch": 0.7260717914915822,
+      "grad_norm": 0.0,
+      "learning_rate": 3.684171787381582e-06,
+      "loss": 0.8253,
+      "step": 26437
+    },
+    {
+      "epoch": 0.7260992557194255,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6834821631183506e-06,
+      "loss": 0.9348,
+      "step": 26438
+    },
+    {
+      "epoch": 0.7261267199472686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6827925888338543e-06,
+      "loss": 0.8465,
+      "step": 26439
+    },
+    {
+      "epoch": 0.7261541841751119,
+      "grad_norm": 0.0,
+      "learning_rate": 3.682103064533554e-06,
+      "loss": 0.8018,
+      "step": 26440
+    },
+    {
+      "epoch": 0.7261816484029552,
+      "grad_norm": 0.0,
+      "learning_rate": 3.681413590222899e-06,
+      "loss": 0.8411,
+      "step": 26441
+    },
+    {
+      "epoch": 0.7262091126307983,
+      "grad_norm": 0.0,
+      "learning_rate": 3.680724165907347e-06,
+      "loss": 0.8282,
+      "step": 26442
+    },
+    {
+      "epoch": 0.7262365768586416,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6800347915923518e-06,
+      "loss": 0.8246,
+      "step": 26443
+    },
+    {
+      "epoch": 0.7262640410864849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.679345467283374e-06,
+      "loss": 0.841,
+      "step": 26444
+    },
+    {
+      "epoch": 0.7262915053143281,
+      "grad_norm": 0.0,
+      "learning_rate": 3.67865619298586e-06,
+      "loss": 0.8063,
+      "step": 26445
+    },
+    {
+      "epoch": 0.7263189695421713,
+      "grad_norm": 0.0,
+      "learning_rate": 3.67796696870527e-06,
+      "loss": 0.7978,
+      "step": 26446
+    },
+    {
+      "epoch": 0.7263464337700145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6772777944470505e-06,
+      "loss": 0.8234,
+      "step": 26447
+    },
+    {
+      "epoch": 0.7263738979978578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6765886702166586e-06,
+      "loss": 0.812,
+      "step": 26448
+    },
+    {
+      "epoch": 0.7264013622257011,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6758995960195466e-06,
+      "loss": 0.8186,
+      "step": 26449
+    },
+    {
+      "epoch": 0.7264288264535442,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6752105718611664e-06,
+      "loss": 0.8359,
+      "step": 26450
+    },
+    {
+      "epoch": 0.7264562906813875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6745215977469704e-06,
+      "loss": 0.7925,
+      "step": 26451
+    },
+    {
+      "epoch": 0.7264837549092307,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6738326736824126e-06,
+      "loss": 0.8803,
+      "step": 26452
+    },
+    {
+      "epoch": 0.726511219137074,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6731437996729425e-06,
+      "loss": 0.8106,
+      "step": 26453
+    },
+    {
+      "epoch": 0.7265386833649172,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6724549757240046e-06,
+      "loss": 0.7986,
+      "step": 26454
+    },
+    {
+      "epoch": 0.7265661475927604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.671766201841055e-06,
+      "loss": 0.8254,
+      "step": 26455
+    },
+    {
+      "epoch": 0.7265936118206037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.671077478029542e-06,
+      "loss": 0.7748,
+      "step": 26456
+    },
+    {
+      "epoch": 0.726621076048447,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6703888042949166e-06,
+      "loss": 0.8768,
+      "step": 26457
+    },
+    {
+      "epoch": 0.7266485402762901,
+      "grad_norm": 0.0,
+      "learning_rate": 3.669700180642629e-06,
+      "loss": 0.7811,
+      "step": 26458
+    },
+    {
+      "epoch": 0.7266760045041334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6690116070781225e-06,
+      "loss": 0.7965,
+      "step": 26459
+    },
+    {
+      "epoch": 0.7267034687319766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6683230836068484e-06,
+      "loss": 0.8358,
+      "step": 26460
+    },
+    {
+      "epoch": 0.7267309329598198,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6676346102342586e-06,
+      "loss": 0.9621,
+      "step": 26461
+    },
+    {
+      "epoch": 0.7267583971876631,
+      "grad_norm": 0.0,
+      "learning_rate": 3.666946186965794e-06,
+      "loss": 0.8307,
+      "step": 26462
+    },
+    {
+      "epoch": 0.7267858614155063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6662578138069028e-06,
+      "loss": 0.8257,
+      "step": 26463
+    },
+    {
+      "epoch": 0.7268133256433496,
+      "grad_norm": 0.0,
+      "learning_rate": 3.665569490763038e-06,
+      "loss": 0.8705,
+      "step": 26464
+    },
+    {
+      "epoch": 0.7268407898711927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6648812178396364e-06,
+      "loss": 0.8852,
+      "step": 26465
+    },
+    {
+      "epoch": 0.726868254099036,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6641929950421497e-06,
+      "loss": 0.8642,
+      "step": 26466
+    },
+    {
+      "epoch": 0.7268957183268793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6635048223760217e-06,
+      "loss": 0.7818,
+      "step": 26467
+    },
+    {
+      "epoch": 0.7269231825547224,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6628166998467018e-06,
+      "loss": 0.8183,
+      "step": 26468
+    },
+    {
+      "epoch": 0.7269506467825657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.662128627459628e-06,
+      "loss": 0.7262,
+      "step": 26469
+    },
+    {
+      "epoch": 0.726978111010409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.66144060522025e-06,
+      "loss": 0.8866,
+      "step": 26470
+    },
+    {
+      "epoch": 0.7270055752382522,
+      "grad_norm": 0.0,
+      "learning_rate": 3.660752633134006e-06,
+      "loss": 0.8698,
+      "step": 26471
+    },
+    {
+      "epoch": 0.7270330394660954,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6600647112063435e-06,
+      "loss": 0.9138,
+      "step": 26472
+    },
+    {
+      "epoch": 0.7270605036939386,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6593768394427043e-06,
+      "loss": 0.8331,
+      "step": 26473
+    },
+    {
+      "epoch": 0.7270879679217819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6586890178485314e-06,
+      "loss": 0.9154,
+      "step": 26474
+    },
+    {
+      "epoch": 0.7271154321496252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6580012464292712e-06,
+      "loss": 0.8597,
+      "step": 26475
+    },
+    {
+      "epoch": 0.7271428963774683,
+      "grad_norm": 0.0,
+      "learning_rate": 3.657313525190358e-06,
+      "loss": 0.8367,
+      "step": 26476
+    },
+    {
+      "epoch": 0.7271703606053116,
+      "grad_norm": 0.0,
+      "learning_rate": 3.656625854137241e-06,
+      "loss": 0.7987,
+      "step": 26477
+    },
+    {
+      "epoch": 0.7271978248331549,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6559382332753536e-06,
+      "loss": 0.8557,
+      "step": 26478
+    },
+    {
+      "epoch": 0.727225289060998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6552506626101415e-06,
+      "loss": 0.8616,
+      "step": 26479
+    },
+    {
+      "epoch": 0.7272527532888413,
+      "grad_norm": 0.0,
+      "learning_rate": 3.654563142147043e-06,
+      "loss": 0.7825,
+      "step": 26480
+    },
+    {
+      "epoch": 0.7272802175166845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6538756718915016e-06,
+      "loss": 0.757,
+      "step": 26481
+    },
+    {
+      "epoch": 0.7273076817445278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6531882518489513e-06,
+      "loss": 0.8514,
+      "step": 26482
+    },
+    {
+      "epoch": 0.727335145972371,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6525008820248344e-06,
+      "loss": 0.8906,
+      "step": 26483
+    },
+    {
+      "epoch": 0.7273626102002142,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6518135624245886e-06,
+      "loss": 0.9044,
+      "step": 26484
+    },
+    {
+      "epoch": 0.7273900744280575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6511262930536575e-06,
+      "loss": 0.7556,
+      "step": 26485
+    },
+    {
+      "epoch": 0.7274175386559006,
+      "grad_norm": 0.0,
+      "learning_rate": 3.65043907391747e-06,
+      "loss": 0.8375,
+      "step": 26486
+    },
+    {
+      "epoch": 0.7274450028837439,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6497519050214724e-06,
+      "loss": 0.7899,
+      "step": 26487
+    },
+    {
+      "epoch": 0.7274724671115872,
+      "grad_norm": 0.0,
+      "learning_rate": 3.649064786371094e-06,
+      "loss": 0.8908,
+      "step": 26488
+    },
+    {
+      "epoch": 0.7274999313394304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.648377717971774e-06,
+      "loss": 0.9088,
+      "step": 26489
+    },
+    {
+      "epoch": 0.7275273955672736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6476906998289506e-06,
+      "loss": 0.9586,
+      "step": 26490
+    },
+    {
+      "epoch": 0.7275548597951169,
+      "grad_norm": 0.0,
+      "learning_rate": 3.64700373194806e-06,
+      "loss": 0.7476,
+      "step": 26491
+    },
+    {
+      "epoch": 0.7275823240229601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6463168143345394e-06,
+      "loss": 0.7994,
+      "step": 26492
+    },
+    {
+      "epoch": 0.7276097882508034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6456299469938173e-06,
+      "loss": 0.8006,
+      "step": 26493
+    },
+    {
+      "epoch": 0.7276372524786465,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6449431299313365e-06,
+      "loss": 0.8568,
+      "step": 26494
+    },
+    {
+      "epoch": 0.7276647167064898,
+      "grad_norm": 0.0,
+      "learning_rate": 3.644256363152524e-06,
+      "loss": 0.744,
+      "step": 26495
+    },
+    {
+      "epoch": 0.7276921809343331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.643569646662818e-06,
+      "loss": 0.8935,
+      "step": 26496
+    },
+    {
+      "epoch": 0.7277196451621762,
+      "grad_norm": 0.0,
+      "learning_rate": 3.642882980467649e-06,
+      "loss": 0.8499,
+      "step": 26497
+    },
+    {
+      "epoch": 0.7277471093900195,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6421963645724545e-06,
+      "loss": 0.844,
+      "step": 26498
+    },
+    {
+      "epoch": 0.7277745736178627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6415097989826674e-06,
+      "loss": 0.8108,
+      "step": 26499
+    },
+    {
+      "epoch": 0.727802037845706,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6408232837037148e-06,
+      "loss": 0.924,
+      "step": 26500
+    },
+    {
+      "epoch": 0.7278295020735492,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6401368187410344e-06,
+      "loss": 0.8362,
+      "step": 26501
+    },
+    {
+      "epoch": 0.7278569663013924,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6394504041000534e-06,
+      "loss": 0.873,
+      "step": 26502
+    },
+    {
+      "epoch": 0.7278844305292357,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6387640397862036e-06,
+      "loss": 0.8346,
+      "step": 26503
+    },
+    {
+      "epoch": 0.727911894757079,
+      "grad_norm": 0.0,
+      "learning_rate": 3.638077725804916e-06,
+      "loss": 0.8585,
+      "step": 26504
+    },
+    {
+      "epoch": 0.7279393589849221,
+      "grad_norm": 0.0,
+      "learning_rate": 3.637391462161627e-06,
+      "loss": 0.8174,
+      "step": 26505
+    },
+    {
+      "epoch": 0.7279668232127654,
+      "grad_norm": 0.0,
+      "learning_rate": 3.636705248861756e-06,
+      "loss": 0.8396,
+      "step": 26506
+    },
+    {
+      "epoch": 0.7279942874406086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6360190859107392e-06,
+      "loss": 0.7922,
+      "step": 26507
+    },
+    {
+      "epoch": 0.7280217516684518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6353329733140042e-06,
+      "loss": 0.9138,
+      "step": 26508
+    },
+    {
+      "epoch": 0.7280492158962951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6346469110769834e-06,
+      "loss": 0.8235,
+      "step": 26509
+    },
+    {
+      "epoch": 0.7280766801241383,
+      "grad_norm": 0.0,
+      "learning_rate": 3.633960899205099e-06,
+      "loss": 0.8821,
+      "step": 26510
+    },
+    {
+      "epoch": 0.7281041443519816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6332749377037857e-06,
+      "loss": 0.8699,
+      "step": 26511
+    },
+    {
+      "epoch": 0.7281316085798247,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6325890265784625e-06,
+      "loss": 0.7219,
+      "step": 26512
+    },
+    {
+      "epoch": 0.728159072807668,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6319031658345628e-06,
+      "loss": 0.8425,
+      "step": 26513
+    },
+    {
+      "epoch": 0.7281865370355113,
+      "grad_norm": 0.0,
+      "learning_rate": 3.631217355477511e-06,
+      "loss": 0.8806,
+      "step": 26514
+    },
+    {
+      "epoch": 0.7282140012633544,
+      "grad_norm": 0.0,
+      "learning_rate": 3.630531595512735e-06,
+      "loss": 0.864,
+      "step": 26515
+    },
+    {
+      "epoch": 0.7282414654911977,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6298458859456643e-06,
+      "loss": 0.9121,
+      "step": 26516
+    },
+    {
+      "epoch": 0.728268929719041,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6291602267817193e-06,
+      "loss": 0.7815,
+      "step": 26517
+    },
+    {
+      "epoch": 0.7282963939468842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.628474618026324e-06,
+      "loss": 0.8762,
+      "step": 26518
+    },
+    {
+      "epoch": 0.7283238581747274,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6277890596849054e-06,
+      "loss": 0.8172,
+      "step": 26519
+    },
+    {
+      "epoch": 0.7283513224025706,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6271035517628885e-06,
+      "loss": 0.8702,
+      "step": 26520
+    },
+    {
+      "epoch": 0.7283787866304139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6264180942656955e-06,
+      "loss": 0.7648,
+      "step": 26521
+    },
+    {
+      "epoch": 0.7284062508582572,
+      "grad_norm": 0.0,
+      "learning_rate": 3.625732687198753e-06,
+      "loss": 0.7978,
+      "step": 26522
+    },
+    {
+      "epoch": 0.7284337150861003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.625047330567485e-06,
+      "loss": 0.7202,
+      "step": 26523
+    },
+    {
+      "epoch": 0.7284611793139436,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6243620243773094e-06,
+      "loss": 0.8643,
+      "step": 26524
+    },
+    {
+      "epoch": 0.7284886435417868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6236767686336536e-06,
+      "loss": 0.7909,
+      "step": 26525
+    },
+    {
+      "epoch": 0.72851610776963,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6229915633419344e-06,
+      "loss": 0.8613,
+      "step": 26526
+    },
+    {
+      "epoch": 0.7285435719974733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6223064085075756e-06,
+      "loss": 0.8217,
+      "step": 26527
+    },
+    {
+      "epoch": 0.7285710362253165,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6216213041360025e-06,
+      "loss": 0.9196,
+      "step": 26528
+    },
+    {
+      "epoch": 0.7285985004531598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6209362502326295e-06,
+      "loss": 0.8272,
+      "step": 26529
+    },
+    {
+      "epoch": 0.728625964681003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6202512468028795e-06,
+      "loss": 0.7979,
+      "step": 26530
+    },
+    {
+      "epoch": 0.7286534289088462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6195662938521736e-06,
+      "loss": 0.8068,
+      "step": 26531
+    },
+    {
+      "epoch": 0.7286808931366895,
+      "grad_norm": 0.0,
+      "learning_rate": 3.61888139138593e-06,
+      "loss": 0.8239,
+      "step": 26532
+    },
+    {
+      "epoch": 0.7287083573645327,
+      "grad_norm": 0.0,
+      "learning_rate": 3.618196539409572e-06,
+      "loss": 0.8889,
+      "step": 26533
+    },
+    {
+      "epoch": 0.7287358215923759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.617511737928515e-06,
+      "loss": 0.8918,
+      "step": 26534
+    },
+    {
+      "epoch": 0.7287632858202192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6168269869481743e-06,
+      "loss": 0.8413,
+      "step": 26535
+    },
+    {
+      "epoch": 0.7287907500480624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6161422864739715e-06,
+      "loss": 0.8565,
+      "step": 26536
+    },
+    {
+      "epoch": 0.7288182142759057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6154576365113237e-06,
+      "loss": 0.8326,
+      "step": 26537
+    },
+    {
+      "epoch": 0.7288456785037488,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6147730370656476e-06,
+      "loss": 0.8104,
+      "step": 26538
+    },
+    {
+      "epoch": 0.7288731427315921,
+      "grad_norm": 0.0,
+      "learning_rate": 3.614088488142361e-06,
+      "loss": 0.7988,
+      "step": 26539
+    },
+    {
+      "epoch": 0.7289006069594354,
+      "grad_norm": 0.0,
+      "learning_rate": 3.613403989746883e-06,
+      "loss": 0.8458,
+      "step": 26540
+    },
+    {
+      "epoch": 0.7289280711872785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6127195418846274e-06,
+      "loss": 0.7067,
+      "step": 26541
+    },
+    {
+      "epoch": 0.7289555354151218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6120351445610046e-06,
+      "loss": 0.7912,
+      "step": 26542
+    },
+    {
+      "epoch": 0.7289829996429651,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6113507977814353e-06,
+      "loss": 0.8203,
+      "step": 26543
+    },
+    {
+      "epoch": 0.7290104638708083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.610666501551332e-06,
+      "loss": 0.8327,
+      "step": 26544
+    },
+    {
+      "epoch": 0.7290379280986515,
+      "grad_norm": 0.0,
+      "learning_rate": 3.609982255876111e-06,
+      "loss": 0.8232,
+      "step": 26545
+    },
+    {
+      "epoch": 0.7290653923264947,
+      "grad_norm": 0.0,
+      "learning_rate": 3.609298060761189e-06,
+      "loss": 0.9291,
+      "step": 26546
+    },
+    {
+      "epoch": 0.729092856554338,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6086139162119727e-06,
+      "loss": 0.8421,
+      "step": 26547
+    },
+    {
+      "epoch": 0.7291203207821813,
+      "grad_norm": 0.0,
+      "learning_rate": 3.60792982223388e-06,
+      "loss": 0.9169,
+      "step": 26548
+    },
+    {
+      "epoch": 0.7291477850100244,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6072457788323244e-06,
+      "loss": 0.7976,
+      "step": 26549
+    },
+    {
+      "epoch": 0.7291752492378677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6065617860127135e-06,
+      "loss": 0.8344,
+      "step": 26550
+    },
+    {
+      "epoch": 0.7292027134657109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.605877843780463e-06,
+      "loss": 0.8636,
+      "step": 26551
+    },
+    {
+      "epoch": 0.7292301776935541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6051939521409864e-06,
+      "loss": 0.8642,
+      "step": 26552
+    },
+    {
+      "epoch": 0.7292576419213974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6045101110996893e-06,
+      "loss": 0.7913,
+      "step": 26553
+    },
+    {
+      "epoch": 0.7292851061492406,
+      "grad_norm": 0.0,
+      "learning_rate": 3.603826320661986e-06,
+      "loss": 0.7927,
+      "step": 26554
+    },
+    {
+      "epoch": 0.7293125703770839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6031425808332853e-06,
+      "loss": 0.9324,
+      "step": 26555
+    },
+    {
+      "epoch": 0.7293400346049271,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6024588916190027e-06,
+      "loss": 0.7837,
+      "step": 26556
+    },
+    {
+      "epoch": 0.7293674988327703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.601775253024541e-06,
+      "loss": 0.8184,
+      "step": 26557
+    },
+    {
+      "epoch": 0.7293949630606136,
+      "grad_norm": 0.0,
+      "learning_rate": 3.601091665055314e-06,
+      "loss": 0.8723,
+      "step": 26558
+    },
+    {
+      "epoch": 0.7294224272884567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6004081277167256e-06,
+      "loss": 0.8408,
+      "step": 26559
+    },
+    {
+      "epoch": 0.7294498915163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.599724641014186e-06,
+      "loss": 0.8607,
+      "step": 26560
+    },
+    {
+      "epoch": 0.7294773557441433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5990412049531053e-06,
+      "loss": 0.8085,
+      "step": 26561
+    },
+    {
+      "epoch": 0.7295048199719865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.59835781953889e-06,
+      "loss": 0.7209,
+      "step": 26562
+    },
+    {
+      "epoch": 0.7295322841998297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5976744847769474e-06,
+      "loss": 0.925,
+      "step": 26563
+    },
+    {
+      "epoch": 0.7295597484276729,
+      "grad_norm": 0.0,
+      "learning_rate": 3.596991200672688e-06,
+      "loss": 0.8538,
+      "step": 26564
+    },
+    {
+      "epoch": 0.7295872126555162,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5963079672315136e-06,
+      "loss": 0.8798,
+      "step": 26565
+    },
+    {
+      "epoch": 0.7296146768833595,
+      "grad_norm": 0.0,
+      "learning_rate": 3.595624784458829e-06,
+      "loss": 0.8447,
+      "step": 26566
+    },
+    {
+      "epoch": 0.7296421411112026,
+      "grad_norm": 0.0,
+      "learning_rate": 3.594941652360041e-06,
+      "loss": 0.7774,
+      "step": 26567
+    },
+    {
+      "epoch": 0.7296696053390459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5942585709405563e-06,
+      "loss": 0.9911,
+      "step": 26568
+    },
+    {
+      "epoch": 0.7296970695668892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.593575540205778e-06,
+      "loss": 0.8378,
+      "step": 26569
+    },
+    {
+      "epoch": 0.7297245337947323,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5928925601611163e-06,
+      "loss": 0.812,
+      "step": 26570
+    },
+    {
+      "epoch": 0.7297519980225756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.592209630811967e-06,
+      "loss": 0.9083,
+      "step": 26571
+    },
+    {
+      "epoch": 0.7297794622504188,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5915267521637364e-06,
+      "loss": 0.7896,
+      "step": 26572
+    },
+    {
+      "epoch": 0.7298069264782621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5908439242218317e-06,
+      "loss": 0.7644,
+      "step": 26573
+    },
+    {
+      "epoch": 0.7298343907061053,
+      "grad_norm": 0.0,
+      "learning_rate": 3.590161146991651e-06,
+      "loss": 0.8085,
+      "step": 26574
+    },
+    {
+      "epoch": 0.7298618549339485,
+      "grad_norm": 0.0,
+      "learning_rate": 3.589478420478597e-06,
+      "loss": 0.8231,
+      "step": 26575
+    },
+    {
+      "epoch": 0.7298893191617918,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5887957446880763e-06,
+      "loss": 0.8665,
+      "step": 26576
+    },
+    {
+      "epoch": 0.729916783389635,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5881131196254837e-06,
+      "loss": 0.7785,
+      "step": 26577
+    },
+    {
+      "epoch": 0.7299442476174782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5874305452962234e-06,
+      "loss": 0.7753,
+      "step": 26578
+    },
+    {
+      "epoch": 0.7299717118453215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.586748021705697e-06,
+      "loss": 0.9153,
+      "step": 26579
+    },
+    {
+      "epoch": 0.7299991760731647,
+      "grad_norm": 0.0,
+      "learning_rate": 3.586065548859309e-06,
+      "loss": 0.8117,
+      "step": 26580
+    },
+    {
+      "epoch": 0.730026640301008,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5853831267624508e-06,
+      "loss": 0.8813,
+      "step": 26581
+    },
+    {
+      "epoch": 0.7300541045288512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.584700755420528e-06,
+      "loss": 0.7522,
+      "step": 26582
+    },
+    {
+      "epoch": 0.7300815687566944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5840184348389362e-06,
+      "loss": 0.8741,
+      "step": 26583
+    },
+    {
+      "epoch": 0.7301090329845377,
+      "grad_norm": 0.0,
+      "learning_rate": 3.583336165023076e-06,
+      "loss": 0.8203,
+      "step": 26584
+    },
+    {
+      "epoch": 0.7301364972123808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.582653945978345e-06,
+      "loss": 0.8126,
+      "step": 26585
+    },
+    {
+      "epoch": 0.7301639614402241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5819717777101416e-06,
+      "loss": 0.8583,
+      "step": 26586
+    },
+    {
+      "epoch": 0.7301914256680674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5812896602238633e-06,
+      "loss": 0.8875,
+      "step": 26587
+    },
+    {
+      "epoch": 0.7302188898959106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.580607593524912e-06,
+      "loss": 0.8065,
+      "step": 26588
+    },
+    {
+      "epoch": 0.7302463541237538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5799255776186802e-06,
+      "loss": 0.7778,
+      "step": 26589
+    },
+    {
+      "epoch": 0.730273818351597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.57924361251056e-06,
+      "loss": 0.8577,
+      "step": 26590
+    },
+    {
+      "epoch": 0.7303012825794403,
+      "grad_norm": 0.0,
+      "learning_rate": 3.578561698205951e-06,
+      "loss": 0.8341,
+      "step": 26591
+    },
+    {
+      "epoch": 0.7303287468072835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5778798347102507e-06,
+      "loss": 0.8914,
+      "step": 26592
+    },
+    {
+      "epoch": 0.7303562110351267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.577198022028855e-06,
+      "loss": 0.7926,
+      "step": 26593
+    },
+    {
+      "epoch": 0.73038367526297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.576516260167153e-06,
+      "loss": 0.9359,
+      "step": 26594
+    },
+    {
+      "epoch": 0.7304111394908133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5758345491305434e-06,
+      "loss": 0.8416,
+      "step": 26595
+    },
+    {
+      "epoch": 0.7304386037186564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.575152888924419e-06,
+      "loss": 0.7937,
+      "step": 26596
+    },
+    {
+      "epoch": 0.7304660679464997,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5744712795541767e-06,
+      "loss": 0.8403,
+      "step": 26597
+    },
+    {
+      "epoch": 0.7304935321743429,
+      "grad_norm": 0.0,
+      "learning_rate": 3.573789721025205e-06,
+      "loss": 0.8011,
+      "step": 26598
+    },
+    {
+      "epoch": 0.7305209964021862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5731082133428995e-06,
+      "loss": 0.8755,
+      "step": 26599
+    },
+    {
+      "epoch": 0.7305484606300294,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5724267565126493e-06,
+      "loss": 0.8642,
+      "step": 26600
+    },
+    {
+      "epoch": 0.7305759248578726,
+      "grad_norm": 0.0,
+      "learning_rate": 3.571745350539848e-06,
+      "loss": 0.8456,
+      "step": 26601
+    },
+    {
+      "epoch": 0.7306033890857159,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5710639954298874e-06,
+      "loss": 0.8503,
+      "step": 26602
+    },
+    {
+      "epoch": 0.730630853313559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5703826911881588e-06,
+      "loss": 0.8109,
+      "step": 26603
+    },
+    {
+      "epoch": 0.7306583175414023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5697014378200566e-06,
+      "loss": 0.9419,
+      "step": 26604
+    },
+    {
+      "epoch": 0.7306857817692456,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5690202353309644e-06,
+      "loss": 0.8279,
+      "step": 26605
+    },
+    {
+      "epoch": 0.7307132459970888,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5683390837262776e-06,
+      "loss": 0.8709,
+      "step": 26606
+    },
+    {
+      "epoch": 0.730740710224932,
+      "grad_norm": 0.0,
+      "learning_rate": 3.567657983011381e-06,
+      "loss": 0.6941,
+      "step": 26607
+    },
+    {
+      "epoch": 0.7307681744527753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.566976933191665e-06,
+      "loss": 0.9007,
+      "step": 26608
+    },
+    {
+      "epoch": 0.7307956386806185,
+      "grad_norm": 0.0,
+      "learning_rate": 3.56629593427252e-06,
+      "loss": 0.7956,
+      "step": 26609
+    },
+    {
+      "epoch": 0.7308231029084618,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5656149862593337e-06,
+      "loss": 0.851,
+      "step": 26610
+    },
+    {
+      "epoch": 0.7308505671363049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.564934089157498e-06,
+      "loss": 0.8078,
+      "step": 26611
+    },
+    {
+      "epoch": 0.7308780313641482,
+      "grad_norm": 0.0,
+      "learning_rate": 3.564253242972392e-06,
+      "loss": 0.8264,
+      "step": 26612
+    },
+    {
+      "epoch": 0.7309054955919915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5635724477094114e-06,
+      "loss": 0.799,
+      "step": 26613
+    },
+    {
+      "epoch": 0.7309329598198346,
+      "grad_norm": 0.0,
+      "learning_rate": 3.562891703373936e-06,
+      "loss": 0.7449,
+      "step": 26614
+    },
+    {
+      "epoch": 0.7309604240476779,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5622110099713536e-06,
+      "loss": 0.9326,
+      "step": 26615
+    },
+    {
+      "epoch": 0.7309878882755211,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5615303675070524e-06,
+      "loss": 0.8386,
+      "step": 26616
+    },
+    {
+      "epoch": 0.7310153525033644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5608497759864204e-06,
+      "loss": 0.7563,
+      "step": 26617
+    },
+    {
+      "epoch": 0.7310428167312076,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5601692354148354e-06,
+      "loss": 0.8273,
+      "step": 26618
+    },
+    {
+      "epoch": 0.7310702809590508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5594887457976866e-06,
+      "loss": 0.9016,
+      "step": 26619
+    },
+    {
+      "epoch": 0.7310977451868941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5588083071403566e-06,
+      "loss": 0.8051,
+      "step": 26620
+    },
+    {
+      "epoch": 0.7311252094147374,
+      "grad_norm": 0.0,
+      "learning_rate": 3.558127919448234e-06,
+      "loss": 0.8459,
+      "step": 26621
+    },
+    {
+      "epoch": 0.7311526736425805,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5574475827266963e-06,
+      "loss": 0.783,
+      "step": 26622
+    },
+    {
+      "epoch": 0.7311801378704238,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5567672969811317e-06,
+      "loss": 0.8281,
+      "step": 26623
+    },
+    {
+      "epoch": 0.731207602098267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5560870622169162e-06,
+      "loss": 0.7998,
+      "step": 26624
+    },
+    {
+      "epoch": 0.7312350663261102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.555406878439437e-06,
+      "loss": 0.9439,
+      "step": 26625
+    },
+    {
+      "epoch": 0.7312625305539535,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5547267456540746e-06,
+      "loss": 0.8458,
+      "step": 26626
+    },
+    {
+      "epoch": 0.7312899947817967,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5540466638662108e-06,
+      "loss": 0.9107,
+      "step": 26627
+    },
+    {
+      "epoch": 0.73131745900964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.55336663308123e-06,
+      "loss": 0.8459,
+      "step": 26628
+    },
+    {
+      "epoch": 0.7313449232374831,
+      "grad_norm": 0.0,
+      "learning_rate": 3.552686653304506e-06,
+      "loss": 0.8619,
+      "step": 26629
+    },
+    {
+      "epoch": 0.7313723874653264,
+      "grad_norm": 0.0,
+      "learning_rate": 3.552006724541426e-06,
+      "loss": 0.8575,
+      "step": 26630
+    },
+    {
+      "epoch": 0.7313998516931697,
+      "grad_norm": 0.0,
+      "learning_rate": 3.551326846797363e-06,
+      "loss": 0.8525,
+      "step": 26631
+    },
+    {
+      "epoch": 0.7314273159210128,
+      "grad_norm": 0.0,
+      "learning_rate": 3.550647020077701e-06,
+      "loss": 0.9208,
+      "step": 26632
+    },
+    {
+      "epoch": 0.7314547801488561,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5499672443878162e-06,
+      "loss": 0.8383,
+      "step": 26633
+    },
+    {
+      "epoch": 0.7314822443766994,
+      "grad_norm": 0.0,
+      "learning_rate": 3.549287519733091e-06,
+      "loss": 0.8447,
+      "step": 26634
+    },
+    {
+      "epoch": 0.7315097086045426,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5486078461189045e-06,
+      "loss": 0.8568,
+      "step": 26635
+    },
+    {
+      "epoch": 0.7315371728323858,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5479282235506273e-06,
+      "loss": 0.8609,
+      "step": 26636
+    },
+    {
+      "epoch": 0.731564637060229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.547248652033646e-06,
+      "loss": 0.787,
+      "step": 26637
+    },
+    {
+      "epoch": 0.7315921012880723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5465691315733286e-06,
+      "loss": 0.7356,
+      "step": 26638
+    },
+    {
+      "epoch": 0.7316195655159156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5458896621750573e-06,
+      "loss": 0.9016,
+      "step": 26639
+    },
+    {
+      "epoch": 0.7316470297437587,
+      "grad_norm": 0.0,
+      "learning_rate": 3.545210243844206e-06,
+      "loss": 0.7823,
+      "step": 26640
+    },
+    {
+      "epoch": 0.731674493971602,
+      "grad_norm": 0.0,
+      "learning_rate": 3.544530876586155e-06,
+      "loss": 0.8437,
+      "step": 26641
+    },
+    {
+      "epoch": 0.7317019581994452,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5438515604062727e-06,
+      "loss": 0.7966,
+      "step": 26642
+    },
+    {
+      "epoch": 0.7317294224272884,
+      "grad_norm": 0.0,
+      "learning_rate": 3.543172295309939e-06,
+      "loss": 0.8427,
+      "step": 26643
+    },
+    {
+      "epoch": 0.7317568866551317,
+      "grad_norm": 0.0,
+      "learning_rate": 3.542493081302526e-06,
+      "loss": 0.7118,
+      "step": 26644
+    },
+    {
+      "epoch": 0.7317843508829749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.541813918389413e-06,
+      "loss": 0.8105,
+      "step": 26645
+    },
+    {
+      "epoch": 0.7318118151108182,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5411348065759664e-06,
+      "loss": 0.734,
+      "step": 26646
+    },
+    {
+      "epoch": 0.7318392793386614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.540455745867567e-06,
+      "loss": 0.8461,
+      "step": 26647
+    },
+    {
+      "epoch": 0.7318667435665046,
+      "grad_norm": 0.0,
+      "learning_rate": 3.53977673626958e-06,
+      "loss": 0.8446,
+      "step": 26648
+    },
+    {
+      "epoch": 0.7318942077943479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5390977777873824e-06,
+      "loss": 0.8253,
+      "step": 26649
+    },
+    {
+      "epoch": 0.731921672022191,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5384188704263455e-06,
+      "loss": 0.7158,
+      "step": 26650
+    },
+    {
+      "epoch": 0.7319491362500343,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5377400141918417e-06,
+      "loss": 0.7762,
+      "step": 26651
+    },
+    {
+      "epoch": 0.7319766004778776,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5370612090892453e-06,
+      "loss": 0.8679,
+      "step": 26652
+    },
+    {
+      "epoch": 0.7320040647057208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.536382455123921e-06,
+      "loss": 0.8787,
+      "step": 26653
+    },
+    {
+      "epoch": 0.732031528933564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.535703752301246e-06,
+      "loss": 0.8615,
+      "step": 26654
+    },
+    {
+      "epoch": 0.7320589931614072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5350251006265833e-06,
+      "loss": 0.8173,
+      "step": 26655
+    },
+    {
+      "epoch": 0.7320864573892505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5343465001053056e-06,
+      "loss": 0.7343,
+      "step": 26656
+    },
+    {
+      "epoch": 0.7321139216170938,
+      "grad_norm": 0.0,
+      "learning_rate": 3.533667950742784e-06,
+      "loss": 0.844,
+      "step": 26657
+    },
+    {
+      "epoch": 0.7321413858449369,
+      "grad_norm": 0.0,
+      "learning_rate": 3.53298945254439e-06,
+      "loss": 0.7767,
+      "step": 26658
+    },
+    {
+      "epoch": 0.7321688500727802,
+      "grad_norm": 0.0,
+      "learning_rate": 3.532311005515484e-06,
+      "loss": 0.9543,
+      "step": 26659
+    },
+    {
+      "epoch": 0.7321963143006235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5316326096614394e-06,
+      "loss": 0.8421,
+      "step": 26660
+    },
+    {
+      "epoch": 0.7322237785284667,
+      "grad_norm": 0.0,
+      "learning_rate": 3.530954264987626e-06,
+      "loss": 0.8525,
+      "step": 26661
+    },
+    {
+      "epoch": 0.7322512427563099,
+      "grad_norm": 0.0,
+      "learning_rate": 3.530275971499405e-06,
+      "loss": 0.8867,
+      "step": 26662
+    },
+    {
+      "epoch": 0.7322787069841531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5295977292021476e-06,
+      "loss": 0.8363,
+      "step": 26663
+    },
+    {
+      "epoch": 0.7323061712119964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.528919538101222e-06,
+      "loss": 0.8354,
+      "step": 26664
+    },
+    {
+      "epoch": 0.7323336354398396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.528241398201988e-06,
+      "loss": 0.8267,
+      "step": 26665
+    },
+    {
+      "epoch": 0.7323610996676828,
+      "grad_norm": 0.0,
+      "learning_rate": 3.527563309509816e-06,
+      "loss": 0.8477,
+      "step": 26666
+    },
+    {
+      "epoch": 0.7323885638955261,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5268852720300697e-06,
+      "loss": 0.6863,
+      "step": 26667
+    },
+    {
+      "epoch": 0.7324160281233693,
+      "grad_norm": 0.0,
+      "learning_rate": 3.526207285768114e-06,
+      "loss": 0.843,
+      "step": 26668
+    },
+    {
+      "epoch": 0.7324434923512125,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5255293507293177e-06,
+      "loss": 0.9133,
+      "step": 26669
+    },
+    {
+      "epoch": 0.7324709565790558,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5248514669190416e-06,
+      "loss": 0.7728,
+      "step": 26670
+    },
+    {
+      "epoch": 0.732498420806899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.524173634342646e-06,
+      "loss": 0.8661,
+      "step": 26671
+    },
+    {
+      "epoch": 0.7325258850347423,
+      "grad_norm": 0.0,
+      "learning_rate": 3.523495853005495e-06,
+      "loss": 0.8808,
+      "step": 26672
+    },
+    {
+      "epoch": 0.7325533492625855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5228181229129553e-06,
+      "loss": 0.7983,
+      "step": 26673
+    },
+    {
+      "epoch": 0.7325808134904287,
+      "grad_norm": 0.0,
+      "learning_rate": 3.522140444070388e-06,
+      "loss": 0.8044,
+      "step": 26674
+    },
+    {
+      "epoch": 0.732608277718272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5214628164831534e-06,
+      "loss": 0.7564,
+      "step": 26675
+    },
+    {
+      "epoch": 0.7326357419461151,
+      "grad_norm": 0.0,
+      "learning_rate": 3.520785240156619e-06,
+      "loss": 0.8364,
+      "step": 26676
+    },
+    {
+      "epoch": 0.7326632061739584,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5201077150961405e-06,
+      "loss": 0.8153,
+      "step": 26677
+    },
+    {
+      "epoch": 0.7326906704018017,
+      "grad_norm": 0.0,
+      "learning_rate": 3.519430241307078e-06,
+      "loss": 0.7477,
+      "step": 26678
+    },
+    {
+      "epoch": 0.7327181346296449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.518752818794793e-06,
+      "loss": 0.8728,
+      "step": 26679
+    },
+    {
+      "epoch": 0.7327455988574881,
+      "grad_norm": 0.0,
+      "learning_rate": 3.518075447564646e-06,
+      "loss": 0.8461,
+      "step": 26680
+    },
+    {
+      "epoch": 0.7327730630853314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.517398127621997e-06,
+      "loss": 0.6843,
+      "step": 26681
+    },
+    {
+      "epoch": 0.7328005273131746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.516720858972208e-06,
+      "loss": 0.9727,
+      "step": 26682
+    },
+    {
+      "epoch": 0.7328279915410179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.516043641620632e-06,
+      "loss": 0.7974,
+      "step": 26683
+    },
+    {
+      "epoch": 0.732855455768861,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5153664755726314e-06,
+      "loss": 0.79,
+      "step": 26684
+    },
+    {
+      "epoch": 0.7328829199967043,
+      "grad_norm": 0.0,
+      "learning_rate": 3.514689360833565e-06,
+      "loss": 0.9038,
+      "step": 26685
+    },
+    {
+      "epoch": 0.7329103842245476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5140122974087863e-06,
+      "loss": 0.784,
+      "step": 26686
+    },
+    {
+      "epoch": 0.7329378484523907,
+      "grad_norm": 0.0,
+      "learning_rate": 3.513335285303654e-06,
+      "loss": 0.8376,
+      "step": 26687
+    },
+    {
+      "epoch": 0.732965312680234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.512658324523529e-06,
+      "loss": 0.8394,
+      "step": 26688
+    },
+    {
+      "epoch": 0.7329927769080772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5119814150737618e-06,
+      "loss": 0.8004,
+      "step": 26689
+    },
+    {
+      "epoch": 0.7330202411359205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5113045569597105e-06,
+      "loss": 0.7879,
+      "step": 26690
+    },
+    {
+      "epoch": 0.7330477053637637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5106277501867305e-06,
+      "loss": 0.8318,
+      "step": 26691
+    },
+    {
+      "epoch": 0.7330751695916069,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5099509947601775e-06,
+      "loss": 0.8331,
+      "step": 26692
+    },
+    {
+      "epoch": 0.7331026338194502,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5092742906854095e-06,
+      "loss": 0.8788,
+      "step": 26693
+    },
+    {
+      "epoch": 0.7331300980472935,
+      "grad_norm": 0.0,
+      "learning_rate": 3.508597637967778e-06,
+      "loss": 0.8561,
+      "step": 26694
+    },
+    {
+      "epoch": 0.7331575622751366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5079210366126327e-06,
+      "loss": 0.8639,
+      "step": 26695
+    },
+    {
+      "epoch": 0.7331850265029799,
+      "grad_norm": 0.0,
+      "learning_rate": 3.507244486625332e-06,
+      "loss": 0.8902,
+      "step": 26696
+    },
+    {
+      "epoch": 0.7332124907308231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.506567988011227e-06,
+      "loss": 0.8328,
+      "step": 26697
+    },
+    {
+      "epoch": 0.7332399549586663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.505891540775672e-06,
+      "loss": 0.7013,
+      "step": 26698
+    },
+    {
+      "epoch": 0.7332674191865096,
+      "grad_norm": 0.0,
+      "learning_rate": 3.505215144924018e-06,
+      "loss": 0.8475,
+      "step": 26699
+    },
+    {
+      "epoch": 0.7332948834143528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.504538800461621e-06,
+      "loss": 0.8851,
+      "step": 26700
+    },
+    {
+      "epoch": 0.7333223476421961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5038625073938306e-06,
+      "loss": 0.8469,
+      "step": 26701
+    },
+    {
+      "epoch": 0.7333498118700392,
+      "grad_norm": 0.0,
+      "learning_rate": 3.503186265725991e-06,
+      "loss": 0.8052,
+      "step": 26702
+    },
+    {
+      "epoch": 0.7333772760978825,
+      "grad_norm": 0.0,
+      "learning_rate": 3.50251007546346e-06,
+      "loss": 0.8131,
+      "step": 26703
+    },
+    {
+      "epoch": 0.7334047403257258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.501833936611585e-06,
+      "loss": 0.932,
+      "step": 26704
+    },
+    {
+      "epoch": 0.733432204553569,
+      "grad_norm": 0.0,
+      "learning_rate": 3.501157849175717e-06,
+      "loss": 0.8469,
+      "step": 26705
+    },
+    {
+      "epoch": 0.7334596687814122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5004818131612096e-06,
+      "loss": 0.9378,
+      "step": 26706
+    },
+    {
+      "epoch": 0.7334871330092555,
+      "grad_norm": 0.0,
+      "learning_rate": 3.499805828573404e-06,
+      "loss": 0.7987,
+      "step": 26707
+    },
+    {
+      "epoch": 0.7335145972370987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4991298954176523e-06,
+      "loss": 0.8152,
+      "step": 26708
+    },
+    {
+      "epoch": 0.7335420614649419,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4984540136993074e-06,
+      "loss": 0.9219,
+      "step": 26709
+    },
+    {
+      "epoch": 0.7335695256927851,
+      "grad_norm": 0.0,
+      "learning_rate": 3.497778183423709e-06,
+      "loss": 0.8586,
+      "step": 26710
+    },
+    {
+      "epoch": 0.7335969899206284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.497102404596211e-06,
+      "loss": 0.8408,
+      "step": 26711
+    },
+    {
+      "epoch": 0.7336244541484717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4964266772221558e-06,
+      "loss": 0.813,
+      "step": 26712
+    },
+    {
+      "epoch": 0.7336519183763148,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4957510013068907e-06,
+      "loss": 0.8387,
+      "step": 26713
+    },
+    {
+      "epoch": 0.7336793826041581,
+      "grad_norm": 0.0,
+      "learning_rate": 3.495075376855763e-06,
+      "loss": 0.9152,
+      "step": 26714
+    },
+    {
+      "epoch": 0.7337068468320013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.49439980387412e-06,
+      "loss": 0.8557,
+      "step": 26715
+    },
+    {
+      "epoch": 0.7337343110598445,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4937242823673077e-06,
+      "loss": 0.8183,
+      "step": 26716
+    },
+    {
+      "epoch": 0.7337617752876878,
+      "grad_norm": 0.0,
+      "learning_rate": 3.493048812340667e-06,
+      "loss": 0.8589,
+      "step": 26717
+    },
+    {
+      "epoch": 0.733789239515531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4923733937995476e-06,
+      "loss": 0.8662,
+      "step": 26718
+    },
+    {
+      "epoch": 0.7338167037433743,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4916980267492872e-06,
+      "loss": 0.9279,
+      "step": 26719
+    },
+    {
+      "epoch": 0.7338441679712175,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4910227111952335e-06,
+      "loss": 0.7814,
+      "step": 26720
+    },
+    {
+      "epoch": 0.7338716321990607,
+      "grad_norm": 0.0,
+      "learning_rate": 3.490347447142729e-06,
+      "loss": 0.9135,
+      "step": 26721
+    },
+    {
+      "epoch": 0.733899096426904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.489672234597118e-06,
+      "loss": 0.8268,
+      "step": 26722
+    },
+    {
+      "epoch": 0.7339265606547472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4889970735637447e-06,
+      "loss": 0.7951,
+      "step": 26723
+    },
+    {
+      "epoch": 0.7339540248825904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4883219640479448e-06,
+      "loss": 0.7937,
+      "step": 26724
+    },
+    {
+      "epoch": 0.7339814891104337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4876469060550677e-06,
+      "loss": 0.8497,
+      "step": 26725
+    },
+    {
+      "epoch": 0.7340089533382769,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4869718995904477e-06,
+      "loss": 0.7819,
+      "step": 26726
+    },
+    {
+      "epoch": 0.7340364175661201,
+      "grad_norm": 0.0,
+      "learning_rate": 3.48629694465943e-06,
+      "loss": 0.8097,
+      "step": 26727
+    },
+    {
+      "epoch": 0.7340638817939633,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4856220412673523e-06,
+      "loss": 0.8493,
+      "step": 26728
+    },
+    {
+      "epoch": 0.7340913460218066,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4849471894195607e-06,
+      "loss": 0.8219,
+      "step": 26729
+    },
+    {
+      "epoch": 0.7341188102496499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.484272389121387e-06,
+      "loss": 0.8079,
+      "step": 26730
+    },
+    {
+      "epoch": 0.734146274477493,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4835976403781734e-06,
+      "loss": 0.7619,
+      "step": 26731
+    },
+    {
+      "epoch": 0.7341737387053363,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4829229431952606e-06,
+      "loss": 0.785,
+      "step": 26732
+    },
+    {
+      "epoch": 0.7342012029331796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.482248297577989e-06,
+      "loss": 0.7843,
+      "step": 26733
+    },
+    {
+      "epoch": 0.7342286671610228,
+      "grad_norm": 0.0,
+      "learning_rate": 3.48157370353169e-06,
+      "loss": 0.8497,
+      "step": 26734
+    },
+    {
+      "epoch": 0.734256131388866,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4808991610617083e-06,
+      "loss": 0.7993,
+      "step": 26735
+    },
+    {
+      "epoch": 0.7342835956167092,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4802246701733746e-06,
+      "loss": 0.8795,
+      "step": 26736
+    },
+    {
+      "epoch": 0.7343110598445525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4795502308720287e-06,
+      "loss": 0.7381,
+      "step": 26737
+    },
+    {
+      "epoch": 0.7343385240723957,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4788758431630077e-06,
+      "loss": 0.8307,
+      "step": 26738
+    },
+    {
+      "epoch": 0.7343659883002389,
+      "grad_norm": 0.0,
+      "learning_rate": 3.478201507051646e-06,
+      "loss": 0.9292,
+      "step": 26739
+    },
+    {
+      "epoch": 0.7343934525280822,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4775272225432844e-06,
+      "loss": 0.8157,
+      "step": 26740
+    },
+    {
+      "epoch": 0.7344209167559254,
+      "grad_norm": 0.0,
+      "learning_rate": 3.47685298964325e-06,
+      "loss": 0.8416,
+      "step": 26741
+    },
+    {
+      "epoch": 0.7344483809837686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.476178808356886e-06,
+      "loss": 0.8429,
+      "step": 26742
+    },
+    {
+      "epoch": 0.7344758452116119,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4755046786895173e-06,
+      "loss": 0.8336,
+      "step": 26743
+    },
+    {
+      "epoch": 0.7345033094394551,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4748306006464837e-06,
+      "loss": 0.8531,
+      "step": 26744
+    },
+    {
+      "epoch": 0.7345307736672984,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4741565742331186e-06,
+      "loss": 0.8241,
+      "step": 26745
+    },
+    {
+      "epoch": 0.7345582378951416,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4734825994547537e-06,
+      "loss": 0.7923,
+      "step": 26746
+    },
+    {
+      "epoch": 0.7345857021229848,
+      "grad_norm": 0.0,
+      "learning_rate": 3.472808676316727e-06,
+      "loss": 0.9312,
+      "step": 26747
+    },
+    {
+      "epoch": 0.7346131663508281,
+      "grad_norm": 0.0,
+      "learning_rate": 3.472134804824363e-06,
+      "loss": 0.8995,
+      "step": 26748
+    },
+    {
+      "epoch": 0.7346406305786712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4714609849829996e-06,
+      "loss": 0.729,
+      "step": 26749
+    },
+    {
+      "epoch": 0.7346680948065145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4707872167979627e-06,
+      "loss": 0.9248,
+      "step": 26750
+    },
+    {
+      "epoch": 0.7346955590343578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4701135002745868e-06,
+      "loss": 0.8871,
+      "step": 26751
+    },
+    {
+      "epoch": 0.734723023262201,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4694398354182026e-06,
+      "loss": 0.813,
+      "step": 26752
+    },
+    {
+      "epoch": 0.7347504874900442,
+      "grad_norm": 0.0,
+      "learning_rate": 3.468766222234142e-06,
+      "loss": 0.8523,
+      "step": 26753
+    },
+    {
+      "epoch": 0.7347779517178874,
+      "grad_norm": 0.0,
+      "learning_rate": 3.468092660727731e-06,
+      "loss": 0.7215,
+      "step": 26754
+    },
+    {
+      "epoch": 0.7348054159457307,
+      "grad_norm": 0.0,
+      "learning_rate": 3.467419150904301e-06,
+      "loss": 0.7803,
+      "step": 26755
+    },
+    {
+      "epoch": 0.734832880173574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4667456927691812e-06,
+      "loss": 0.8173,
+      "step": 26756
+    },
+    {
+      "epoch": 0.7348603444014171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.466072286327703e-06,
+      "loss": 0.9049,
+      "step": 26757
+    },
+    {
+      "epoch": 0.7348878086292604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4653989315851887e-06,
+      "loss": 0.8717,
+      "step": 26758
+    },
+    {
+      "epoch": 0.7349152728571037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4647256285469722e-06,
+      "loss": 0.7307,
+      "step": 26759
+    },
+    {
+      "epoch": 0.7349427370849468,
+      "grad_norm": 0.0,
+      "learning_rate": 3.464052377218375e-06,
+      "loss": 0.8419,
+      "step": 26760
+    },
+    {
+      "epoch": 0.7349702013127901,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4633791776047276e-06,
+      "loss": 0.7922,
+      "step": 26761
+    },
+    {
+      "epoch": 0.7349976655406333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.462706029711357e-06,
+      "loss": 0.805,
+      "step": 26762
+    },
+    {
+      "epoch": 0.7350251297684766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4620329335435876e-06,
+      "loss": 0.8042,
+      "step": 26763
+    },
+    {
+      "epoch": 0.7350525939963198,
+      "grad_norm": 0.0,
+      "learning_rate": 3.461359889106749e-06,
+      "loss": 0.7429,
+      "step": 26764
+    },
+    {
+      "epoch": 0.735080058224163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4606868964061613e-06,
+      "loss": 0.8279,
+      "step": 26765
+    },
+    {
+      "epoch": 0.7351075224520063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4600139554471544e-06,
+      "loss": 0.7765,
+      "step": 26766
+    },
+    {
+      "epoch": 0.7351349866798494,
+      "grad_norm": 0.0,
+      "learning_rate": 3.459341066235048e-06,
+      "loss": 0.8112,
+      "step": 26767
+    },
+    {
+      "epoch": 0.7351624509076927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.458668228775168e-06,
+      "loss": 0.7631,
+      "step": 26768
+    },
+    {
+      "epoch": 0.735189915135536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.45799544307284e-06,
+      "loss": 0.8579,
+      "step": 26769
+    },
+    {
+      "epoch": 0.7352173793633792,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4573227091333882e-06,
+      "loss": 0.7535,
+      "step": 26770
+    },
+    {
+      "epoch": 0.7352448435912224,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4566500269621304e-06,
+      "loss": 0.9587,
+      "step": 26771
+    },
+    {
+      "epoch": 0.7352723078190657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4559773965643926e-06,
+      "loss": 0.8267,
+      "step": 26772
+    },
+    {
+      "epoch": 0.7352997720469089,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4553048179455007e-06,
+      "loss": 0.7497,
+      "step": 26773
+    },
+    {
+      "epoch": 0.7353272362747522,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4546322911107676e-06,
+      "loss": 0.7778,
+      "step": 26774
+    },
+    {
+      "epoch": 0.7353547005025953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4539598160655195e-06,
+      "loss": 0.8755,
+      "step": 26775
+    },
+    {
+      "epoch": 0.7353821647304386,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4532873928150812e-06,
+      "loss": 0.8286,
+      "step": 26776
+    },
+    {
+      "epoch": 0.7354096289582819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4526150213647645e-06,
+      "loss": 0.8561,
+      "step": 26777
+    },
+    {
+      "epoch": 0.735437093186125,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4519427017198947e-06,
+      "loss": 0.7677,
+      "step": 26778
+    },
+    {
+      "epoch": 0.7354645574139683,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4512704338857907e-06,
+      "loss": 0.7279,
+      "step": 26779
+    },
+    {
+      "epoch": 0.7354920216418115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.450598217867772e-06,
+      "loss": 0.7983,
+      "step": 26780
+    },
+    {
+      "epoch": 0.7355194858696548,
+      "grad_norm": 0.0,
+      "learning_rate": 3.44992605367116e-06,
+      "loss": 0.8922,
+      "step": 26781
+    },
+    {
+      "epoch": 0.735546950097498,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4492539413012704e-06,
+      "loss": 0.8379,
+      "step": 26782
+    },
+    {
+      "epoch": 0.7355744143253412,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4485818807634175e-06,
+      "loss": 0.965,
+      "step": 26783
+    },
+    {
+      "epoch": 0.7356018785531845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4479098720629234e-06,
+      "loss": 0.8053,
+      "step": 26784
+    },
+    {
+      "epoch": 0.7356293427810278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4472379152051037e-06,
+      "loss": 0.7348,
+      "step": 26785
+    },
+    {
+      "epoch": 0.7356568070088709,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4465660101952757e-06,
+      "loss": 0.8371,
+      "step": 26786
+    },
+    {
+      "epoch": 0.7356842712367142,
+      "grad_norm": 0.0,
+      "learning_rate": 3.445894157038757e-06,
+      "loss": 0.739,
+      "step": 26787
+    },
+    {
+      "epoch": 0.7357117354645574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.445222355740865e-06,
+      "loss": 0.7808,
+      "step": 26788
+    },
+    {
+      "epoch": 0.7357391996924006,
+      "grad_norm": 0.0,
+      "learning_rate": 3.44455060630691e-06,
+      "loss": 0.8383,
+      "step": 26789
+    },
+    {
+      "epoch": 0.7357666639202439,
+      "grad_norm": 0.0,
+      "learning_rate": 3.443878908742213e-06,
+      "loss": 0.9604,
+      "step": 26790
+    },
+    {
+      "epoch": 0.7357941281480871,
+      "grad_norm": 0.0,
+      "learning_rate": 3.443207263052082e-06,
+      "loss": 0.7581,
+      "step": 26791
+    },
+    {
+      "epoch": 0.7358215923759304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4425356692418353e-06,
+      "loss": 0.9066,
+      "step": 26792
+    },
+    {
+      "epoch": 0.7358490566037735,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4418641273167874e-06,
+      "loss": 0.7249,
+      "step": 26793
+    },
+    {
+      "epoch": 0.7358765208316168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4411926372822535e-06,
+      "loss": 0.8276,
+      "step": 26794
+    },
+    {
+      "epoch": 0.7359039850594601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4405211991435406e-06,
+      "loss": 0.7352,
+      "step": 26795
+    },
+    {
+      "epoch": 0.7359314492873033,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4398498129059644e-06,
+      "loss": 0.8668,
+      "step": 26796
+    },
+    {
+      "epoch": 0.7359589135151465,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4391784785748416e-06,
+      "loss": 0.9275,
+      "step": 26797
+    },
+    {
+      "epoch": 0.7359863777429898,
+      "grad_norm": 0.0,
+      "learning_rate": 3.438507196155477e-06,
+      "loss": 0.7264,
+      "step": 26798
+    },
+    {
+      "epoch": 0.736013841970833,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4378359656531846e-06,
+      "loss": 0.8134,
+      "step": 26799
+    },
+    {
+      "epoch": 0.7360413061986762,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4371647870732795e-06,
+      "loss": 0.8156,
+      "step": 26800
+    },
+    {
+      "epoch": 0.7360687704265194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.436493660421065e-06,
+      "loss": 0.8675,
+      "step": 26801
+    },
+    {
+      "epoch": 0.7360962346543627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4358225857018556e-06,
+      "loss": 0.8591,
+      "step": 26802
+    },
+    {
+      "epoch": 0.736123698882206,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4351515629209607e-06,
+      "loss": 0.7423,
+      "step": 26803
+    },
+    {
+      "epoch": 0.7361511631100491,
+      "grad_norm": 0.0,
+      "learning_rate": 3.43448059208369e-06,
+      "loss": 0.8318,
+      "step": 26804
+    },
+    {
+      "epoch": 0.7361786273378924,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4338096731953553e-06,
+      "loss": 0.7816,
+      "step": 26805
+    },
+    {
+      "epoch": 0.7362060915657356,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4331388062612615e-06,
+      "loss": 0.8635,
+      "step": 26806
+    },
+    {
+      "epoch": 0.7362335557935789,
+      "grad_norm": 0.0,
+      "learning_rate": 3.432467991286714e-06,
+      "loss": 0.7907,
+      "step": 26807
+    },
+    {
+      "epoch": 0.7362610200214221,
+      "grad_norm": 0.0,
+      "learning_rate": 3.431797228277024e-06,
+      "loss": 0.8443,
+      "step": 26808
+    },
+    {
+      "epoch": 0.7362884842492653,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4311265172374987e-06,
+      "loss": 0.7579,
+      "step": 26809
+    },
+    {
+      "epoch": 0.7363159484771086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4304558581734437e-06,
+      "loss": 0.8475,
+      "step": 26810
+    },
+    {
+      "epoch": 0.7363434127049518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4297852510901685e-06,
+      "loss": 0.7778,
+      "step": 26811
+    },
+    {
+      "epoch": 0.736370876932795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.42911469599298e-06,
+      "loss": 0.832,
+      "step": 26812
+    },
+    {
+      "epoch": 0.7363983411606383,
+      "grad_norm": 0.0,
+      "learning_rate": 3.428444192887177e-06,
+      "loss": 0.8168,
+      "step": 26813
+    },
+    {
+      "epoch": 0.7364258053884815,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4277737417780743e-06,
+      "loss": 0.7603,
+      "step": 26814
+    },
+    {
+      "epoch": 0.7364532696163247,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4271033426709664e-06,
+      "loss": 0.7289,
+      "step": 26815
+    },
+    {
+      "epoch": 0.736480733844168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.426432995571165e-06,
+      "loss": 0.7851,
+      "step": 26816
+    },
+    {
+      "epoch": 0.7365081980720112,
+      "grad_norm": 0.0,
+      "learning_rate": 3.425762700483971e-06,
+      "loss": 0.8836,
+      "step": 26817
+    },
+    {
+      "epoch": 0.7365356622998545,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4250924574146926e-06,
+      "loss": 0.7798,
+      "step": 26818
+    },
+    {
+      "epoch": 0.7365631265276976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4244222663686256e-06,
+      "loss": 0.8111,
+      "step": 26819
+    },
+    {
+      "epoch": 0.7365905907555409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4237521273510776e-06,
+      "loss": 0.8665,
+      "step": 26820
+    },
+    {
+      "epoch": 0.7366180549833842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4230820403673526e-06,
+      "loss": 0.9445,
+      "step": 26821
+    },
+    {
+      "epoch": 0.7366455192112273,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4224120054227485e-06,
+      "loss": 0.7381,
+      "step": 26822
+    },
+    {
+      "epoch": 0.7366729834390706,
+      "grad_norm": 0.0,
+      "learning_rate": 3.421742022522567e-06,
+      "loss": 0.7614,
+      "step": 26823
+    },
+    {
+      "epoch": 0.7367004476669139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.421072091672114e-06,
+      "loss": 0.8022,
+      "step": 26824
+    },
+    {
+      "epoch": 0.7367279118947571,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4204022128766845e-06,
+      "loss": 0.7138,
+      "step": 26825
+    },
+    {
+      "epoch": 0.7367553761226003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.419732386141581e-06,
+      "loss": 0.886,
+      "step": 26826
+    },
+    {
+      "epoch": 0.7367828403504435,
+      "grad_norm": 0.0,
+      "learning_rate": 3.419062611472105e-06,
+      "loss": 0.8409,
+      "step": 26827
+    },
+    {
+      "epoch": 0.7368103045782868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.418392888873553e-06,
+      "loss": 0.8614,
+      "step": 26828
+    },
+    {
+      "epoch": 0.7368377688061301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.41772321835123e-06,
+      "loss": 0.7925,
+      "step": 26829
+    },
+    {
+      "epoch": 0.7368652330339732,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4170535999104306e-06,
+      "loss": 0.9695,
+      "step": 26830
+    },
+    {
+      "epoch": 0.7368926972618165,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4163840335564503e-06,
+      "loss": 0.8454,
+      "step": 26831
+    },
+    {
+      "epoch": 0.7369201614896597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4157145192945896e-06,
+      "loss": 0.8345,
+      "step": 26832
+    },
+    {
+      "epoch": 0.7369476257175029,
+      "grad_norm": 0.0,
+      "learning_rate": 3.415045057130145e-06,
+      "loss": 0.8932,
+      "step": 26833
+    },
+    {
+      "epoch": 0.7369750899453462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4143756470684163e-06,
+      "loss": 0.8172,
+      "step": 26834
+    },
+    {
+      "epoch": 0.7370025541731894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.413706289114701e-06,
+      "loss": 0.8167,
+      "step": 26835
+    },
+    {
+      "epoch": 0.7370300184010327,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4130369832742894e-06,
+      "loss": 0.7661,
+      "step": 26836
+    },
+    {
+      "epoch": 0.7370574826288759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.412367729552484e-06,
+      "loss": 0.8363,
+      "step": 26837
+    },
+    {
+      "epoch": 0.7370849468567191,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4116985279545745e-06,
+      "loss": 0.8562,
+      "step": 26838
+    },
+    {
+      "epoch": 0.7371124110845624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4110293784858573e-06,
+      "loss": 0.825,
+      "step": 26839
+    },
+    {
+      "epoch": 0.7371398753124055,
+      "grad_norm": 0.0,
+      "learning_rate": 3.410360281151628e-06,
+      "loss": 0.833,
+      "step": 26840
+    },
+    {
+      "epoch": 0.7371673395402488,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4096912359571843e-06,
+      "loss": 0.8241,
+      "step": 26841
+    },
+    {
+      "epoch": 0.7371948037680921,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4090222429078134e-06,
+      "loss": 0.8228,
+      "step": 26842
+    },
+    {
+      "epoch": 0.7372222679959353,
+      "grad_norm": 0.0,
+      "learning_rate": 3.408353302008811e-06,
+      "loss": 0.8823,
+      "step": 26843
+    },
+    {
+      "epoch": 0.7372497322237785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4076844132654717e-06,
+      "loss": 0.8456,
+      "step": 26844
+    },
+    {
+      "epoch": 0.7372771964516217,
+      "grad_norm": 0.0,
+      "learning_rate": 3.407015576683089e-06,
+      "loss": 0.8336,
+      "step": 26845
+    },
+    {
+      "epoch": 0.737304660679465,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4063467922669503e-06,
+      "loss": 0.7227,
+      "step": 26846
+    },
+    {
+      "epoch": 0.7373321249073083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.405678060022354e-06,
+      "loss": 0.7859,
+      "step": 26847
+    },
+    {
+      "epoch": 0.7373595891351514,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4050093799545827e-06,
+      "loss": 0.8492,
+      "step": 26848
+    },
+    {
+      "epoch": 0.7373870533629947,
+      "grad_norm": 0.0,
+      "learning_rate": 3.404340752068932e-06,
+      "loss": 0.8253,
+      "step": 26849
+    },
+    {
+      "epoch": 0.737414517590838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4036721763706925e-06,
+      "loss": 0.9098,
+      "step": 26850
+    },
+    {
+      "epoch": 0.7374419818186811,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4030036528651543e-06,
+      "loss": 0.7656,
+      "step": 26851
+    },
+    {
+      "epoch": 0.7374694460465244,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4023351815576057e-06,
+      "loss": 0.8467,
+      "step": 26852
+    },
+    {
+      "epoch": 0.7374969102743676,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4016667624533404e-06,
+      "loss": 0.7924,
+      "step": 26853
+    },
+    {
+      "epoch": 0.7375243745022109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.400998395557644e-06,
+      "loss": 0.9237,
+      "step": 26854
+    },
+    {
+      "epoch": 0.7375518387300541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.400330080875801e-06,
+      "loss": 0.8115,
+      "step": 26855
+    },
+    {
+      "epoch": 0.7375793029578973,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3996618184131026e-06,
+      "loss": 0.8658,
+      "step": 26856
+    },
+    {
+      "epoch": 0.7376067671857406,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3989936081748374e-06,
+      "loss": 0.762,
+      "step": 26857
+    },
+    {
+      "epoch": 0.7376342314135838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.398325450166291e-06,
+      "loss": 0.8378,
+      "step": 26858
+    },
+    {
+      "epoch": 0.737661695641427,
+      "grad_norm": 0.0,
+      "learning_rate": 3.397657344392755e-06,
+      "loss": 0.8778,
+      "step": 26859
+    },
+    {
+      "epoch": 0.7376891598692703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.396989290859507e-06,
+      "loss": 0.7618,
+      "step": 26860
+    },
+    {
+      "epoch": 0.7377166240971135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.396321289571842e-06,
+      "loss": 0.8115,
+      "step": 26861
+    },
+    {
+      "epoch": 0.7377440883249567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3956533405350365e-06,
+      "loss": 0.8715,
+      "step": 26862
+    },
+    {
+      "epoch": 0.7377715525528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.394985443754382e-06,
+      "loss": 0.8729,
+      "step": 26863
+    },
+    {
+      "epoch": 0.7377990167806432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3943175992351595e-06,
+      "loss": 0.8186,
+      "step": 26864
+    },
+    {
+      "epoch": 0.7378264810084865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3936498069826586e-06,
+      "loss": 0.9221,
+      "step": 26865
+    },
+    {
+      "epoch": 0.7378539452363296,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3929820670021564e-06,
+      "loss": 0.9531,
+      "step": 26866
+    },
+    {
+      "epoch": 0.7378814094641729,
+      "grad_norm": 0.0,
+      "learning_rate": 3.392314379298939e-06,
+      "loss": 0.854,
+      "step": 26867
+    },
+    {
+      "epoch": 0.7379088736920162,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3916467438782907e-06,
+      "loss": 0.8257,
+      "step": 26868
+    },
+    {
+      "epoch": 0.7379363379198594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3909791607454957e-06,
+      "loss": 0.7316,
+      "step": 26869
+    },
+    {
+      "epoch": 0.7379638021477026,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3903116299058303e-06,
+      "loss": 0.7538,
+      "step": 26870
+    },
+    {
+      "epoch": 0.7379912663755459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.389644151364584e-06,
+      "loss": 0.7816,
+      "step": 26871
+    },
+    {
+      "epoch": 0.7380187306033891,
+      "grad_norm": 0.0,
+      "learning_rate": 3.388976725127029e-06,
+      "loss": 0.7842,
+      "step": 26872
+    },
+    {
+      "epoch": 0.7380461948312323,
+      "grad_norm": 0.0,
+      "learning_rate": 3.388309351198452e-06,
+      "loss": 0.8625,
+      "step": 26873
+    },
+    {
+      "epoch": 0.7380736590590755,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3876420295841316e-06,
+      "loss": 0.7749,
+      "step": 26874
+    },
+    {
+      "epoch": 0.7381011232869188,
+      "grad_norm": 0.0,
+      "learning_rate": 3.38697476028935e-06,
+      "loss": 0.7507,
+      "step": 26875
+    },
+    {
+      "epoch": 0.7381285875147621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.386307543319388e-06,
+      "loss": 0.8976,
+      "step": 26876
+    },
+    {
+      "epoch": 0.7381560517426052,
+      "grad_norm": 0.0,
+      "learning_rate": 3.38564037867952e-06,
+      "loss": 0.8599,
+      "step": 26877
+    },
+    {
+      "epoch": 0.7381835159704485,
+      "grad_norm": 0.0,
+      "learning_rate": 3.384973266375031e-06,
+      "loss": 0.8512,
+      "step": 26878
+    },
+    {
+      "epoch": 0.7382109801982917,
+      "grad_norm": 0.0,
+      "learning_rate": 3.384306206411192e-06,
+      "loss": 0.8057,
+      "step": 26879
+    },
+    {
+      "epoch": 0.738238444426135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.383639198793286e-06,
+      "loss": 0.7579,
+      "step": 26880
+    },
+    {
+      "epoch": 0.7382659086539782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.382972243526589e-06,
+      "loss": 0.8961,
+      "step": 26881
+    },
+    {
+      "epoch": 0.7382933728818214,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3823053406163775e-06,
+      "loss": 0.7782,
+      "step": 26882
+    },
+    {
+      "epoch": 0.7383208371096647,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3816384900679346e-06,
+      "loss": 0.8776,
+      "step": 26883
+    },
+    {
+      "epoch": 0.738348301337508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3809716918865278e-06,
+      "loss": 0.8867,
+      "step": 26884
+    },
+    {
+      "epoch": 0.7383757655653511,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3803049460774396e-06,
+      "loss": 0.9557,
+      "step": 26885
+    },
+    {
+      "epoch": 0.7384032297931944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.37963825264594e-06,
+      "loss": 0.7957,
+      "step": 26886
+    },
+    {
+      "epoch": 0.7384306940210376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.378971611597308e-06,
+      "loss": 0.8985,
+      "step": 26887
+    },
+    {
+      "epoch": 0.7384581582488808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3783050229368164e-06,
+      "loss": 0.8373,
+      "step": 26888
+    },
+    {
+      "epoch": 0.7384856224767241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3776384866697444e-06,
+      "loss": 0.826,
+      "step": 26889
+    },
+    {
+      "epoch": 0.7385130867045673,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3769720028013574e-06,
+      "loss": 0.8743,
+      "step": 26890
+    },
+    {
+      "epoch": 0.7385405509324106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.376305571336935e-06,
+      "loss": 0.8404,
+      "step": 26891
+    },
+    {
+      "epoch": 0.7385680151602537,
+      "grad_norm": 0.0,
+      "learning_rate": 3.375639192281748e-06,
+      "loss": 0.7878,
+      "step": 26892
+    },
+    {
+      "epoch": 0.738595479388097,
+      "grad_norm": 0.0,
+      "learning_rate": 3.374972865641073e-06,
+      "loss": 0.6515,
+      "step": 26893
+    },
+    {
+      "epoch": 0.7386229436159403,
+      "grad_norm": 0.0,
+      "learning_rate": 3.374306591420179e-06,
+      "loss": 0.8708,
+      "step": 26894
+    },
+    {
+      "epoch": 0.7386504078437834,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3736403696243336e-06,
+      "loss": 0.8333,
+      "step": 26895
+    },
+    {
+      "epoch": 0.7386778720716267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.372974200258814e-06,
+      "loss": 0.7829,
+      "step": 26896
+    },
+    {
+      "epoch": 0.73870533629947,
+      "grad_norm": 0.0,
+      "learning_rate": 3.372308083328889e-06,
+      "loss": 0.7528,
+      "step": 26897
+    },
+    {
+      "epoch": 0.7387328005273132,
+      "grad_norm": 0.0,
+      "learning_rate": 3.371642018839829e-06,
+      "loss": 0.7361,
+      "step": 26898
+    },
+    {
+      "epoch": 0.7387602647551564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3709760067969055e-06,
+      "loss": 0.8054,
+      "step": 26899
+    },
+    {
+      "epoch": 0.7387877289829996,
+      "grad_norm": 0.0,
+      "learning_rate": 3.370310047205392e-06,
+      "loss": 0.8806,
+      "step": 26900
+    },
+    {
+      "epoch": 0.7388151932108429,
+      "grad_norm": 0.0,
+      "learning_rate": 3.369644140070548e-06,
+      "loss": 0.8098,
+      "step": 26901
+    },
+    {
+      "epoch": 0.7388426574386862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.368978285397653e-06,
+      "loss": 0.7538,
+      "step": 26902
+    },
+    {
+      "epoch": 0.7388701216665293,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3683124831919665e-06,
+      "loss": 0.8298,
+      "step": 26903
+    },
+    {
+      "epoch": 0.7388975858943726,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3676467334587604e-06,
+      "loss": 0.8974,
+      "step": 26904
+    },
+    {
+      "epoch": 0.7389250501222158,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3669810362033027e-06,
+      "loss": 0.8345,
+      "step": 26905
+    },
+    {
+      "epoch": 0.738952514350059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.366315391430863e-06,
+      "loss": 0.8568,
+      "step": 26906
+    },
+    {
+      "epoch": 0.7389799785779023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.365649799146704e-06,
+      "loss": 0.856,
+      "step": 26907
+    },
+    {
+      "epoch": 0.7390074428057455,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3649842593560924e-06,
+      "loss": 0.8087,
+      "step": 26908
+    },
+    {
+      "epoch": 0.7390349070335888,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3643187720642987e-06,
+      "loss": 0.9477,
+      "step": 26909
+    },
+    {
+      "epoch": 0.739062371261432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.363653337276582e-06,
+      "loss": 0.8008,
+      "step": 26910
+    },
+    {
+      "epoch": 0.7390898354892752,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3629879549982113e-06,
+      "loss": 0.7944,
+      "step": 26911
+    },
+    {
+      "epoch": 0.7391172997171185,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3623226252344543e-06,
+      "loss": 0.9541,
+      "step": 26912
+    },
+    {
+      "epoch": 0.7391447639449616,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3616573479905677e-06,
+      "loss": 0.7965,
+      "step": 26913
+    },
+    {
+      "epoch": 0.7391722281728049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.36099212327182e-06,
+      "loss": 0.9621,
+      "step": 26914
+    },
+    {
+      "epoch": 0.7391996924006482,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3603269510834745e-06,
+      "loss": 0.7657,
+      "step": 26915
+    },
+    {
+      "epoch": 0.7392271566284914,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3596618314307937e-06,
+      "loss": 0.734,
+      "step": 26916
+    },
+    {
+      "epoch": 0.7392546208563346,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3589967643190448e-06,
+      "loss": 0.8453,
+      "step": 26917
+    },
+    {
+      "epoch": 0.7392820850841778,
+      "grad_norm": 0.0,
+      "learning_rate": 3.358331749753486e-06,
+      "loss": 0.7543,
+      "step": 26918
+    },
+    {
+      "epoch": 0.7393095493120211,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3576667877393754e-06,
+      "loss": 0.8334,
+      "step": 26919
+    },
+    {
+      "epoch": 0.7393370135398644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.357001878281978e-06,
+      "loss": 0.8835,
+      "step": 26920
+    },
+    {
+      "epoch": 0.7393644777677075,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3563370213865552e-06,
+      "loss": 0.8828,
+      "step": 26921
+    },
+    {
+      "epoch": 0.7393919419955508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3556722170583667e-06,
+      "loss": 0.8307,
+      "step": 26922
+    },
+    {
+      "epoch": 0.7394194062233941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3550074653026745e-06,
+      "loss": 0.875,
+      "step": 26923
+    },
+    {
+      "epoch": 0.7394468704512372,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3543427661247397e-06,
+      "loss": 0.8188,
+      "step": 26924
+    },
+    {
+      "epoch": 0.7394743346790805,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3536781195298163e-06,
+      "loss": 0.8506,
+      "step": 26925
+    },
+    {
+      "epoch": 0.7395017989069237,
+      "grad_norm": 0.0,
+      "learning_rate": 3.35301352552317e-06,
+      "loss": 0.7581,
+      "step": 26926
+    },
+    {
+      "epoch": 0.739529263134767,
+      "grad_norm": 0.0,
+      "learning_rate": 3.352348984110051e-06,
+      "loss": 0.79,
+      "step": 26927
+    },
+    {
+      "epoch": 0.7395567273626102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3516844952957227e-06,
+      "loss": 0.7433,
+      "step": 26928
+    },
+    {
+      "epoch": 0.7395841915904534,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3510200590854425e-06,
+      "loss": 0.8209,
+      "step": 26929
+    },
+    {
+      "epoch": 0.7396116558182967,
+      "grad_norm": 0.0,
+      "learning_rate": 3.350355675484469e-06,
+      "loss": 0.731,
+      "step": 26930
+    },
+    {
+      "epoch": 0.7396391200461399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.349691344498055e-06,
+      "loss": 0.7715,
+      "step": 26931
+    },
+    {
+      "epoch": 0.7396665842739831,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3490270661314584e-06,
+      "loss": 0.8654,
+      "step": 26932
+    },
+    {
+      "epoch": 0.7396940485018264,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3483628403899404e-06,
+      "loss": 0.773,
+      "step": 26933
+    },
+    {
+      "epoch": 0.7397215127296696,
+      "grad_norm": 0.0,
+      "learning_rate": 3.347698667278748e-06,
+      "loss": 0.8322,
+      "step": 26934
+    },
+    {
+      "epoch": 0.7397489769575128,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3470345468031406e-06,
+      "loss": 0.7444,
+      "step": 26935
+    },
+    {
+      "epoch": 0.7397764411853561,
+      "grad_norm": 0.0,
+      "learning_rate": 3.346370478968377e-06,
+      "loss": 0.8963,
+      "step": 26936
+    },
+    {
+      "epoch": 0.7398039054131993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3457064637797034e-06,
+      "loss": 0.641,
+      "step": 26937
+    },
+    {
+      "epoch": 0.7398313696410426,
+      "grad_norm": 0.0,
+      "learning_rate": 3.345042501242378e-06,
+      "loss": 0.8457,
+      "step": 26938
+    },
+    {
+      "epoch": 0.7398588338688857,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3443785913616543e-06,
+      "loss": 0.8076,
+      "step": 26939
+    },
+    {
+      "epoch": 0.739886298096729,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3437147341427858e-06,
+      "loss": 0.8932,
+      "step": 26940
+    },
+    {
+      "epoch": 0.7399137623245723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3430509295910274e-06,
+      "loss": 0.7458,
+      "step": 26941
+    },
+    {
+      "epoch": 0.7399412265524155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.342387177711629e-06,
+      "loss": 0.7995,
+      "step": 26942
+    },
+    {
+      "epoch": 0.7399686907802587,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3417234785098397e-06,
+      "loss": 0.808,
+      "step": 26943
+    },
+    {
+      "epoch": 0.7399961550081019,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3410598319909127e-06,
+      "loss": 0.8608,
+      "step": 26944
+    },
+    {
+      "epoch": 0.7400236192359452,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3403962381600997e-06,
+      "loss": 0.8097,
+      "step": 26945
+    },
+    {
+      "epoch": 0.7400510834637885,
+      "grad_norm": 0.0,
+      "learning_rate": 3.339732697022651e-06,
+      "loss": 0.8967,
+      "step": 26946
+    },
+    {
+      "epoch": 0.7400785476916316,
+      "grad_norm": 0.0,
+      "learning_rate": 3.339069208583817e-06,
+      "loss": 0.8547,
+      "step": 26947
+    },
+    {
+      "epoch": 0.7401060119194749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.338405772848852e-06,
+      "loss": 0.8327,
+      "step": 26948
+    },
+    {
+      "epoch": 0.7401334761473182,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3377423898229964e-06,
+      "loss": 0.9939,
+      "step": 26949
+    },
+    {
+      "epoch": 0.7401609403751613,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3370790595115077e-06,
+      "loss": 0.8304,
+      "step": 26950
+    },
+    {
+      "epoch": 0.7401884046030046,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3364157819196274e-06,
+      "loss": 0.8972,
+      "step": 26951
+    },
+    {
+      "epoch": 0.7402158688308478,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3357525570526063e-06,
+      "loss": 0.8354,
+      "step": 26952
+    },
+    {
+      "epoch": 0.740243333058691,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3350893849156952e-06,
+      "loss": 0.825,
+      "step": 26953
+    },
+    {
+      "epoch": 0.7402707972865343,
+      "grad_norm": 0.0,
+      "learning_rate": 3.334426265514137e-06,
+      "loss": 0.8559,
+      "step": 26954
+    },
+    {
+      "epoch": 0.7402982615143775,
+      "grad_norm": 0.0,
+      "learning_rate": 3.333763198853178e-06,
+      "loss": 0.7861,
+      "step": 26955
+    },
+    {
+      "epoch": 0.7403257257422208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3331001849380684e-06,
+      "loss": 0.845,
+      "step": 26956
+    },
+    {
+      "epoch": 0.7403531899700639,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3324372237740552e-06,
+      "loss": 0.8231,
+      "step": 26957
+    },
+    {
+      "epoch": 0.7403806541979072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3317743153663783e-06,
+      "loss": 0.7777,
+      "step": 26958
+    },
+    {
+      "epoch": 0.7404081184257505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3311114597202896e-06,
+      "loss": 0.9429,
+      "step": 26959
+    },
+    {
+      "epoch": 0.7404355826535937,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3304486568410264e-06,
+      "loss": 0.875,
+      "step": 26960
+    },
+    {
+      "epoch": 0.7404630468814369,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3297859067338378e-06,
+      "loss": 0.9004,
+      "step": 26961
+    },
+    {
+      "epoch": 0.7404905111092802,
+      "grad_norm": 0.0,
+      "learning_rate": 3.329123209403967e-06,
+      "loss": 0.8405,
+      "step": 26962
+    },
+    {
+      "epoch": 0.7405179753371234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.328460564856657e-06,
+      "loss": 0.8531,
+      "step": 26963
+    },
+    {
+      "epoch": 0.7405454395649667,
+      "grad_norm": 0.0,
+      "learning_rate": 3.327797973097151e-06,
+      "loss": 0.906,
+      "step": 26964
+    },
+    {
+      "epoch": 0.7405729037928098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.327135434130696e-06,
+      "loss": 0.7788,
+      "step": 26965
+    },
+    {
+      "epoch": 0.7406003680206531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.326472947962529e-06,
+      "loss": 0.8662,
+      "step": 26966
+    },
+    {
+      "epoch": 0.7406278322484964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3258105145978916e-06,
+      "loss": 0.8433,
+      "step": 26967
+    },
+    {
+      "epoch": 0.7406552964763395,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3251481340420256e-06,
+      "loss": 0.872,
+      "step": 26968
+    },
+    {
+      "epoch": 0.7406827607041828,
+      "grad_norm": 0.0,
+      "learning_rate": 3.324485806300173e-06,
+      "loss": 0.8443,
+      "step": 26969
+    },
+    {
+      "epoch": 0.740710224932026,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3238235313775744e-06,
+      "loss": 0.8258,
+      "step": 26970
+    },
+    {
+      "epoch": 0.7407376891598693,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3231613092794747e-06,
+      "loss": 0.8069,
+      "step": 26971
+    },
+    {
+      "epoch": 0.7407651533877125,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3224991400111052e-06,
+      "loss": 0.8324,
+      "step": 26972
+    },
+    {
+      "epoch": 0.7407926176155557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3218370235777085e-06,
+      "loss": 0.7636,
+      "step": 26973
+    },
+    {
+      "epoch": 0.740820081843399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.321174959984528e-06,
+      "loss": 0.7881,
+      "step": 26974
+    },
+    {
+      "epoch": 0.7408475460712423,
+      "grad_norm": 0.0,
+      "learning_rate": 3.320512949236795e-06,
+      "loss": 0.8732,
+      "step": 26975
+    },
+    {
+      "epoch": 0.7408750102990854,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3198509913397504e-06,
+      "loss": 0.7378,
+      "step": 26976
+    },
+    {
+      "epoch": 0.7409024745269287,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3191890862986366e-06,
+      "loss": 0.7595,
+      "step": 26977
+    },
+    {
+      "epoch": 0.7409299387547719,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3185272341186824e-06,
+      "loss": 0.7917,
+      "step": 26978
+    },
+    {
+      "epoch": 0.7409574029826151,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3178654348051296e-06,
+      "loss": 0.8104,
+      "step": 26979
+    },
+    {
+      "epoch": 0.7409848672104584,
+      "grad_norm": 0.0,
+      "learning_rate": 3.317203688363214e-06,
+      "loss": 0.8243,
+      "step": 26980
+    },
+    {
+      "epoch": 0.7410123314383016,
+      "grad_norm": 0.0,
+      "learning_rate": 3.316541994798175e-06,
+      "loss": 0.8429,
+      "step": 26981
+    },
+    {
+      "epoch": 0.7410397956661449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3158803541152397e-06,
+      "loss": 0.8535,
+      "step": 26982
+    },
+    {
+      "epoch": 0.741067259893988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3152187663196534e-06,
+      "loss": 0.9523,
+      "step": 26983
+    },
+    {
+      "epoch": 0.7410947241218313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3145572314166406e-06,
+      "loss": 0.7424,
+      "step": 26984
+    },
+    {
+      "epoch": 0.7411221883496746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.313895749411442e-06,
+      "loss": 0.8909,
+      "step": 26985
+    },
+    {
+      "epoch": 0.7411496525775177,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3132343203092897e-06,
+      "loss": 0.8568,
+      "step": 26986
+    },
+    {
+      "epoch": 0.741177116805361,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3125729441154188e-06,
+      "loss": 0.8625,
+      "step": 26987
+    },
+    {
+      "epoch": 0.7412045810332043,
+      "grad_norm": 0.0,
+      "learning_rate": 3.31191162083506e-06,
+      "loss": 0.9326,
+      "step": 26988
+    },
+    {
+      "epoch": 0.7412320452610475,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3112503504734504e-06,
+      "loss": 0.8855,
+      "step": 26989
+    },
+    {
+      "epoch": 0.7412595094888907,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3105891330358185e-06,
+      "loss": 0.8139,
+      "step": 26990
+    },
+    {
+      "epoch": 0.7412869737167339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3099279685273946e-06,
+      "loss": 0.8179,
+      "step": 26991
+    },
+    {
+      "epoch": 0.7413144379445772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3092668569534112e-06,
+      "loss": 0.8162,
+      "step": 26992
+    },
+    {
+      "epoch": 0.7413419021724205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.308605798319101e-06,
+      "loss": 0.8141,
+      "step": 26993
+    },
+    {
+      "epoch": 0.7413693664002636,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3079447926296936e-06,
+      "loss": 0.7734,
+      "step": 26994
+    },
+    {
+      "epoch": 0.7413968306281069,
+      "grad_norm": 0.0,
+      "learning_rate": 3.307283839890422e-06,
+      "loss": 0.7494,
+      "step": 26995
+    },
+    {
+      "epoch": 0.7414242948559501,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3066229401065098e-06,
+      "loss": 0.7545,
+      "step": 26996
+    },
+    {
+      "epoch": 0.7414517590837933,
+      "grad_norm": 0.0,
+      "learning_rate": 3.305962093283194e-06,
+      "loss": 0.7635,
+      "step": 26997
+    },
+    {
+      "epoch": 0.7414792233116366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.305301299425695e-06,
+      "loss": 0.8134,
+      "step": 26998
+    },
+    {
+      "epoch": 0.7415066875394798,
+      "grad_norm": 0.0,
+      "learning_rate": 3.304640558539245e-06,
+      "loss": 0.788,
+      "step": 26999
+    },
+    {
+      "epoch": 0.7415341517673231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3039798706290737e-06,
+      "loss": 0.7831,
+      "step": 27000
+    },
+    {
+      "epoch": 0.7415616159951663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3033192357004096e-06,
+      "loss": 0.8433,
+      "step": 27001
+    },
+    {
+      "epoch": 0.7415890802230095,
+      "grad_norm": 0.0,
+      "learning_rate": 3.302658653758475e-06,
+      "loss": 0.7831,
+      "step": 27002
+    },
+    {
+      "epoch": 0.7416165444508528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3019981248084996e-06,
+      "loss": 0.8624,
+      "step": 27003
+    },
+    {
+      "epoch": 0.741644008678696,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3013376488557094e-06,
+      "loss": 0.8246,
+      "step": 27004
+    },
+    {
+      "epoch": 0.7416714729065392,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3006772259053333e-06,
+      "loss": 0.8777,
+      "step": 27005
+    },
+    {
+      "epoch": 0.7416989371343825,
+      "grad_norm": 0.0,
+      "learning_rate": 3.300016855962591e-06,
+      "loss": 0.8822,
+      "step": 27006
+    },
+    {
+      "epoch": 0.7417264013622257,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2993565390327142e-06,
+      "loss": 0.8409,
+      "step": 27007
+    },
+    {
+      "epoch": 0.741753865590069,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2986962751209207e-06,
+      "loss": 0.8108,
+      "step": 27008
+    },
+    {
+      "epoch": 0.7417813298179121,
+      "grad_norm": 0.0,
+      "learning_rate": 3.298036064232438e-06,
+      "loss": 0.9118,
+      "step": 27009
+    },
+    {
+      "epoch": 0.7418087940457554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.297375906372491e-06,
+      "loss": 0.8672,
+      "step": 27010
+    },
+    {
+      "epoch": 0.7418362582735987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.296715801546301e-06,
+      "loss": 0.7356,
+      "step": 27011
+    },
+    {
+      "epoch": 0.7418637225014418,
+      "grad_norm": 0.0,
+      "learning_rate": 3.296055749759096e-06,
+      "loss": 0.8206,
+      "step": 27012
+    },
+    {
+      "epoch": 0.7418911867292851,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2953957510160904e-06,
+      "loss": 0.879,
+      "step": 27013
+    },
+    {
+      "epoch": 0.7419186509571284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.294735805322515e-06,
+      "loss": 0.9169,
+      "step": 27014
+    },
+    {
+      "epoch": 0.7419461151849716,
+      "grad_norm": 0.0,
+      "learning_rate": 3.294075912683584e-06,
+      "loss": 0.8189,
+      "step": 27015
+    },
+    {
+      "epoch": 0.7419735794128148,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2934160731045216e-06,
+      "loss": 0.8047,
+      "step": 27016
+    },
+    {
+      "epoch": 0.742001043640658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2927562865905506e-06,
+      "loss": 0.8322,
+      "step": 27017
+    },
+    {
+      "epoch": 0.7420285078685013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.292096553146892e-06,
+      "loss": 0.8056,
+      "step": 27018
+    },
+    {
+      "epoch": 0.7420559720963446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2914368727787603e-06,
+      "loss": 0.8655,
+      "step": 27019
+    },
+    {
+      "epoch": 0.7420834363241877,
+      "grad_norm": 0.0,
+      "learning_rate": 3.290777245491379e-06,
+      "loss": 0.9004,
+      "step": 27020
+    },
+    {
+      "epoch": 0.742110900552031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2901176712899706e-06,
+      "loss": 0.814,
+      "step": 27021
+    },
+    {
+      "epoch": 0.7421383647798742,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2894581501797464e-06,
+      "loss": 0.823,
+      "step": 27022
+    },
+    {
+      "epoch": 0.7421658290077174,
+      "grad_norm": 0.0,
+      "learning_rate": 3.28879868216593e-06,
+      "loss": 0.7837,
+      "step": 27023
+    },
+    {
+      "epoch": 0.7421932932355607,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2881392672537403e-06,
+      "loss": 0.8721,
+      "step": 27024
+    },
+    {
+      "epoch": 0.7422207574634039,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2874799054483897e-06,
+      "loss": 0.8318,
+      "step": 27025
+    },
+    {
+      "epoch": 0.7422482216912472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2868205967550983e-06,
+      "loss": 0.8021,
+      "step": 27026
+    },
+    {
+      "epoch": 0.7422756859190904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2861613411790825e-06,
+      "loss": 0.8645,
+      "step": 27027
+    },
+    {
+      "epoch": 0.7423031501469336,
+      "grad_norm": 0.0,
+      "learning_rate": 3.285502138725559e-06,
+      "loss": 0.8017,
+      "step": 27028
+    },
+    {
+      "epoch": 0.7423306143747769,
+      "grad_norm": 0.0,
+      "learning_rate": 3.284842989399747e-06,
+      "loss": 0.8031,
+      "step": 27029
+    },
+    {
+      "epoch": 0.74235807860262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2841838932068583e-06,
+      "loss": 0.8291,
+      "step": 27030
+    },
+    {
+      "epoch": 0.7423855428304633,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2835248501521044e-06,
+      "loss": 0.8733,
+      "step": 27031
+    },
+    {
+      "epoch": 0.7424130070583066,
+      "grad_norm": 0.0,
+      "learning_rate": 3.282865860240705e-06,
+      "loss": 0.8405,
+      "step": 27032
+    },
+    {
+      "epoch": 0.7424404712861498,
+      "grad_norm": 0.0,
+      "learning_rate": 3.282206923477871e-06,
+      "loss": 0.7953,
+      "step": 27033
+    },
+    {
+      "epoch": 0.742467935513993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.281548039868819e-06,
+      "loss": 0.8297,
+      "step": 27034
+    },
+    {
+      "epoch": 0.7424953997418362,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2808892094187616e-06,
+      "loss": 0.8123,
+      "step": 27035
+    },
+    {
+      "epoch": 0.7425228639696795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.280230432132915e-06,
+      "loss": 0.8712,
+      "step": 27036
+    },
+    {
+      "epoch": 0.7425503281975228,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2795717080164847e-06,
+      "loss": 0.7771,
+      "step": 27037
+    },
+    {
+      "epoch": 0.7425777924253659,
+      "grad_norm": 0.0,
+      "learning_rate": 3.278913037074689e-06,
+      "loss": 0.7854,
+      "step": 27038
+    },
+    {
+      "epoch": 0.7426052566532092,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2782544193127343e-06,
+      "loss": 0.8158,
+      "step": 27039
+    },
+    {
+      "epoch": 0.7426327208810525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2775958547358334e-06,
+      "loss": 0.9127,
+      "step": 27040
+    },
+    {
+      "epoch": 0.7426601851088956,
+      "grad_norm": 0.0,
+      "learning_rate": 3.276937343349199e-06,
+      "loss": 0.8561,
+      "step": 27041
+    },
+    {
+      "epoch": 0.7426876493367389,
+      "grad_norm": 0.0,
+      "learning_rate": 3.276278885158043e-06,
+      "loss": 0.7864,
+      "step": 27042
+    },
+    {
+      "epoch": 0.7427151135645821,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2756204801675704e-06,
+      "loss": 0.7569,
+      "step": 27043
+    },
+    {
+      "epoch": 0.7427425777924254,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2749621283829934e-06,
+      "loss": 0.8246,
+      "step": 27044
+    },
+    {
+      "epoch": 0.7427700420202686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.274303829809524e-06,
+      "loss": 0.8254,
+      "step": 27045
+    },
+    {
+      "epoch": 0.7427975062481118,
+      "grad_norm": 0.0,
+      "learning_rate": 3.273645584452365e-06,
+      "loss": 0.8883,
+      "step": 27046
+    },
+    {
+      "epoch": 0.7428249704759551,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2729873923167275e-06,
+      "loss": 0.7455,
+      "step": 27047
+    },
+    {
+      "epoch": 0.7428524347037982,
+      "grad_norm": 0.0,
+      "learning_rate": 3.272329253407823e-06,
+      "loss": 0.8412,
+      "step": 27048
+    },
+    {
+      "epoch": 0.7428798989316415,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2716711677308523e-06,
+      "loss": 0.7856,
+      "step": 27049
+    },
+    {
+      "epoch": 0.7429073631594848,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2710131352910257e-06,
+      "loss": 0.9538,
+      "step": 27050
+    },
+    {
+      "epoch": 0.742934827387328,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2703551560935497e-06,
+      "loss": 0.9045,
+      "step": 27051
+    },
+    {
+      "epoch": 0.7429622916151712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2696972301436302e-06,
+      "loss": 0.9392,
+      "step": 27052
+    },
+    {
+      "epoch": 0.7429897558430145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2690393574464765e-06,
+      "loss": 0.7832,
+      "step": 27053
+    },
+    {
+      "epoch": 0.7430172200708577,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2683815380072905e-06,
+      "loss": 0.7668,
+      "step": 27054
+    },
+    {
+      "epoch": 0.743044684298701,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2677237718312748e-06,
+      "loss": 0.8522,
+      "step": 27055
+    },
+    {
+      "epoch": 0.7430721485265441,
+      "grad_norm": 0.0,
+      "learning_rate": 3.267066058923636e-06,
+      "loss": 0.8928,
+      "step": 27056
+    },
+    {
+      "epoch": 0.7430996127543874,
+      "grad_norm": 0.0,
+      "learning_rate": 3.266408399289579e-06,
+      "loss": 0.8721,
+      "step": 27057
+    },
+    {
+      "epoch": 0.7431270769822307,
+      "grad_norm": 0.0,
+      "learning_rate": 3.265750792934307e-06,
+      "loss": 0.8534,
+      "step": 27058
+    },
+    {
+      "epoch": 0.7431545412100738,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2650932398630233e-06,
+      "loss": 0.6961,
+      "step": 27059
+    },
+    {
+      "epoch": 0.7431820054379171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2644357400809332e-06,
+      "loss": 0.8242,
+      "step": 27060
+    },
+    {
+      "epoch": 0.7432094696657603,
+      "grad_norm": 0.0,
+      "learning_rate": 3.263778293593234e-06,
+      "loss": 0.7487,
+      "step": 27061
+    },
+    {
+      "epoch": 0.7432369338936036,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2631209004051335e-06,
+      "loss": 0.6999,
+      "step": 27062
+    },
+    {
+      "epoch": 0.7432643981214468,
+      "grad_norm": 0.0,
+      "learning_rate": 3.262463560521827e-06,
+      "loss": 0.8099,
+      "step": 27063
+    },
+    {
+      "epoch": 0.74329186234929,
+      "grad_norm": 0.0,
+      "learning_rate": 3.261806273948519e-06,
+      "loss": 0.8007,
+      "step": 27064
+    },
+    {
+      "epoch": 0.7433193265771333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.261149040690409e-06,
+      "loss": 0.7526,
+      "step": 27065
+    },
+    {
+      "epoch": 0.7433467908049766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2604918607527023e-06,
+      "loss": 0.8626,
+      "step": 27066
+    },
+    {
+      "epoch": 0.7433742550328197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.259834734140591e-06,
+      "loss": 0.8532,
+      "step": 27067
+    },
+    {
+      "epoch": 0.743401719260663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2591776608592784e-06,
+      "loss": 0.8554,
+      "step": 27068
+    },
+    {
+      "epoch": 0.7434291834885062,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2585206409139656e-06,
+      "loss": 0.7911,
+      "step": 27069
+    },
+    {
+      "epoch": 0.7434566477163495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2578636743098467e-06,
+      "loss": 0.7991,
+      "step": 27070
+    },
+    {
+      "epoch": 0.7434841119441927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2572067610521253e-06,
+      "loss": 0.9107,
+      "step": 27071
+    },
+    {
+      "epoch": 0.7435115761720359,
+      "grad_norm": 0.0,
+      "learning_rate": 3.256549901145992e-06,
+      "loss": 0.7666,
+      "step": 27072
+    },
+    {
+      "epoch": 0.7435390403998792,
+      "grad_norm": 0.0,
+      "learning_rate": 3.255893094596648e-06,
+      "loss": 0.762,
+      "step": 27073
+    },
+    {
+      "epoch": 0.7435665046277224,
+      "grad_norm": 0.0,
+      "learning_rate": 3.255236341409291e-06,
+      "loss": 0.8294,
+      "step": 27074
+    },
+    {
+      "epoch": 0.7435939688555656,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2545796415891164e-06,
+      "loss": 0.8492,
+      "step": 27075
+    },
+    {
+      "epoch": 0.7436214330834089,
+      "grad_norm": 0.0,
+      "learning_rate": 3.253922995141321e-06,
+      "loss": 0.778,
+      "step": 27076
+    },
+    {
+      "epoch": 0.743648897311252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2532664020711026e-06,
+      "loss": 0.902,
+      "step": 27077
+    },
+    {
+      "epoch": 0.7436763615390953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2526098623836542e-06,
+      "loss": 0.7555,
+      "step": 27078
+    },
+    {
+      "epoch": 0.7437038257669386,
+      "grad_norm": 0.0,
+      "learning_rate": 3.251953376084167e-06,
+      "loss": 0.8785,
+      "step": 27079
+    },
+    {
+      "epoch": 0.7437312899947818,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2512969431778385e-06,
+      "loss": 0.7641,
+      "step": 27080
+    },
+    {
+      "epoch": 0.743758754222625,
+      "grad_norm": 0.0,
+      "learning_rate": 3.250640563669862e-06,
+      "loss": 0.75,
+      "step": 27081
+    },
+    {
+      "epoch": 0.7437862184504682,
+      "grad_norm": 0.0,
+      "learning_rate": 3.249984237565433e-06,
+      "loss": 0.8203,
+      "step": 27082
+    },
+    {
+      "epoch": 0.7438136826783115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2493279648697463e-06,
+      "loss": 0.8204,
+      "step": 27083
+    },
+    {
+      "epoch": 0.7438411469061548,
+      "grad_norm": 0.0,
+      "learning_rate": 3.248671745587989e-06,
+      "loss": 0.8939,
+      "step": 27084
+    },
+    {
+      "epoch": 0.7438686111339979,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2480155797253553e-06,
+      "loss": 0.7615,
+      "step": 27085
+    },
+    {
+      "epoch": 0.7438960753618412,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2473594672870413e-06,
+      "loss": 0.9093,
+      "step": 27086
+    },
+    {
+      "epoch": 0.7439235395896845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.246703408278231e-06,
+      "loss": 0.8535,
+      "step": 27087
+    },
+    {
+      "epoch": 0.7439510038175277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.24604740270412e-06,
+      "loss": 0.8314,
+      "step": 27088
+    },
+    {
+      "epoch": 0.7439784680453709,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2453914505699015e-06,
+      "loss": 0.7887,
+      "step": 27089
+    },
+    {
+      "epoch": 0.7440059322732141,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2447355518807588e-06,
+      "loss": 0.8117,
+      "step": 27090
+    },
+    {
+      "epoch": 0.7440333965010574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.244079706641886e-06,
+      "loss": 0.6866,
+      "step": 27091
+    },
+    {
+      "epoch": 0.7440608607289007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.24342391485847e-06,
+      "loss": 0.8639,
+      "step": 27092
+    },
+    {
+      "epoch": 0.7440883249567438,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2427681765357054e-06,
+      "loss": 0.8404,
+      "step": 27093
+    },
+    {
+      "epoch": 0.7441157891845871,
+      "grad_norm": 0.0,
+      "learning_rate": 3.242112491678774e-06,
+      "loss": 0.884,
+      "step": 27094
+    },
+    {
+      "epoch": 0.7441432534124303,
+      "grad_norm": 0.0,
+      "learning_rate": 3.241456860292869e-06,
+      "loss": 0.7659,
+      "step": 27095
+    },
+    {
+      "epoch": 0.7441707176402735,
+      "grad_norm": 0.0,
+      "learning_rate": 3.240801282383171e-06,
+      "loss": 0.8217,
+      "step": 27096
+    },
+    {
+      "epoch": 0.7441981818681168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2401457579548725e-06,
+      "loss": 0.8399,
+      "step": 27097
+    },
+    {
+      "epoch": 0.74422564609596,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2394902870131596e-06,
+      "loss": 0.8157,
+      "step": 27098
+    },
+    {
+      "epoch": 0.7442531103238033,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2388348695632176e-06,
+      "loss": 0.9013,
+      "step": 27099
+    },
+    {
+      "epoch": 0.7442805745516465,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2381795056102337e-06,
+      "loss": 0.8411,
+      "step": 27100
+    },
+    {
+      "epoch": 0.7443080387794897,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2375241951593948e-06,
+      "loss": 0.9162,
+      "step": 27101
+    },
+    {
+      "epoch": 0.744335503007333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.236868938215886e-06,
+      "loss": 0.858,
+      "step": 27102
+    },
+    {
+      "epoch": 0.7443629672351761,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2362137347848854e-06,
+      "loss": 0.7864,
+      "step": 27103
+    },
+    {
+      "epoch": 0.7443904314630194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.235558584871582e-06,
+      "loss": 0.7458,
+      "step": 27104
+    },
+    {
+      "epoch": 0.7444178956908627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2349034884811593e-06,
+      "loss": 0.8812,
+      "step": 27105
+    },
+    {
+      "epoch": 0.7444453599187059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2342484456188005e-06,
+      "loss": 0.7854,
+      "step": 27106
+    },
+    {
+      "epoch": 0.7444728241465491,
+      "grad_norm": 0.0,
+      "learning_rate": 3.233593456289693e-06,
+      "loss": 0.836,
+      "step": 27107
+    },
+    {
+      "epoch": 0.7445002883743923,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2329385204990116e-06,
+      "loss": 0.7824,
+      "step": 27108
+    },
+    {
+      "epoch": 0.7445277526022356,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2322836382519407e-06,
+      "loss": 0.822,
+      "step": 27109
+    },
+    {
+      "epoch": 0.7445552168300789,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2316288095536672e-06,
+      "loss": 0.806,
+      "step": 27110
+    },
+    {
+      "epoch": 0.744582681057922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.230974034409365e-06,
+      "loss": 0.8649,
+      "step": 27111
+    },
+    {
+      "epoch": 0.7446101452857653,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2303193128242194e-06,
+      "loss": 0.8954,
+      "step": 27112
+    },
+    {
+      "epoch": 0.7446376095136086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.229664644803412e-06,
+      "loss": 0.8081,
+      "step": 27113
+    },
+    {
+      "epoch": 0.7446650737414517,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2290100303521176e-06,
+      "loss": 0.8716,
+      "step": 27114
+    },
+    {
+      "epoch": 0.744692537969295,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2283554694755192e-06,
+      "loss": 0.805,
+      "step": 27115
+    },
+    {
+      "epoch": 0.7447200021971382,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2277009621787946e-06,
+      "loss": 0.8851,
+      "step": 27116
+    },
+    {
+      "epoch": 0.7447474664249815,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2270465084671267e-06,
+      "loss": 0.8748,
+      "step": 27117
+    },
+    {
+      "epoch": 0.7447749306528247,
+      "grad_norm": 0.0,
+      "learning_rate": 3.226392108345687e-06,
+      "loss": 0.9047,
+      "step": 27118
+    },
+    {
+      "epoch": 0.7448023948806679,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2257377618196606e-06,
+      "loss": 0.7815,
+      "step": 27119
+    },
+    {
+      "epoch": 0.7448298591085112,
+      "grad_norm": 0.0,
+      "learning_rate": 3.225083468894217e-06,
+      "loss": 0.7992,
+      "step": 27120
+    },
+    {
+      "epoch": 0.7448573233363543,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2244292295745383e-06,
+      "loss": 0.8859,
+      "step": 27121
+    },
+    {
+      "epoch": 0.7448847875641976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2237750438657987e-06,
+      "loss": 0.8203,
+      "step": 27122
+    },
+    {
+      "epoch": 0.7449122517920409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.223120911773178e-06,
+      "loss": 0.9229,
+      "step": 27123
+    },
+    {
+      "epoch": 0.7449397160198841,
+      "grad_norm": 0.0,
+      "learning_rate": 3.222466833301848e-06,
+      "loss": 0.8222,
+      "step": 27124
+    },
+    {
+      "epoch": 0.7449671802477273,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2218128084569888e-06,
+      "loss": 0.7523,
+      "step": 27125
+    },
+    {
+      "epoch": 0.7449946444755706,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2211588372437742e-06,
+      "loss": 0.8668,
+      "step": 27126
+    },
+    {
+      "epoch": 0.7450221087034138,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2205049196673723e-06,
+      "loss": 0.8468,
+      "step": 27127
+    },
+    {
+      "epoch": 0.7450495729312571,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2198510557329608e-06,
+      "loss": 0.8122,
+      "step": 27128
+    },
+    {
+      "epoch": 0.7450770371591002,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2191972454457156e-06,
+      "loss": 0.8362,
+      "step": 27129
+    },
+    {
+      "epoch": 0.7451045013869435,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2185434888108104e-06,
+      "loss": 0.85,
+      "step": 27130
+    },
+    {
+      "epoch": 0.7451319656147868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2178897858334133e-06,
+      "loss": 0.8126,
+      "step": 27131
+    },
+    {
+      "epoch": 0.74515942984263,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2172361365186998e-06,
+      "loss": 0.8709,
+      "step": 27132
+    },
+    {
+      "epoch": 0.7451868940704732,
+      "grad_norm": 0.0,
+      "learning_rate": 3.216582540871841e-06,
+      "loss": 0.8409,
+      "step": 27133
+    },
+    {
+      "epoch": 0.7452143582983164,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2159289988980124e-06,
+      "loss": 0.8015,
+      "step": 27134
+    },
+    {
+      "epoch": 0.7452418225261597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.215275510602378e-06,
+      "loss": 0.8026,
+      "step": 27135
+    },
+    {
+      "epoch": 0.745269286754003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.214622075990115e-06,
+      "loss": 0.8443,
+      "step": 27136
+    },
+    {
+      "epoch": 0.7452967509818461,
+      "grad_norm": 0.0,
+      "learning_rate": 3.213968695066387e-06,
+      "loss": 0.7816,
+      "step": 27137
+    },
+    {
+      "epoch": 0.7453242152096894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.213315367836368e-06,
+      "loss": 0.8509,
+      "step": 27138
+    },
+    {
+      "epoch": 0.7453516794375327,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2126620943052267e-06,
+      "loss": 0.7447,
+      "step": 27139
+    },
+    {
+      "epoch": 0.7453791436653758,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2120088744781307e-06,
+      "loss": 0.8707,
+      "step": 27140
+    },
+    {
+      "epoch": 0.7454066078932191,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2113557083602543e-06,
+      "loss": 0.824,
+      "step": 27141
+    },
+    {
+      "epoch": 0.7454340721210623,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2107025959567616e-06,
+      "loss": 0.7042,
+      "step": 27142
+    },
+    {
+      "epoch": 0.7454615363489056,
+      "grad_norm": 0.0,
+      "learning_rate": 3.210049537272816e-06,
+      "loss": 0.8824,
+      "step": 27143
+    },
+    {
+      "epoch": 0.7454890005767488,
+      "grad_norm": 0.0,
+      "learning_rate": 3.209396532313589e-06,
+      "loss": 0.7998,
+      "step": 27144
+    },
+    {
+      "epoch": 0.745516464804592,
+      "grad_norm": 0.0,
+      "learning_rate": 3.208743581084247e-06,
+      "loss": 0.8646,
+      "step": 27145
+    },
+    {
+      "epoch": 0.7455439290324353,
+      "grad_norm": 0.0,
+      "learning_rate": 3.208090683589956e-06,
+      "loss": 0.8169,
+      "step": 27146
+    },
+    {
+      "epoch": 0.7455713932602784,
+      "grad_norm": 0.0,
+      "learning_rate": 3.207437839835883e-06,
+      "loss": 0.7909,
+      "step": 27147
+    },
+    {
+      "epoch": 0.7455988574881217,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2067850498271957e-06,
+      "loss": 0.8679,
+      "step": 27148
+    },
+    {
+      "epoch": 0.745626321715965,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2061323135690536e-06,
+      "loss": 0.8219,
+      "step": 27149
+    },
+    {
+      "epoch": 0.7456537859438082,
+      "grad_norm": 0.0,
+      "learning_rate": 3.205479631066628e-06,
+      "loss": 0.8957,
+      "step": 27150
+    },
+    {
+      "epoch": 0.7456812501716514,
+      "grad_norm": 0.0,
+      "learning_rate": 3.204827002325076e-06,
+      "loss": 0.8782,
+      "step": 27151
+    },
+    {
+      "epoch": 0.7457087143994947,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2041744273495644e-06,
+      "loss": 0.7686,
+      "step": 27152
+    },
+    {
+      "epoch": 0.7457361786273379,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2035219061452572e-06,
+      "loss": 0.8353,
+      "step": 27153
+    },
+    {
+      "epoch": 0.7457636428551812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.202869438717321e-06,
+      "loss": 0.7938,
+      "step": 27154
+    },
+    {
+      "epoch": 0.7457911070830243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2022170250709117e-06,
+      "loss": 0.8724,
+      "step": 27155
+    },
+    {
+      "epoch": 0.7458185713108676,
+      "grad_norm": 0.0,
+      "learning_rate": 3.201564665211194e-06,
+      "loss": 0.7447,
+      "step": 27156
+    },
+    {
+      "epoch": 0.7458460355387109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2009123591433342e-06,
+      "loss": 0.6865,
+      "step": 27157
+    },
+    {
+      "epoch": 0.745873499766554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.200260106872486e-06,
+      "loss": 0.8969,
+      "step": 27158
+    },
+    {
+      "epoch": 0.7459009639943973,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1996079084038134e-06,
+      "loss": 0.8433,
+      "step": 27159
+    },
+    {
+      "epoch": 0.7459284282222405,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1989557637424806e-06,
+      "loss": 0.8279,
+      "step": 27160
+    },
+    {
+      "epoch": 0.7459558924500838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.198303672893641e-06,
+      "loss": 0.8532,
+      "step": 27161
+    },
+    {
+      "epoch": 0.745983356677927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1976516358624577e-06,
+      "loss": 0.7656,
+      "step": 27162
+    },
+    {
+      "epoch": 0.7460108209057702,
+      "grad_norm": 0.0,
+      "learning_rate": 3.196999652654089e-06,
+      "loss": 0.7992,
+      "step": 27163
+    },
+    {
+      "epoch": 0.7460382851336135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.196347723273695e-06,
+      "loss": 0.8666,
+      "step": 27164
+    },
+    {
+      "epoch": 0.7460657493614568,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1956958477264367e-06,
+      "loss": 0.8333,
+      "step": 27165
+    },
+    {
+      "epoch": 0.7460932135892999,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1950440260174677e-06,
+      "loss": 0.8489,
+      "step": 27166
+    },
+    {
+      "epoch": 0.7461206778171432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1943922581519437e-06,
+      "loss": 0.8263,
+      "step": 27167
+    },
+    {
+      "epoch": 0.7461481420449864,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1937405441350244e-06,
+      "loss": 0.8088,
+      "step": 27168
+    },
+    {
+      "epoch": 0.7461756062728296,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1930888839718665e-06,
+      "loss": 0.8279,
+      "step": 27169
+    },
+    {
+      "epoch": 0.7462030705006729,
+      "grad_norm": 0.0,
+      "learning_rate": 3.192437277667626e-06,
+      "loss": 0.8543,
+      "step": 27170
+    },
+    {
+      "epoch": 0.7462305347285161,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1917857252274587e-06,
+      "loss": 0.7197,
+      "step": 27171
+    },
+    {
+      "epoch": 0.7462579989563594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1911342266565226e-06,
+      "loss": 0.8593,
+      "step": 27172
+    },
+    {
+      "epoch": 0.7462854631842025,
+      "grad_norm": 0.0,
+      "learning_rate": 3.190482781959967e-06,
+      "loss": 0.7744,
+      "step": 27173
+    },
+    {
+      "epoch": 0.7463129274120458,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1898313911429533e-06,
+      "loss": 0.754,
+      "step": 27174
+    },
+    {
+      "epoch": 0.7463403916398891,
+      "grad_norm": 0.0,
+      "learning_rate": 3.189180054210629e-06,
+      "loss": 0.8442,
+      "step": 27175
+    },
+    {
+      "epoch": 0.7463678558677322,
+      "grad_norm": 0.0,
+      "learning_rate": 3.188528771168149e-06,
+      "loss": 0.8387,
+      "step": 27176
+    },
+    {
+      "epoch": 0.7463953200955755,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1878775420206685e-06,
+      "loss": 0.7795,
+      "step": 27177
+    },
+    {
+      "epoch": 0.7464227843234188,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1872263667733427e-06,
+      "loss": 0.7959,
+      "step": 27178
+    },
+    {
+      "epoch": 0.746450248551262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1865752454313172e-06,
+      "loss": 0.8869,
+      "step": 27179
+    },
+    {
+      "epoch": 0.7464777127791052,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1859241779997485e-06,
+      "loss": 0.8292,
+      "step": 27180
+    },
+    {
+      "epoch": 0.7465051770069484,
+      "grad_norm": 0.0,
+      "learning_rate": 3.18527316448379e-06,
+      "loss": 0.877,
+      "step": 27181
+    },
+    {
+      "epoch": 0.7465326412347917,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1846222048885867e-06,
+      "loss": 0.9272,
+      "step": 27182
+    },
+    {
+      "epoch": 0.746560105462635,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1839712992192927e-06,
+      "loss": 0.8336,
+      "step": 27183
+    },
+    {
+      "epoch": 0.7465875696904781,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1833204474810607e-06,
+      "loss": 0.6886,
+      "step": 27184
+    },
+    {
+      "epoch": 0.7466150339183214,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1826696496790353e-06,
+      "loss": 0.8772,
+      "step": 27185
+    },
+    {
+      "epoch": 0.7466424981461646,
+      "grad_norm": 0.0,
+      "learning_rate": 3.182018905818368e-06,
+      "loss": 0.8037,
+      "step": 27186
+    },
+    {
+      "epoch": 0.7466699623740078,
+      "grad_norm": 0.0,
+      "learning_rate": 3.181368215904209e-06,
+      "loss": 0.8374,
+      "step": 27187
+    },
+    {
+      "epoch": 0.7466974266018511,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1807175799417056e-06,
+      "loss": 0.8717,
+      "step": 27188
+    },
+    {
+      "epoch": 0.7467248908296943,
+      "grad_norm": 0.0,
+      "learning_rate": 3.180066997936009e-06,
+      "loss": 0.7935,
+      "step": 27189
+    },
+    {
+      "epoch": 0.7467523550575376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.179416469892265e-06,
+      "loss": 0.8842,
+      "step": 27190
+    },
+    {
+      "epoch": 0.7467798192853808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.178765995815616e-06,
+      "loss": 0.8267,
+      "step": 27191
+    },
+    {
+      "epoch": 0.746807283513224,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1781155757112125e-06,
+      "loss": 0.8623,
+      "step": 27192
+    },
+    {
+      "epoch": 0.7468347477410673,
+      "grad_norm": 0.0,
+      "learning_rate": 3.177465209584202e-06,
+      "loss": 0.8759,
+      "step": 27193
+    },
+    {
+      "epoch": 0.7468622119689104,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1768148974397282e-06,
+      "loss": 0.8827,
+      "step": 27194
+    },
+    {
+      "epoch": 0.7468896761967537,
+      "grad_norm": 0.0,
+      "learning_rate": 3.176164639282943e-06,
+      "loss": 0.7835,
+      "step": 27195
+    },
+    {
+      "epoch": 0.746917140424597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1755144351189813e-06,
+      "loss": 0.7906,
+      "step": 27196
+    },
+    {
+      "epoch": 0.7469446046524402,
+      "grad_norm": 0.0,
+      "learning_rate": 3.174864284952994e-06,
+      "loss": 0.8184,
+      "step": 27197
+    },
+    {
+      "epoch": 0.7469720688802834,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1742141887901266e-06,
+      "loss": 0.8837,
+      "step": 27198
+    },
+    {
+      "epoch": 0.7469995331081266,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1735641466355192e-06,
+      "loss": 0.8717,
+      "step": 27199
+    },
+    {
+      "epoch": 0.7470269973359699,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1729141584943158e-06,
+      "loss": 0.9444,
+      "step": 27200
+    },
+    {
+      "epoch": 0.7470544615638132,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1722642243716627e-06,
+      "loss": 0.7486,
+      "step": 27201
+    },
+    {
+      "epoch": 0.7470819257916563,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1716143442726965e-06,
+      "loss": 0.9133,
+      "step": 27202
+    },
+    {
+      "epoch": 0.7471093900194996,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1709645182025637e-06,
+      "loss": 0.8119,
+      "step": 27203
+    },
+    {
+      "epoch": 0.7471368542473429,
+      "grad_norm": 0.0,
+      "learning_rate": 3.170314746166404e-06,
+      "loss": 0.7964,
+      "step": 27204
+    },
+    {
+      "epoch": 0.747164318475186,
+      "grad_norm": 0.0,
+      "learning_rate": 3.169665028169363e-06,
+      "loss": 0.8182,
+      "step": 27205
+    },
+    {
+      "epoch": 0.7471917827030293,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1690153642165756e-06,
+      "loss": 0.9347,
+      "step": 27206
+    },
+    {
+      "epoch": 0.7472192469308725,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1683657543131876e-06,
+      "loss": 0.7999,
+      "step": 27207
+    },
+    {
+      "epoch": 0.7472467111587158,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1677161984643323e-06,
+      "loss": 0.7886,
+      "step": 27208
+    },
+    {
+      "epoch": 0.747274175386559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.167066696675154e-06,
+      "loss": 0.7357,
+      "step": 27209
+    },
+    {
+      "epoch": 0.7473016396144022,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1664172489507904e-06,
+      "loss": 0.8022,
+      "step": 27210
+    },
+    {
+      "epoch": 0.7473291038422455,
+      "grad_norm": 0.0,
+      "learning_rate": 3.16576785529638e-06,
+      "loss": 0.8771,
+      "step": 27211
+    },
+    {
+      "epoch": 0.7473565680700887,
+      "grad_norm": 0.0,
+      "learning_rate": 3.165118515717063e-06,
+      "loss": 0.8113,
+      "step": 27212
+    },
+    {
+      "epoch": 0.7473840322979319,
+      "grad_norm": 0.0,
+      "learning_rate": 3.164469230217978e-06,
+      "loss": 0.9621,
+      "step": 27213
+    },
+    {
+      "epoch": 0.7474114965257752,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1638199988042608e-06,
+      "loss": 0.7453,
+      "step": 27214
+    },
+    {
+      "epoch": 0.7474389607536184,
+      "grad_norm": 0.0,
+      "learning_rate": 3.163170821481044e-06,
+      "loss": 0.8786,
+      "step": 27215
+    },
+    {
+      "epoch": 0.7474664249814617,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1625216982534678e-06,
+      "loss": 0.8898,
+      "step": 27216
+    },
+    {
+      "epoch": 0.7474938892093049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.16187262912667e-06,
+      "loss": 0.7869,
+      "step": 27217
+    },
+    {
+      "epoch": 0.7475213534371481,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1612236141057828e-06,
+      "loss": 0.8281,
+      "step": 27218
+    },
+    {
+      "epoch": 0.7475488176649914,
+      "grad_norm": 0.0,
+      "learning_rate": 3.160574653195948e-06,
+      "loss": 0.8137,
+      "step": 27219
+    },
+    {
+      "epoch": 0.7475762818928345,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1599257464022924e-06,
+      "loss": 0.7879,
+      "step": 27220
+    },
+    {
+      "epoch": 0.7476037461206778,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1592768937299547e-06,
+      "loss": 0.8657,
+      "step": 27221
+    },
+    {
+      "epoch": 0.7476312103485211,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1586280951840697e-06,
+      "loss": 0.7446,
+      "step": 27222
+    },
+    {
+      "epoch": 0.7476586745763643,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1579793507697674e-06,
+      "loss": 0.8619,
+      "step": 27223
+    },
+    {
+      "epoch": 0.7476861388042075,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1573306604921817e-06,
+      "loss": 0.83,
+      "step": 27224
+    },
+    {
+      "epoch": 0.7477136030320507,
+      "grad_norm": 0.0,
+      "learning_rate": 3.15668202435645e-06,
+      "loss": 0.9183,
+      "step": 27225
+    },
+    {
+      "epoch": 0.747741067259894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1560334423676986e-06,
+      "loss": 0.7445,
+      "step": 27226
+    },
+    {
+      "epoch": 0.7477685314877373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1553849145310613e-06,
+      "loss": 0.8243,
+      "step": 27227
+    },
+    {
+      "epoch": 0.7477959957155804,
+      "grad_norm": 0.0,
+      "learning_rate": 3.154736440851669e-06,
+      "loss": 0.9442,
+      "step": 27228
+    },
+    {
+      "epoch": 0.7478234599434237,
+      "grad_norm": 0.0,
+      "learning_rate": 3.154088021334658e-06,
+      "loss": 0.819,
+      "step": 27229
+    },
+    {
+      "epoch": 0.747850924171267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.153439655985151e-06,
+      "loss": 0.8074,
+      "step": 27230
+    },
+    {
+      "epoch": 0.7478783883991101,
+      "grad_norm": 0.0,
+      "learning_rate": 3.152791344808284e-06,
+      "loss": 0.8642,
+      "step": 27231
+    },
+    {
+      "epoch": 0.7479058526269534,
+      "grad_norm": 0.0,
+      "learning_rate": 3.152143087809182e-06,
+      "loss": 0.8797,
+      "step": 27232
+    },
+    {
+      "epoch": 0.7479333168547966,
+      "grad_norm": 0.0,
+      "learning_rate": 3.151494884992975e-06,
+      "loss": 0.8604,
+      "step": 27233
+    },
+    {
+      "epoch": 0.7479607810826399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1508467363647943e-06,
+      "loss": 0.8186,
+      "step": 27234
+    },
+    {
+      "epoch": 0.7479882453104831,
+      "grad_norm": 0.0,
+      "learning_rate": 3.150198641929768e-06,
+      "loss": 0.7066,
+      "step": 27235
+    },
+    {
+      "epoch": 0.7480157095383263,
+      "grad_norm": 0.0,
+      "learning_rate": 3.149550601693022e-06,
+      "loss": 0.7446,
+      "step": 27236
+    },
+    {
+      "epoch": 0.7480431737661696,
+      "grad_norm": 0.0,
+      "learning_rate": 3.148902615659689e-06,
+      "loss": 0.8449,
+      "step": 27237
+    },
+    {
+      "epoch": 0.7480706379940127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1482546838348925e-06,
+      "loss": 0.8239,
+      "step": 27238
+    },
+    {
+      "epoch": 0.748098102221856,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1476068062237565e-06,
+      "loss": 0.7743,
+      "step": 27239
+    },
+    {
+      "epoch": 0.7481255664496993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1469589828314096e-06,
+      "loss": 0.8427,
+      "step": 27240
+    },
+    {
+      "epoch": 0.7481530306775425,
+      "grad_norm": 0.0,
+      "learning_rate": 3.146311213662977e-06,
+      "loss": 0.7743,
+      "step": 27241
+    },
+    {
+      "epoch": 0.7481804949053857,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1456634987235857e-06,
+      "loss": 0.8067,
+      "step": 27242
+    },
+    {
+      "epoch": 0.748207959133229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1450158380183624e-06,
+      "loss": 0.7932,
+      "step": 27243
+    },
+    {
+      "epoch": 0.7482354233610722,
+      "grad_norm": 0.0,
+      "learning_rate": 3.144368231552426e-06,
+      "loss": 0.855,
+      "step": 27244
+    },
+    {
+      "epoch": 0.7482628875889155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.143720679330905e-06,
+      "loss": 0.7723,
+      "step": 27245
+    },
+    {
+      "epoch": 0.7482903518167586,
+      "grad_norm": 0.0,
+      "learning_rate": 3.143073181358923e-06,
+      "loss": 0.7452,
+      "step": 27246
+    },
+    {
+      "epoch": 0.7483178160446019,
+      "grad_norm": 0.0,
+      "learning_rate": 3.142425737641601e-06,
+      "loss": 0.8631,
+      "step": 27247
+    },
+    {
+      "epoch": 0.7483452802724452,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1417783481840615e-06,
+      "loss": 0.869,
+      "step": 27248
+    },
+    {
+      "epoch": 0.7483727445002883,
+      "grad_norm": 0.0,
+      "learning_rate": 3.141131012991433e-06,
+      "loss": 0.8325,
+      "step": 27249
+    },
+    {
+      "epoch": 0.7484002087281316,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1404837320688287e-06,
+      "loss": 0.8048,
+      "step": 27250
+    },
+    {
+      "epoch": 0.7484276729559748,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1398365054213732e-06,
+      "loss": 0.9618,
+      "step": 27251
+    },
+    {
+      "epoch": 0.7484551371838181,
+      "grad_norm": 0.0,
+      "learning_rate": 3.13918933305419e-06,
+      "loss": 0.8644,
+      "step": 27252
+    },
+    {
+      "epoch": 0.7484826014116613,
+      "grad_norm": 0.0,
+      "learning_rate": 3.138542214972401e-06,
+      "loss": 0.8105,
+      "step": 27253
+    },
+    {
+      "epoch": 0.7485100656395045,
+      "grad_norm": 0.0,
+      "learning_rate": 3.137895151181124e-06,
+      "loss": 0.7646,
+      "step": 27254
+    },
+    {
+      "epoch": 0.7485375298673478,
+      "grad_norm": 0.0,
+      "learning_rate": 3.137248141685476e-06,
+      "loss": 0.7999,
+      "step": 27255
+    },
+    {
+      "epoch": 0.7485649940951911,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1366011864905786e-06,
+      "loss": 0.7525,
+      "step": 27256
+    },
+    {
+      "epoch": 0.7485924583230342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.135954285601551e-06,
+      "loss": 0.8115,
+      "step": 27257
+    },
+    {
+      "epoch": 0.7486199225508775,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1353074390235126e-06,
+      "loss": 0.8569,
+      "step": 27258
+    },
+    {
+      "epoch": 0.7486473867787207,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1346606467615805e-06,
+      "loss": 0.7564,
+      "step": 27259
+    },
+    {
+      "epoch": 0.748674851006564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1340139088208766e-06,
+      "loss": 0.9143,
+      "step": 27260
+    },
+    {
+      "epoch": 0.7487023152344072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1333672252065104e-06,
+      "loss": 0.7165,
+      "step": 27261
+    },
+    {
+      "epoch": 0.7487297794622504,
+      "grad_norm": 0.0,
+      "learning_rate": 3.132720595923605e-06,
+      "loss": 0.847,
+      "step": 27262
+    },
+    {
+      "epoch": 0.7487572436900937,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1320740209772726e-06,
+      "loss": 0.7536,
+      "step": 27263
+    },
+    {
+      "epoch": 0.7487847079179369,
+      "grad_norm": 0.0,
+      "learning_rate": 3.13142750037263e-06,
+      "loss": 0.8085,
+      "step": 27264
+    },
+    {
+      "epoch": 0.7488121721457801,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1307810341147947e-06,
+      "loss": 0.7292,
+      "step": 27265
+    },
+    {
+      "epoch": 0.7488396363736234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1301346222088833e-06,
+      "loss": 0.8188,
+      "step": 27266
+    },
+    {
+      "epoch": 0.7488671006014666,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1294882646600043e-06,
+      "loss": 0.84,
+      "step": 27267
+    },
+    {
+      "epoch": 0.7488945648293098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1288419614732755e-06,
+      "loss": 0.8352,
+      "step": 27268
+    },
+    {
+      "epoch": 0.7489220290571531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.128195712653812e-06,
+      "loss": 0.8117,
+      "step": 27269
+    },
+    {
+      "epoch": 0.7489494932849963,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1275495182067285e-06,
+      "loss": 0.8924,
+      "step": 27270
+    },
+    {
+      "epoch": 0.7489769575128395,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1269033781371315e-06,
+      "loss": 0.8199,
+      "step": 27271
+    },
+    {
+      "epoch": 0.7490044217406827,
+      "grad_norm": 0.0,
+      "learning_rate": 3.126257292450141e-06,
+      "loss": 0.7769,
+      "step": 27272
+    },
+    {
+      "epoch": 0.749031885968526,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1256112611508615e-06,
+      "loss": 0.8447,
+      "step": 27273
+    },
+    {
+      "epoch": 0.7490593501963693,
+      "grad_norm": 0.0,
+      "learning_rate": 3.12496528424441e-06,
+      "loss": 0.9131,
+      "step": 27274
+    },
+    {
+      "epoch": 0.7490868144242124,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1243193617358947e-06,
+      "loss": 0.8697,
+      "step": 27275
+    },
+    {
+      "epoch": 0.7491142786520557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.123673493630429e-06,
+      "loss": 0.8656,
+      "step": 27276
+    },
+    {
+      "epoch": 0.749141742879899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.123027679933126e-06,
+      "loss": 0.8274,
+      "step": 27277
+    },
+    {
+      "epoch": 0.7491692071077422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.122381920649091e-06,
+      "loss": 0.7635,
+      "step": 27278
+    },
+    {
+      "epoch": 0.7491966713355854,
+      "grad_norm": 0.0,
+      "learning_rate": 3.121736215783431e-06,
+      "loss": 0.8297,
+      "step": 27279
+    },
+    {
+      "epoch": 0.7492241355634286,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1210905653412595e-06,
+      "loss": 0.7913,
+      "step": 27280
+    },
+    {
+      "epoch": 0.7492515997912719,
+      "grad_norm": 0.0,
+      "learning_rate": 3.120444969327683e-06,
+      "loss": 0.8269,
+      "step": 27281
+    },
+    {
+      "epoch": 0.7492790640191151,
+      "grad_norm": 0.0,
+      "learning_rate": 3.119799427747812e-06,
+      "loss": 0.8026,
+      "step": 27282
+    },
+    {
+      "epoch": 0.7493065282469583,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1191539406067516e-06,
+      "loss": 0.798,
+      "step": 27283
+    },
+    {
+      "epoch": 0.7493339924748016,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1185085079096146e-06,
+      "loss": 0.7975,
+      "step": 27284
+    },
+    {
+      "epoch": 0.7493614567026448,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1178631296615003e-06,
+      "loss": 0.9003,
+      "step": 27285
+    },
+    {
+      "epoch": 0.749388920930488,
+      "grad_norm": 0.0,
+      "learning_rate": 3.117217805867522e-06,
+      "loss": 0.7384,
+      "step": 27286
+    },
+    {
+      "epoch": 0.7494163851583313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.11657253653278e-06,
+      "loss": 0.8189,
+      "step": 27287
+    },
+    {
+      "epoch": 0.7494438493861745,
+      "grad_norm": 0.0,
+      "learning_rate": 3.115927321662382e-06,
+      "loss": 0.8338,
+      "step": 27288
+    },
+    {
+      "epoch": 0.7494713136140178,
+      "grad_norm": 0.0,
+      "learning_rate": 3.115282161261434e-06,
+      "loss": 0.8635,
+      "step": 27289
+    },
+    {
+      "epoch": 0.749498777841861,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1146370553350435e-06,
+      "loss": 0.7669,
+      "step": 27290
+    },
+    {
+      "epoch": 0.7495262420697042,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1139920038883077e-06,
+      "loss": 0.8949,
+      "step": 27291
+    },
+    {
+      "epoch": 0.7495537062975475,
+      "grad_norm": 0.0,
+      "learning_rate": 3.113347006926336e-06,
+      "loss": 0.8406,
+      "step": 27292
+    },
+    {
+      "epoch": 0.7495811705253906,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1127020644542326e-06,
+      "loss": 0.8337,
+      "step": 27293
+    },
+    {
+      "epoch": 0.7496086347532339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.112057176477096e-06,
+      "loss": 0.8384,
+      "step": 27294
+    },
+    {
+      "epoch": 0.7496360989810772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1114123430000296e-06,
+      "loss": 0.7699,
+      "step": 27295
+    },
+    {
+      "epoch": 0.7496635632089204,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1107675640281423e-06,
+      "loss": 0.8488,
+      "step": 27296
+    },
+    {
+      "epoch": 0.7496910274367636,
+      "grad_norm": 0.0,
+      "learning_rate": 3.110122839566526e-06,
+      "loss": 0.8141,
+      "step": 27297
+    },
+    {
+      "epoch": 0.7497184916646068,
+      "grad_norm": 0.0,
+      "learning_rate": 3.109478169620287e-06,
+      "loss": 0.7949,
+      "step": 27298
+    },
+    {
+      "epoch": 0.7497459558924501,
+      "grad_norm": 0.0,
+      "learning_rate": 3.108833554194526e-06,
+      "loss": 0.828,
+      "step": 27299
+    },
+    {
+      "epoch": 0.7497734201202934,
+      "grad_norm": 0.0,
+      "learning_rate": 3.108188993294343e-06,
+      "loss": 0.8233,
+      "step": 27300
+    },
+    {
+      "epoch": 0.7498008843481365,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1075444869248428e-06,
+      "loss": 0.8184,
+      "step": 27301
+    },
+    {
+      "epoch": 0.7498283485759798,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1069000350911192e-06,
+      "loss": 0.7602,
+      "step": 27302
+    },
+    {
+      "epoch": 0.7498558128038231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.10625563779827e-06,
+      "loss": 0.7935,
+      "step": 27303
+    },
+    {
+      "epoch": 0.7498832770316662,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1056112950513962e-06,
+      "loss": 0.8573,
+      "step": 27304
+    },
+    {
+      "epoch": 0.7499107412595095,
+      "grad_norm": 0.0,
+      "learning_rate": 3.104967006855597e-06,
+      "loss": 0.8422,
+      "step": 27305
+    },
+    {
+      "epoch": 0.7499382054873527,
+      "grad_norm": 0.0,
+      "learning_rate": 3.104322773215969e-06,
+      "loss": 0.7737,
+      "step": 27306
+    },
+    {
+      "epoch": 0.749965669715196,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103678594137611e-06,
+      "loss": 0.8156,
+      "step": 27307
+    },
+    {
+      "epoch": 0.7499931339430392,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103034469625622e-06,
+      "loss": 0.8138,
+      "step": 27308
+    },
+    {
+      "epoch": 0.7500205981708824,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1023903996850935e-06,
+      "loss": 0.8128,
+      "step": 27309
+    },
+    {
+      "epoch": 0.7500480623987257,
+      "grad_norm": 0.0,
+      "learning_rate": 3.101746384321126e-06,
+      "loss": 0.8155,
+      "step": 27310
+    },
+    {
+      "epoch": 0.7500755266265688,
+      "grad_norm": 0.0,
+      "learning_rate": 3.101102423538811e-06,
+      "loss": 0.9385,
+      "step": 27311
+    },
+    {
+      "epoch": 0.7501029908544121,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1004585173432457e-06,
+      "loss": 0.7705,
+      "step": 27312
+    },
+    {
+      "epoch": 0.7501304550822554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.099814665739529e-06,
+      "loss": 0.8128,
+      "step": 27313
+    },
+    {
+      "epoch": 0.7501579193100986,
+      "grad_norm": 0.0,
+      "learning_rate": 3.099170868732747e-06,
+      "loss": 0.8007,
+      "step": 27314
+    },
+    {
+      "epoch": 0.7501853835379418,
+      "grad_norm": 0.0,
+      "learning_rate": 3.098527126327999e-06,
+      "loss": 0.8501,
+      "step": 27315
+    },
+    {
+      "epoch": 0.7502128477657851,
+      "grad_norm": 0.0,
+      "learning_rate": 3.097883438530378e-06,
+      "loss": 0.8423,
+      "step": 27316
+    },
+    {
+      "epoch": 0.7502403119936283,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0972398053449805e-06,
+      "loss": 0.8321,
+      "step": 27317
+    },
+    {
+      "epoch": 0.7502677762214716,
+      "grad_norm": 0.0,
+      "learning_rate": 3.096596226776891e-06,
+      "loss": 0.8056,
+      "step": 27318
+    },
+    {
+      "epoch": 0.7502952404493147,
+      "grad_norm": 0.0,
+      "learning_rate": 3.095952702831211e-06,
+      "loss": 0.7735,
+      "step": 27319
+    },
+    {
+      "epoch": 0.750322704677158,
+      "grad_norm": 0.0,
+      "learning_rate": 3.095309233513023e-06,
+      "loss": 0.7978,
+      "step": 27320
+    },
+    {
+      "epoch": 0.7503501689050013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.094665818827424e-06,
+      "loss": 0.7973,
+      "step": 27321
+    },
+    {
+      "epoch": 0.7503776331328444,
+      "grad_norm": 0.0,
+      "learning_rate": 3.094022458779503e-06,
+      "loss": 0.8972,
+      "step": 27322
+    },
+    {
+      "epoch": 0.7504050973606877,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0933791533743517e-06,
+      "loss": 0.8383,
+      "step": 27323
+    },
+    {
+      "epoch": 0.7504325615885309,
+      "grad_norm": 0.0,
+      "learning_rate": 3.09273590261706e-06,
+      "loss": 0.7673,
+      "step": 27324
+    },
+    {
+      "epoch": 0.7504600258163742,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0920927065127205e-06,
+      "loss": 0.8195,
+      "step": 27325
+    },
+    {
+      "epoch": 0.7504874900442174,
+      "grad_norm": 0.0,
+      "learning_rate": 3.09144956506642e-06,
+      "loss": 0.8272,
+      "step": 27326
+    },
+    {
+      "epoch": 0.7505149542720606,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0908064782832424e-06,
+      "loss": 0.7523,
+      "step": 27327
+    },
+    {
+      "epoch": 0.7505424184999039,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0901634461682816e-06,
+      "loss": 0.7419,
+      "step": 27328
+    },
+    {
+      "epoch": 0.7505698827277472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0895204687266233e-06,
+      "loss": 0.8833,
+      "step": 27329
+    },
+    {
+      "epoch": 0.7505973469555903,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0888775459633568e-06,
+      "loss": 0.8078,
+      "step": 27330
+    },
+    {
+      "epoch": 0.7506248111834336,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0882346778835704e-06,
+      "loss": 0.7221,
+      "step": 27331
+    },
+    {
+      "epoch": 0.7506522754112768,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0875918644923464e-06,
+      "loss": 0.8301,
+      "step": 27332
+    },
+    {
+      "epoch": 0.75067973963912,
+      "grad_norm": 0.0,
+      "learning_rate": 3.086949105794772e-06,
+      "loss": 0.837,
+      "step": 27333
+    },
+    {
+      "epoch": 0.7507072038669633,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0863064017959376e-06,
+      "loss": 0.7358,
+      "step": 27334
+    },
+    {
+      "epoch": 0.7507346680948065,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0856637525009223e-06,
+      "loss": 0.7968,
+      "step": 27335
+    },
+    {
+      "epoch": 0.7507621323226498,
+      "grad_norm": 0.0,
+      "learning_rate": 3.085021157914814e-06,
+      "loss": 0.8034,
+      "step": 27336
+    },
+    {
+      "epoch": 0.7507895965504929,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0843786180427004e-06,
+      "loss": 0.8497,
+      "step": 27337
+    },
+    {
+      "epoch": 0.7508170607783362,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0837361328896578e-06,
+      "loss": 0.824,
+      "step": 27338
+    },
+    {
+      "epoch": 0.7508445250061795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0830937024607745e-06,
+      "loss": 0.7439,
+      "step": 27339
+    },
+    {
+      "epoch": 0.7508719892340227,
+      "grad_norm": 0.0,
+      "learning_rate": 3.082451326761133e-06,
+      "loss": 0.8732,
+      "step": 27340
+    },
+    {
+      "epoch": 0.7508994534618659,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0818090057958195e-06,
+      "loss": 0.8546,
+      "step": 27341
+    },
+    {
+      "epoch": 0.7509269176897092,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0811667395699096e-06,
+      "loss": 0.7974,
+      "step": 27342
+    },
+    {
+      "epoch": 0.7509543819175524,
+      "grad_norm": 0.0,
+      "learning_rate": 3.080524528088492e-06,
+      "loss": 0.8116,
+      "step": 27343
+    },
+    {
+      "epoch": 0.7509818461453956,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0798823713566416e-06,
+      "loss": 0.8392,
+      "step": 27344
+    },
+    {
+      "epoch": 0.7510093103732388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.079240269379442e-06,
+      "loss": 0.8575,
+      "step": 27345
+    },
+    {
+      "epoch": 0.7510367746010821,
+      "grad_norm": 0.0,
+      "learning_rate": 3.078598222161975e-06,
+      "loss": 0.7604,
+      "step": 27346
+    },
+    {
+      "epoch": 0.7510642388289254,
+      "grad_norm": 0.0,
+      "learning_rate": 3.077956229709319e-06,
+      "loss": 0.7502,
+      "step": 27347
+    },
+    {
+      "epoch": 0.7510917030567685,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0773142920265552e-06,
+      "loss": 0.777,
+      "step": 27348
+    },
+    {
+      "epoch": 0.7511191672846118,
+      "grad_norm": 0.0,
+      "learning_rate": 3.076672409118766e-06,
+      "loss": 0.9134,
+      "step": 27349
+    },
+    {
+      "epoch": 0.751146631512455,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0760305809910263e-06,
+      "loss": 0.9205,
+      "step": 27350
+    },
+    {
+      "epoch": 0.7511740957402983,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0753888076484105e-06,
+      "loss": 0.8503,
+      "step": 27351
+    },
+    {
+      "epoch": 0.7512015599681415,
+      "grad_norm": 0.0,
+      "learning_rate": 3.074747089096002e-06,
+      "loss": 0.8145,
+      "step": 27352
+    },
+    {
+      "epoch": 0.7512290241959847,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0741054253388767e-06,
+      "loss": 0.8593,
+      "step": 27353
+    },
+    {
+      "epoch": 0.751256488423828,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0734638163821117e-06,
+      "loss": 0.7711,
+      "step": 27354
+    },
+    {
+      "epoch": 0.7512839526516712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.072822262230788e-06,
+      "loss": 0.7364,
+      "step": 27355
+    },
+    {
+      "epoch": 0.7513114168795144,
+      "grad_norm": 0.0,
+      "learning_rate": 3.072180762889975e-06,
+      "loss": 0.7406,
+      "step": 27356
+    },
+    {
+      "epoch": 0.7513388811073577,
+      "grad_norm": 0.0,
+      "learning_rate": 3.07153931836475e-06,
+      "loss": 0.8149,
+      "step": 27357
+    },
+    {
+      "epoch": 0.7513663453352009,
+      "grad_norm": 0.0,
+      "learning_rate": 3.070897928660194e-06,
+      "loss": 0.7666,
+      "step": 27358
+    },
+    {
+      "epoch": 0.7513938095630441,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0702565937813744e-06,
+      "loss": 0.7904,
+      "step": 27359
+    },
+    {
+      "epoch": 0.7514212737908874,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0696153137333685e-06,
+      "loss": 0.8088,
+      "step": 27360
+    },
+    {
+      "epoch": 0.7514487380187306,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0689740885212537e-06,
+      "loss": 0.8759,
+      "step": 27361
+    },
+    {
+      "epoch": 0.7514762022465739,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0683329181500986e-06,
+      "loss": 0.7625,
+      "step": 27362
+    },
+    {
+      "epoch": 0.751503666474417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.067691802624978e-06,
+      "loss": 0.89,
+      "step": 27363
+    },
+    {
+      "epoch": 0.7515311307022603,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0670507419509645e-06,
+      "loss": 0.7759,
+      "step": 27364
+    },
+    {
+      "epoch": 0.7515585949301036,
+      "grad_norm": 0.0,
+      "learning_rate": 3.066409736133136e-06,
+      "loss": 0.7978,
+      "step": 27365
+    },
+    {
+      "epoch": 0.7515860591579467,
+      "grad_norm": 0.0,
+      "learning_rate": 3.065768785176555e-06,
+      "loss": 0.8427,
+      "step": 27366
+    },
+    {
+      "epoch": 0.75161352338579,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0651278890863003e-06,
+      "loss": 0.8152,
+      "step": 27367
+    },
+    {
+      "epoch": 0.7516409876136333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0644870478674373e-06,
+      "loss": 0.8213,
+      "step": 27368
+    },
+    {
+      "epoch": 0.7516684518414765,
+      "grad_norm": 0.0,
+      "learning_rate": 3.06384626152504e-06,
+      "loss": 0.9111,
+      "step": 27369
+    },
+    {
+      "epoch": 0.7516959160693197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0632055300641773e-06,
+      "loss": 0.8566,
+      "step": 27370
+    },
+    {
+      "epoch": 0.7517233802971629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0625648534899188e-06,
+      "loss": 0.8417,
+      "step": 27371
+    },
+    {
+      "epoch": 0.7517508445250062,
+      "grad_norm": 0.0,
+      "learning_rate": 3.061924231807338e-06,
+      "loss": 0.8387,
+      "step": 27372
+    },
+    {
+      "epoch": 0.7517783087528495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0612836650214983e-06,
+      "loss": 0.8549,
+      "step": 27373
+    },
+    {
+      "epoch": 0.7518057729806926,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0606431531374723e-06,
+      "loss": 0.7965,
+      "step": 27374
+    },
+    {
+      "epoch": 0.7518332372085359,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0600026961603225e-06,
+      "loss": 0.869,
+      "step": 27375
+    },
+    {
+      "epoch": 0.7518607014363791,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0593622940951194e-06,
+      "loss": 0.853,
+      "step": 27376
+    },
+    {
+      "epoch": 0.7518881656642223,
+      "grad_norm": 0.0,
+      "learning_rate": 3.058721946946931e-06,
+      "loss": 0.8406,
+      "step": 27377
+    },
+    {
+      "epoch": 0.7519156298920656,
+      "grad_norm": 0.0,
+      "learning_rate": 3.058081654720827e-06,
+      "loss": 0.8224,
+      "step": 27378
+    },
+    {
+      "epoch": 0.7519430941199088,
+      "grad_norm": 0.0,
+      "learning_rate": 3.057441417421866e-06,
+      "loss": 0.8989,
+      "step": 27379
+    },
+    {
+      "epoch": 0.7519705583477521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0568012350551177e-06,
+      "loss": 0.8613,
+      "step": 27380
+    },
+    {
+      "epoch": 0.7519980225755953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0561611076256483e-06,
+      "loss": 0.8265,
+      "step": 27381
+    },
+    {
+      "epoch": 0.7520254868034385,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0555210351385246e-06,
+      "loss": 0.876,
+      "step": 27382
+    },
+    {
+      "epoch": 0.7520529510312818,
+      "grad_norm": 0.0,
+      "learning_rate": 3.054881017598805e-06,
+      "loss": 0.8878,
+      "step": 27383
+    },
+    {
+      "epoch": 0.752080415259125,
+      "grad_norm": 0.0,
+      "learning_rate": 3.054241055011561e-06,
+      "loss": 0.8654,
+      "step": 27384
+    },
+    {
+      "epoch": 0.7521078794869682,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0536011473818483e-06,
+      "loss": 0.8615,
+      "step": 27385
+    },
+    {
+      "epoch": 0.7521353437148115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0529612947147347e-06,
+      "loss": 0.7418,
+      "step": 27386
+    },
+    {
+      "epoch": 0.7521628079426547,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0523214970152825e-06,
+      "loss": 0.7768,
+      "step": 27387
+    },
+    {
+      "epoch": 0.7521902721704979,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0516817542885537e-06,
+      "loss": 0.8532,
+      "step": 27388
+    },
+    {
+      "epoch": 0.7522177363983411,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0510420665396134e-06,
+      "loss": 0.7148,
+      "step": 27389
+    },
+    {
+      "epoch": 0.7522452006261844,
+      "grad_norm": 0.0,
+      "learning_rate": 3.050402433773516e-06,
+      "loss": 0.7453,
+      "step": 27390
+    },
+    {
+      "epoch": 0.7522726648540277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0497628559953307e-06,
+      "loss": 0.8984,
+      "step": 27391
+    },
+    {
+      "epoch": 0.7523001290818708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.049123333210111e-06,
+      "loss": 0.7141,
+      "step": 27392
+    },
+    {
+      "epoch": 0.7523275933097141,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0484838654229197e-06,
+      "loss": 0.8362,
+      "step": 27393
+    },
+    {
+      "epoch": 0.7523550575375574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.047844452638817e-06,
+      "loss": 0.848,
+      "step": 27394
+    },
+    {
+      "epoch": 0.7523825217654005,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0472050948628617e-06,
+      "loss": 0.9348,
+      "step": 27395
+    },
+    {
+      "epoch": 0.7524099859932438,
+      "grad_norm": 0.0,
+      "learning_rate": 3.046565792100117e-06,
+      "loss": 0.8542,
+      "step": 27396
+    },
+    {
+      "epoch": 0.752437450221087,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0459265443556328e-06,
+      "loss": 0.7997,
+      "step": 27397
+    },
+    {
+      "epoch": 0.7524649144489303,
+      "grad_norm": 0.0,
+      "learning_rate": 3.045287351634475e-06,
+      "loss": 0.8258,
+      "step": 27398
+    },
+    {
+      "epoch": 0.7524923786767735,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0446482139416956e-06,
+      "loss": 0.8295,
+      "step": 27399
+    },
+    {
+      "epoch": 0.7525198429046167,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0440091312823527e-06,
+      "loss": 0.7763,
+      "step": 27400
+    },
+    {
+      "epoch": 0.75254730713246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.043370103661504e-06,
+      "loss": 0.7863,
+      "step": 27401
+    },
+    {
+      "epoch": 0.7525747713603032,
+      "grad_norm": 0.0,
+      "learning_rate": 3.042731131084209e-06,
+      "loss": 0.7321,
+      "step": 27402
+    },
+    {
+      "epoch": 0.7526022355881464,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0420922135555173e-06,
+      "loss": 0.7723,
+      "step": 27403
+    },
+    {
+      "epoch": 0.7526296998159897,
+      "grad_norm": 0.0,
+      "learning_rate": 3.041453351080488e-06,
+      "loss": 0.8373,
+      "step": 27404
+    },
+    {
+      "epoch": 0.7526571640438329,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0408145436641743e-06,
+      "loss": 0.9046,
+      "step": 27405
+    },
+    {
+      "epoch": 0.7526846282716761,
+      "grad_norm": 0.0,
+      "learning_rate": 3.040175791311635e-06,
+      "loss": 0.8027,
+      "step": 27406
+    },
+    {
+      "epoch": 0.7527120924995194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0395370940279177e-06,
+      "loss": 0.9094,
+      "step": 27407
+    },
+    {
+      "epoch": 0.7527395567273626,
+      "grad_norm": 0.0,
+      "learning_rate": 3.038898451818082e-06,
+      "loss": 0.8316,
+      "step": 27408
+    },
+    {
+      "epoch": 0.7527670209552059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0382598646871753e-06,
+      "loss": 0.8621,
+      "step": 27409
+    },
+    {
+      "epoch": 0.752794485183049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.037621332640254e-06,
+      "loss": 0.9087,
+      "step": 27410
+    },
+    {
+      "epoch": 0.7528219494108923,
+      "grad_norm": 0.0,
+      "learning_rate": 3.036982855682369e-06,
+      "loss": 0.729,
+      "step": 27411
+    },
+    {
+      "epoch": 0.7528494136387356,
+      "grad_norm": 0.0,
+      "learning_rate": 3.036344433818572e-06,
+      "loss": 0.836,
+      "step": 27412
+    },
+    {
+      "epoch": 0.7528768778665788,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0357060670539197e-06,
+      "loss": 0.7923,
+      "step": 27413
+    },
+    {
+      "epoch": 0.752904342094422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0350677553934582e-06,
+      "loss": 0.8208,
+      "step": 27414
+    },
+    {
+      "epoch": 0.7529318063222652,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0344294988422364e-06,
+      "loss": 0.8203,
+      "step": 27415
+    },
+    {
+      "epoch": 0.7529592705501085,
+      "grad_norm": 0.0,
+      "learning_rate": 3.033791297405305e-06,
+      "loss": 0.8076,
+      "step": 27416
+    },
+    {
+      "epoch": 0.7529867347779517,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0331531510877166e-06,
+      "loss": 0.8644,
+      "step": 27417
+    },
+    {
+      "epoch": 0.7530141990057949,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0325150598945187e-06,
+      "loss": 0.8201,
+      "step": 27418
+    },
+    {
+      "epoch": 0.7530416632336382,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0318770238307605e-06,
+      "loss": 0.798,
+      "step": 27419
+    },
+    {
+      "epoch": 0.7530691274614815,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0312390429014936e-06,
+      "loss": 0.8903,
+      "step": 27420
+    },
+    {
+      "epoch": 0.7530965916893246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.03060111711176e-06,
+      "loss": 0.8201,
+      "step": 27421
+    },
+    {
+      "epoch": 0.7531240559171679,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0299632464666127e-06,
+      "loss": 0.8714,
+      "step": 27422
+    },
+    {
+      "epoch": 0.7531515201450111,
+      "grad_norm": 0.0,
+      "learning_rate": 3.029325430971093e-06,
+      "loss": 0.8131,
+      "step": 27423
+    },
+    {
+      "epoch": 0.7531789843728544,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0286876706302517e-06,
+      "loss": 0.8085,
+      "step": 27424
+    },
+    {
+      "epoch": 0.7532064486006976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0280499654491324e-06,
+      "loss": 0.8001,
+      "step": 27425
+    },
+    {
+      "epoch": 0.7532339128285408,
+      "grad_norm": 0.0,
+      "learning_rate": 3.027412315432787e-06,
+      "loss": 0.8009,
+      "step": 27426
+    },
+    {
+      "epoch": 0.7532613770563841,
+      "grad_norm": 0.0,
+      "learning_rate": 3.026774720586252e-06,
+      "loss": 0.8248,
+      "step": 27427
+    },
+    {
+      "epoch": 0.7532888412842272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.026137180914578e-06,
+      "loss": 0.9021,
+      "step": 27428
+    },
+    {
+      "epoch": 0.7533163055120705,
+      "grad_norm": 0.0,
+      "learning_rate": 3.025499696422807e-06,
+      "loss": 0.9055,
+      "step": 27429
+    },
+    {
+      "epoch": 0.7533437697399138,
+      "grad_norm": 0.0,
+      "learning_rate": 3.024862267115987e-06,
+      "loss": 0.7144,
+      "step": 27430
+    },
+    {
+      "epoch": 0.753371233967757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.024224892999159e-06,
+      "loss": 0.8897,
+      "step": 27431
+    },
+    {
+      "epoch": 0.7533986981956002,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0235875740773622e-06,
+      "loss": 0.8703,
+      "step": 27432
+    },
+    {
+      "epoch": 0.7534261624234435,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0229503103556434e-06,
+      "loss": 0.7262,
+      "step": 27433
+    },
+    {
+      "epoch": 0.7534536266512867,
+      "grad_norm": 0.0,
+      "learning_rate": 3.022313101839045e-06,
+      "loss": 0.8674,
+      "step": 27434
+    },
+    {
+      "epoch": 0.75348109087913,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0216759485326073e-06,
+      "loss": 0.7479,
+      "step": 27435
+    },
+    {
+      "epoch": 0.7535085551069731,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0210388504413733e-06,
+      "loss": 0.7692,
+      "step": 27436
+    },
+    {
+      "epoch": 0.7535360193348164,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0204018075703854e-06,
+      "loss": 0.8003,
+      "step": 27437
+    },
+    {
+      "epoch": 0.7535634835626597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0197648199246833e-06,
+      "loss": 0.8087,
+      "step": 27438
+    },
+    {
+      "epoch": 0.7535909477905028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0191278875093024e-06,
+      "loss": 0.8421,
+      "step": 27439
+    },
+    {
+      "epoch": 0.7536184120183461,
+      "grad_norm": 0.0,
+      "learning_rate": 3.018491010329285e-06,
+      "loss": 0.8495,
+      "step": 27440
+    },
+    {
+      "epoch": 0.7536458762461893,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0178541883896718e-06,
+      "loss": 0.7695,
+      "step": 27441
+    },
+    {
+      "epoch": 0.7536733404740326,
+      "grad_norm": 0.0,
+      "learning_rate": 3.017217421695501e-06,
+      "loss": 0.8881,
+      "step": 27442
+    },
+    {
+      "epoch": 0.7537008047018758,
+      "grad_norm": 0.0,
+      "learning_rate": 3.016580710251813e-06,
+      "loss": 0.7479,
+      "step": 27443
+    },
+    {
+      "epoch": 0.753728268929719,
+      "grad_norm": 0.0,
+      "learning_rate": 3.015944054063642e-06,
+      "loss": 0.8998,
+      "step": 27444
+    },
+    {
+      "epoch": 0.7537557331575623,
+      "grad_norm": 0.0,
+      "learning_rate": 3.015307453136026e-06,
+      "loss": 0.7394,
+      "step": 27445
+    },
+    {
+      "epoch": 0.7537831973854056,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0146709074740053e-06,
+      "loss": 0.8321,
+      "step": 27446
+    },
+    {
+      "epoch": 0.7538106616132487,
+      "grad_norm": 0.0,
+      "learning_rate": 3.014034417082613e-06,
+      "loss": 0.8463,
+      "step": 27447
+    },
+    {
+      "epoch": 0.753838125841092,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0133979819668856e-06,
+      "loss": 0.8561,
+      "step": 27448
+    },
+    {
+      "epoch": 0.7538655900689352,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0127616021318627e-06,
+      "loss": 0.7629,
+      "step": 27449
+    },
+    {
+      "epoch": 0.7538930542967784,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0121252775825736e-06,
+      "loss": 0.7926,
+      "step": 27450
+    },
+    {
+      "epoch": 0.7539205185246217,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0114890083240556e-06,
+      "loss": 0.7767,
+      "step": 27451
+    },
+    {
+      "epoch": 0.7539479827524649,
+      "grad_norm": 0.0,
+      "learning_rate": 3.010852794361344e-06,
+      "loss": 0.805,
+      "step": 27452
+    },
+    {
+      "epoch": 0.7539754469803082,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0102166356994767e-06,
+      "loss": 0.7938,
+      "step": 27453
+    },
+    {
+      "epoch": 0.7540029112081513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.009580532343478e-06,
+      "loss": 0.795,
+      "step": 27454
+    },
+    {
+      "epoch": 0.7540303754359946,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0089444842983908e-06,
+      "loss": 0.747,
+      "step": 27455
+    },
+    {
+      "epoch": 0.7540578396638379,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0083084915692384e-06,
+      "loss": 0.8367,
+      "step": 27456
+    },
+    {
+      "epoch": 0.754085303891681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0076725541610584e-06,
+      "loss": 0.76,
+      "step": 27457
+    },
+    {
+      "epoch": 0.7541127681195243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0070366720788814e-06,
+      "loss": 0.8691,
+      "step": 27458
+    },
+    {
+      "epoch": 0.7541402323473676,
+      "grad_norm": 0.0,
+      "learning_rate": 3.006400845327739e-06,
+      "loss": 0.7695,
+      "step": 27459
+    },
+    {
+      "epoch": 0.7541676965752108,
+      "grad_norm": 0.0,
+      "learning_rate": 3.005765073912662e-06,
+      "loss": 0.7932,
+      "step": 27460
+    },
+    {
+      "epoch": 0.754195160803054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.005129357838684e-06,
+      "loss": 0.8318,
+      "step": 27461
+    },
+    {
+      "epoch": 0.7542226250308972,
+      "grad_norm": 0.0,
+      "learning_rate": 3.004493697110833e-06,
+      "loss": 0.7832,
+      "step": 27462
+    },
+    {
+      "epoch": 0.7542500892587405,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0038580917341343e-06,
+      "loss": 0.8558,
+      "step": 27463
+    },
+    {
+      "epoch": 0.7542775534865838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0032225417136196e-06,
+      "loss": 0.6993,
+      "step": 27464
+    },
+    {
+      "epoch": 0.7543050177144269,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0025870470543194e-06,
+      "loss": 0.8047,
+      "step": 27465
+    },
+    {
+      "epoch": 0.7543324819422702,
+      "grad_norm": 0.0,
+      "learning_rate": 3.001951607761261e-06,
+      "loss": 0.8468,
+      "step": 27466
+    },
+    {
+      "epoch": 0.7543599461701135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0013162238394745e-06,
+      "loss": 0.7634,
+      "step": 27467
+    },
+    {
+      "epoch": 0.7543874103979566,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0006808952939824e-06,
+      "loss": 0.8099,
+      "step": 27468
+    },
+    {
+      "epoch": 0.7544148746257999,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0000456221298148e-06,
+      "loss": 0.9891,
+      "step": 27469
+    },
+    {
+      "epoch": 0.7544423388536431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.999410404352e-06,
+      "loss": 0.9573,
+      "step": 27470
+    },
+    {
+      "epoch": 0.7544698030814864,
+      "grad_norm": 0.0,
+      "learning_rate": 2.998775241965559e-06,
+      "loss": 0.8509,
+      "step": 27471
+    },
+    {
+      "epoch": 0.7544972673093296,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9981401349755213e-06,
+      "loss": 0.904,
+      "step": 27472
+    },
+    {
+      "epoch": 0.7545247315371728,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9975050833869136e-06,
+      "loss": 0.7902,
+      "step": 27473
+    },
+    {
+      "epoch": 0.7545521957650161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.996870087204754e-06,
+      "loss": 0.7442,
+      "step": 27474
+    },
+    {
+      "epoch": 0.7545796599928593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9962351464340722e-06,
+      "loss": 0.8277,
+      "step": 27475
+    },
+    {
+      "epoch": 0.7546071242207025,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9956002610798906e-06,
+      "loss": 0.7823,
+      "step": 27476
+    },
+    {
+      "epoch": 0.7546345884485458,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9949654311472365e-06,
+      "loss": 0.7976,
+      "step": 27477
+    },
+    {
+      "epoch": 0.754662052676389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9943306566411257e-06,
+      "loss": 0.8298,
+      "step": 27478
+    },
+    {
+      "epoch": 0.7546895169042322,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9936959375665873e-06,
+      "loss": 0.8543,
+      "step": 27479
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9930612739286393e-06,
+      "loss": 0.8147,
+      "step": 27480
+    },
+    {
+      "epoch": 0.7547444453599187,
+      "grad_norm": 0.0,
+      "learning_rate": 2.992426665732303e-06,
+      "loss": 0.7561,
+      "step": 27481
+    },
+    {
+      "epoch": 0.754771909587762,
+      "grad_norm": 0.0,
+      "learning_rate": 2.991792112982601e-06,
+      "loss": 0.7854,
+      "step": 27482
+    },
+    {
+      "epoch": 0.7547993738156051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9911576156845554e-06,
+      "loss": 0.7781,
+      "step": 27483
+    },
+    {
+      "epoch": 0.7548268380434484,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9905231738431862e-06,
+      "loss": 0.8032,
+      "step": 27484
+    },
+    {
+      "epoch": 0.7548543022712917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.989888787463515e-06,
+      "loss": 0.8465,
+      "step": 27485
+    },
+    {
+      "epoch": 0.7548817664991349,
+      "grad_norm": 0.0,
+      "learning_rate": 2.98925445655056e-06,
+      "loss": 0.7836,
+      "step": 27486
+    },
+    {
+      "epoch": 0.7549092307269781,
+      "grad_norm": 0.0,
+      "learning_rate": 2.988620181109336e-06,
+      "loss": 0.6665,
+      "step": 27487
+    },
+    {
+      "epoch": 0.7549366949548213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.987985961144865e-06,
+      "loss": 0.8502,
+      "step": 27488
+    },
+    {
+      "epoch": 0.7549641591826646,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9873517966621655e-06,
+      "loss": 0.8296,
+      "step": 27489
+    },
+    {
+      "epoch": 0.7549916234105079,
+      "grad_norm": 0.0,
+      "learning_rate": 2.986717687666255e-06,
+      "loss": 0.7938,
+      "step": 27490
+    },
+    {
+      "epoch": 0.755019087638351,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9860836341621537e-06,
+      "loss": 0.805,
+      "step": 27491
+    },
+    {
+      "epoch": 0.7550465518661943,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9854496361548722e-06,
+      "loss": 0.7662,
+      "step": 27492
+    },
+    {
+      "epoch": 0.7550740160940376,
+      "grad_norm": 0.0,
+      "learning_rate": 2.984815693649431e-06,
+      "loss": 0.7169,
+      "step": 27493
+    },
+    {
+      "epoch": 0.7551014803218807,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9841818066508487e-06,
+      "loss": 0.8635,
+      "step": 27494
+    },
+    {
+      "epoch": 0.755128944549724,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9835479751641338e-06,
+      "loss": 0.8249,
+      "step": 27495
+    },
+    {
+      "epoch": 0.7551564087775672,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9829141991943087e-06,
+      "loss": 0.8013,
+      "step": 27496
+    },
+    {
+      "epoch": 0.7551838730054105,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9822804787463812e-06,
+      "loss": 0.7582,
+      "step": 27497
+    },
+    {
+      "epoch": 0.7552113372332537,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9816468138253695e-06,
+      "loss": 0.8355,
+      "step": 27498
+    },
+    {
+      "epoch": 0.7552388014610969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.981013204436287e-06,
+      "loss": 0.7743,
+      "step": 27499
+    },
+    {
+      "epoch": 0.7552662656889402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9803796505841465e-06,
+      "loss": 0.8495,
+      "step": 27500
+    },
+    {
+      "epoch": 0.7552937299167833,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9797461522739645e-06,
+      "loss": 0.7483,
+      "step": 27501
+    },
+    {
+      "epoch": 0.7553211941446266,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9791127095107463e-06,
+      "loss": 0.8234,
+      "step": 27502
+    },
+    {
+      "epoch": 0.7553486583724699,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9784793222995122e-06,
+      "loss": 0.8652,
+      "step": 27503
+    },
+    {
+      "epoch": 0.7553761226003131,
+      "grad_norm": 0.0,
+      "learning_rate": 2.977845990645266e-06,
+      "loss": 0.8329,
+      "step": 27504
+    },
+    {
+      "epoch": 0.7554035868281563,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9772127145530225e-06,
+      "loss": 0.8929,
+      "step": 27505
+    },
+    {
+      "epoch": 0.7554310510559996,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9765794940277936e-06,
+      "loss": 0.8726,
+      "step": 27506
+    },
+    {
+      "epoch": 0.7554585152838428,
+      "grad_norm": 0.0,
+      "learning_rate": 2.975946329074587e-06,
+      "loss": 0.8345,
+      "step": 27507
+    },
+    {
+      "epoch": 0.7554859795116861,
+      "grad_norm": 0.0,
+      "learning_rate": 2.975313219698417e-06,
+      "loss": 0.9006,
+      "step": 27508
+    },
+    {
+      "epoch": 0.7555134437395292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9746801659042877e-06,
+      "loss": 0.9138,
+      "step": 27509
+    },
+    {
+      "epoch": 0.7555409079673725,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9740471676972126e-06,
+      "loss": 0.8132,
+      "step": 27510
+    },
+    {
+      "epoch": 0.7555683721952158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9734142250821953e-06,
+      "loss": 0.846,
+      "step": 27511
+    },
+    {
+      "epoch": 0.7555958364230589,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9727813380642458e-06,
+      "loss": 0.767,
+      "step": 27512
+    },
+    {
+      "epoch": 0.7556233006509022,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9721485066483723e-06,
+      "loss": 0.8647,
+      "step": 27513
+    },
+    {
+      "epoch": 0.7556507648787454,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9715157308395858e-06,
+      "loss": 0.8056,
+      "step": 27514
+    },
+    {
+      "epoch": 0.7556782291065887,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9708830106428864e-06,
+      "loss": 0.7411,
+      "step": 27515
+    },
+    {
+      "epoch": 0.7557056933344319,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9702503460632838e-06,
+      "loss": 0.8353,
+      "step": 27516
+    },
+    {
+      "epoch": 0.7557331575622751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.969617737105783e-06,
+      "loss": 0.8682,
+      "step": 27517
+    },
+    {
+      "epoch": 0.7557606217901184,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9689851837753935e-06,
+      "loss": 0.8141,
+      "step": 27518
+    },
+    {
+      "epoch": 0.7557880860179617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.968352686077113e-06,
+      "loss": 0.841,
+      "step": 27519
+    },
+    {
+      "epoch": 0.7558155502458048,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9677202440159536e-06,
+      "loss": 0.846,
+      "step": 27520
+    },
+    {
+      "epoch": 0.7558430144736481,
+      "grad_norm": 0.0,
+      "learning_rate": 2.967087857596912e-06,
+      "loss": 0.8341,
+      "step": 27521
+    },
+    {
+      "epoch": 0.7558704787014913,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9664555268249962e-06,
+      "loss": 0.8278,
+      "step": 27522
+    },
+    {
+      "epoch": 0.7558979429293345,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9658232517052077e-06,
+      "loss": 0.79,
+      "step": 27523
+    },
+    {
+      "epoch": 0.7559254071571778,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9651910322425514e-06,
+      "loss": 0.7664,
+      "step": 27524
+    },
+    {
+      "epoch": 0.755952871385021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.964558868442031e-06,
+      "loss": 0.8654,
+      "step": 27525
+    },
+    {
+      "epoch": 0.7559803356128643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.963926760308644e-06,
+      "loss": 0.7839,
+      "step": 27526
+    },
+    {
+      "epoch": 0.7560077998407074,
+      "grad_norm": 0.0,
+      "learning_rate": 2.963294707847396e-06,
+      "loss": 0.7741,
+      "step": 27527
+    },
+    {
+      "epoch": 0.7560352640685507,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9626627110632843e-06,
+      "loss": 0.8401,
+      "step": 27528
+    },
+    {
+      "epoch": 0.756062728296394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9620307699613103e-06,
+      "loss": 0.7877,
+      "step": 27529
+    },
+    {
+      "epoch": 0.7560901925242371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9613988845464746e-06,
+      "loss": 0.8513,
+      "step": 27530
+    },
+    {
+      "epoch": 0.7561176567520804,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9607670548237787e-06,
+      "loss": 0.8604,
+      "step": 27531
+    },
+    {
+      "epoch": 0.7561451209799237,
+      "grad_norm": 0.0,
+      "learning_rate": 2.960135280798223e-06,
+      "loss": 0.8045,
+      "step": 27532
+    },
+    {
+      "epoch": 0.7561725852077669,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9595035624748003e-06,
+      "loss": 0.8366,
+      "step": 27533
+    },
+    {
+      "epoch": 0.7562000494356101,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9588718998585166e-06,
+      "loss": 0.7926,
+      "step": 27534
+    },
+    {
+      "epoch": 0.7562275136634533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.958240292954363e-06,
+      "loss": 0.8763,
+      "step": 27535
+    },
+    {
+      "epoch": 0.7562549778912966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9576087417673395e-06,
+      "loss": 0.7937,
+      "step": 27536
+    },
+    {
+      "epoch": 0.7562824421191399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9569772463024425e-06,
+      "loss": 0.845,
+      "step": 27537
+    },
+    {
+      "epoch": 0.756309906346983,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9563458065646744e-06,
+      "loss": 0.8497,
+      "step": 27538
+    },
+    {
+      "epoch": 0.7563373705748263,
+      "grad_norm": 0.0,
+      "learning_rate": 2.955714422559023e-06,
+      "loss": 0.8473,
+      "step": 27539
+    },
+    {
+      "epoch": 0.7563648348026695,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9550830942904884e-06,
+      "loss": 0.7861,
+      "step": 27540
+    },
+    {
+      "epoch": 0.7563922990305127,
+      "grad_norm": 0.0,
+      "learning_rate": 2.954451821764065e-06,
+      "loss": 0.8623,
+      "step": 27541
+    },
+    {
+      "epoch": 0.756419763258356,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9538206049847505e-06,
+      "loss": 0.7134,
+      "step": 27542
+    },
+    {
+      "epoch": 0.7564472274861992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9531894439575338e-06,
+      "loss": 0.7825,
+      "step": 27543
+    },
+    {
+      "epoch": 0.7564746917140425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9525583386874157e-06,
+      "loss": 0.9092,
+      "step": 27544
+    },
+    {
+      "epoch": 0.7565021559418857,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9519272891793828e-06,
+      "loss": 0.866,
+      "step": 27545
+    },
+    {
+      "epoch": 0.7565296201697289,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9512962954384304e-06,
+      "loss": 0.8338,
+      "step": 27546
+    },
+    {
+      "epoch": 0.7565570843975722,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9506653574695523e-06,
+      "loss": 0.8192,
+      "step": 27547
+    },
+    {
+      "epoch": 0.7565845486254154,
+      "grad_norm": 0.0,
+      "learning_rate": 2.950034475277741e-06,
+      "loss": 0.9006,
+      "step": 27548
+    },
+    {
+      "epoch": 0.7566120128532586,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9494036488679913e-06,
+      "loss": 0.8345,
+      "step": 27549
+    },
+    {
+      "epoch": 0.7566394770811019,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9487728782452873e-06,
+      "loss": 0.82,
+      "step": 27550
+    },
+    {
+      "epoch": 0.7566669413089451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.948142163414627e-06,
+      "loss": 0.8281,
+      "step": 27551
+    },
+    {
+      "epoch": 0.7566944055367884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.947511504380994e-06,
+      "loss": 0.8377,
+      "step": 27552
+    },
+    {
+      "epoch": 0.7567218697646315,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9468809011493825e-06,
+      "loss": 0.8379,
+      "step": 27553
+    },
+    {
+      "epoch": 0.7567493339924748,
+      "grad_norm": 0.0,
+      "learning_rate": 2.946250353724781e-06,
+      "loss": 0.7989,
+      "step": 27554
+    },
+    {
+      "epoch": 0.7567767982203181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9456198621121824e-06,
+      "loss": 0.887,
+      "step": 27555
+    },
+    {
+      "epoch": 0.7568042624481612,
+      "grad_norm": 0.0,
+      "learning_rate": 2.944989426316569e-06,
+      "loss": 0.8463,
+      "step": 27556
+    },
+    {
+      "epoch": 0.7568317266760045,
+      "grad_norm": 0.0,
+      "learning_rate": 2.944359046342932e-06,
+      "loss": 0.8315,
+      "step": 27557
+    },
+    {
+      "epoch": 0.7568591909038478,
+      "grad_norm": 0.0,
+      "learning_rate": 2.943728722196263e-06,
+      "loss": 0.819,
+      "step": 27558
+    },
+    {
+      "epoch": 0.756886655131691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.943098453881542e-06,
+      "loss": 0.8089,
+      "step": 27559
+    },
+    {
+      "epoch": 0.7569141193595342,
+      "grad_norm": 0.0,
+      "learning_rate": 2.94246824140376e-06,
+      "loss": 0.8334,
+      "step": 27560
+    },
+    {
+      "epoch": 0.7569415835873774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9418380847679053e-06,
+      "loss": 0.8046,
+      "step": 27561
+    },
+    {
+      "epoch": 0.7569690478152207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9412079839789585e-06,
+      "loss": 0.7894,
+      "step": 27562
+    },
+    {
+      "epoch": 0.756996512043064,
+      "grad_norm": 0.0,
+      "learning_rate": 2.94057793904191e-06,
+      "loss": 0.8274,
+      "step": 27563
+    },
+    {
+      "epoch": 0.7570239762709071,
+      "grad_norm": 0.0,
+      "learning_rate": 2.939947949961742e-06,
+      "loss": 0.8391,
+      "step": 27564
+    },
+    {
+      "epoch": 0.7570514404987504,
+      "grad_norm": 0.0,
+      "learning_rate": 2.939318016743441e-06,
+      "loss": 0.8405,
+      "step": 27565
+    },
+    {
+      "epoch": 0.7570789047265936,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9386881393919943e-06,
+      "loss": 0.8385,
+      "step": 27566
+    },
+    {
+      "epoch": 0.7571063689544368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.938058317912382e-06,
+      "loss": 0.8779,
+      "step": 27567
+    },
+    {
+      "epoch": 0.7571338331822801,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9374285523095857e-06,
+      "loss": 0.8658,
+      "step": 27568
+    },
+    {
+      "epoch": 0.7571612974101233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9367988425885895e-06,
+      "loss": 0.8082,
+      "step": 27569
+    },
+    {
+      "epoch": 0.7571887616379666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9361691887543764e-06,
+      "loss": 0.8181,
+      "step": 27570
+    },
+    {
+      "epoch": 0.7572162258658098,
+      "grad_norm": 0.0,
+      "learning_rate": 2.93553959081193e-06,
+      "loss": 0.8152,
+      "step": 27571
+    },
+    {
+      "epoch": 0.757243690093653,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9349100487662307e-06,
+      "loss": 0.8303,
+      "step": 27572
+    },
+    {
+      "epoch": 0.7572711543214963,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9342805626222626e-06,
+      "loss": 0.7793,
+      "step": 27573
+    },
+    {
+      "epoch": 0.7572986185493394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9336511323850037e-06,
+      "loss": 0.814,
+      "step": 27574
+    },
+    {
+      "epoch": 0.7573260827771827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9330217580594313e-06,
+      "loss": 0.8733,
+      "step": 27575
+    },
+    {
+      "epoch": 0.757353547005026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.932392439650529e-06,
+      "loss": 0.7976,
+      "step": 27576
+    },
+    {
+      "epoch": 0.7573810112328692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.931763177163274e-06,
+      "loss": 0.8814,
+      "step": 27577
+    },
+    {
+      "epoch": 0.7574084754607124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9311339706026474e-06,
+      "loss": 0.7632,
+      "step": 27578
+    },
+    {
+      "epoch": 0.7574359396885556,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9305048199736296e-06,
+      "loss": 0.821,
+      "step": 27579
+    },
+    {
+      "epoch": 0.7574634039163989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.929875725281194e-06,
+      "loss": 0.8115,
+      "step": 27580
+    },
+    {
+      "epoch": 0.7574908681442422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9292466865303205e-06,
+      "loss": 0.9404,
+      "step": 27581
+    },
+    {
+      "epoch": 0.7575183323720853,
+      "grad_norm": 0.0,
+      "learning_rate": 2.928617703725988e-06,
+      "loss": 0.9026,
+      "step": 27582
+    },
+    {
+      "epoch": 0.7575457965999286,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9279887768731695e-06,
+      "loss": 0.7679,
+      "step": 27583
+    },
+    {
+      "epoch": 0.7575732608277719,
+      "grad_norm": 0.0,
+      "learning_rate": 2.927359905976842e-06,
+      "loss": 0.799,
+      "step": 27584
+    },
+    {
+      "epoch": 0.757600725055615,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9267310910419867e-06,
+      "loss": 0.808,
+      "step": 27585
+    },
+    {
+      "epoch": 0.7576281892834583,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9261023320735714e-06,
+      "loss": 0.823,
+      "step": 27586
+    },
+    {
+      "epoch": 0.7576556535113015,
+      "grad_norm": 0.0,
+      "learning_rate": 2.925473629076575e-06,
+      "loss": 0.7239,
+      "step": 27587
+    },
+    {
+      "epoch": 0.7576831177391448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.924844982055972e-06,
+      "loss": 0.8125,
+      "step": 27588
+    },
+    {
+      "epoch": 0.757710581966988,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9242163910167354e-06,
+      "loss": 0.7178,
+      "step": 27589
+    },
+    {
+      "epoch": 0.7577380461948312,
+      "grad_norm": 0.0,
+      "learning_rate": 2.923587855963843e-06,
+      "loss": 0.9081,
+      "step": 27590
+    },
+    {
+      "epoch": 0.7577655104226745,
+      "grad_norm": 0.0,
+      "learning_rate": 2.922959376902265e-06,
+      "loss": 0.9313,
+      "step": 27591
+    },
+    {
+      "epoch": 0.7577929746505176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9223309538369692e-06,
+      "loss": 0.9143,
+      "step": 27592
+    },
+    {
+      "epoch": 0.7578204388783609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9217025867729343e-06,
+      "loss": 0.7993,
+      "step": 27593
+    },
+    {
+      "epoch": 0.7578479031062042,
+      "grad_norm": 0.0,
+      "learning_rate": 2.921074275715129e-06,
+      "loss": 0.8662,
+      "step": 27594
+    },
+    {
+      "epoch": 0.7578753673340474,
+      "grad_norm": 0.0,
+      "learning_rate": 2.920446020668527e-06,
+      "loss": 0.8079,
+      "step": 27595
+    },
+    {
+      "epoch": 0.7579028315618906,
+      "grad_norm": 0.0,
+      "learning_rate": 2.919817821638098e-06,
+      "loss": 0.8115,
+      "step": 27596
+    },
+    {
+      "epoch": 0.7579302957897339,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9191896786288165e-06,
+      "loss": 0.8426,
+      "step": 27597
+    },
+    {
+      "epoch": 0.7579577600175771,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9185615916456488e-06,
+      "loss": 0.8087,
+      "step": 27598
+    },
+    {
+      "epoch": 0.7579852242454204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9179335606935623e-06,
+      "loss": 0.8515,
+      "step": 27599
+    },
+    {
+      "epoch": 0.7580126884732635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.917305585777528e-06,
+      "loss": 0.8758,
+      "step": 27600
+    },
+    {
+      "epoch": 0.7580401527011068,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9166776669025166e-06,
+      "loss": 0.9335,
+      "step": 27601
+    },
+    {
+      "epoch": 0.7580676169289501,
+      "grad_norm": 0.0,
+      "learning_rate": 2.916049804073494e-06,
+      "loss": 0.828,
+      "step": 27602
+    },
+    {
+      "epoch": 0.7580950811567932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9154219972954334e-06,
+      "loss": 0.8797,
+      "step": 27603
+    },
+    {
+      "epoch": 0.7581225453846365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.914794246573295e-06,
+      "loss": 0.8281,
+      "step": 27604
+    },
+    {
+      "epoch": 0.7581500096124797,
+      "grad_norm": 0.0,
+      "learning_rate": 2.914166551912049e-06,
+      "loss": 0.7114,
+      "step": 27605
+    },
+    {
+      "epoch": 0.758177473840323,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9135389133166647e-06,
+      "loss": 0.748,
+      "step": 27606
+    },
+    {
+      "epoch": 0.7582049380681662,
+      "grad_norm": 0.0,
+      "learning_rate": 2.912911330792103e-06,
+      "loss": 0.854,
+      "step": 27607
+    },
+    {
+      "epoch": 0.7582324022960094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9122838043433323e-06,
+      "loss": 0.7802,
+      "step": 27608
+    },
+    {
+      "epoch": 0.7582598665238527,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9116563339753202e-06,
+      "loss": 0.7421,
+      "step": 27609
+    },
+    {
+      "epoch": 0.758287330751696,
+      "grad_norm": 0.0,
+      "learning_rate": 2.911028919693025e-06,
+      "loss": 0.8881,
+      "step": 27610
+    },
+    {
+      "epoch": 0.7583147949795391,
+      "grad_norm": 0.0,
+      "learning_rate": 2.910401561501417e-06,
+      "loss": 0.7589,
+      "step": 27611
+    },
+    {
+      "epoch": 0.7583422592073824,
+      "grad_norm": 0.0,
+      "learning_rate": 2.909774259405457e-06,
+      "loss": 0.7133,
+      "step": 27612
+    },
+    {
+      "epoch": 0.7583697234352256,
+      "grad_norm": 0.0,
+      "learning_rate": 2.909147013410113e-06,
+      "loss": 0.8333,
+      "step": 27613
+    },
+    {
+      "epoch": 0.7583971876630689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9085198235203406e-06,
+      "loss": 0.7764,
+      "step": 27614
+    },
+    {
+      "epoch": 0.7584246518909121,
+      "grad_norm": 0.0,
+      "learning_rate": 2.907892689741111e-06,
+      "loss": 0.9188,
+      "step": 27615
+    },
+    {
+      "epoch": 0.7584521161187553,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9072656120773768e-06,
+      "loss": 0.942,
+      "step": 27616
+    },
+    {
+      "epoch": 0.7584795803465986,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9066385905341054e-06,
+      "loss": 0.8124,
+      "step": 27617
+    },
+    {
+      "epoch": 0.7585070445744417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9060116251162573e-06,
+      "loss": 0.8959,
+      "step": 27618
+    },
+    {
+      "epoch": 0.758534508802285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.905384715828793e-06,
+      "loss": 0.9329,
+      "step": 27619
+    },
+    {
+      "epoch": 0.7585619730301283,
+      "grad_norm": 0.0,
+      "learning_rate": 2.904757862676676e-06,
+      "loss": 0.8172,
+      "step": 27620
+    },
+    {
+      "epoch": 0.7585894372579715,
+      "grad_norm": 0.0,
+      "learning_rate": 2.90413106566486e-06,
+      "loss": 0.8327,
+      "step": 27621
+    },
+    {
+      "epoch": 0.7586169014858147,
+      "grad_norm": 0.0,
+      "learning_rate": 2.903504324798312e-06,
+      "loss": 0.8271,
+      "step": 27622
+    },
+    {
+      "epoch": 0.758644365713658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9028776400819826e-06,
+      "loss": 0.7918,
+      "step": 27623
+    },
+    {
+      "epoch": 0.7586718299415012,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9022510115208346e-06,
+      "loss": 0.8277,
+      "step": 27624
+    },
+    {
+      "epoch": 0.7586992941693445,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9016244391198256e-06,
+      "loss": 0.805,
+      "step": 27625
+    },
+    {
+      "epoch": 0.7587267583971876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9009979228839167e-06,
+      "loss": 0.791,
+      "step": 27626
+    },
+    {
+      "epoch": 0.7587542226250309,
+      "grad_norm": 0.0,
+      "learning_rate": 2.90037146281806e-06,
+      "loss": 0.8127,
+      "step": 27627
+    },
+    {
+      "epoch": 0.7587816868528742,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8997450589272136e-06,
+      "loss": 0.8983,
+      "step": 27628
+    },
+    {
+      "epoch": 0.7588091510807173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8991187112163353e-06,
+      "loss": 0.8188,
+      "step": 27629
+    },
+    {
+      "epoch": 0.7588366153085606,
+      "grad_norm": 0.0,
+      "learning_rate": 2.898492419690383e-06,
+      "loss": 0.8363,
+      "step": 27630
+    },
+    {
+      "epoch": 0.7588640795364038,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8978661843543067e-06,
+      "loss": 0.8754,
+      "step": 27631
+    },
+    {
+      "epoch": 0.7588915437642471,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8972400052130666e-06,
+      "loss": 0.8679,
+      "step": 27632
+    },
+    {
+      "epoch": 0.7589190079920903,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8966138822716127e-06,
+      "loss": 0.7635,
+      "step": 27633
+    },
+    {
+      "epoch": 0.7589464722199335,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8959878155349007e-06,
+      "loss": 0.7458,
+      "step": 27634
+    },
+    {
+      "epoch": 0.7589739364477768,
+      "grad_norm": 0.0,
+      "learning_rate": 2.895361805007886e-06,
+      "loss": 0.985,
+      "step": 27635
+    },
+    {
+      "epoch": 0.75900140067562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89473585069552e-06,
+      "loss": 0.794,
+      "step": 27636
+    },
+    {
+      "epoch": 0.7590288649034632,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8941099526027593e-06,
+      "loss": 0.8063,
+      "step": 27637
+    },
+    {
+      "epoch": 0.7590563291313065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89348411073455e-06,
+      "loss": 0.7917,
+      "step": 27638
+    },
+    {
+      "epoch": 0.7590837933591497,
+      "grad_norm": 0.0,
+      "learning_rate": 2.892858325095851e-06,
+      "loss": 0.7894,
+      "step": 27639
+    },
+    {
+      "epoch": 0.7591112575869929,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8922325956916064e-06,
+      "loss": 0.7939,
+      "step": 27640
+    },
+    {
+      "epoch": 0.7591387218148362,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8916069225267705e-06,
+      "loss": 0.8519,
+      "step": 27641
+    },
+    {
+      "epoch": 0.7591661860426794,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8909813056062954e-06,
+      "loss": 0.8021,
+      "step": 27642
+    },
+    {
+      "epoch": 0.7591936502705227,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8903557449351285e-06,
+      "loss": 0.8138,
+      "step": 27643
+    },
+    {
+      "epoch": 0.7592211144983658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.889730240518225e-06,
+      "loss": 0.83,
+      "step": 27644
+    },
+    {
+      "epoch": 0.7592485787262091,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8891047923605265e-06,
+      "loss": 0.8103,
+      "step": 27645
+    },
+    {
+      "epoch": 0.7592760429540524,
+      "grad_norm": 0.0,
+      "learning_rate": 2.888479400466988e-06,
+      "loss": 0.8299,
+      "step": 27646
+    },
+    {
+      "epoch": 0.7593035071818955,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8878540648425536e-06,
+      "loss": 0.7758,
+      "step": 27647
+    },
+    {
+      "epoch": 0.7593309714097388,
+      "grad_norm": 0.0,
+      "learning_rate": 2.887228785492172e-06,
+      "loss": 0.8198,
+      "step": 27648
+    },
+    {
+      "epoch": 0.7593584356375821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.886603562420791e-06,
+      "loss": 0.8984,
+      "step": 27649
+    },
+    {
+      "epoch": 0.7593858998654253,
+      "grad_norm": 0.0,
+      "learning_rate": 2.885978395633362e-06,
+      "loss": 0.8523,
+      "step": 27650
+    },
+    {
+      "epoch": 0.7594133640932685,
+      "grad_norm": 0.0,
+      "learning_rate": 2.885353285134824e-06,
+      "loss": 0.8009,
+      "step": 27651
+    },
+    {
+      "epoch": 0.7594408283211117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.884728230930127e-06,
+      "loss": 0.8223,
+      "step": 27652
+    },
+    {
+      "epoch": 0.759468292548955,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8841032330242156e-06,
+      "loss": 0.7837,
+      "step": 27653
+    },
+    {
+      "epoch": 0.7594957567767983,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8834782914220407e-06,
+      "loss": 0.7309,
+      "step": 27654
+    },
+    {
+      "epoch": 0.7595232210046414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8828534061285374e-06,
+      "loss": 0.7593,
+      "step": 27655
+    },
+    {
+      "epoch": 0.7595506852324847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8822285771486593e-06,
+      "loss": 0.8139,
+      "step": 27656
+    },
+    {
+      "epoch": 0.759578149460328,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8816038044873416e-06,
+      "loss": 0.9471,
+      "step": 27657
+    },
+    {
+      "epoch": 0.7596056136881711,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8809790881495326e-06,
+      "loss": 0.8015,
+      "step": 27658
+    },
+    {
+      "epoch": 0.7596330779160144,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8803544281401753e-06,
+      "loss": 0.8218,
+      "step": 27659
+    },
+    {
+      "epoch": 0.7596605421438576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8797298244642104e-06,
+      "loss": 0.8302,
+      "step": 27660
+    },
+    {
+      "epoch": 0.7596880063717009,
+      "grad_norm": 0.0,
+      "learning_rate": 2.879105277126585e-06,
+      "loss": 0.8405,
+      "step": 27661
+    },
+    {
+      "epoch": 0.7597154705995441,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8784807861322337e-06,
+      "loss": 0.8038,
+      "step": 27662
+    },
+    {
+      "epoch": 0.7597429348273873,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8778563514861048e-06,
+      "loss": 0.7145,
+      "step": 27663
+    },
+    {
+      "epoch": 0.7597703990552306,
+      "grad_norm": 0.0,
+      "learning_rate": 2.877231973193132e-06,
+      "loss": 0.8936,
+      "step": 27664
+    },
+    {
+      "epoch": 0.7597978632830737,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8766076512582587e-06,
+      "loss": 0.7752,
+      "step": 27665
+    },
+    {
+      "epoch": 0.759825327510917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.875983385686425e-06,
+      "loss": 0.8086,
+      "step": 27666
+    },
+    {
+      "epoch": 0.7598527917387603,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8753591764825697e-06,
+      "loss": 0.8074,
+      "step": 27667
+    },
+    {
+      "epoch": 0.7598802559666035,
+      "grad_norm": 0.0,
+      "learning_rate": 2.874735023651636e-06,
+      "loss": 0.8602,
+      "step": 27668
+    },
+    {
+      "epoch": 0.7599077201944467,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8741109271985556e-06,
+      "loss": 0.8945,
+      "step": 27669
+    },
+    {
+      "epoch": 0.75993518442229,
+      "grad_norm": 0.0,
+      "learning_rate": 2.873486887128274e-06,
+      "loss": 0.8156,
+      "step": 27670
+    },
+    {
+      "epoch": 0.7599626486501332,
+      "grad_norm": 0.0,
+      "learning_rate": 2.87286290344572e-06,
+      "loss": 0.8189,
+      "step": 27671
+    },
+    {
+      "epoch": 0.7599901128779765,
+      "grad_norm": 0.0,
+      "learning_rate": 2.872238976155837e-06,
+      "loss": 0.6889,
+      "step": 27672
+    },
+    {
+      "epoch": 0.7600175771058196,
+      "grad_norm": 0.0,
+      "learning_rate": 2.871615105263562e-06,
+      "loss": 0.7216,
+      "step": 27673
+    },
+    {
+      "epoch": 0.7600450413336629,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8709912907738278e-06,
+      "loss": 0.8498,
+      "step": 27674
+    },
+    {
+      "epoch": 0.7600725055615062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8703675326915715e-06,
+      "loss": 0.8549,
+      "step": 27675
+    },
+    {
+      "epoch": 0.7600999697893494,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8697438310217286e-06,
+      "loss": 0.858,
+      "step": 27676
+    },
+    {
+      "epoch": 0.7601274340171926,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8691201857692343e-06,
+      "loss": 0.853,
+      "step": 27677
+    },
+    {
+      "epoch": 0.7601548982450358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8684965969390276e-06,
+      "loss": 0.784,
+      "step": 27678
+    },
+    {
+      "epoch": 0.7601823624728791,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8678730645360374e-06,
+      "loss": 0.9064,
+      "step": 27679
+    },
+    {
+      "epoch": 0.7602098267007223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.867249588565195e-06,
+      "loss": 0.9232,
+      "step": 27680
+    },
+    {
+      "epoch": 0.7602372909285655,
+      "grad_norm": 0.0,
+      "learning_rate": 2.866626169031438e-06,
+      "loss": 0.7942,
+      "step": 27681
+    },
+    {
+      "epoch": 0.7602647551564088,
+      "grad_norm": 0.0,
+      "learning_rate": 2.866002805939696e-06,
+      "loss": 0.869,
+      "step": 27682
+    },
+    {
+      "epoch": 0.7602922193842521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8653794992949047e-06,
+      "loss": 0.8785,
+      "step": 27683
+    },
+    {
+      "epoch": 0.7603196836120952,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8647562491019943e-06,
+      "loss": 0.7624,
+      "step": 27684
+    },
+    {
+      "epoch": 0.7603471478399385,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8641330553658996e-06,
+      "loss": 0.9255,
+      "step": 27685
+    },
+    {
+      "epoch": 0.7603746120677817,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8635099180915447e-06,
+      "loss": 0.9555,
+      "step": 27686
+    },
+    {
+      "epoch": 0.760402076295625,
+      "grad_norm": 0.0,
+      "learning_rate": 2.862886837283867e-06,
+      "loss": 0.8206,
+      "step": 27687
+    },
+    {
+      "epoch": 0.7604295405234682,
+      "grad_norm": 0.0,
+      "learning_rate": 2.862263812947791e-06,
+      "loss": 0.7895,
+      "step": 27688
+    },
+    {
+      "epoch": 0.7604570047513114,
+      "grad_norm": 0.0,
+      "learning_rate": 2.861640845088247e-06,
+      "loss": 0.7269,
+      "step": 27689
+    },
+    {
+      "epoch": 0.7604844689791547,
+      "grad_norm": 0.0,
+      "learning_rate": 2.861017933710167e-06,
+      "loss": 0.8265,
+      "step": 27690
+    },
+    {
+      "epoch": 0.7605119332069978,
+      "grad_norm": 0.0,
+      "learning_rate": 2.860395078818481e-06,
+      "loss": 0.6732,
+      "step": 27691
+    },
+    {
+      "epoch": 0.7605393974348411,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8597722804181128e-06,
+      "loss": 0.8045,
+      "step": 27692
+    },
+    {
+      "epoch": 0.7605668616626844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8591495385139912e-06,
+      "loss": 0.7844,
+      "step": 27693
+    },
+    {
+      "epoch": 0.7605943258905276,
+      "grad_norm": 0.0,
+      "learning_rate": 2.858526853111048e-06,
+      "loss": 0.9191,
+      "step": 27694
+    },
+    {
+      "epoch": 0.7606217901183708,
+      "grad_norm": 0.0,
+      "learning_rate": 2.857904224214203e-06,
+      "loss": 0.8713,
+      "step": 27695
+    },
+    {
+      "epoch": 0.7606492543462141,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8572816518283866e-06,
+      "loss": 0.7811,
+      "step": 27696
+    },
+    {
+      "epoch": 0.7606767185740573,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8566591359585273e-06,
+      "loss": 0.8305,
+      "step": 27697
+    },
+    {
+      "epoch": 0.7607041828019006,
+      "grad_norm": 0.0,
+      "learning_rate": 2.856036676609546e-06,
+      "loss": 0.8568,
+      "step": 27698
+    },
+    {
+      "epoch": 0.7607316470297437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8554142737863687e-06,
+      "loss": 0.8411,
+      "step": 27699
+    },
+    {
+      "epoch": 0.760759111257587,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8547919274939217e-06,
+      "loss": 0.77,
+      "step": 27700
+    },
+    {
+      "epoch": 0.7607865754854303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8541696377371287e-06,
+      "loss": 0.8217,
+      "step": 27701
+    },
+    {
+      "epoch": 0.7608140397132734,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8535474045209167e-06,
+      "loss": 0.8243,
+      "step": 27702
+    },
+    {
+      "epoch": 0.7608415039411167,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8529252278502063e-06,
+      "loss": 0.8298,
+      "step": 27703
+    },
+    {
+      "epoch": 0.7608689681689599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.852303107729917e-06,
+      "loss": 0.867,
+      "step": 27704
+    },
+    {
+      "epoch": 0.7608964323968032,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8516810441649732e-06,
+      "loss": 0.8008,
+      "step": 27705
+    },
+    {
+      "epoch": 0.7609238966246464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8510590371602985e-06,
+      "loss": 0.884,
+      "step": 27706
+    },
+    {
+      "epoch": 0.7609513608524896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8504370867208144e-06,
+      "loss": 0.7469,
+      "step": 27707
+    },
+    {
+      "epoch": 0.7609788250803329,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8498151928514408e-06,
+      "loss": 0.7576,
+      "step": 27708
+    },
+    {
+      "epoch": 0.7610062893081762,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8491933555571028e-06,
+      "loss": 0.8331,
+      "step": 27709
+    },
+    {
+      "epoch": 0.7610337535360193,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8485715748427132e-06,
+      "loss": 0.9287,
+      "step": 27710
+    },
+    {
+      "epoch": 0.7610612177638626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8479498507131997e-06,
+      "loss": 0.8278,
+      "step": 27711
+    },
+    {
+      "epoch": 0.7610886819917058,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8473281831734733e-06,
+      "loss": 0.9374,
+      "step": 27712
+    },
+    {
+      "epoch": 0.761116146219549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.846706572228457e-06,
+      "loss": 0.8792,
+      "step": 27713
+    },
+    {
+      "epoch": 0.7611436104473923,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8460850178830703e-06,
+      "loss": 0.7547,
+      "step": 27714
+    },
+    {
+      "epoch": 0.7611710746752355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.845463520142232e-06,
+      "loss": 0.7873,
+      "step": 27715
+    },
+    {
+      "epoch": 0.7611985389030788,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8448420790108557e-06,
+      "loss": 0.8712,
+      "step": 27716
+    },
+    {
+      "epoch": 0.7612260031309219,
+      "grad_norm": 0.0,
+      "learning_rate": 2.844220694493859e-06,
+      "loss": 0.9418,
+      "step": 27717
+    },
+    {
+      "epoch": 0.7612534673587652,
+      "grad_norm": 0.0,
+      "learning_rate": 2.843599366596165e-06,
+      "loss": 0.7253,
+      "step": 27718
+    },
+    {
+      "epoch": 0.7612809315866085,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8429780953226806e-06,
+      "loss": 0.8319,
+      "step": 27719
+    },
+    {
+      "epoch": 0.7613083958144516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.842356880678325e-06,
+      "loss": 0.7513,
+      "step": 27720
+    },
+    {
+      "epoch": 0.7613358600422949,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8417357226680198e-06,
+      "loss": 0.7677,
+      "step": 27721
+    },
+    {
+      "epoch": 0.7613633242701382,
+      "grad_norm": 0.0,
+      "learning_rate": 2.84111462129667e-06,
+      "loss": 0.7471,
+      "step": 27722
+    },
+    {
+      "epoch": 0.7613907884979814,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8404935765691944e-06,
+      "loss": 0.835,
+      "step": 27723
+    },
+    {
+      "epoch": 0.7614182527258246,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8398725884905064e-06,
+      "loss": 0.8744,
+      "step": 27724
+    },
+    {
+      "epoch": 0.7614457169536678,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8392516570655213e-06,
+      "loss": 0.8852,
+      "step": 27725
+    },
+    {
+      "epoch": 0.7614731811815111,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8386307822991533e-06,
+      "loss": 0.8103,
+      "step": 27726
+    },
+    {
+      "epoch": 0.7615006454093544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.838009964196312e-06,
+      "loss": 0.8585,
+      "step": 27727
+    },
+    {
+      "epoch": 0.7615281096371975,
+      "grad_norm": 0.0,
+      "learning_rate": 2.837389202761909e-06,
+      "loss": 0.8001,
+      "step": 27728
+    },
+    {
+      "epoch": 0.7615555738650408,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836768498000856e-06,
+      "loss": 0.7741,
+      "step": 27729
+    },
+    {
+      "epoch": 0.761583038092884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836147849918065e-06,
+      "loss": 0.7981,
+      "step": 27730
+    },
+    {
+      "epoch": 0.7616105023207272,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8355272585184466e-06,
+      "loss": 0.8632,
+      "step": 27731
+    },
+    {
+      "epoch": 0.7616379665485705,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8349067238069138e-06,
+      "loss": 0.8932,
+      "step": 27732
+    },
+    {
+      "epoch": 0.7616654307764137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8342862457883756e-06,
+      "loss": 0.7913,
+      "step": 27733
+    },
+    {
+      "epoch": 0.761692895004257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.833665824467742e-06,
+      "loss": 0.8341,
+      "step": 27734
+    },
+    {
+      "epoch": 0.7617203592321002,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8330454598499167e-06,
+      "loss": 0.718,
+      "step": 27735
+    },
+    {
+      "epoch": 0.7617478234599434,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8324251519398115e-06,
+      "loss": 0.8013,
+      "step": 27736
+    },
+    {
+      "epoch": 0.7617752876877867,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8318049007423356e-06,
+      "loss": 0.7496,
+      "step": 27737
+    },
+    {
+      "epoch": 0.7618027519156298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.831184706262399e-06,
+      "loss": 0.819,
+      "step": 27738
+    },
+    {
+      "epoch": 0.7618302161434731,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8305645685049022e-06,
+      "loss": 0.8449,
+      "step": 27739
+    },
+    {
+      "epoch": 0.7618576803713164,
+      "grad_norm": 0.0,
+      "learning_rate": 2.829944487474756e-06,
+      "loss": 0.8394,
+      "step": 27740
+    },
+    {
+      "epoch": 0.7618851445991596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.829324463176867e-06,
+      "loss": 0.7603,
+      "step": 27741
+    },
+    {
+      "epoch": 0.7619126088270028,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8287044956161424e-06,
+      "loss": 0.6887,
+      "step": 27742
+    },
+    {
+      "epoch": 0.761940073054846,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8280845847974837e-06,
+      "loss": 0.7876,
+      "step": 27743
+    },
+    {
+      "epoch": 0.7619675372826893,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8274647307258007e-06,
+      "loss": 0.8584,
+      "step": 27744
+    },
+    {
+      "epoch": 0.7619950015105326,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8268449334059922e-06,
+      "loss": 0.7842,
+      "step": 27745
+    },
+    {
+      "epoch": 0.7620224657383757,
+      "grad_norm": 0.0,
+      "learning_rate": 2.826225192842965e-06,
+      "loss": 0.8427,
+      "step": 27746
+    },
+    {
+      "epoch": 0.762049929966219,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8256055090416224e-06,
+      "loss": 0.7896,
+      "step": 27747
+    },
+    {
+      "epoch": 0.7620773941940623,
+      "grad_norm": 0.0,
+      "learning_rate": 2.824985882006869e-06,
+      "loss": 0.8276,
+      "step": 27748
+    },
+    {
+      "epoch": 0.7621048584219055,
+      "grad_norm": 0.0,
+      "learning_rate": 2.824366311743606e-06,
+      "loss": 0.8992,
+      "step": 27749
+    },
+    {
+      "epoch": 0.7621323226497487,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8237467982567402e-06,
+      "loss": 0.795,
+      "step": 27750
+    },
+    {
+      "epoch": 0.7621597868775919,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8231273415511682e-06,
+      "loss": 0.9674,
+      "step": 27751
+    },
+    {
+      "epoch": 0.7621872511054352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8225079416317913e-06,
+      "loss": 0.9125,
+      "step": 27752
+    },
+    {
+      "epoch": 0.7622147153332784,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8218885985035107e-06,
+      "loss": 0.8404,
+      "step": 27753
+    },
+    {
+      "epoch": 0.7622421795611216,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8212693121712286e-06,
+      "loss": 0.7599,
+      "step": 27754
+    },
+    {
+      "epoch": 0.7622696437889649,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8206500826398444e-06,
+      "loss": 0.8583,
+      "step": 27755
+    },
+    {
+      "epoch": 0.7622971080168081,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8200309099142597e-06,
+      "loss": 0.7975,
+      "step": 27756
+    },
+    {
+      "epoch": 0.7623245722446513,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8194117939993705e-06,
+      "loss": 0.7551,
+      "step": 27757
+    },
+    {
+      "epoch": 0.7623520364724946,
+      "grad_norm": 0.0,
+      "learning_rate": 2.818792734900079e-06,
+      "loss": 0.7922,
+      "step": 27758
+    },
+    {
+      "epoch": 0.7623795007003378,
+      "grad_norm": 0.0,
+      "learning_rate": 2.818173732621278e-06,
+      "loss": 0.8084,
+      "step": 27759
+    },
+    {
+      "epoch": 0.762406964928181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.817554787167868e-06,
+      "loss": 0.8365,
+      "step": 27760
+    },
+    {
+      "epoch": 0.7624344291560243,
+      "grad_norm": 0.0,
+      "learning_rate": 2.816935898544747e-06,
+      "loss": 0.7526,
+      "step": 27761
+    },
+    {
+      "epoch": 0.7624618933838675,
+      "grad_norm": 0.0,
+      "learning_rate": 2.816317066756814e-06,
+      "loss": 0.8372,
+      "step": 27762
+    },
+    {
+      "epoch": 0.7624893576117108,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8156982918089616e-06,
+      "loss": 0.7598,
+      "step": 27763
+    },
+    {
+      "epoch": 0.7625168218395539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8150795737060853e-06,
+      "loss": 0.9178,
+      "step": 27764
+    },
+    {
+      "epoch": 0.7625442860673972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8144609124530843e-06,
+      "loss": 0.8759,
+      "step": 27765
+    },
+    {
+      "epoch": 0.7625717502952405,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8138423080548537e-06,
+      "loss": 0.8289,
+      "step": 27766
+    },
+    {
+      "epoch": 0.7625992145230837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.813223760516285e-06,
+      "loss": 0.8065,
+      "step": 27767
+    },
+    {
+      "epoch": 0.7626266787509269,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8126052698422755e-06,
+      "loss": 0.7376,
+      "step": 27768
+    },
+    {
+      "epoch": 0.7626541429787701,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8119868360377146e-06,
+      "loss": 0.9191,
+      "step": 27769
+    },
+    {
+      "epoch": 0.7626816072066134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8113684591074986e-06,
+      "loss": 0.8957,
+      "step": 27770
+    },
+    {
+      "epoch": 0.7627090714344567,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8107501390565195e-06,
+      "loss": 0.8946,
+      "step": 27771
+    },
+    {
+      "epoch": 0.7627365356622998,
+      "grad_norm": 0.0,
+      "learning_rate": 2.81013187588967e-06,
+      "loss": 0.7457,
+      "step": 27772
+    },
+    {
+      "epoch": 0.7627639998901431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8095136696118452e-06,
+      "loss": 0.8688,
+      "step": 27773
+    },
+    {
+      "epoch": 0.7627914641179864,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8088955202279312e-06,
+      "loss": 0.798,
+      "step": 27774
+    },
+    {
+      "epoch": 0.7628189283458295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8082774277428236e-06,
+      "loss": 0.8426,
+      "step": 27775
+    },
+    {
+      "epoch": 0.7628463925736728,
+      "grad_norm": 0.0,
+      "learning_rate": 2.807659392161408e-06,
+      "loss": 0.8674,
+      "step": 27776
+    },
+    {
+      "epoch": 0.762873856801516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.807041413488578e-06,
+      "loss": 0.7668,
+      "step": 27777
+    },
+    {
+      "epoch": 0.7629013210293593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8064234917292222e-06,
+      "loss": 0.8117,
+      "step": 27778
+    },
+    {
+      "epoch": 0.7629287852572025,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8058056268882294e-06,
+      "loss": 0.8046,
+      "step": 27779
+    },
+    {
+      "epoch": 0.7629562494850457,
+      "grad_norm": 0.0,
+      "learning_rate": 2.805187818970494e-06,
+      "loss": 0.8076,
+      "step": 27780
+    },
+    {
+      "epoch": 0.762983713712889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8045700679808952e-06,
+      "loss": 0.8268,
+      "step": 27781
+    },
+    {
+      "epoch": 0.7630111779407321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.803952373924329e-06,
+      "loss": 0.7012,
+      "step": 27782
+    },
+    {
+      "epoch": 0.7630386421685754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.803334736805675e-06,
+      "loss": 0.8271,
+      "step": 27783
+    },
+    {
+      "epoch": 0.7630661063964187,
+      "grad_norm": 0.0,
+      "learning_rate": 2.802717156629825e-06,
+      "loss": 0.8963,
+      "step": 27784
+    },
+    {
+      "epoch": 0.7630935706242619,
+      "grad_norm": 0.0,
+      "learning_rate": 2.802099633401666e-06,
+      "loss": 0.8006,
+      "step": 27785
+    },
+    {
+      "epoch": 0.7631210348521051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.801482167126084e-06,
+      "loss": 0.761,
+      "step": 27786
+    },
+    {
+      "epoch": 0.7631484990799484,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8008647578079615e-06,
+      "loss": 0.7881,
+      "step": 27787
+    },
+    {
+      "epoch": 0.7631759633077916,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8002474054521854e-06,
+      "loss": 0.9339,
+      "step": 27788
+    },
+    {
+      "epoch": 0.7632034275356349,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7996301100636414e-06,
+      "loss": 0.8473,
+      "step": 27789
+    },
+    {
+      "epoch": 0.763230891763478,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7990128716472153e-06,
+      "loss": 0.8606,
+      "step": 27790
+    },
+    {
+      "epoch": 0.7632583559913213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7983956902077857e-06,
+      "loss": 0.7571,
+      "step": 27791
+    },
+    {
+      "epoch": 0.7632858202191646,
+      "grad_norm": 0.0,
+      "learning_rate": 2.797778565750243e-06,
+      "loss": 0.9166,
+      "step": 27792
+    },
+    {
+      "epoch": 0.7633132844470077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7971614982794624e-06,
+      "loss": 0.8779,
+      "step": 27793
+    },
+    {
+      "epoch": 0.763340748674851,
+      "grad_norm": 0.0,
+      "learning_rate": 2.79654448780033e-06,
+      "loss": 0.7873,
+      "step": 27794
+    },
+    {
+      "epoch": 0.7633682129026942,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7959275343177283e-06,
+      "loss": 0.7667,
+      "step": 27795
+    },
+    {
+      "epoch": 0.7633956771305375,
+      "grad_norm": 0.0,
+      "learning_rate": 2.795310637836539e-06,
+      "loss": 0.8937,
+      "step": 27796
+    },
+    {
+      "epoch": 0.7634231413583807,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7946937983616452e-06,
+      "loss": 0.8261,
+      "step": 27797
+    },
+    {
+      "epoch": 0.7634506055862239,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7940770158979225e-06,
+      "loss": 0.787,
+      "step": 27798
+    },
+    {
+      "epoch": 0.7634780698140672,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7934602904502562e-06,
+      "loss": 0.6921,
+      "step": 27799
+    },
+    {
+      "epoch": 0.7635055340419105,
+      "grad_norm": 0.0,
+      "learning_rate": 2.792843622023521e-06,
+      "loss": 0.7729,
+      "step": 27800
+    },
+    {
+      "epoch": 0.7635329982697536,
+      "grad_norm": 0.0,
+      "learning_rate": 2.792227010622598e-06,
+      "loss": 0.8438,
+      "step": 27801
+    },
+    {
+      "epoch": 0.7635604624975969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7916104562523683e-06,
+      "loss": 0.8033,
+      "step": 27802
+    },
+    {
+      "epoch": 0.7635879267254401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7909939589177114e-06,
+      "loss": 0.7184,
+      "step": 27803
+    },
+    {
+      "epoch": 0.7636153909532833,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7903775186234992e-06,
+      "loss": 0.8257,
+      "step": 27804
+    },
+    {
+      "epoch": 0.7636428551811266,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7897611353746136e-06,
+      "loss": 0.8504,
+      "step": 27805
+    },
+    {
+      "epoch": 0.7636703194089698,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7891448091759333e-06,
+      "loss": 0.8765,
+      "step": 27806
+    },
+    {
+      "epoch": 0.7636977836368131,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7885285400323302e-06,
+      "loss": 0.7694,
+      "step": 27807
+    },
+    {
+      "epoch": 0.7637252478646562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.787912327948682e-06,
+      "loss": 0.7968,
+      "step": 27808
+    },
+    {
+      "epoch": 0.7637527120924995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7872961729298686e-06,
+      "loss": 0.8846,
+      "step": 27809
+    },
+    {
+      "epoch": 0.7637801763203428,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7866800749807586e-06,
+      "loss": 0.8597,
+      "step": 27810
+    },
+    {
+      "epoch": 0.763807640548186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7860640341062307e-06,
+      "loss": 0.8613,
+      "step": 27811
+    },
+    {
+      "epoch": 0.7638351047760292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7854480503111568e-06,
+      "loss": 0.7733,
+      "step": 27812
+    },
+    {
+      "epoch": 0.7638625690038725,
+      "grad_norm": 0.0,
+      "learning_rate": 2.784832123600414e-06,
+      "loss": 0.8134,
+      "step": 27813
+    },
+    {
+      "epoch": 0.7638900332317157,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7842162539788774e-06,
+      "loss": 0.8811,
+      "step": 27814
+    },
+    {
+      "epoch": 0.763917497459559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7836004414514175e-06,
+      "loss": 0.7446,
+      "step": 27815
+    },
+    {
+      "epoch": 0.7639449616874021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7829846860229017e-06,
+      "loss": 0.8189,
+      "step": 27816
+    },
+    {
+      "epoch": 0.7639724259152454,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7823689876982072e-06,
+      "loss": 0.7277,
+      "step": 27817
+    },
+    {
+      "epoch": 0.7639998901430887,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7817533464822054e-06,
+      "loss": 0.7284,
+      "step": 27818
+    },
+    {
+      "epoch": 0.7640273543709318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7811377623797663e-06,
+      "loss": 0.7692,
+      "step": 27819
+    },
+    {
+      "epoch": 0.7640548185987751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7805222353957628e-06,
+      "loss": 0.7779,
+      "step": 27820
+    },
+    {
+      "epoch": 0.7640822828266183,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7799067655350666e-06,
+      "loss": 0.752,
+      "step": 27821
+    },
+    {
+      "epoch": 0.7641097470544616,
+      "grad_norm": 0.0,
+      "learning_rate": 2.779291352802541e-06,
+      "loss": 0.8392,
+      "step": 27822
+    },
+    {
+      "epoch": 0.7641372112823048,
+      "grad_norm": 0.0,
+      "learning_rate": 2.778675997203064e-06,
+      "loss": 0.7939,
+      "step": 27823
+    },
+    {
+      "epoch": 0.764164675510148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.778060698741496e-06,
+      "loss": 0.8204,
+      "step": 27824
+    },
+    {
+      "epoch": 0.7641921397379913,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7774454574227083e-06,
+      "loss": 0.7683,
+      "step": 27825
+    },
+    {
+      "epoch": 0.7642196039658345,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7768302732515707e-06,
+      "loss": 0.7938,
+      "step": 27826
+    },
+    {
+      "epoch": 0.7642470681936777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7762151462329534e-06,
+      "loss": 0.8293,
+      "step": 27827
+    },
+    {
+      "epoch": 0.764274532421521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.775600076371716e-06,
+      "loss": 0.8938,
+      "step": 27828
+    },
+    {
+      "epoch": 0.7643019966493642,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7749850636727295e-06,
+      "loss": 0.9542,
+      "step": 27829
+    },
+    {
+      "epoch": 0.7643294608772074,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7743701081408638e-06,
+      "loss": 0.8624,
+      "step": 27830
+    },
+    {
+      "epoch": 0.7643569251050507,
+      "grad_norm": 0.0,
+      "learning_rate": 2.773755209780977e-06,
+      "loss": 0.8358,
+      "step": 27831
+    },
+    {
+      "epoch": 0.7643843893328939,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7731403685979384e-06,
+      "loss": 0.7983,
+      "step": 27832
+    },
+    {
+      "epoch": 0.7644118535607372,
+      "grad_norm": 0.0,
+      "learning_rate": 2.772525584596616e-06,
+      "loss": 0.8204,
+      "step": 27833
+    },
+    {
+      "epoch": 0.7644393177885803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7719108577818664e-06,
+      "loss": 0.8567,
+      "step": 27834
+    },
+    {
+      "epoch": 0.7644667820164236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.771296188158559e-06,
+      "loss": 0.7701,
+      "step": 27835
+    },
+    {
+      "epoch": 0.7644942462442669,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7706815757315563e-06,
+      "loss": 0.8325,
+      "step": 27836
+    },
+    {
+      "epoch": 0.76452171047211,
+      "grad_norm": 0.0,
+      "learning_rate": 2.770067020505721e-06,
+      "loss": 0.664,
+      "step": 27837
+    },
+    {
+      "epoch": 0.7645491746999533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7694525224859194e-06,
+      "loss": 0.7973,
+      "step": 27838
+    },
+    {
+      "epoch": 0.7645766389277966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.768838081677011e-06,
+      "loss": 0.8207,
+      "step": 27839
+    },
+    {
+      "epoch": 0.7646041031556398,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7682236980838527e-06,
+      "loss": 0.927,
+      "step": 27840
+    },
+    {
+      "epoch": 0.764631567383483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7676093717113097e-06,
+      "loss": 0.8631,
+      "step": 27841
+    },
+    {
+      "epoch": 0.7646590316113262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.766995102564244e-06,
+      "loss": 0.7557,
+      "step": 27842
+    },
+    {
+      "epoch": 0.7646864958391695,
+      "grad_norm": 0.0,
+      "learning_rate": 2.766380890647514e-06,
+      "loss": 0.8386,
+      "step": 27843
+    },
+    {
+      "epoch": 0.7647139600670128,
+      "grad_norm": 0.0,
+      "learning_rate": 2.76576673596598e-06,
+      "loss": 0.7801,
+      "step": 27844
+    },
+    {
+      "epoch": 0.7647414242948559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7651526385245054e-06,
+      "loss": 0.8607,
+      "step": 27845
+    },
+    {
+      "epoch": 0.7647688885226992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7645385983279428e-06,
+      "loss": 0.8251,
+      "step": 27846
+    },
+    {
+      "epoch": 0.7647963527505424,
+      "grad_norm": 0.0,
+      "learning_rate": 2.763924615381157e-06,
+      "loss": 0.8186,
+      "step": 27847
+    },
+    {
+      "epoch": 0.7648238169783856,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7633106896889995e-06,
+      "loss": 0.8539,
+      "step": 27848
+    },
+    {
+      "epoch": 0.7648512812062289,
+      "grad_norm": 0.0,
+      "learning_rate": 2.76269682125633e-06,
+      "loss": 0.7804,
+      "step": 27849
+    },
+    {
+      "epoch": 0.7648787454340721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.762083010088007e-06,
+      "loss": 0.8285,
+      "step": 27850
+    },
+    {
+      "epoch": 0.7649062096619154,
+      "grad_norm": 0.0,
+      "learning_rate": 2.76146925618889e-06,
+      "loss": 0.8186,
+      "step": 27851
+    },
+    {
+      "epoch": 0.7649336738897586,
+      "grad_norm": 0.0,
+      "learning_rate": 2.76085555956383e-06,
+      "loss": 0.7592,
+      "step": 27852
+    },
+    {
+      "epoch": 0.7649611381176018,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7602419202176844e-06,
+      "loss": 0.8333,
+      "step": 27853
+    },
+    {
+      "epoch": 0.7649886023454451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7596283381553123e-06,
+      "loss": 0.8445,
+      "step": 27854
+    },
+    {
+      "epoch": 0.7650160665732882,
+      "grad_norm": 0.0,
+      "learning_rate": 2.759014813381562e-06,
+      "loss": 0.9339,
+      "step": 27855
+    },
+    {
+      "epoch": 0.7650435308011315,
+      "grad_norm": 0.0,
+      "learning_rate": 2.758401345901295e-06,
+      "loss": 0.7494,
+      "step": 27856
+    },
+    {
+      "epoch": 0.7650709950289748,
+      "grad_norm": 0.0,
+      "learning_rate": 2.757787935719358e-06,
+      "loss": 0.8714,
+      "step": 27857
+    },
+    {
+      "epoch": 0.765098459256818,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7571745828406094e-06,
+      "loss": 0.8676,
+      "step": 27858
+    },
+    {
+      "epoch": 0.7651259234846612,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7565612872698997e-06,
+      "loss": 0.8759,
+      "step": 27859
+    },
+    {
+      "epoch": 0.7651533877125045,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7559480490120836e-06,
+      "loss": 0.8761,
+      "step": 27860
+    },
+    {
+      "epoch": 0.7651808519403477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7553348680720126e-06,
+      "loss": 0.8729,
+      "step": 27861
+    },
+    {
+      "epoch": 0.765208316168191,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7547217444545413e-06,
+      "loss": 0.959,
+      "step": 27862
+    },
+    {
+      "epoch": 0.7652357803960341,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7541086781645187e-06,
+      "loss": 0.7861,
+      "step": 27863
+    },
+    {
+      "epoch": 0.7652632446238774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7534956692067915e-06,
+      "loss": 0.8416,
+      "step": 27864
+    },
+    {
+      "epoch": 0.7652907088517207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.752882717586214e-06,
+      "loss": 0.9175,
+      "step": 27865
+    },
+    {
+      "epoch": 0.7653181730795638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.752269823307635e-06,
+      "loss": 0.8186,
+      "step": 27866
+    },
+    {
+      "epoch": 0.7653456373074071,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7516569863759046e-06,
+      "loss": 0.7599,
+      "step": 27867
+    },
+    {
+      "epoch": 0.7653731015352503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.751044206795875e-06,
+      "loss": 0.873,
+      "step": 27868
+    },
+    {
+      "epoch": 0.7654005657630936,
+      "grad_norm": 0.0,
+      "learning_rate": 2.750431484572389e-06,
+      "loss": 0.917,
+      "step": 27869
+    },
+    {
+      "epoch": 0.7654280299909368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.749818819710297e-06,
+      "loss": 0.8196,
+      "step": 27870
+    },
+    {
+      "epoch": 0.76545549421878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7492062122144505e-06,
+      "loss": 0.9092,
+      "step": 27871
+    },
+    {
+      "epoch": 0.7654829584466233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.74859366208969e-06,
+      "loss": 0.765,
+      "step": 27872
+    },
+    {
+      "epoch": 0.7655104226744666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7479811693408655e-06,
+      "loss": 0.8641,
+      "step": 27873
+    },
+    {
+      "epoch": 0.7655378869023097,
+      "grad_norm": 0.0,
+      "learning_rate": 2.747368733972825e-06,
+      "loss": 0.7144,
+      "step": 27874
+    },
+    {
+      "epoch": 0.765565351130153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.746756355990411e-06,
+      "loss": 0.7933,
+      "step": 27875
+    },
+    {
+      "epoch": 0.7655928153579962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7461440353984693e-06,
+      "loss": 0.8106,
+      "step": 27876
+    },
+    {
+      "epoch": 0.7656202795858394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.745531772201846e-06,
+      "loss": 0.841,
+      "step": 27877
+    },
+    {
+      "epoch": 0.7656477438136827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7449195664053885e-06,
+      "loss": 0.7819,
+      "step": 27878
+    },
+    {
+      "epoch": 0.7656752080415259,
+      "grad_norm": 0.0,
+      "learning_rate": 2.744307418013935e-06,
+      "loss": 0.8651,
+      "step": 27879
+    },
+    {
+      "epoch": 0.7657026722693692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.743695327032334e-06,
+      "loss": 0.8612,
+      "step": 27880
+    },
+    {
+      "epoch": 0.7657301364972123,
+      "grad_norm": 0.0,
+      "learning_rate": 2.743083293465425e-06,
+      "loss": 0.7845,
+      "step": 27881
+    },
+    {
+      "epoch": 0.7657576007250556,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7424713173180507e-06,
+      "loss": 0.8085,
+      "step": 27882
+    },
+    {
+      "epoch": 0.7657850649528989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.741859398595055e-06,
+      "loss": 0.768,
+      "step": 27883
+    },
+    {
+      "epoch": 0.765812529180742,
+      "grad_norm": 0.0,
+      "learning_rate": 2.741247537301279e-06,
+      "loss": 0.818,
+      "step": 27884
+    },
+    {
+      "epoch": 0.7658399934085853,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7406357334415646e-06,
+      "loss": 0.8567,
+      "step": 27885
+    },
+    {
+      "epoch": 0.7658674576364286,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7400239870207556e-06,
+      "loss": 0.8521,
+      "step": 27886
+    },
+    {
+      "epoch": 0.7658949218642718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7394122980436886e-06,
+      "loss": 0.7591,
+      "step": 27887
+    },
+    {
+      "epoch": 0.765922386092115,
+      "grad_norm": 0.0,
+      "learning_rate": 2.738800666515201e-06,
+      "loss": 0.7042,
+      "step": 27888
+    },
+    {
+      "epoch": 0.7659498503199582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7381890924401354e-06,
+      "loss": 0.8213,
+      "step": 27889
+    },
+    {
+      "epoch": 0.7659773145478015,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7375775758233303e-06,
+      "loss": 0.7443,
+      "step": 27890
+    },
+    {
+      "epoch": 0.7660047787756448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.736966116669625e-06,
+      "loss": 0.8782,
+      "step": 27891
+    },
+    {
+      "epoch": 0.7660322430034879,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7363547149838597e-06,
+      "loss": 0.8188,
+      "step": 27892
+    },
+    {
+      "epoch": 0.7660597072313312,
+      "grad_norm": 0.0,
+      "learning_rate": 2.735743370770868e-06,
+      "loss": 0.8237,
+      "step": 27893
+    },
+    {
+      "epoch": 0.7660871714591744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.73513208403549e-06,
+      "loss": 0.8446,
+      "step": 27894
+    },
+    {
+      "epoch": 0.7661146356870177,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7345208547825584e-06,
+      "loss": 0.8348,
+      "step": 27895
+    },
+    {
+      "epoch": 0.7661420999148609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.733909683016913e-06,
+      "loss": 0.8729,
+      "step": 27896
+    },
+    {
+      "epoch": 0.7661695641427041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7332985687433888e-06,
+      "loss": 0.8072,
+      "step": 27897
+    },
+    {
+      "epoch": 0.7661970283705474,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7326875119668227e-06,
+      "loss": 0.7918,
+      "step": 27898
+    },
+    {
+      "epoch": 0.7662244925983906,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7320765126920467e-06,
+      "loss": 0.8174,
+      "step": 27899
+    },
+    {
+      "epoch": 0.7662519568262338,
+      "grad_norm": 0.0,
+      "learning_rate": 2.731465570923896e-06,
+      "loss": 0.8508,
+      "step": 27900
+    },
+    {
+      "epoch": 0.7662794210540771,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7308546866672046e-06,
+      "loss": 0.8566,
+      "step": 27901
+    },
+    {
+      "epoch": 0.7663068852819203,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7302438599268113e-06,
+      "loss": 0.8246,
+      "step": 27902
+    },
+    {
+      "epoch": 0.7663343495097635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7296330907075407e-06,
+      "loss": 0.7623,
+      "step": 27903
+    },
+    {
+      "epoch": 0.7663618137376068,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7290223790142323e-06,
+      "loss": 0.8477,
+      "step": 27904
+    },
+    {
+      "epoch": 0.76638927796545,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7284117248517117e-06,
+      "loss": 0.8067,
+      "step": 27905
+    },
+    {
+      "epoch": 0.7664167421932933,
+      "grad_norm": 0.0,
+      "learning_rate": 2.727801128224815e-06,
+      "loss": 0.8283,
+      "step": 27906
+    },
+    {
+      "epoch": 0.7664442064211364,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7271905891383734e-06,
+      "loss": 0.7135,
+      "step": 27907
+    },
+    {
+      "epoch": 0.7664716706489797,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7265801075972165e-06,
+      "loss": 0.7275,
+      "step": 27908
+    },
+    {
+      "epoch": 0.766499134876823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7259696836061745e-06,
+      "loss": 0.7783,
+      "step": 27909
+    },
+    {
+      "epoch": 0.7665265991046661,
+      "grad_norm": 0.0,
+      "learning_rate": 2.725359317170082e-06,
+      "loss": 0.7917,
+      "step": 27910
+    },
+    {
+      "epoch": 0.7665540633325094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7247490082937644e-06,
+      "loss": 0.8591,
+      "step": 27911
+    },
+    {
+      "epoch": 0.7665815275603527,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7241387569820477e-06,
+      "loss": 0.7999,
+      "step": 27912
+    },
+    {
+      "epoch": 0.7666089917881959,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7235285632397635e-06,
+      "loss": 0.8342,
+      "step": 27913
+    },
+    {
+      "epoch": 0.7666364560160391,
+      "grad_norm": 0.0,
+      "learning_rate": 2.72291842707174e-06,
+      "loss": 0.7488,
+      "step": 27914
+    },
+    {
+      "epoch": 0.7666639202438823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.722308348482807e-06,
+      "loss": 0.8232,
+      "step": 27915
+    },
+    {
+      "epoch": 0.7666913844717256,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7216983274777877e-06,
+      "loss": 0.7698,
+      "step": 27916
+    },
+    {
+      "epoch": 0.7667188486995689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7210883640615105e-06,
+      "loss": 0.8363,
+      "step": 27917
+    },
+    {
+      "epoch": 0.766746312927412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.720478458238803e-06,
+      "loss": 0.7842,
+      "step": 27918
+    },
+    {
+      "epoch": 0.7667737771552553,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7198686100144888e-06,
+      "loss": 0.7885,
+      "step": 27919
+    },
+    {
+      "epoch": 0.7668012413830985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.719258819393393e-06,
+      "loss": 0.8288,
+      "step": 27920
+    },
+    {
+      "epoch": 0.7668287056109417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7186490863803462e-06,
+      "loss": 0.8531,
+      "step": 27921
+    },
+    {
+      "epoch": 0.766856169838785,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7180394109801633e-06,
+      "loss": 0.8379,
+      "step": 27922
+    },
+    {
+      "epoch": 0.7668836340666282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7174297931976746e-06,
+      "loss": 0.862,
+      "step": 27923
+    },
+    {
+      "epoch": 0.7669110982944715,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7168202330377023e-06,
+      "loss": 0.8668,
+      "step": 27924
+    },
+    {
+      "epoch": 0.7669385625223147,
+      "grad_norm": 0.0,
+      "learning_rate": 2.716210730505069e-06,
+      "loss": 0.7683,
+      "step": 27925
+    },
+    {
+      "epoch": 0.7669660267501579,
+      "grad_norm": 0.0,
+      "learning_rate": 2.715601285604602e-06,
+      "loss": 0.7939,
+      "step": 27926
+    },
+    {
+      "epoch": 0.7669934909780012,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7149918983411183e-06,
+      "loss": 0.9644,
+      "step": 27927
+    },
+    {
+      "epoch": 0.7670209552058443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.714382568719438e-06,
+      "loss": 0.8262,
+      "step": 27928
+    },
+    {
+      "epoch": 0.7670484194336876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.713773296744385e-06,
+      "loss": 0.7226,
+      "step": 27929
+    },
+    {
+      "epoch": 0.7670758836615309,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7131640824207794e-06,
+      "loss": 0.8331,
+      "step": 27930
+    },
+    {
+      "epoch": 0.7671033478893741,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7125549257534434e-06,
+      "loss": 0.9025,
+      "step": 27931
+    },
+    {
+      "epoch": 0.7671308121172173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7119458267471955e-06,
+      "loss": 0.7365,
+      "step": 27932
+    },
+    {
+      "epoch": 0.7671582763450605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.711336785406857e-06,
+      "loss": 0.8988,
+      "step": 27933
+    },
+    {
+      "epoch": 0.7671857405729038,
+      "grad_norm": 0.0,
+      "learning_rate": 2.710727801737244e-06,
+      "loss": 0.7862,
+      "step": 27934
+    },
+    {
+      "epoch": 0.7672132048007471,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7101188757431783e-06,
+      "loss": 0.8315,
+      "step": 27935
+    },
+    {
+      "epoch": 0.7672406690285902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7095100074294733e-06,
+      "loss": 0.8786,
+      "step": 27936
+    },
+    {
+      "epoch": 0.7672681332564335,
+      "grad_norm": 0.0,
+      "learning_rate": 2.708901196800948e-06,
+      "loss": 0.8211,
+      "step": 27937
+    },
+    {
+      "epoch": 0.7672955974842768,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7082924438624225e-06,
+      "loss": 0.9128,
+      "step": 27938
+    },
+    {
+      "epoch": 0.76732306171212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7076837486187134e-06,
+      "loss": 0.6716,
+      "step": 27939
+    },
+    {
+      "epoch": 0.7673505259399632,
+      "grad_norm": 0.0,
+      "learning_rate": 2.707075111074633e-06,
+      "loss": 0.7405,
+      "step": 27940
+    },
+    {
+      "epoch": 0.7673779901678064,
+      "grad_norm": 0.0,
+      "learning_rate": 2.706466531234999e-06,
+      "loss": 0.7727,
+      "step": 27941
+    },
+    {
+      "epoch": 0.7674054543956497,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7058580091046304e-06,
+      "loss": 0.8535,
+      "step": 27942
+    },
+    {
+      "epoch": 0.7674329186234929,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7052495446883355e-06,
+      "loss": 0.7568,
+      "step": 27943
+    },
+    {
+      "epoch": 0.7674603828513361,
+      "grad_norm": 0.0,
+      "learning_rate": 2.704641137990931e-06,
+      "loss": 0.7679,
+      "step": 27944
+    },
+    {
+      "epoch": 0.7674878470791794,
+      "grad_norm": 0.0,
+      "learning_rate": 2.704032789017236e-06,
+      "loss": 0.7674,
+      "step": 27945
+    },
+    {
+      "epoch": 0.7675153113070226,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7034244977720558e-06,
+      "loss": 0.8004,
+      "step": 27946
+    },
+    {
+      "epoch": 0.7675427755348658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7028162642602084e-06,
+      "loss": 0.7426,
+      "step": 27947
+    },
+    {
+      "epoch": 0.7675702397627091,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7022080884865055e-06,
+      "loss": 0.856,
+      "step": 27948
+    },
+    {
+      "epoch": 0.7675977039905523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.701599970455758e-06,
+      "loss": 0.8244,
+      "step": 27949
+    },
+    {
+      "epoch": 0.7676251682183955,
+      "grad_norm": 0.0,
+      "learning_rate": 2.700991910172782e-06,
+      "loss": 0.8093,
+      "step": 27950
+    },
+    {
+      "epoch": 0.7676526324462388,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7003839076423853e-06,
+      "loss": 0.8002,
+      "step": 27951
+    },
+    {
+      "epoch": 0.767680096674082,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6997759628693766e-06,
+      "loss": 0.9266,
+      "step": 27952
+    },
+    {
+      "epoch": 0.7677075609019253,
+      "grad_norm": 0.0,
+      "learning_rate": 2.699168075858568e-06,
+      "loss": 0.8586,
+      "step": 27953
+    },
+    {
+      "epoch": 0.7677350251297684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6985602466147688e-06,
+      "loss": 0.6983,
+      "step": 27954
+    },
+    {
+      "epoch": 0.7677624893576117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6979524751427898e-06,
+      "loss": 0.8125,
+      "step": 27955
+    },
+    {
+      "epoch": 0.767789953585455,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6973447614474393e-06,
+      "loss": 0.7899,
+      "step": 27956
+    },
+    {
+      "epoch": 0.7678174178132982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.696737105533529e-06,
+      "loss": 0.7755,
+      "step": 27957
+    },
+    {
+      "epoch": 0.7678448820411414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6961295074058604e-06,
+      "loss": 0.7963,
+      "step": 27958
+    },
+    {
+      "epoch": 0.7678723462689846,
+      "grad_norm": 0.0,
+      "learning_rate": 2.695521967069248e-06,
+      "loss": 0.7555,
+      "step": 27959
+    },
+    {
+      "epoch": 0.7678998104968279,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6949144845284912e-06,
+      "loss": 0.8096,
+      "step": 27960
+    },
+    {
+      "epoch": 0.7679272747246711,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6943070597884027e-06,
+      "loss": 0.6792,
+      "step": 27961
+    },
+    {
+      "epoch": 0.7679547389525143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.693699692853785e-06,
+      "loss": 0.8799,
+      "step": 27962
+    },
+    {
+      "epoch": 0.7679822031803576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6930923837294488e-06,
+      "loss": 0.7247,
+      "step": 27963
+    },
+    {
+      "epoch": 0.7680096674082009,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6924851324201926e-06,
+      "loss": 0.7537,
+      "step": 27964
+    },
+    {
+      "epoch": 0.768037131636044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.691877938930826e-06,
+      "loss": 0.8041,
+      "step": 27965
+    },
+    {
+      "epoch": 0.7680645958638873,
+      "grad_norm": 0.0,
+      "learning_rate": 2.691270803266154e-06,
+      "loss": 0.8589,
+      "step": 27966
+    },
+    {
+      "epoch": 0.7680920600917305,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6906637254309762e-06,
+      "loss": 0.8019,
+      "step": 27967
+    },
+    {
+      "epoch": 0.7681195243195738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6900567054300973e-06,
+      "loss": 0.8065,
+      "step": 27968
+    },
+    {
+      "epoch": 0.768146988547417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6894497432683254e-06,
+      "loss": 0.8812,
+      "step": 27969
+    },
+    {
+      "epoch": 0.7681744527752602,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6888428389504563e-06,
+      "loss": 0.8504,
+      "step": 27970
+    },
+    {
+      "epoch": 0.7682019170031035,
+      "grad_norm": 0.0,
+      "learning_rate": 2.688235992481294e-06,
+      "loss": 0.8844,
+      "step": 27971
+    },
+    {
+      "epoch": 0.7682293812309466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.68762920386564e-06,
+      "loss": 0.8487,
+      "step": 27972
+    },
+    {
+      "epoch": 0.7682568454587899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.687022473108298e-06,
+      "loss": 0.7554,
+      "step": 27973
+    },
+    {
+      "epoch": 0.7682843096866332,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6864158002140693e-06,
+      "loss": 0.8684,
+      "step": 27974
+    },
+    {
+      "epoch": 0.7683117739144764,
+      "grad_norm": 0.0,
+      "learning_rate": 2.685809185187751e-06,
+      "loss": 0.7975,
+      "step": 27975
+    },
+    {
+      "epoch": 0.7683392381423196,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6852026280341413e-06,
+      "loss": 0.7802,
+      "step": 27976
+    },
+    {
+      "epoch": 0.7683667023701629,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6845961287580424e-06,
+      "loss": 0.8457,
+      "step": 27977
+    },
+    {
+      "epoch": 0.7683941665980061,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6839896873642526e-06,
+      "loss": 0.7596,
+      "step": 27978
+    },
+    {
+      "epoch": 0.7684216308258494,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6833833038575697e-06,
+      "loss": 0.8194,
+      "step": 27979
+    },
+    {
+      "epoch": 0.7684490950536925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.682776978242795e-06,
+      "loss": 0.8309,
+      "step": 27980
+    },
+    {
+      "epoch": 0.7684765592815358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6821707105247206e-06,
+      "loss": 0.8171,
+      "step": 27981
+    },
+    {
+      "epoch": 0.7685040235093791,
+      "grad_norm": 0.0,
+      "learning_rate": 2.681564500708147e-06,
+      "loss": 0.8136,
+      "step": 27982
+    },
+    {
+      "epoch": 0.7685314877372222,
+      "grad_norm": 0.0,
+      "learning_rate": 2.680958348797872e-06,
+      "loss": 0.8761,
+      "step": 27983
+    },
+    {
+      "epoch": 0.7685589519650655,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6803522547986873e-06,
+      "loss": 0.7812,
+      "step": 27984
+    },
+    {
+      "epoch": 0.7685864161929087,
+      "grad_norm": 0.0,
+      "learning_rate": 2.679746218715391e-06,
+      "loss": 0.8182,
+      "step": 27985
+    },
+    {
+      "epoch": 0.768613880420752,
+      "grad_norm": 0.0,
+      "learning_rate": 2.67914024055278e-06,
+      "loss": 0.884,
+      "step": 27986
+    },
+    {
+      "epoch": 0.7686413446485952,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6785343203156454e-06,
+      "loss": 0.8768,
+      "step": 27987
+    },
+    {
+      "epoch": 0.7686688088764384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.677928458008783e-06,
+      "loss": 0.8043,
+      "step": 27988
+    },
+    {
+      "epoch": 0.7686962731042817,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6773226536369867e-06,
+      "loss": 0.7403,
+      "step": 27989
+    },
+    {
+      "epoch": 0.768723737332125,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6767169072050537e-06,
+      "loss": 0.737,
+      "step": 27990
+    },
+    {
+      "epoch": 0.7687512015599681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6761112187177697e-06,
+      "loss": 0.9019,
+      "step": 27991
+    },
+    {
+      "epoch": 0.7687786657878114,
+      "grad_norm": 0.0,
+      "learning_rate": 2.675505588179934e-06,
+      "loss": 0.8464,
+      "step": 27992
+    },
+    {
+      "epoch": 0.7688061300156546,
+      "grad_norm": 0.0,
+      "learning_rate": 2.674900015596332e-06,
+      "loss": 0.7593,
+      "step": 27993
+    },
+    {
+      "epoch": 0.7688335942434978,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6742945009717593e-06,
+      "loss": 0.8209,
+      "step": 27994
+    },
+    {
+      "epoch": 0.7688610584713411,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6736890443110063e-06,
+      "loss": 0.855,
+      "step": 27995
+    },
+    {
+      "epoch": 0.7688885226991843,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6730836456188623e-06,
+      "loss": 0.8464,
+      "step": 27996
+    },
+    {
+      "epoch": 0.7689159869270276,
+      "grad_norm": 0.0,
+      "learning_rate": 2.67247830490012e-06,
+      "loss": 0.8168,
+      "step": 27997
+    },
+    {
+      "epoch": 0.7689434511548707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.67187302215957e-06,
+      "loss": 0.7791,
+      "step": 27998
+    },
+    {
+      "epoch": 0.768970915382714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6712677974019994e-06,
+      "loss": 0.7954,
+      "step": 27999
+    },
+    {
+      "epoch": 0.7689983796105573,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6706626306321935e-06,
+      "loss": 0.804,
+      "step": 28000
+    },
+    {
+      "epoch": 0.7690258438384004,
+      "grad_norm": 0.0,
+      "learning_rate": 2.670057521854945e-06,
+      "loss": 0.842,
+      "step": 28001
+    },
+    {
+      "epoch": 0.7690533080662437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6694524710750404e-06,
+      "loss": 0.8376,
+      "step": 28002
+    },
+    {
+      "epoch": 0.769080772294087,
+      "grad_norm": 0.0,
+      "learning_rate": 2.668847478297266e-06,
+      "loss": 0.9367,
+      "step": 28003
+    },
+    {
+      "epoch": 0.7691082365219302,
+      "grad_norm": 0.0,
+      "learning_rate": 2.668242543526415e-06,
+      "loss": 0.8456,
+      "step": 28004
+    },
+    {
+      "epoch": 0.7691357007497734,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6676376667672655e-06,
+      "loss": 0.8022,
+      "step": 28005
+    },
+    {
+      "epoch": 0.7691631649776166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6670328480246065e-06,
+      "loss": 0.9194,
+      "step": 28006
+    },
+    {
+      "epoch": 0.7691906292054599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.666428087303228e-06,
+      "loss": 0.767,
+      "step": 28007
+    },
+    {
+      "epoch": 0.7692180934333032,
+      "grad_norm": 0.0,
+      "learning_rate": 2.665823384607907e-06,
+      "loss": 0.7804,
+      "step": 28008
+    },
+    {
+      "epoch": 0.7692455576611463,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6652187399434336e-06,
+      "loss": 0.8317,
+      "step": 28009
+    },
+    {
+      "epoch": 0.7692730218889896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.664614153314593e-06,
+      "loss": 0.8279,
+      "step": 28010
+    },
+    {
+      "epoch": 0.7693004861168328,
+      "grad_norm": 0.0,
+      "learning_rate": 2.664009624726164e-06,
+      "loss": 0.8235,
+      "step": 28011
+    },
+    {
+      "epoch": 0.769327950344676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.663405154182933e-06,
+      "loss": 0.8086,
+      "step": 28012
+    },
+    {
+      "epoch": 0.7693554145725193,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6628007416896815e-06,
+      "loss": 0.856,
+      "step": 28013
+    },
+    {
+      "epoch": 0.7693828788003625,
+      "grad_norm": 0.0,
+      "learning_rate": 2.662196387251196e-06,
+      "loss": 0.6968,
+      "step": 28014
+    },
+    {
+      "epoch": 0.7694103430282058,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6615920908722525e-06,
+      "loss": 0.8442,
+      "step": 28015
+    },
+    {
+      "epoch": 0.769437807256049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.660987852557637e-06,
+      "loss": 0.7725,
+      "step": 28016
+    },
+    {
+      "epoch": 0.7694652714838922,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6603836723121266e-06,
+      "loss": 0.8609,
+      "step": 28017
+    },
+    {
+      "epoch": 0.7694927357117355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6597795501405034e-06,
+      "loss": 0.7949,
+      "step": 28018
+    },
+    {
+      "epoch": 0.7695201999395787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.659175486047547e-06,
+      "loss": 0.852,
+      "step": 28019
+    },
+    {
+      "epoch": 0.7695476641674219,
+      "grad_norm": 0.0,
+      "learning_rate": 2.658571480038038e-06,
+      "loss": 0.9631,
+      "step": 28020
+    },
+    {
+      "epoch": 0.7695751283952652,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6579675321167565e-06,
+      "loss": 0.8217,
+      "step": 28021
+    },
+    {
+      "epoch": 0.7696025926231084,
+      "grad_norm": 0.0,
+      "learning_rate": 2.657363642288482e-06,
+      "loss": 0.7773,
+      "step": 28022
+    },
+    {
+      "epoch": 0.7696300568509516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6567598105579895e-06,
+      "loss": 0.7576,
+      "step": 28023
+    },
+    {
+      "epoch": 0.7696575210787948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6561560369300564e-06,
+      "loss": 0.7623,
+      "step": 28024
+    },
+    {
+      "epoch": 0.7696849853066381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6555523214094614e-06,
+      "loss": 0.9023,
+      "step": 28025
+    },
+    {
+      "epoch": 0.7697124495344814,
+      "grad_norm": 0.0,
+      "learning_rate": 2.65494866400098e-06,
+      "loss": 0.7753,
+      "step": 28026
+    },
+    {
+      "epoch": 0.7697399137623245,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6543450647093915e-06,
+      "loss": 0.7259,
+      "step": 28027
+    },
+    {
+      "epoch": 0.7697673779901678,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6537415235394724e-06,
+      "loss": 0.8339,
+      "step": 28028
+    },
+    {
+      "epoch": 0.7697948422180111,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6531380404959937e-06,
+      "loss": 0.8568,
+      "step": 28029
+    },
+    {
+      "epoch": 0.7698223064458543,
+      "grad_norm": 0.0,
+      "learning_rate": 2.652534615583735e-06,
+      "loss": 0.8673,
+      "step": 28030
+    },
+    {
+      "epoch": 0.7698497706736975,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6519312488074666e-06,
+      "loss": 0.8511,
+      "step": 28031
+    },
+    {
+      "epoch": 0.7698772349015407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.651327940171964e-06,
+      "loss": 0.8308,
+      "step": 28032
+    },
+    {
+      "epoch": 0.769904699129384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6507246896820004e-06,
+      "loss": 0.7822,
+      "step": 28033
+    },
+    {
+      "epoch": 0.7699321633572273,
+      "grad_norm": 0.0,
+      "learning_rate": 2.650121497342354e-06,
+      "loss": 0.8872,
+      "step": 28034
+    },
+    {
+      "epoch": 0.7699596275850704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6495183631577904e-06,
+      "loss": 0.7832,
+      "step": 28035
+    },
+    {
+      "epoch": 0.7699870918129137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.648915287133085e-06,
+      "loss": 0.806,
+      "step": 28036
+    },
+    {
+      "epoch": 0.7700145560407569,
+      "grad_norm": 0.0,
+      "learning_rate": 2.648312269273008e-06,
+      "loss": 0.8163,
+      "step": 28037
+    },
+    {
+      "epoch": 0.7700420202686001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.647709309582336e-06,
+      "loss": 0.8909,
+      "step": 28038
+    },
+    {
+      "epoch": 0.7700694844964434,
+      "grad_norm": 0.0,
+      "learning_rate": 2.647106408065836e-06,
+      "loss": 0.8979,
+      "step": 28039
+    },
+    {
+      "epoch": 0.7700969487242866,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6465035647282753e-06,
+      "loss": 0.7369,
+      "step": 28040
+    },
+    {
+      "epoch": 0.7701244129521299,
+      "grad_norm": 0.0,
+      "learning_rate": 2.645900779574427e-06,
+      "loss": 0.8279,
+      "step": 28041
+    },
+    {
+      "epoch": 0.7701518771799731,
+      "grad_norm": 0.0,
+      "learning_rate": 2.64529805260906e-06,
+      "loss": 0.806,
+      "step": 28042
+    },
+    {
+      "epoch": 0.7701793414078163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6446953838369437e-06,
+      "loss": 0.7417,
+      "step": 28043
+    },
+    {
+      "epoch": 0.7702068056356596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6440927732628476e-06,
+      "loss": 0.7768,
+      "step": 28044
+    },
+    {
+      "epoch": 0.7702342698635027,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6434902208915414e-06,
+      "loss": 0.8857,
+      "step": 28045
+    },
+    {
+      "epoch": 0.770261734091346,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6428877267277863e-06,
+      "loss": 0.8005,
+      "step": 28046
+    },
+    {
+      "epoch": 0.7702891983191893,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6422852907763573e-06,
+      "loss": 0.745,
+      "step": 28047
+    },
+    {
+      "epoch": 0.7703166625470325,
+      "grad_norm": 0.0,
+      "learning_rate": 2.641682913042014e-06,
+      "loss": 0.9249,
+      "step": 28048
+    },
+    {
+      "epoch": 0.7703441267748757,
+      "grad_norm": 0.0,
+      "learning_rate": 2.641080593529527e-06,
+      "loss": 0.8332,
+      "step": 28049
+    },
+    {
+      "epoch": 0.770371591002719,
+      "grad_norm": 0.0,
+      "learning_rate": 2.640478332243659e-06,
+      "loss": 0.8188,
+      "step": 28050
+    },
+    {
+      "epoch": 0.7703990552305622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.639876129189182e-06,
+      "loss": 0.7897,
+      "step": 28051
+    },
+    {
+      "epoch": 0.7704265194584055,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6392739843708513e-06,
+      "loss": 0.7852,
+      "step": 28052
+    },
+    {
+      "epoch": 0.7704539836862486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6386718977934368e-06,
+      "loss": 0.773,
+      "step": 28053
+    },
+    {
+      "epoch": 0.7704814479140919,
+      "grad_norm": 0.0,
+      "learning_rate": 2.638069869461706e-06,
+      "loss": 0.7413,
+      "step": 28054
+    },
+    {
+      "epoch": 0.7705089121419352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.637467899380414e-06,
+      "loss": 0.7889,
+      "step": 28055
+    },
+    {
+      "epoch": 0.7705363763697783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.636865987554328e-06,
+      "loss": 0.8494,
+      "step": 28056
+    },
+    {
+      "epoch": 0.7705638405976216,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6362641339882133e-06,
+      "loss": 0.8415,
+      "step": 28057
+    },
+    {
+      "epoch": 0.7705913048254648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6356623386868264e-06,
+      "loss": 0.781,
+      "step": 28058
+    },
+    {
+      "epoch": 0.7706187690533081,
+      "grad_norm": 0.0,
+      "learning_rate": 2.635060601654932e-06,
+      "loss": 0.844,
+      "step": 28059
+    },
+    {
+      "epoch": 0.7706462332811513,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6344589228972906e-06,
+      "loss": 0.9494,
+      "step": 28060
+    },
+    {
+      "epoch": 0.7706736975089945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.633857302418663e-06,
+      "loss": 0.6859,
+      "step": 28061
+    },
+    {
+      "epoch": 0.7707011617368378,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6332557402238134e-06,
+      "loss": 0.8223,
+      "step": 28062
+    },
+    {
+      "epoch": 0.7707286259646811,
+      "grad_norm": 0.0,
+      "learning_rate": 2.632654236317498e-06,
+      "loss": 0.7776,
+      "step": 28063
+    },
+    {
+      "epoch": 0.7707560901925242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6320527907044733e-06,
+      "loss": 0.7422,
+      "step": 28064
+    },
+    {
+      "epoch": 0.7707835544203675,
+      "grad_norm": 0.0,
+      "learning_rate": 2.631451403389501e-06,
+      "loss": 0.7353,
+      "step": 28065
+    },
+    {
+      "epoch": 0.7708110186482107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6308500743773392e-06,
+      "loss": 0.8653,
+      "step": 28066
+    },
+    {
+      "epoch": 0.7708384828760539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6302488036727457e-06,
+      "loss": 0.8413,
+      "step": 28067
+    },
+    {
+      "epoch": 0.7708659471038972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6296475912804786e-06,
+      "loss": 0.8347,
+      "step": 28068
+    },
+    {
+      "epoch": 0.7708934113317404,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6290464372052983e-06,
+      "loss": 0.8369,
+      "step": 28069
+    },
+    {
+      "epoch": 0.7709208755595837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.628445341451955e-06,
+      "loss": 0.7858,
+      "step": 28070
+    },
+    {
+      "epoch": 0.7709483397874268,
+      "grad_norm": 0.0,
+      "learning_rate": 2.62784430402521e-06,
+      "loss": 0.8419,
+      "step": 28071
+    },
+    {
+      "epoch": 0.7709758040152701,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6272433249298135e-06,
+      "loss": 0.9664,
+      "step": 28072
+    },
+    {
+      "epoch": 0.7710032682431134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.626642404170524e-06,
+      "loss": 0.7903,
+      "step": 28073
+    },
+    {
+      "epoch": 0.7710307324709565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.626041541752097e-06,
+      "loss": 0.8602,
+      "step": 28074
+    },
+    {
+      "epoch": 0.7710581966987998,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6254407376792867e-06,
+      "loss": 0.7297,
+      "step": 28075
+    },
+    {
+      "epoch": 0.7710856609266431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6248399919568446e-06,
+      "loss": 0.8899,
+      "step": 28076
+    },
+    {
+      "epoch": 0.7711131251544863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6242393045895244e-06,
+      "loss": 0.8225,
+      "step": 28077
+    },
+    {
+      "epoch": 0.7711405893823295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.623638675582083e-06,
+      "loss": 0.9033,
+      "step": 28078
+    },
+    {
+      "epoch": 0.7711680536101727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.623038104939267e-06,
+      "loss": 0.8468,
+      "step": 28079
+    },
+    {
+      "epoch": 0.771195517838016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6224375926658307e-06,
+      "loss": 0.8673,
+      "step": 28080
+    },
+    {
+      "epoch": 0.7712229820658593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.62183713876653e-06,
+      "loss": 0.7257,
+      "step": 28081
+    },
+    {
+      "epoch": 0.7712504462937024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6212367432461084e-06,
+      "loss": 0.8417,
+      "step": 28082
+    },
+    {
+      "epoch": 0.7712779105215457,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6206364061093194e-06,
+      "loss": 0.7444,
+      "step": 28083
+    },
+    {
+      "epoch": 0.7713053747493889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.620036127360914e-06,
+      "loss": 0.8076,
+      "step": 28084
+    },
+    {
+      "epoch": 0.7713328389772321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.619435907005642e-06,
+      "loss": 0.8052,
+      "step": 28085
+    },
+    {
+      "epoch": 0.7713603032050754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.618835745048254e-06,
+      "loss": 0.7717,
+      "step": 28086
+    },
+    {
+      "epoch": 0.7713877674329186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6182356414934973e-06,
+      "loss": 0.927,
+      "step": 28087
+    },
+    {
+      "epoch": 0.7714152316607619,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6176355963461164e-06,
+      "loss": 0.7551,
+      "step": 28088
+    },
+    {
+      "epoch": 0.7714426958886051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.617035609610863e-06,
+      "loss": 0.8394,
+      "step": 28089
+    },
+    {
+      "epoch": 0.7714701601164483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.616435681292483e-06,
+      "loss": 0.7869,
+      "step": 28090
+    },
+    {
+      "epoch": 0.7714976243442916,
+      "grad_norm": 0.0,
+      "learning_rate": 2.615835811395724e-06,
+      "loss": 0.7349,
+      "step": 28091
+    },
+    {
+      "epoch": 0.7715250885721348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.615235999925333e-06,
+      "loss": 0.8087,
+      "step": 28092
+    },
+    {
+      "epoch": 0.771552552799978,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6146362468860573e-06,
+      "loss": 0.8759,
+      "step": 28093
+    },
+    {
+      "epoch": 0.7715800170278213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6140365522826383e-06,
+      "loss": 0.8949,
+      "step": 28094
+    },
+    {
+      "epoch": 0.7716074812556645,
+      "grad_norm": 0.0,
+      "learning_rate": 2.613436916119826e-06,
+      "loss": 0.8286,
+      "step": 28095
+    },
+    {
+      "epoch": 0.7716349454835078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6128373384023596e-06,
+      "loss": 0.8129,
+      "step": 28096
+    },
+    {
+      "epoch": 0.7716624097113509,
+      "grad_norm": 0.0,
+      "learning_rate": 2.612237819134985e-06,
+      "loss": 0.6896,
+      "step": 28097
+    },
+    {
+      "epoch": 0.7716898739391942,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6116383583224503e-06,
+      "loss": 0.9207,
+      "step": 28098
+    },
+    {
+      "epoch": 0.7717173381670375,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6110389559694902e-06,
+      "loss": 0.8534,
+      "step": 28099
+    },
+    {
+      "epoch": 0.7717448023948806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.610439612080854e-06,
+      "loss": 0.8392,
+      "step": 28100
+    },
+    {
+      "epoch": 0.7717722666227239,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6098403266612815e-06,
+      "loss": 0.8566,
+      "step": 28101
+    },
+    {
+      "epoch": 0.7717997308505672,
+      "grad_norm": 0.0,
+      "learning_rate": 2.609241099715517e-06,
+      "loss": 0.9015,
+      "step": 28102
+    },
+    {
+      "epoch": 0.7718271950784104,
+      "grad_norm": 0.0,
+      "learning_rate": 2.608641931248298e-06,
+      "loss": 0.778,
+      "step": 28103
+    },
+    {
+      "epoch": 0.7718546593062536,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6080428212643692e-06,
+      "loss": 0.7998,
+      "step": 28104
+    },
+    {
+      "epoch": 0.7718821235340968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6074437697684673e-06,
+      "loss": 0.8506,
+      "step": 28105
+    },
+    {
+      "epoch": 0.7719095877619401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6068447767653328e-06,
+      "loss": 0.8585,
+      "step": 28106
+    },
+    {
+      "epoch": 0.7719370519897834,
+      "grad_norm": 0.0,
+      "learning_rate": 2.606245842259707e-06,
+      "loss": 0.8958,
+      "step": 28107
+    },
+    {
+      "epoch": 0.7719645162176265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.605646966256328e-06,
+      "loss": 0.8125,
+      "step": 28108
+    },
+    {
+      "epoch": 0.7719919804454698,
+      "grad_norm": 0.0,
+      "learning_rate": 2.605048148759933e-06,
+      "loss": 0.8333,
+      "step": 28109
+    },
+    {
+      "epoch": 0.772019444673313,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6044493897752667e-06,
+      "loss": 0.8911,
+      "step": 28110
+    },
+    {
+      "epoch": 0.7720469089011562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6038506893070604e-06,
+      "loss": 0.8699,
+      "step": 28111
+    },
+    {
+      "epoch": 0.7720743731289995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.60325204736005e-06,
+      "loss": 0.7643,
+      "step": 28112
+    },
+    {
+      "epoch": 0.7721018373568427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6026534639389745e-06,
+      "loss": 0.8263,
+      "step": 28113
+    },
+    {
+      "epoch": 0.772129301584686,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6020549390485695e-06,
+      "loss": 0.7823,
+      "step": 28114
+    },
+    {
+      "epoch": 0.7721567658125292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6014564726935722e-06,
+      "loss": 0.9094,
+      "step": 28115
+    },
+    {
+      "epoch": 0.7721842300403724,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6008580648787195e-06,
+      "loss": 0.8125,
+      "step": 28116
+    },
+    {
+      "epoch": 0.7722116942682157,
+      "grad_norm": 0.0,
+      "learning_rate": 2.600259715608742e-06,
+      "loss": 0.7461,
+      "step": 28117
+    },
+    {
+      "epoch": 0.7722391584960588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5996614248883755e-06,
+      "loss": 0.8055,
+      "step": 28118
+    },
+    {
+      "epoch": 0.7722666227239021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5990631927223574e-06,
+      "loss": 0.8474,
+      "step": 28119
+    },
+    {
+      "epoch": 0.7722940869517454,
+      "grad_norm": 0.0,
+      "learning_rate": 2.598465019115416e-06,
+      "loss": 0.8322,
+      "step": 28120
+    },
+    {
+      "epoch": 0.7723215511795886,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5978669040722857e-06,
+      "loss": 0.7989,
+      "step": 28121
+    },
+    {
+      "epoch": 0.7723490154074318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5972688475977027e-06,
+      "loss": 0.8876,
+      "step": 28122
+    },
+    {
+      "epoch": 0.772376479635275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5966708496963933e-06,
+      "loss": 0.7533,
+      "step": 28123
+    },
+    {
+      "epoch": 0.7724039438631183,
+      "grad_norm": 0.0,
+      "learning_rate": 2.596072910373092e-06,
+      "loss": 0.7536,
+      "step": 28124
+    },
+    {
+      "epoch": 0.7724314080909616,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5954750296325294e-06,
+      "loss": 0.7718,
+      "step": 28125
+    },
+    {
+      "epoch": 0.7724588723188047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5948772074794405e-06,
+      "loss": 0.829,
+      "step": 28126
+    },
+    {
+      "epoch": 0.772486336546648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.594279443918547e-06,
+      "loss": 0.8563,
+      "step": 28127
+    },
+    {
+      "epoch": 0.7725138007744913,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5936817389545878e-06,
+      "loss": 0.8063,
+      "step": 28128
+    },
+    {
+      "epoch": 0.7725412650023344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.593084092592284e-06,
+      "loss": 0.833,
+      "step": 28129
+    },
+    {
+      "epoch": 0.7725687292301777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.592486504836368e-06,
+      "loss": 0.7924,
+      "step": 28130
+    },
+    {
+      "epoch": 0.7725961934580209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5918889756915677e-06,
+      "loss": 0.8862,
+      "step": 28131
+    },
+    {
+      "epoch": 0.7726236576858642,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5912915051626118e-06,
+      "loss": 0.8223,
+      "step": 28132
+    },
+    {
+      "epoch": 0.7726511219137074,
+      "grad_norm": 0.0,
+      "learning_rate": 2.590694093254228e-06,
+      "loss": 0.7812,
+      "step": 28133
+    },
+    {
+      "epoch": 0.7726785861415506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.590096739971145e-06,
+      "loss": 0.8938,
+      "step": 28134
+    },
+    {
+      "epoch": 0.7727060503693939,
+      "grad_norm": 0.0,
+      "learning_rate": 2.589499445318088e-06,
+      "loss": 0.8765,
+      "step": 28135
+    },
+    {
+      "epoch": 0.772733514597237,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5889022092997774e-06,
+      "loss": 0.8424,
+      "step": 28136
+    },
+    {
+      "epoch": 0.7727609788250803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5883050319209446e-06,
+      "loss": 0.7737,
+      "step": 28137
+    },
+    {
+      "epoch": 0.7727884430529236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5877079131863125e-06,
+      "loss": 0.7806,
+      "step": 28138
+    },
+    {
+      "epoch": 0.7728159072807668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.587110853100606e-06,
+      "loss": 0.8642,
+      "step": 28139
+    },
+    {
+      "epoch": 0.77284337150861,
+      "grad_norm": 0.0,
+      "learning_rate": 2.586513851668554e-06,
+      "loss": 0.8086,
+      "step": 28140
+    },
+    {
+      "epoch": 0.7728708357364533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5859169088948734e-06,
+      "loss": 0.7516,
+      "step": 28141
+    },
+    {
+      "epoch": 0.7728982999642965,
+      "grad_norm": 0.0,
+      "learning_rate": 2.58532002478429e-06,
+      "loss": 0.7892,
+      "step": 28142
+    },
+    {
+      "epoch": 0.7729257641921398,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5847231993415288e-06,
+      "loss": 0.8802,
+      "step": 28143
+    },
+    {
+      "epoch": 0.7729532284199829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5841264325713078e-06,
+      "loss": 0.8832,
+      "step": 28144
+    },
+    {
+      "epoch": 0.7729806926478262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.583529724478351e-06,
+      "loss": 0.8291,
+      "step": 28145
+    },
+    {
+      "epoch": 0.7730081568756695,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5829330750673833e-06,
+      "loss": 0.8943,
+      "step": 28146
+    },
+    {
+      "epoch": 0.7730356211035126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5823364843431187e-06,
+      "loss": 0.875,
+      "step": 28147
+    },
+    {
+      "epoch": 0.7730630853313559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5817399523102803e-06,
+      "loss": 0.9633,
+      "step": 28148
+    },
+    {
+      "epoch": 0.7730905495591991,
+      "grad_norm": 0.0,
+      "learning_rate": 2.58114347897359e-06,
+      "loss": 0.7016,
+      "step": 28149
+    },
+    {
+      "epoch": 0.7731180137870424,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5805470643377694e-06,
+      "loss": 0.8615,
+      "step": 28150
+    },
+    {
+      "epoch": 0.7731454780148856,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5799507084075315e-06,
+      "loss": 0.8937,
+      "step": 28151
+    },
+    {
+      "epoch": 0.7731729422427288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5793544111876e-06,
+      "loss": 0.9008,
+      "step": 28152
+    },
+    {
+      "epoch": 0.7732004064705721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.578758172682689e-06,
+      "loss": 0.778,
+      "step": 28153
+    },
+    {
+      "epoch": 0.7732278706984154,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5781619928975177e-06,
+      "loss": 0.7952,
+      "step": 28154
+    },
+    {
+      "epoch": 0.7732553349262585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5775658718368034e-06,
+      "loss": 0.7928,
+      "step": 28155
+    },
+    {
+      "epoch": 0.7732827991541018,
+      "grad_norm": 0.0,
+      "learning_rate": 2.576969809505263e-06,
+      "loss": 0.7824,
+      "step": 28156
+    },
+    {
+      "epoch": 0.773310263381945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.576373805907616e-06,
+      "loss": 0.8089,
+      "step": 28157
+    },
+    {
+      "epoch": 0.7733377276097883,
+      "grad_norm": 0.0,
+      "learning_rate": 2.575777861048572e-06,
+      "loss": 0.927,
+      "step": 28158
+    },
+    {
+      "epoch": 0.7733651918376315,
+      "grad_norm": 0.0,
+      "learning_rate": 2.575181974932852e-06,
+      "loss": 0.8367,
+      "step": 28159
+    },
+    {
+      "epoch": 0.7733926560654747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5745861475651668e-06,
+      "loss": 0.7778,
+      "step": 28160
+    },
+    {
+      "epoch": 0.773420120293318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5739903789502307e-06,
+      "loss": 0.8779,
+      "step": 28161
+    },
+    {
+      "epoch": 0.7734475845211611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.57339466909276e-06,
+      "loss": 0.774,
+      "step": 28162
+    },
+    {
+      "epoch": 0.7734750487490044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5727990179974705e-06,
+      "loss": 0.7382,
+      "step": 28163
+    },
+    {
+      "epoch": 0.7735025129768477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.57220342566907e-06,
+      "loss": 0.7714,
+      "step": 28164
+    },
+    {
+      "epoch": 0.7735299772046909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5716078921122724e-06,
+      "loss": 0.8281,
+      "step": 28165
+    },
+    {
+      "epoch": 0.7735574414325341,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5710124173317897e-06,
+      "loss": 0.7837,
+      "step": 28166
+    },
+    {
+      "epoch": 0.7735849056603774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.570417001332338e-06,
+      "loss": 0.894,
+      "step": 28167
+    },
+    {
+      "epoch": 0.7736123698882206,
+      "grad_norm": 0.0,
+      "learning_rate": 2.569821644118623e-06,
+      "loss": 0.8049,
+      "step": 28168
+    },
+    {
+      "epoch": 0.7736398341160639,
+      "grad_norm": 0.0,
+      "learning_rate": 2.569226345695359e-06,
+      "loss": 0.8277,
+      "step": 28169
+    },
+    {
+      "epoch": 0.773667298343907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.568631106067253e-06,
+      "loss": 0.8741,
+      "step": 28170
+    },
+    {
+      "epoch": 0.7736947625717503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5680359252390153e-06,
+      "loss": 0.886,
+      "step": 28171
+    },
+    {
+      "epoch": 0.7737222267995936,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5674408032153565e-06,
+      "loss": 0.7758,
+      "step": 28172
+    },
+    {
+      "epoch": 0.7737496910274367,
+      "grad_norm": 0.0,
+      "learning_rate": 2.566845740000985e-06,
+      "loss": 0.8201,
+      "step": 28173
+    },
+    {
+      "epoch": 0.77377715525528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.566250735600613e-06,
+      "loss": 0.8239,
+      "step": 28174
+    },
+    {
+      "epoch": 0.7738046194831232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5656557900189437e-06,
+      "loss": 0.7992,
+      "step": 28175
+    },
+    {
+      "epoch": 0.7738320837109665,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5650609032606832e-06,
+      "loss": 0.8187,
+      "step": 28176
+    },
+    {
+      "epoch": 0.7738595479388097,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5644660753305407e-06,
+      "loss": 0.8314,
+      "step": 28177
+    },
+    {
+      "epoch": 0.7738870121666529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5638713062332233e-06,
+      "loss": 0.7794,
+      "step": 28178
+    },
+    {
+      "epoch": 0.7739144763944962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.563276595973436e-06,
+      "loss": 0.8517,
+      "step": 28179
+    },
+    {
+      "epoch": 0.7739419406223395,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5626819445558846e-06,
+      "loss": 0.8242,
+      "step": 28180
+    },
+    {
+      "epoch": 0.7739694048501826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5620873519852784e-06,
+      "loss": 0.7983,
+      "step": 28181
+    },
+    {
+      "epoch": 0.7739968690780259,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5614928182663156e-06,
+      "loss": 0.865,
+      "step": 28182
+    },
+    {
+      "epoch": 0.7740243333058691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5608983434037047e-06,
+      "loss": 0.9156,
+      "step": 28183
+    },
+    {
+      "epoch": 0.7740517975337123,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5603039274021456e-06,
+      "loss": 0.8887,
+      "step": 28184
+    },
+    {
+      "epoch": 0.7740792617615556,
+      "grad_norm": 0.0,
+      "learning_rate": 2.559709570266343e-06,
+      "loss": 0.9138,
+      "step": 28185
+    },
+    {
+      "epoch": 0.7741067259893988,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5591152720010013e-06,
+      "loss": 0.9172,
+      "step": 28186
+    },
+    {
+      "epoch": 0.7741341902172421,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5585210326108245e-06,
+      "loss": 0.7924,
+      "step": 28187
+    },
+    {
+      "epoch": 0.7741616544450852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5579268521005095e-06,
+      "loss": 0.8047,
+      "step": 28188
+    },
+    {
+      "epoch": 0.7741891186729285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.55733273047476e-06,
+      "loss": 0.77,
+      "step": 28189
+    },
+    {
+      "epoch": 0.7742165829007718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5567386677382788e-06,
+      "loss": 0.7779,
+      "step": 28190
+    },
+    {
+      "epoch": 0.7742440471286149,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5561446638957622e-06,
+      "loss": 0.7827,
+      "step": 28191
+    },
+    {
+      "epoch": 0.7742715113564582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5555507189519125e-06,
+      "loss": 0.7712,
+      "step": 28192
+    },
+    {
+      "epoch": 0.7742989755843015,
+      "grad_norm": 0.0,
+      "learning_rate": 2.554956832911432e-06,
+      "loss": 0.8035,
+      "step": 28193
+    },
+    {
+      "epoch": 0.7743264398121447,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5543630057790137e-06,
+      "loss": 0.8654,
+      "step": 28194
+    },
+    {
+      "epoch": 0.7743539040399879,
+      "grad_norm": 0.0,
+      "learning_rate": 2.553769237559359e-06,
+      "loss": 0.7821,
+      "step": 28195
+    },
+    {
+      "epoch": 0.7743813682678311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.553175528257168e-06,
+      "loss": 0.8956,
+      "step": 28196
+    },
+    {
+      "epoch": 0.7744088324956744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5525818778771362e-06,
+      "loss": 0.8712,
+      "step": 28197
+    },
+    {
+      "epoch": 0.7744362967235177,
+      "grad_norm": 0.0,
+      "learning_rate": 2.551988286423964e-06,
+      "loss": 0.7838,
+      "step": 28198
+    },
+    {
+      "epoch": 0.7744637609513608,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5513947539023465e-06,
+      "loss": 0.8075,
+      "step": 28199
+    },
+    {
+      "epoch": 0.7744912251792041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5508012803169756e-06,
+      "loss": 0.8194,
+      "step": 28200
+    },
+    {
+      "epoch": 0.7745186894070473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5502078656725503e-06,
+      "loss": 0.7543,
+      "step": 28201
+    },
+    {
+      "epoch": 0.7745461536348905,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5496145099737657e-06,
+      "loss": 0.7632,
+      "step": 28202
+    },
+    {
+      "epoch": 0.7745736178627338,
+      "grad_norm": 0.0,
+      "learning_rate": 2.549021213225318e-06,
+      "loss": 0.7937,
+      "step": 28203
+    },
+    {
+      "epoch": 0.774601082090577,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5484279754318995e-06,
+      "loss": 0.8601,
+      "step": 28204
+    },
+    {
+      "epoch": 0.7746285463184203,
+      "grad_norm": 0.0,
+      "learning_rate": 2.54783479659821e-06,
+      "loss": 0.8161,
+      "step": 28205
+    },
+    {
+      "epoch": 0.7746560105462635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5472416767289345e-06,
+      "loss": 0.7068,
+      "step": 28206
+    },
+    {
+      "epoch": 0.7746834747741067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.546648615828772e-06,
+      "loss": 0.7569,
+      "step": 28207
+    },
+    {
+      "epoch": 0.77471093900195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5460556139024106e-06,
+      "loss": 0.8553,
+      "step": 28208
+    },
+    {
+      "epoch": 0.7747384032297931,
+      "grad_norm": 0.0,
+      "learning_rate": 2.545462670954545e-06,
+      "loss": 0.846,
+      "step": 28209
+    },
+    {
+      "epoch": 0.7747658674576364,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5448697869898654e-06,
+      "loss": 0.816,
+      "step": 28210
+    },
+    {
+      "epoch": 0.7747933316854797,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5442769620130668e-06,
+      "loss": 0.8311,
+      "step": 28211
+    },
+    {
+      "epoch": 0.7748207959133229,
+      "grad_norm": 0.0,
+      "learning_rate": 2.543684196028835e-06,
+      "loss": 0.8497,
+      "step": 28212
+    },
+    {
+      "epoch": 0.7748482601411661,
+      "grad_norm": 0.0,
+      "learning_rate": 2.543091489041861e-06,
+      "loss": 0.7314,
+      "step": 28213
+    },
+    {
+      "epoch": 0.7748757243690093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5424988410568394e-06,
+      "loss": 0.8352,
+      "step": 28214
+    },
+    {
+      "epoch": 0.7749031885968526,
+      "grad_norm": 0.0,
+      "learning_rate": 2.541906252078453e-06,
+      "loss": 0.8147,
+      "step": 28215
+    },
+    {
+      "epoch": 0.7749306528246959,
+      "grad_norm": 0.0,
+      "learning_rate": 2.541313722111396e-06,
+      "loss": 0.8063,
+      "step": 28216
+    },
+    {
+      "epoch": 0.774958117052539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5407212511603507e-06,
+      "loss": 0.8598,
+      "step": 28217
+    },
+    {
+      "epoch": 0.7749855812803823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.540128839230008e-06,
+      "loss": 0.7558,
+      "step": 28218
+    },
+    {
+      "epoch": 0.7750130455082256,
+      "grad_norm": 0.0,
+      "learning_rate": 2.539536486325057e-06,
+      "loss": 0.9163,
+      "step": 28219
+    },
+    {
+      "epoch": 0.7750405097360688,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5389441924501825e-06,
+      "loss": 0.8689,
+      "step": 28220
+    },
+    {
+      "epoch": 0.775067973963912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.538351957610071e-06,
+      "loss": 0.7947,
+      "step": 28221
+    },
+    {
+      "epoch": 0.7750954381917552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5377597818094134e-06,
+      "loss": 0.7876,
+      "step": 28222
+    },
+    {
+      "epoch": 0.7751229024195985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.53716766505289e-06,
+      "loss": 0.8203,
+      "step": 28223
+    },
+    {
+      "epoch": 0.7751503666474417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5365756073451843e-06,
+      "loss": 0.7794,
+      "step": 28224
+    },
+    {
+      "epoch": 0.7751778308752849,
+      "grad_norm": 0.0,
+      "learning_rate": 2.535983608690984e-06,
+      "loss": 0.8609,
+      "step": 28225
+    },
+    {
+      "epoch": 0.7752052951031282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.535391669094971e-06,
+      "loss": 0.7924,
+      "step": 28226
+    },
+    {
+      "epoch": 0.7752327593309714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5347997885618327e-06,
+      "loss": 0.7785,
+      "step": 28227
+    },
+    {
+      "epoch": 0.7752602235588146,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5342079670962518e-06,
+      "loss": 0.8479,
+      "step": 28228
+    },
+    {
+      "epoch": 0.7752876877866579,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5336162047029066e-06,
+      "loss": 0.7704,
+      "step": 28229
+    },
+    {
+      "epoch": 0.7753151520145011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.533024501386483e-06,
+      "loss": 0.8389,
+      "step": 28230
+    },
+    {
+      "epoch": 0.7753426162423444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.532432857151663e-06,
+      "loss": 0.732,
+      "step": 28231
+    },
+    {
+      "epoch": 0.7753700804701876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.531841272003126e-06,
+      "loss": 0.8653,
+      "step": 28232
+    },
+    {
+      "epoch": 0.7753975446980308,
+      "grad_norm": 0.0,
+      "learning_rate": 2.531249745945553e-06,
+      "loss": 0.7571,
+      "step": 28233
+    },
+    {
+      "epoch": 0.7754250089258741,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5306582789836287e-06,
+      "loss": 0.811,
+      "step": 28234
+    },
+    {
+      "epoch": 0.7754524731537172,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5300668711220256e-06,
+      "loss": 0.8552,
+      "step": 28235
+    },
+    {
+      "epoch": 0.7754799373815605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5294755223654266e-06,
+      "loss": 0.8536,
+      "step": 28236
+    },
+    {
+      "epoch": 0.7755074016094038,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5288842327185127e-06,
+      "loss": 0.8227,
+      "step": 28237
+    },
+    {
+      "epoch": 0.775534865837247,
+      "grad_norm": 0.0,
+      "learning_rate": 2.528293002185962e-06,
+      "loss": 0.8769,
+      "step": 28238
+    },
+    {
+      "epoch": 0.7755623300650902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.52770183077245e-06,
+      "loss": 0.8396,
+      "step": 28239
+    },
+    {
+      "epoch": 0.7755897942929334,
+      "grad_norm": 0.0,
+      "learning_rate": 2.527110718482657e-06,
+      "loss": 0.8115,
+      "step": 28240
+    },
+    {
+      "epoch": 0.7756172585207767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5265196653212567e-06,
+      "loss": 0.8573,
+      "step": 28241
+    },
+    {
+      "epoch": 0.77564472274862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5259286712929275e-06,
+      "loss": 0.8127,
+      "step": 28242
+    },
+    {
+      "epoch": 0.7756721869764631,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5253377364023456e-06,
+      "loss": 0.8414,
+      "step": 28243
+    },
+    {
+      "epoch": 0.7756996512043064,
+      "grad_norm": 0.0,
+      "learning_rate": 2.524746860654188e-06,
+      "loss": 0.8049,
+      "step": 28244
+    },
+    {
+      "epoch": 0.7757271154321497,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5241560440531277e-06,
+      "loss": 0.863,
+      "step": 28245
+    },
+    {
+      "epoch": 0.7757545796599928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5235652866038442e-06,
+      "loss": 0.9006,
+      "step": 28246
+    },
+    {
+      "epoch": 0.7757820438878361,
+      "grad_norm": 0.0,
+      "learning_rate": 2.522974588311008e-06,
+      "loss": 0.7906,
+      "step": 28247
+    },
+    {
+      "epoch": 0.7758095081156793,
+      "grad_norm": 0.0,
+      "learning_rate": 2.52238394917929e-06,
+      "loss": 0.8291,
+      "step": 28248
+    },
+    {
+      "epoch": 0.7758369723435226,
+      "grad_norm": 0.0,
+      "learning_rate": 2.521793369213367e-06,
+      "loss": 0.7463,
+      "step": 28249
+    },
+    {
+      "epoch": 0.7758644365713658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.521202848417912e-06,
+      "loss": 0.8631,
+      "step": 28250
+    },
+    {
+      "epoch": 0.775891900799209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.520612386797596e-06,
+      "loss": 0.7928,
+      "step": 28251
+    },
+    {
+      "epoch": 0.7759193650270523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.520021984357096e-06,
+      "loss": 0.8647,
+      "step": 28252
+    },
+    {
+      "epoch": 0.7759468292548956,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5194316411010767e-06,
+      "loss": 0.8434,
+      "step": 28253
+    },
+    {
+      "epoch": 0.7759742934827387,
+      "grad_norm": 0.0,
+      "learning_rate": 2.518841357034212e-06,
+      "loss": 0.8432,
+      "step": 28254
+    },
+    {
+      "epoch": 0.776001757710582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5182511321611746e-06,
+      "loss": 0.8371,
+      "step": 28255
+    },
+    {
+      "epoch": 0.7760292219384252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.517660966486629e-06,
+      "loss": 0.8524,
+      "step": 28256
+    },
+    {
+      "epoch": 0.7760566861662684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5170708600152493e-06,
+      "loss": 0.7969,
+      "step": 28257
+    },
+    {
+      "epoch": 0.7760841503941117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.516480812751706e-06,
+      "loss": 0.8683,
+      "step": 28258
+    },
+    {
+      "epoch": 0.7761116146219549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.515890824700663e-06,
+      "loss": 0.7809,
+      "step": 28259
+    },
+    {
+      "epoch": 0.7761390788497982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5153008958667903e-06,
+      "loss": 0.7922,
+      "step": 28260
+    },
+    {
+      "epoch": 0.7761665430776413,
+      "grad_norm": 0.0,
+      "learning_rate": 2.514711026254756e-06,
+      "loss": 0.7807,
+      "step": 28261
+    },
+    {
+      "epoch": 0.7761940073054846,
+      "grad_norm": 0.0,
+      "learning_rate": 2.514121215869231e-06,
+      "loss": 0.7742,
+      "step": 28262
+    },
+    {
+      "epoch": 0.7762214715333279,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5135314647148754e-06,
+      "loss": 0.8219,
+      "step": 28263
+    },
+    {
+      "epoch": 0.776248935761171,
+      "grad_norm": 0.0,
+      "learning_rate": 2.512941772796362e-06,
+      "loss": 0.8624,
+      "step": 28264
+    },
+    {
+      "epoch": 0.7762763999890143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5123521401183503e-06,
+      "loss": 0.8407,
+      "step": 28265
+    },
+    {
+      "epoch": 0.7763038642168576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5117625666855095e-06,
+      "loss": 0.8818,
+      "step": 28266
+    },
+    {
+      "epoch": 0.7763313284447008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5111730525025034e-06,
+      "loss": 0.8736,
+      "step": 28267
+    },
+    {
+      "epoch": 0.776358792672544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5105835975739976e-06,
+      "loss": 0.7838,
+      "step": 28268
+    },
+    {
+      "epoch": 0.7763862569003872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.509994201904655e-06,
+      "loss": 0.975,
+      "step": 28269
+    },
+    {
+      "epoch": 0.7764137211282305,
+      "grad_norm": 0.0,
+      "learning_rate": 2.509404865499143e-06,
+      "loss": 0.8829,
+      "step": 28270
+    },
+    {
+      "epoch": 0.7764411853560738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5088155883621212e-06,
+      "loss": 0.8487,
+      "step": 28271
+    },
+    {
+      "epoch": 0.7764686495839169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.508226370498249e-06,
+      "loss": 0.7409,
+      "step": 28272
+    },
+    {
+      "epoch": 0.7764961138117602,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5076372119121917e-06,
+      "loss": 0.915,
+      "step": 28273
+    },
+    {
+      "epoch": 0.7765235780396034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.507048112608611e-06,
+      "loss": 0.8066,
+      "step": 28274
+    },
+    {
+      "epoch": 0.7765510422674466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.506459072592171e-06,
+      "loss": 0.812,
+      "step": 28275
+    },
+    {
+      "epoch": 0.7765785064952899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5058700918675274e-06,
+      "loss": 0.7994,
+      "step": 28276
+    },
+    {
+      "epoch": 0.7766059707231331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.505281170439342e-06,
+      "loss": 0.8978,
+      "step": 28277
+    },
+    {
+      "epoch": 0.7766334349509764,
+      "grad_norm": 0.0,
+      "learning_rate": 2.504692308312275e-06,
+      "loss": 0.8308,
+      "step": 28278
+    },
+    {
+      "epoch": 0.7766608991788196,
+      "grad_norm": 0.0,
+      "learning_rate": 2.504103505490989e-06,
+      "loss": 0.827,
+      "step": 28279
+    },
+    {
+      "epoch": 0.7766883634066628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.503514761980137e-06,
+      "loss": 0.7832,
+      "step": 28280
+    },
+    {
+      "epoch": 0.7767158276345061,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5029260777843824e-06,
+      "loss": 0.8255,
+      "step": 28281
+    },
+    {
+      "epoch": 0.7767432918623492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.502337452908379e-06,
+      "loss": 0.8356,
+      "step": 28282
+    },
+    {
+      "epoch": 0.7767707560901925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.501748887356784e-06,
+      "loss": 0.9479,
+      "step": 28283
+    },
+    {
+      "epoch": 0.7767982203180358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5011603811342577e-06,
+      "loss": 0.7674,
+      "step": 28284
+    },
+    {
+      "epoch": 0.776825684545879,
+      "grad_norm": 0.0,
+      "learning_rate": 2.500571934245456e-06,
+      "loss": 0.8727,
+      "step": 28285
+    },
+    {
+      "epoch": 0.7768531487737222,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4999835466950363e-06,
+      "loss": 0.7005,
+      "step": 28286
+    },
+    {
+      "epoch": 0.7768806130015654,
+      "grad_norm": 0.0,
+      "learning_rate": 2.499395218487648e-06,
+      "loss": 0.8053,
+      "step": 28287
+    },
+    {
+      "epoch": 0.7769080772294087,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4988069496279545e-06,
+      "loss": 0.8007,
+      "step": 28288
+    },
+    {
+      "epoch": 0.776935541457252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4982187401206025e-06,
+      "loss": 0.7474,
+      "step": 28289
+    },
+    {
+      "epoch": 0.7769630056850951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4976305899702503e-06,
+      "loss": 0.7819,
+      "step": 28290
+    },
+    {
+      "epoch": 0.7769904699129384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.49704249918155e-06,
+      "loss": 0.8572,
+      "step": 28291
+    },
+    {
+      "epoch": 0.7770179341407817,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4964544677591574e-06,
+      "loss": 0.7738,
+      "step": 28292
+    },
+    {
+      "epoch": 0.7770453983686249,
+      "grad_norm": 0.0,
+      "learning_rate": 2.495866495707725e-06,
+      "loss": 0.8324,
+      "step": 28293
+    },
+    {
+      "epoch": 0.7770728625964681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4952785830319014e-06,
+      "loss": 0.7828,
+      "step": 28294
+    },
+    {
+      "epoch": 0.7771003268243113,
+      "grad_norm": 0.0,
+      "learning_rate": 2.494690729736344e-06,
+      "loss": 0.87,
+      "step": 28295
+    },
+    {
+      "epoch": 0.7771277910521546,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4941029358256976e-06,
+      "loss": 0.814,
+      "step": 28296
+    },
+    {
+      "epoch": 0.7771552552799978,
+      "grad_norm": 0.0,
+      "learning_rate": 2.493515201304616e-06,
+      "loss": 0.8304,
+      "step": 28297
+    },
+    {
+      "epoch": 0.777182719507841,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4929275261777507e-06,
+      "loss": 0.8471,
+      "step": 28298
+    },
+    {
+      "epoch": 0.7772101837356843,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4923399104497525e-06,
+      "loss": 0.7352,
+      "step": 28299
+    },
+    {
+      "epoch": 0.7772376479635275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4917523541252675e-06,
+      "loss": 0.8016,
+      "step": 28300
+    },
+    {
+      "epoch": 0.7772651121913707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.491164857208945e-06,
+      "loss": 0.8971,
+      "step": 28301
+    },
+    {
+      "epoch": 0.777292576419214,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4905774197054356e-06,
+      "loss": 0.7565,
+      "step": 28302
+    },
+    {
+      "epoch": 0.7773200406470572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.489990041619389e-06,
+      "loss": 0.9149,
+      "step": 28303
+    },
+    {
+      "epoch": 0.7773475048749005,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4894027229554474e-06,
+      "loss": 0.7483,
+      "step": 28304
+    },
+    {
+      "epoch": 0.7773749691027437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4888154637182637e-06,
+      "loss": 0.8568,
+      "step": 28305
+    },
+    {
+      "epoch": 0.7774024333305869,
+      "grad_norm": 0.0,
+      "learning_rate": 2.48822826391248e-06,
+      "loss": 0.8429,
+      "step": 28306
+    },
+    {
+      "epoch": 0.7774298975584302,
+      "grad_norm": 0.0,
+      "learning_rate": 2.487641123542742e-06,
+      "loss": 0.8411,
+      "step": 28307
+    },
+    {
+      "epoch": 0.7774573617862733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.487054042613699e-06,
+      "loss": 0.8074,
+      "step": 28308
+    },
+    {
+      "epoch": 0.7774848260141166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4864670211299937e-06,
+      "loss": 0.8311,
+      "step": 28309
+    },
+    {
+      "epoch": 0.7775122902419599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.485880059096275e-06,
+      "loss": 0.7658,
+      "step": 28310
+    },
+    {
+      "epoch": 0.7775397544698031,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4852931565171847e-06,
+      "loss": 0.897,
+      "step": 28311
+    },
+    {
+      "epoch": 0.7775672186976463,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4847063133973614e-06,
+      "loss": 0.792,
+      "step": 28312
+    },
+    {
+      "epoch": 0.7775946829254895,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4841195297414546e-06,
+      "loss": 0.8528,
+      "step": 28313
+    },
+    {
+      "epoch": 0.7776221471533328,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4835328055541043e-06,
+      "loss": 0.7957,
+      "step": 28314
+    },
+    {
+      "epoch": 0.777649611381176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4829461408399548e-06,
+      "loss": 0.8065,
+      "step": 28315
+    },
+    {
+      "epoch": 0.7776770756090192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.482359535603647e-06,
+      "loss": 0.8098,
+      "step": 28316
+    },
+    {
+      "epoch": 0.7777045398368625,
+      "grad_norm": 0.0,
+      "learning_rate": 2.481772989849824e-06,
+      "loss": 0.8095,
+      "step": 28317
+    },
+    {
+      "epoch": 0.7777320040647058,
+      "grad_norm": 0.0,
+      "learning_rate": 2.481186503583124e-06,
+      "loss": 0.8006,
+      "step": 28318
+    },
+    {
+      "epoch": 0.7777594682925489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4806000768081917e-06,
+      "loss": 0.7877,
+      "step": 28319
+    },
+    {
+      "epoch": 0.7777869325203922,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4800137095296616e-06,
+      "loss": 0.8326,
+      "step": 28320
+    },
+    {
+      "epoch": 0.7778143967482354,
+      "grad_norm": 0.0,
+      "learning_rate": 2.479427401752176e-06,
+      "loss": 0.8552,
+      "step": 28321
+    },
+    {
+      "epoch": 0.7778418609760787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4788411534803735e-06,
+      "loss": 0.8909,
+      "step": 28322
+    },
+    {
+      "epoch": 0.7778693252039219,
+      "grad_norm": 0.0,
+      "learning_rate": 2.478254964718897e-06,
+      "loss": 0.8508,
+      "step": 28323
+    },
+    {
+      "epoch": 0.7778967894317651,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4776688354723764e-06,
+      "loss": 0.8535,
+      "step": 28324
+    },
+    {
+      "epoch": 0.7779242536596084,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4770827657454555e-06,
+      "loss": 0.8389,
+      "step": 28325
+    },
+    {
+      "epoch": 0.7779517178874515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4764967555427687e-06,
+      "loss": 0.7625,
+      "step": 28326
+    },
+    {
+      "epoch": 0.7779791821152948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4759108048689572e-06,
+      "loss": 0.8501,
+      "step": 28327
+    },
+    {
+      "epoch": 0.7780066463431381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4753249137286515e-06,
+      "loss": 0.7458,
+      "step": 28328
+    },
+    {
+      "epoch": 0.7780341105709813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4747390821264916e-06,
+      "loss": 0.8004,
+      "step": 28329
+    },
+    {
+      "epoch": 0.7780615747988245,
+      "grad_norm": 0.0,
+      "learning_rate": 2.474153310067109e-06,
+      "loss": 0.775,
+      "step": 28330
+    },
+    {
+      "epoch": 0.7780890390266678,
+      "grad_norm": 0.0,
+      "learning_rate": 2.473567597555141e-06,
+      "loss": 0.859,
+      "step": 28331
+    },
+    {
+      "epoch": 0.778116503254511,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4729819445952207e-06,
+      "loss": 0.8383,
+      "step": 28332
+    },
+    {
+      "epoch": 0.7781439674823543,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4723963511919837e-06,
+      "loss": 0.8478,
+      "step": 28333
+    },
+    {
+      "epoch": 0.7781714317101974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4718108173500655e-06,
+      "loss": 0.8816,
+      "step": 28334
+    },
+    {
+      "epoch": 0.7781988959380407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.471225343074096e-06,
+      "loss": 0.9128,
+      "step": 28335
+    },
+    {
+      "epoch": 0.778226360165884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4706399283687055e-06,
+      "loss": 0.7812,
+      "step": 28336
+    },
+    {
+      "epoch": 0.7782538243937271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4700545732385284e-06,
+      "loss": 0.7527,
+      "step": 28337
+    },
+    {
+      "epoch": 0.7782812886215704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4694692776881957e-06,
+      "loss": 0.8818,
+      "step": 28338
+    },
+    {
+      "epoch": 0.7783087528494136,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4688840417223393e-06,
+      "loss": 0.8379,
+      "step": 28339
+    },
+    {
+      "epoch": 0.7783362170772569,
+      "grad_norm": 0.0,
+      "learning_rate": 2.468298865345593e-06,
+      "loss": 0.8164,
+      "step": 28340
+    },
+    {
+      "epoch": 0.7783636813051001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4677137485625802e-06,
+      "loss": 0.8415,
+      "step": 28341
+    },
+    {
+      "epoch": 0.7783911455329433,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4671286913779347e-06,
+      "loss": 0.8286,
+      "step": 28342
+    },
+    {
+      "epoch": 0.7784186097607866,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4665436937962872e-06,
+      "loss": 0.8834,
+      "step": 28343
+    },
+    {
+      "epoch": 0.7784460739886299,
+      "grad_norm": 0.0,
+      "learning_rate": 2.465958755822261e-06,
+      "loss": 0.8571,
+      "step": 28344
+    },
+    {
+      "epoch": 0.778473538216473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4653738774604875e-06,
+      "loss": 0.8128,
+      "step": 28345
+    },
+    {
+      "epoch": 0.7785010024443163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4647890587155976e-06,
+      "loss": 0.8294,
+      "step": 28346
+    },
+    {
+      "epoch": 0.7785284666721595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4642042995922133e-06,
+      "loss": 0.81,
+      "step": 28347
+    },
+    {
+      "epoch": 0.7785559309000027,
+      "grad_norm": 0.0,
+      "learning_rate": 2.463619600094963e-06,
+      "loss": 0.7573,
+      "step": 28348
+    },
+    {
+      "epoch": 0.778583395127846,
+      "grad_norm": 0.0,
+      "learning_rate": 2.463034960228474e-06,
+      "loss": 0.7702,
+      "step": 28349
+    },
+    {
+      "epoch": 0.7786108593556892,
+      "grad_norm": 0.0,
+      "learning_rate": 2.462450379997375e-06,
+      "loss": 0.7096,
+      "step": 28350
+    },
+    {
+      "epoch": 0.7786383235835325,
+      "grad_norm": 0.0,
+      "learning_rate": 2.461865859406285e-06,
+      "loss": 0.854,
+      "step": 28351
+    },
+    {
+      "epoch": 0.7786657878113756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.461281398459836e-06,
+      "loss": 0.8638,
+      "step": 28352
+    },
+    {
+      "epoch": 0.7786932520392189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4606969971626447e-06,
+      "loss": 0.885,
+      "step": 28353
+    },
+    {
+      "epoch": 0.7787207162670622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.46011265551934e-06,
+      "loss": 0.8173,
+      "step": 28354
+    },
+    {
+      "epoch": 0.7787481804949054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4595283735345445e-06,
+      "loss": 0.77,
+      "step": 28355
+    },
+    {
+      "epoch": 0.7787756447227486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.458944151212882e-06,
+      "loss": 0.7365,
+      "step": 28356
+    },
+    {
+      "epoch": 0.7788031089505919,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4583599885589737e-06,
+      "loss": 0.8433,
+      "step": 28357
+    },
+    {
+      "epoch": 0.7788305731784351,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4577758855774457e-06,
+      "loss": 0.876,
+      "step": 28358
+    },
+    {
+      "epoch": 0.7788580374062783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4571918422729168e-06,
+      "loss": 0.8366,
+      "step": 28359
+    },
+    {
+      "epoch": 0.7788855016341215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.456607858650004e-06,
+      "loss": 0.7784,
+      "step": 28360
+    },
+    {
+      "epoch": 0.7789129658619648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.456023934713333e-06,
+      "loss": 0.8242,
+      "step": 28361
+    },
+    {
+      "epoch": 0.7789404300898081,
+      "grad_norm": 0.0,
+      "learning_rate": 2.455440070467522e-06,
+      "loss": 0.7785,
+      "step": 28362
+    },
+    {
+      "epoch": 0.7789678943176512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4548562659171915e-06,
+      "loss": 0.7614,
+      "step": 28363
+    },
+    {
+      "epoch": 0.7789953585454945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454272521066965e-06,
+      "loss": 0.8193,
+      "step": 28364
+    },
+    {
+      "epoch": 0.7790228227733377,
+      "grad_norm": 0.0,
+      "learning_rate": 2.453688835921453e-06,
+      "loss": 0.7744,
+      "step": 28365
+    },
+    {
+      "epoch": 0.779050287001181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4531052104852795e-06,
+      "loss": 0.7817,
+      "step": 28366
+    },
+    {
+      "epoch": 0.7790777512290242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4525216447630627e-06,
+      "loss": 0.818,
+      "step": 28367
+    },
+    {
+      "epoch": 0.7791052154568674,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4519381387594153e-06,
+      "loss": 0.7841,
+      "step": 28368
+    },
+    {
+      "epoch": 0.7791326796847107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4513546924789577e-06,
+      "loss": 0.7872,
+      "step": 28369
+    },
+    {
+      "epoch": 0.779160143912554,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4507713059263096e-06,
+      "loss": 0.8399,
+      "step": 28370
+    },
+    {
+      "epoch": 0.7791876081403971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.450187979106079e-06,
+      "loss": 0.7574,
+      "step": 28371
+    },
+    {
+      "epoch": 0.7792150723682404,
+      "grad_norm": 0.0,
+      "learning_rate": 2.449604712022886e-06,
+      "loss": 0.7997,
+      "step": 28372
+    },
+    {
+      "epoch": 0.7792425365960836,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4490215046813447e-06,
+      "loss": 0.8018,
+      "step": 28373
+    },
+    {
+      "epoch": 0.7792700008239268,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4484383570860736e-06,
+      "loss": 0.7481,
+      "step": 28374
+    },
+    {
+      "epoch": 0.7792974650517701,
+      "grad_norm": 0.0,
+      "learning_rate": 2.447855269241681e-06,
+      "loss": 0.8646,
+      "step": 28375
+    },
+    {
+      "epoch": 0.7793249292796133,
+      "grad_norm": 0.0,
+      "learning_rate": 2.447272241152785e-06,
+      "loss": 0.8107,
+      "step": 28376
+    },
+    {
+      "epoch": 0.7793523935074566,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4466892728239943e-06,
+      "loss": 0.8168,
+      "step": 28377
+    },
+    {
+      "epoch": 0.7793798577352997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4461063642599247e-06,
+      "loss": 0.9071,
+      "step": 28378
+    },
+    {
+      "epoch": 0.779407321963143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4455235154651867e-06,
+      "loss": 0.8746,
+      "step": 28379
+    },
+    {
+      "epoch": 0.7794347861909863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4449407264443927e-06,
+      "loss": 0.8355,
+      "step": 28380
+    },
+    {
+      "epoch": 0.7794622504188294,
+      "grad_norm": 0.0,
+      "learning_rate": 2.444357997202155e-06,
+      "loss": 0.8046,
+      "step": 28381
+    },
+    {
+      "epoch": 0.7794897146466727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4437753277430852e-06,
+      "loss": 0.8172,
+      "step": 28382
+    },
+    {
+      "epoch": 0.779517178874516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.443192718071794e-06,
+      "loss": 0.8041,
+      "step": 28383
+    },
+    {
+      "epoch": 0.7795446431023592,
+      "grad_norm": 0.0,
+      "learning_rate": 2.442610168192885e-06,
+      "loss": 0.7549,
+      "step": 28384
+    },
+    {
+      "epoch": 0.7795721073302024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.442027678110972e-06,
+      "loss": 0.8237,
+      "step": 28385
+    },
+    {
+      "epoch": 0.7795995715580456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.441445247830664e-06,
+      "loss": 0.8981,
+      "step": 28386
+    },
+    {
+      "epoch": 0.7796270357858889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.440862877356568e-06,
+      "loss": 0.8292,
+      "step": 28387
+    },
+    {
+      "epoch": 0.7796545000137322,
+      "grad_norm": 0.0,
+      "learning_rate": 2.440280566693297e-06,
+      "loss": 0.8913,
+      "step": 28388
+    },
+    {
+      "epoch": 0.7796819642415753,
+      "grad_norm": 0.0,
+      "learning_rate": 2.439698315845451e-06,
+      "loss": 0.8649,
+      "step": 28389
+    },
+    {
+      "epoch": 0.7797094284694186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.439116124817642e-06,
+      "loss": 0.773,
+      "step": 28390
+    },
+    {
+      "epoch": 0.7797368926972618,
+      "grad_norm": 0.0,
+      "learning_rate": 2.438533993614477e-06,
+      "loss": 0.7782,
+      "step": 28391
+    },
+    {
+      "epoch": 0.779764356925105,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4379519222405566e-06,
+      "loss": 0.8394,
+      "step": 28392
+    },
+    {
+      "epoch": 0.7797918211529483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.43736991070049e-06,
+      "loss": 0.8002,
+      "step": 28393
+    },
+    {
+      "epoch": 0.7798192853807915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.436787958998885e-06,
+      "loss": 0.7604,
+      "step": 28394
+    },
+    {
+      "epoch": 0.7798467496086348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.436206067140341e-06,
+      "loss": 0.8531,
+      "step": 28395
+    },
+    {
+      "epoch": 0.779874213836478,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4356242351294647e-06,
+      "loss": 0.9069,
+      "step": 28396
+    },
+    {
+      "epoch": 0.7799016780643212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.435042462970859e-06,
+      "loss": 0.763,
+      "step": 28397
+    },
+    {
+      "epoch": 0.7799291422921645,
+      "grad_norm": 0.0,
+      "learning_rate": 2.434460750669131e-06,
+      "loss": 0.8192,
+      "step": 28398
+    },
+    {
+      "epoch": 0.7799566065200076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4338790982288775e-06,
+      "loss": 0.7719,
+      "step": 28399
+    },
+    {
+      "epoch": 0.7799840707478509,
+      "grad_norm": 0.0,
+      "learning_rate": 2.433297505654706e-06,
+      "loss": 0.877,
+      "step": 28400
+    },
+    {
+      "epoch": 0.7800115349756942,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4327159729512116e-06,
+      "loss": 0.7745,
+      "step": 28401
+    },
+    {
+      "epoch": 0.7800389992035374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4321345001230012e-06,
+      "loss": 0.8707,
+      "step": 28402
+    },
+    {
+      "epoch": 0.7800664634313806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4315530871746738e-06,
+      "loss": 0.8628,
+      "step": 28403
+    },
+    {
+      "epoch": 0.7800939276592238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4309717341108294e-06,
+      "loss": 0.7373,
+      "step": 28404
+    },
+    {
+      "epoch": 0.7801213918870671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4303904409360736e-06,
+      "loss": 0.8551,
+      "step": 28405
+    },
+    {
+      "epoch": 0.7801488561149104,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4298092076549962e-06,
+      "loss": 0.8279,
+      "step": 28406
+    },
+    {
+      "epoch": 0.7801763203427535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.429228034272204e-06,
+      "loss": 0.8961,
+      "step": 28407
+    },
+    {
+      "epoch": 0.7802037845705968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.428646920792289e-06,
+      "loss": 0.8336,
+      "step": 28408
+    },
+    {
+      "epoch": 0.7802312487984401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.428065867219852e-06,
+      "loss": 0.8773,
+      "step": 28409
+    },
+    {
+      "epoch": 0.7802587130262832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4274848735594925e-06,
+      "loss": 0.8271,
+      "step": 28410
+    },
+    {
+      "epoch": 0.7802861772541265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.426903939815808e-06,
+      "loss": 0.8051,
+      "step": 28411
+    },
+    {
+      "epoch": 0.7803136414819697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.42632306599339e-06,
+      "loss": 0.8187,
+      "step": 28412
+    },
+    {
+      "epoch": 0.780341105709813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4257422520968378e-06,
+      "loss": 0.8302,
+      "step": 28413
+    },
+    {
+      "epoch": 0.7803685699376562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.425161498130747e-06,
+      "loss": 0.8344,
+      "step": 28414
+    },
+    {
+      "epoch": 0.7803960341654994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4245808040997156e-06,
+      "loss": 0.8571,
+      "step": 28415
+    },
+    {
+      "epoch": 0.7804234983933427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4240001700083327e-06,
+      "loss": 0.8433,
+      "step": 28416
+    },
+    {
+      "epoch": 0.7804509626211859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4234195958611974e-06,
+      "loss": 0.9224,
+      "step": 28417
+    },
+    {
+      "epoch": 0.7804784268490291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.422839081662898e-06,
+      "loss": 0.8278,
+      "step": 28418
+    },
+    {
+      "epoch": 0.7805058910768724,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4222586274180315e-06,
+      "loss": 0.8395,
+      "step": 28419
+    },
+    {
+      "epoch": 0.7805333553047156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4216782331311908e-06,
+      "loss": 0.8235,
+      "step": 28420
+    },
+    {
+      "epoch": 0.7805608195325588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4210978988069655e-06,
+      "loss": 0.8712,
+      "step": 28421
+    },
+    {
+      "epoch": 0.7805882837604021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.420517624449954e-06,
+      "loss": 0.8024,
+      "step": 28422
+    },
+    {
+      "epoch": 0.7806157479882453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.41993741006474e-06,
+      "loss": 0.8149,
+      "step": 28423
+    },
+    {
+      "epoch": 0.7806432122160886,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4193572556559196e-06,
+      "loss": 0.7041,
+      "step": 28424
+    },
+    {
+      "epoch": 0.7806706764439317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4187771612280785e-06,
+      "loss": 0.8584,
+      "step": 28425
+    },
+    {
+      "epoch": 0.780698140671775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4181971267858085e-06,
+      "loss": 0.8357,
+      "step": 28426
+    },
+    {
+      "epoch": 0.7807256048996183,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4176171523337012e-06,
+      "loss": 0.7792,
+      "step": 28427
+    },
+    {
+      "epoch": 0.7807530691274615,
+      "grad_norm": 0.0,
+      "learning_rate": 2.417037237876343e-06,
+      "loss": 0.7843,
+      "step": 28428
+    },
+    {
+      "epoch": 0.7807805333553047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.416457383418327e-06,
+      "loss": 0.9233,
+      "step": 28429
+    },
+    {
+      "epoch": 0.7808079975831479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.415877588964235e-06,
+      "loss": 0.8276,
+      "step": 28430
+    },
+    {
+      "epoch": 0.7808354618109912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.415297854518659e-06,
+      "loss": 0.8402,
+      "step": 28431
+    },
+    {
+      "epoch": 0.7808629260388344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.414718180086183e-06,
+      "loss": 0.8982,
+      "step": 28432
+    },
+    {
+      "epoch": 0.7808903902666776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4141385656713947e-06,
+      "loss": 0.8229,
+      "step": 28433
+    },
+    {
+      "epoch": 0.7809178544945209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.41355901127888e-06,
+      "loss": 0.8315,
+      "step": 28434
+    },
+    {
+      "epoch": 0.7809453187223642,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4129795169132276e-06,
+      "loss": 0.8737,
+      "step": 28435
+    },
+    {
+      "epoch": 0.7809727829502073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4124000825790182e-06,
+      "loss": 0.8274,
+      "step": 28436
+    },
+    {
+      "epoch": 0.7810002471780506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4118207082808385e-06,
+      "loss": 0.8363,
+      "step": 28437
+    },
+    {
+      "epoch": 0.7810277114058938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4112413940232726e-06,
+      "loss": 0.8854,
+      "step": 28438
+    },
+    {
+      "epoch": 0.781055175633737,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4106621398109066e-06,
+      "loss": 0.8295,
+      "step": 28439
+    },
+    {
+      "epoch": 0.7810826398615803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4100829456483197e-06,
+      "loss": 0.8514,
+      "step": 28440
+    },
+    {
+      "epoch": 0.7811101040894235,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4095038115400995e-06,
+      "loss": 0.8612,
+      "step": 28441
+    },
+    {
+      "epoch": 0.7811375683172668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.408924737490822e-06,
+      "loss": 0.8191,
+      "step": 28442
+    },
+    {
+      "epoch": 0.78116503254511,
+      "grad_norm": 0.0,
+      "learning_rate": 2.408345723505072e-06,
+      "loss": 0.8048,
+      "step": 28443
+    },
+    {
+      "epoch": 0.7811924967729532,
+      "grad_norm": 0.0,
+      "learning_rate": 2.407766769587432e-06,
+      "loss": 0.8228,
+      "step": 28444
+    },
+    {
+      "epoch": 0.7812199610007965,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4071878757424826e-06,
+      "loss": 0.9135,
+      "step": 28445
+    },
+    {
+      "epoch": 0.7812474252286397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.406609041974806e-06,
+      "loss": 0.8278,
+      "step": 28446
+    },
+    {
+      "epoch": 0.7812748894564829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.406030268288978e-06,
+      "loss": 0.844,
+      "step": 28447
+    },
+    {
+      "epoch": 0.7813023536843262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4054515546895817e-06,
+      "loss": 0.8703,
+      "step": 28448
+    },
+    {
+      "epoch": 0.7813298179121694,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4048729011811934e-06,
+      "loss": 0.8159,
+      "step": 28449
+    },
+    {
+      "epoch": 0.7813572821400127,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4042943077683912e-06,
+      "loss": 0.8384,
+      "step": 28450
+    },
+    {
+      "epoch": 0.7813847463678558,
+      "grad_norm": 0.0,
+      "learning_rate": 2.403715774455756e-06,
+      "loss": 0.9329,
+      "step": 28451
+    },
+    {
+      "epoch": 0.7814122105956991,
+      "grad_norm": 0.0,
+      "learning_rate": 2.403137301247863e-06,
+      "loss": 0.8726,
+      "step": 28452
+    },
+    {
+      "epoch": 0.7814396748235424,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4025588881492947e-06,
+      "loss": 0.8157,
+      "step": 28453
+    },
+    {
+      "epoch": 0.7814671390513855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4019805351646186e-06,
+      "loss": 0.7903,
+      "step": 28454
+    },
+    {
+      "epoch": 0.7814946032792288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4014022422984206e-06,
+      "loss": 0.876,
+      "step": 28455
+    },
+    {
+      "epoch": 0.7815220675070721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.400824009555267e-06,
+      "loss": 0.9278,
+      "step": 28456
+    },
+    {
+      "epoch": 0.7815495317349153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4002458369397387e-06,
+      "loss": 0.8676,
+      "step": 28457
+    },
+    {
+      "epoch": 0.7815769959627585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3996677244564113e-06,
+      "loss": 0.7821,
+      "step": 28458
+    },
+    {
+      "epoch": 0.7816044601906017,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3990896721098545e-06,
+      "loss": 0.8558,
+      "step": 28459
+    },
+    {
+      "epoch": 0.781631924418445,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3985116799046436e-06,
+      "loss": 0.7889,
+      "step": 28460
+    },
+    {
+      "epoch": 0.7816593886462883,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3979337478453534e-06,
+      "loss": 0.7594,
+      "step": 28461
+    },
+    {
+      "epoch": 0.7816868528741314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3973558759365565e-06,
+      "loss": 0.7889,
+      "step": 28462
+    },
+    {
+      "epoch": 0.7817143171019747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3967780641828275e-06,
+      "loss": 0.7555,
+      "step": 28463
+    },
+    {
+      "epoch": 0.7817417813298179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.396200312588737e-06,
+      "loss": 0.7028,
+      "step": 28464
+    },
+    {
+      "epoch": 0.7817692455576611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3956226211588506e-06,
+      "loss": 0.7335,
+      "step": 28465
+    },
+    {
+      "epoch": 0.7817967097855044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3950449898977435e-06,
+      "loss": 0.8105,
+      "step": 28466
+    },
+    {
+      "epoch": 0.7818241740133476,
+      "grad_norm": 0.0,
+      "learning_rate": 2.394467418809987e-06,
+      "loss": 0.8584,
+      "step": 28467
+    },
+    {
+      "epoch": 0.7818516382411909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.39388990790015e-06,
+      "loss": 0.8489,
+      "step": 28468
+    },
+    {
+      "epoch": 0.7818791024690341,
+      "grad_norm": 0.0,
+      "learning_rate": 2.393312457172803e-06,
+      "loss": 0.7168,
+      "step": 28469
+    },
+    {
+      "epoch": 0.7819065666968773,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3927350666325167e-06,
+      "loss": 0.8246,
+      "step": 28470
+    },
+    {
+      "epoch": 0.7819340309247206,
+      "grad_norm": 0.0,
+      "learning_rate": 2.392157736283858e-06,
+      "loss": 0.8597,
+      "step": 28471
+    },
+    {
+      "epoch": 0.7819614951525637,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3915804661313902e-06,
+      "loss": 0.8018,
+      "step": 28472
+    },
+    {
+      "epoch": 0.781988959380407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3910032561796858e-06,
+      "loss": 0.91,
+      "step": 28473
+    },
+    {
+      "epoch": 0.7820164236082503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.390426106433311e-06,
+      "loss": 0.9038,
+      "step": 28474
+    },
+    {
+      "epoch": 0.7820438878360935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3898490168968325e-06,
+      "loss": 0.8546,
+      "step": 28475
+    },
+    {
+      "epoch": 0.7820713520639367,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3892719875748193e-06,
+      "loss": 0.7916,
+      "step": 28476
+    },
+    {
+      "epoch": 0.7820988162917799,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3886950184718304e-06,
+      "loss": 0.8847,
+      "step": 28477
+    },
+    {
+      "epoch": 0.7821262805196232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3881181095924364e-06,
+      "loss": 0.7818,
+      "step": 28478
+    },
+    {
+      "epoch": 0.7821537447474665,
+      "grad_norm": 0.0,
+      "learning_rate": 2.387541260941203e-06,
+      "loss": 0.7738,
+      "step": 28479
+    },
+    {
+      "epoch": 0.7821812089753096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.386964472522688e-06,
+      "loss": 0.8835,
+      "step": 28480
+    },
+    {
+      "epoch": 0.7822086732031529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3863877443414607e-06,
+      "loss": 0.8979,
+      "step": 28481
+    },
+    {
+      "epoch": 0.7822361374309962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.385811076402085e-06,
+      "loss": 0.8234,
+      "step": 28482
+    },
+    {
+      "epoch": 0.7822636016588393,
+      "grad_norm": 0.0,
+      "learning_rate": 2.385234468709118e-06,
+      "loss": 0.8637,
+      "step": 28483
+    },
+    {
+      "epoch": 0.7822910658866826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.384657921267126e-06,
+      "loss": 0.866,
+      "step": 28484
+    },
+    {
+      "epoch": 0.7823185301145258,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3840814340806705e-06,
+      "loss": 0.8387,
+      "step": 28485
+    },
+    {
+      "epoch": 0.7823459943423691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.383505007154313e-06,
+      "loss": 0.8433,
+      "step": 28486
+    },
+    {
+      "epoch": 0.7823734585702123,
+      "grad_norm": 0.0,
+      "learning_rate": 2.382928640492617e-06,
+      "loss": 0.7946,
+      "step": 28487
+    },
+    {
+      "epoch": 0.7824009227980555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.38235233410014e-06,
+      "loss": 0.8423,
+      "step": 28488
+    },
+    {
+      "epoch": 0.7824283870258988,
+      "grad_norm": 0.0,
+      "learning_rate": 2.38177608798144e-06,
+      "loss": 0.872,
+      "step": 28489
+    },
+    {
+      "epoch": 0.782455851253742,
+      "grad_norm": 0.0,
+      "learning_rate": 2.381199902141077e-06,
+      "loss": 0.7439,
+      "step": 28490
+    },
+    {
+      "epoch": 0.7824833154815852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3806237765836125e-06,
+      "loss": 0.8558,
+      "step": 28491
+    },
+    {
+      "epoch": 0.7825107797094285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3800477113136042e-06,
+      "loss": 0.8098,
+      "step": 28492
+    },
+    {
+      "epoch": 0.7825382439372717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.379471706335611e-06,
+      "loss": 0.8447,
+      "step": 28493
+    },
+    {
+      "epoch": 0.782565708165115,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3788957616541906e-06,
+      "loss": 0.8165,
+      "step": 28494
+    },
+    {
+      "epoch": 0.7825931723929582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3783198772739e-06,
+      "loss": 0.7353,
+      "step": 28495
+    },
+    {
+      "epoch": 0.7826206366208014,
+      "grad_norm": 0.0,
+      "learning_rate": 2.377744053199292e-06,
+      "loss": 0.7948,
+      "step": 28496
+    },
+    {
+      "epoch": 0.7826481008486447,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3771682894349268e-06,
+      "loss": 0.7844,
+      "step": 28497
+    },
+    {
+      "epoch": 0.7826755650764878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3765925859853576e-06,
+      "loss": 0.8196,
+      "step": 28498
+    },
+    {
+      "epoch": 0.7827030293043311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3760169428551404e-06,
+      "loss": 0.759,
+      "step": 28499
+    },
+    {
+      "epoch": 0.7827304935321744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3754413600488347e-06,
+      "loss": 0.8235,
+      "step": 28500
+    },
+    {
+      "epoch": 0.7827579577600176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3748658375709866e-06,
+      "loss": 0.8956,
+      "step": 28501
+    },
+    {
+      "epoch": 0.7827854219878608,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3742903754261535e-06,
+      "loss": 0.8573,
+      "step": 28502
+    },
+    {
+      "epoch": 0.782812886215704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3737149736188925e-06,
+      "loss": 0.7919,
+      "step": 28503
+    },
+    {
+      "epoch": 0.7828403504435473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.373139632153749e-06,
+      "loss": 0.8042,
+      "step": 28504
+    },
+    {
+      "epoch": 0.7828678146713905,
+      "grad_norm": 0.0,
+      "learning_rate": 2.37256435103528e-06,
+      "loss": 0.8061,
+      "step": 28505
+    },
+    {
+      "epoch": 0.7828952788992337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3719891302680386e-06,
+      "loss": 0.8231,
+      "step": 28506
+    },
+    {
+      "epoch": 0.782922743127077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.371413969856572e-06,
+      "loss": 0.822,
+      "step": 28507
+    },
+    {
+      "epoch": 0.7829502073549203,
+      "grad_norm": 0.0,
+      "learning_rate": 2.370838869805433e-06,
+      "loss": 0.8306,
+      "step": 28508
+    },
+    {
+      "epoch": 0.7829776715827634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3702638301191715e-06,
+      "loss": 0.8084,
+      "step": 28509
+    },
+    {
+      "epoch": 0.7830051358106067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.369688850802341e-06,
+      "loss": 0.7934,
+      "step": 28510
+    },
+    {
+      "epoch": 0.7830326000384499,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3691139318594845e-06,
+      "loss": 0.8752,
+      "step": 28511
+    },
+    {
+      "epoch": 0.7830600642662932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3685390732951584e-06,
+      "loss": 0.788,
+      "step": 28512
+    },
+    {
+      "epoch": 0.7830875284941364,
+      "grad_norm": 0.0,
+      "learning_rate": 2.367964275113904e-06,
+      "loss": 0.8525,
+      "step": 28513
+    },
+    {
+      "epoch": 0.7831149927219796,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3673895373202725e-06,
+      "loss": 0.7877,
+      "step": 28514
+    },
+    {
+      "epoch": 0.7831424569498229,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3668148599188124e-06,
+      "loss": 0.7313,
+      "step": 28515
+    },
+    {
+      "epoch": 0.783169921177666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3662402429140696e-06,
+      "loss": 0.718,
+      "step": 28516
+    },
+    {
+      "epoch": 0.7831973854055093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3656656863105933e-06,
+      "loss": 0.8519,
+      "step": 28517
+    },
+    {
+      "epoch": 0.7832248496333526,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3650911901129248e-06,
+      "loss": 0.777,
+      "step": 28518
+    },
+    {
+      "epoch": 0.7832523138611958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3645167543256155e-06,
+      "loss": 0.8402,
+      "step": 28519
+    },
+    {
+      "epoch": 0.783279778089039,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3639423789532044e-06,
+      "loss": 0.8881,
+      "step": 28520
+    },
+    {
+      "epoch": 0.7833072423168823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3633680640002387e-06,
+      "loss": 0.8592,
+      "step": 28521
+    },
+    {
+      "epoch": 0.7833347065447255,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3627938094712632e-06,
+      "loss": 0.8306,
+      "step": 28522
+    },
+    {
+      "epoch": 0.7833621707725688,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3622196153708244e-06,
+      "loss": 0.7497,
+      "step": 28523
+    },
+    {
+      "epoch": 0.7833896350004119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3616454817034597e-06,
+      "loss": 0.7988,
+      "step": 28524
+    },
+    {
+      "epoch": 0.7834170992282552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3610714084737152e-06,
+      "loss": 0.7596,
+      "step": 28525
+    },
+    {
+      "epoch": 0.7834445634560985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.360497395686132e-06,
+      "loss": 0.796,
+      "step": 28526
+    },
+    {
+      "epoch": 0.7834720276839416,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3599234433452558e-06,
+      "loss": 0.8102,
+      "step": 28527
+    },
+    {
+      "epoch": 0.7834994919117849,
+      "grad_norm": 0.0,
+      "learning_rate": 2.359349551455622e-06,
+      "loss": 0.7709,
+      "step": 28528
+    },
+    {
+      "epoch": 0.7835269561396281,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3587757200217766e-06,
+      "loss": 0.9033,
+      "step": 28529
+    },
+    {
+      "epoch": 0.7835544203674714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3582019490482554e-06,
+      "loss": 0.7961,
+      "step": 28530
+    },
+    {
+      "epoch": 0.7835818845953146,
+      "grad_norm": 0.0,
+      "learning_rate": 2.357628238539601e-06,
+      "loss": 0.8757,
+      "step": 28531
+    },
+    {
+      "epoch": 0.7836093488231578,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3570545885003515e-06,
+      "loss": 0.8618,
+      "step": 28532
+    },
+    {
+      "epoch": 0.7836368130510011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.356480998935047e-06,
+      "loss": 0.9076,
+      "step": 28533
+    },
+    {
+      "epoch": 0.7836642772788444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3559074698482277e-06,
+      "loss": 0.7794,
+      "step": 28534
+    },
+    {
+      "epoch": 0.7836917415066875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.355334001244428e-06,
+      "loss": 0.8078,
+      "step": 28535
+    },
+    {
+      "epoch": 0.7837192057345308,
+      "grad_norm": 0.0,
+      "learning_rate": 2.354760593128189e-06,
+      "loss": 0.7828,
+      "step": 28536
+    },
+    {
+      "epoch": 0.783746669962374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3541872455040427e-06,
+      "loss": 0.7928,
+      "step": 28537
+    },
+    {
+      "epoch": 0.7837741341902172,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3536139583765293e-06,
+      "loss": 0.7611,
+      "step": 28538
+    },
+    {
+      "epoch": 0.7838015984180605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3530407317501834e-06,
+      "loss": 0.7894,
+      "step": 28539
+    },
+    {
+      "epoch": 0.7838290626459037,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3524675656295427e-06,
+      "loss": 0.8867,
+      "step": 28540
+    },
+    {
+      "epoch": 0.783856526873747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.351894460019143e-06,
+      "loss": 0.8324,
+      "step": 28541
+    },
+    {
+      "epoch": 0.7838839911015901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.351321414923513e-06,
+      "loss": 0.8965,
+      "step": 28542
+    },
+    {
+      "epoch": 0.7839114553294334,
+      "grad_norm": 0.0,
+      "learning_rate": 2.350748430347195e-06,
+      "loss": 0.804,
+      "step": 28543
+    },
+    {
+      "epoch": 0.7839389195572767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.350175506294715e-06,
+      "loss": 0.8694,
+      "step": 28544
+    },
+    {
+      "epoch": 0.7839663837851198,
+      "grad_norm": 0.0,
+      "learning_rate": 2.34960264277061e-06,
+      "loss": 0.8431,
+      "step": 28545
+    },
+    {
+      "epoch": 0.7839938480129631,
+      "grad_norm": 0.0,
+      "learning_rate": 2.349029839779412e-06,
+      "loss": 0.8134,
+      "step": 28546
+    },
+    {
+      "epoch": 0.7840213122408064,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3484570973256572e-06,
+      "loss": 0.6877,
+      "step": 28547
+    },
+    {
+      "epoch": 0.7840487764686496,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3478844154138704e-06,
+      "loss": 0.8832,
+      "step": 28548
+    },
+    {
+      "epoch": 0.7840762406964928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.347311794048587e-06,
+      "loss": 0.7237,
+      "step": 28549
+    },
+    {
+      "epoch": 0.784103704924336,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3467392332343353e-06,
+      "loss": 0.7219,
+      "step": 28550
+    },
+    {
+      "epoch": 0.7841311691521793,
+      "grad_norm": 0.0,
+      "learning_rate": 2.346166732975652e-06,
+      "loss": 0.842,
+      "step": 28551
+    },
+    {
+      "epoch": 0.7841586333800226,
+      "grad_norm": 0.0,
+      "learning_rate": 2.345594293277058e-06,
+      "loss": 0.8164,
+      "step": 28552
+    },
+    {
+      "epoch": 0.7841860976078657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3450219141430906e-06,
+      "loss": 0.7352,
+      "step": 28553
+    },
+    {
+      "epoch": 0.784213561835709,
+      "grad_norm": 0.0,
+      "learning_rate": 2.344449595578273e-06,
+      "loss": 0.8031,
+      "step": 28554
+    },
+    {
+      "epoch": 0.7842410260635522,
+      "grad_norm": 0.0,
+      "learning_rate": 2.343877337587135e-06,
+      "loss": 0.7951,
+      "step": 28555
+    },
+    {
+      "epoch": 0.7842684902913954,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3433051401742047e-06,
+      "loss": 0.846,
+      "step": 28556
+    },
+    {
+      "epoch": 0.7842959545192387,
+      "grad_norm": 0.0,
+      "learning_rate": 2.342733003344011e-06,
+      "loss": 0.8223,
+      "step": 28557
+    },
+    {
+      "epoch": 0.7843234187470819,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3421609271010816e-06,
+      "loss": 0.8664,
+      "step": 28558
+    },
+    {
+      "epoch": 0.7843508829749252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.341588911449939e-06,
+      "loss": 0.7512,
+      "step": 28559
+    },
+    {
+      "epoch": 0.7843783472027684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3410169563951144e-06,
+      "loss": 0.9051,
+      "step": 28560
+    },
+    {
+      "epoch": 0.7844058114306116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3404450619411277e-06,
+      "loss": 0.802,
+      "step": 28561
+    },
+    {
+      "epoch": 0.7844332756584549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3398732280925063e-06,
+      "loss": 0.7615,
+      "step": 28562
+    },
+    {
+      "epoch": 0.784460739886298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3393014548537764e-06,
+      "loss": 0.7238,
+      "step": 28563
+    },
+    {
+      "epoch": 0.7844882041141413,
+      "grad_norm": 0.0,
+      "learning_rate": 2.33872974222946e-06,
+      "loss": 0.7875,
+      "step": 28564
+    },
+    {
+      "epoch": 0.7845156683419846,
+      "grad_norm": 0.0,
+      "learning_rate": 2.338158090224084e-06,
+      "loss": 0.8309,
+      "step": 28565
+    },
+    {
+      "epoch": 0.7845431325698278,
+      "grad_norm": 0.0,
+      "learning_rate": 2.337586498842167e-06,
+      "loss": 0.7383,
+      "step": 28566
+    },
+    {
+      "epoch": 0.784570596797671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3370149680882357e-06,
+      "loss": 0.7862,
+      "step": 28567
+    },
+    {
+      "epoch": 0.7845980610255142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3364434979668084e-06,
+      "loss": 0.7886,
+      "step": 28568
+    },
+    {
+      "epoch": 0.7846255252533575,
+      "grad_norm": 0.0,
+      "learning_rate": 2.335872088482408e-06,
+      "loss": 0.8212,
+      "step": 28569
+    },
+    {
+      "epoch": 0.7846529894812008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3353007396395566e-06,
+      "loss": 0.8174,
+      "step": 28570
+    },
+    {
+      "epoch": 0.7846804537090439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3347294514427776e-06,
+      "loss": 0.7444,
+      "step": 28571
+    },
+    {
+      "epoch": 0.7847079179368872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3341582238965867e-06,
+      "loss": 0.8739,
+      "step": 28572
+    },
+    {
+      "epoch": 0.7847353821647305,
+      "grad_norm": 0.0,
+      "learning_rate": 2.333587057005504e-06,
+      "loss": 0.8069,
+      "step": 28573
+    },
+    {
+      "epoch": 0.7847628463925737,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3330159507740513e-06,
+      "loss": 0.8073,
+      "step": 28574
+    },
+    {
+      "epoch": 0.7847903106204169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3324449052067477e-06,
+      "loss": 0.7376,
+      "step": 28575
+    },
+    {
+      "epoch": 0.7848177748482601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3318739203081107e-06,
+      "loss": 0.7517,
+      "step": 28576
+    },
+    {
+      "epoch": 0.7848452390761034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.331302996082655e-06,
+      "loss": 0.7715,
+      "step": 28577
+    },
+    {
+      "epoch": 0.7848727033039467,
+      "grad_norm": 0.0,
+      "learning_rate": 2.330732132534901e-06,
+      "loss": 0.8656,
+      "step": 28578
+    },
+    {
+      "epoch": 0.7849001675317898,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3301613296693648e-06,
+      "loss": 0.8488,
+      "step": 28579
+    },
+    {
+      "epoch": 0.7849276317596331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3295905874905633e-06,
+      "loss": 0.7707,
+      "step": 28580
+    },
+    {
+      "epoch": 0.7849550959874763,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3290199060030117e-06,
+      "loss": 0.8452,
+      "step": 28581
+    },
+    {
+      "epoch": 0.7849825602153195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3284492852112297e-06,
+      "loss": 0.7681,
+      "step": 28582
+    },
+    {
+      "epoch": 0.7850100244431628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3278787251197256e-06,
+      "loss": 0.8768,
+      "step": 28583
+    },
+    {
+      "epoch": 0.785037488671006,
+      "grad_norm": 0.0,
+      "learning_rate": 2.327308225733019e-06,
+      "loss": 0.7739,
+      "step": 28584
+    },
+    {
+      "epoch": 0.7850649528988493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3267377870556196e-06,
+      "loss": 0.8104,
+      "step": 28585
+    },
+    {
+      "epoch": 0.7850924171266925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.326167409092043e-06,
+      "loss": 0.8687,
+      "step": 28586
+    },
+    {
+      "epoch": 0.7851198813545357,
+      "grad_norm": 0.0,
+      "learning_rate": 2.325597091846803e-06,
+      "loss": 0.7205,
+      "step": 28587
+    },
+    {
+      "epoch": 0.785147345582379,
+      "grad_norm": 0.0,
+      "learning_rate": 2.325026835324413e-06,
+      "loss": 0.8151,
+      "step": 28588
+    },
+    {
+      "epoch": 0.7851748098102221,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3244566395293823e-06,
+      "loss": 0.7764,
+      "step": 28589
+    },
+    {
+      "epoch": 0.7852022740380654,
+      "grad_norm": 0.0,
+      "learning_rate": 2.323886504466224e-06,
+      "loss": 0.7182,
+      "step": 28590
+    },
+    {
+      "epoch": 0.7852297382659087,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3233164301394508e-06,
+      "loss": 0.8342,
+      "step": 28591
+    },
+    {
+      "epoch": 0.7852572024937519,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3227464165535695e-06,
+      "loss": 0.7167,
+      "step": 28592
+    },
+    {
+      "epoch": 0.7852846667215951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.322176463713092e-06,
+      "loss": 0.8263,
+      "step": 28593
+    },
+    {
+      "epoch": 0.7853121309494383,
+      "grad_norm": 0.0,
+      "learning_rate": 2.321606571622532e-06,
+      "loss": 0.888,
+      "step": 28594
+    },
+    {
+      "epoch": 0.7853395951772816,
+      "grad_norm": 0.0,
+      "learning_rate": 2.321036740286392e-06,
+      "loss": 0.8524,
+      "step": 28595
+    },
+    {
+      "epoch": 0.7853670594051249,
+      "grad_norm": 0.0,
+      "learning_rate": 2.320466969709183e-06,
+      "loss": 0.8084,
+      "step": 28596
+    },
+    {
+      "epoch": 0.785394523632968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.319897259895415e-06,
+      "loss": 0.7078,
+      "step": 28597
+    },
+    {
+      "epoch": 0.7854219878608113,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3193276108495943e-06,
+      "loss": 0.8211,
+      "step": 28598
+    },
+    {
+      "epoch": 0.7854494520886546,
+      "grad_norm": 0.0,
+      "learning_rate": 2.318758022576233e-06,
+      "loss": 0.7258,
+      "step": 28599
+    },
+    {
+      "epoch": 0.7854769163164977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3181884950798316e-06,
+      "loss": 0.7753,
+      "step": 28600
+    },
+    {
+      "epoch": 0.785504380544341,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3176190283648968e-06,
+      "loss": 0.7229,
+      "step": 28601
+    },
+    {
+      "epoch": 0.7855318447721842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3170496224359363e-06,
+      "loss": 0.9344,
+      "step": 28602
+    },
+    {
+      "epoch": 0.7855593090000275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3164802772974548e-06,
+      "loss": 0.8586,
+      "step": 28603
+    },
+    {
+      "epoch": 0.7855867732278707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.315910992953958e-06,
+      "loss": 0.8587,
+      "step": 28604
+    },
+    {
+      "epoch": 0.7856142374557139,
+      "grad_norm": 0.0,
+      "learning_rate": 2.31534176940995e-06,
+      "loss": 0.8734,
+      "step": 28605
+    },
+    {
+      "epoch": 0.7856417016835572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.314772606669937e-06,
+      "loss": 0.8517,
+      "step": 28606
+    },
+    {
+      "epoch": 0.7856691659114003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.314203504738418e-06,
+      "loss": 0.7977,
+      "step": 28607
+    },
+    {
+      "epoch": 0.7856966301392436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3136344636199006e-06,
+      "loss": 0.875,
+      "step": 28608
+    },
+    {
+      "epoch": 0.7857240943670869,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3130654833188814e-06,
+      "loss": 0.826,
+      "step": 28609
+    },
+    {
+      "epoch": 0.7857515585949301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3124965638398656e-06,
+      "loss": 0.765,
+      "step": 28610
+    },
+    {
+      "epoch": 0.7857790228227733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3119277051873555e-06,
+      "loss": 0.7912,
+      "step": 28611
+    },
+    {
+      "epoch": 0.7858064870506166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3113589073658536e-06,
+      "loss": 0.8969,
+      "step": 28612
+    },
+    {
+      "epoch": 0.7858339512784598,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3107901703798563e-06,
+      "loss": 0.73,
+      "step": 28613
+    },
+    {
+      "epoch": 0.7858614155063031,
+      "grad_norm": 0.0,
+      "learning_rate": 2.310221494233865e-06,
+      "loss": 0.7098,
+      "step": 28614
+    },
+    {
+      "epoch": 0.7858888797341462,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3096528789323834e-06,
+      "loss": 0.8601,
+      "step": 28615
+    },
+    {
+      "epoch": 0.7859163439619895,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3090843244799042e-06,
+      "loss": 0.8127,
+      "step": 28616
+    },
+    {
+      "epoch": 0.7859438081898328,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3085158308809296e-06,
+      "loss": 0.7203,
+      "step": 28617
+    },
+    {
+      "epoch": 0.785971272417676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3079473981399592e-06,
+      "loss": 0.9435,
+      "step": 28618
+    },
+    {
+      "epoch": 0.7859987366455192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.307379026261486e-06,
+      "loss": 0.7235,
+      "step": 28619
+    },
+    {
+      "epoch": 0.7860262008733624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3068107152500095e-06,
+      "loss": 0.8635,
+      "step": 28620
+    },
+    {
+      "epoch": 0.7860536651012057,
+      "grad_norm": 0.0,
+      "learning_rate": 2.306242465110027e-06,
+      "loss": 0.8391,
+      "step": 28621
+    },
+    {
+      "epoch": 0.7860811293290489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3056742758460336e-06,
+      "loss": 0.8002,
+      "step": 28622
+    },
+    {
+      "epoch": 0.7861085935568921,
+      "grad_norm": 0.0,
+      "learning_rate": 2.30510614746253e-06,
+      "loss": 0.7784,
+      "step": 28623
+    },
+    {
+      "epoch": 0.7861360577847354,
+      "grad_norm": 0.0,
+      "learning_rate": 2.304538079964006e-06,
+      "loss": 0.8301,
+      "step": 28624
+    },
+    {
+      "epoch": 0.7861635220125787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3039700733549564e-06,
+      "loss": 0.774,
+      "step": 28625
+    },
+    {
+      "epoch": 0.7861909862404218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3034021276398754e-06,
+      "loss": 0.8565,
+      "step": 28626
+    },
+    {
+      "epoch": 0.7862184504682651,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3028342428232585e-06,
+      "loss": 0.7805,
+      "step": 28627
+    },
+    {
+      "epoch": 0.7862459146961083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.302266418909599e-06,
+      "loss": 0.7552,
+      "step": 28628
+    },
+    {
+      "epoch": 0.7862733789239515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3016986559033884e-06,
+      "loss": 0.7732,
+      "step": 28629
+    },
+    {
+      "epoch": 0.7863008431517948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.301130953809123e-06,
+      "loss": 0.8248,
+      "step": 28630
+    },
+    {
+      "epoch": 0.786328307379638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.300563312631293e-06,
+      "loss": 0.813,
+      "step": 28631
+    },
+    {
+      "epoch": 0.7863557716074813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.299995732374385e-06,
+      "loss": 0.7515,
+      "step": 28632
+    },
+    {
+      "epoch": 0.7863832358353244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2994282130428935e-06,
+      "loss": 0.8493,
+      "step": 28633
+    },
+    {
+      "epoch": 0.7864107000631677,
+      "grad_norm": 0.0,
+      "learning_rate": 2.298860754641309e-06,
+      "loss": 0.8493,
+      "step": 28634
+    },
+    {
+      "epoch": 0.786438164291011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2982933571741208e-06,
+      "loss": 0.7868,
+      "step": 28635
+    },
+    {
+      "epoch": 0.7864656285188542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.297726020645823e-06,
+      "loss": 0.7426,
+      "step": 28636
+    },
+    {
+      "epoch": 0.7864930927466974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2971587450608965e-06,
+      "loss": 0.7576,
+      "step": 28637
+    },
+    {
+      "epoch": 0.7865205569745407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.296591530423834e-06,
+      "loss": 0.8519,
+      "step": 28638
+    },
+    {
+      "epoch": 0.7865480212023839,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2960243767391265e-06,
+      "loss": 0.7559,
+      "step": 28639
+    },
+    {
+      "epoch": 0.7865754854302272,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2954572840112553e-06,
+      "loss": 0.7841,
+      "step": 28640
+    },
+    {
+      "epoch": 0.7866029496580703,
+      "grad_norm": 0.0,
+      "learning_rate": 2.294890252244714e-06,
+      "loss": 0.9486,
+      "step": 28641
+    },
+    {
+      "epoch": 0.7866304138859136,
+      "grad_norm": 0.0,
+      "learning_rate": 2.294323281443981e-06,
+      "loss": 0.8333,
+      "step": 28642
+    },
+    {
+      "epoch": 0.7866578781137569,
+      "grad_norm": 0.0,
+      "learning_rate": 2.293756371613549e-06,
+      "loss": 0.8163,
+      "step": 28643
+    },
+    {
+      "epoch": 0.7866853423416,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2931895227579003e-06,
+      "loss": 0.8047,
+      "step": 28644
+    },
+    {
+      "epoch": 0.7867128065694433,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2926227348815224e-06,
+      "loss": 0.715,
+      "step": 28645
+    },
+    {
+      "epoch": 0.7867402707972866,
+      "grad_norm": 0.0,
+      "learning_rate": 2.292056007988899e-06,
+      "loss": 0.7535,
+      "step": 28646
+    },
+    {
+      "epoch": 0.7867677350251298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.291489342084516e-06,
+      "loss": 0.742,
+      "step": 28647
+    },
+    {
+      "epoch": 0.786795199252973,
+      "grad_norm": 0.0,
+      "learning_rate": 2.290922737172857e-06,
+      "loss": 0.8318,
+      "step": 28648
+    },
+    {
+      "epoch": 0.7868226634808162,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2903561932583996e-06,
+      "loss": 0.7827,
+      "step": 28649
+    },
+    {
+      "epoch": 0.7868501277086595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.289789710345631e-06,
+      "loss": 0.8722,
+      "step": 28650
+    },
+    {
+      "epoch": 0.7868775919365028,
+      "grad_norm": 0.0,
+      "learning_rate": 2.289223288439032e-06,
+      "loss": 0.7585,
+      "step": 28651
+    },
+    {
+      "epoch": 0.7869050561643459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2886569275430857e-06,
+      "loss": 0.879,
+      "step": 28652
+    },
+    {
+      "epoch": 0.7869325203921892,
+      "grad_norm": 0.0,
+      "learning_rate": 2.288090627662276e-06,
+      "loss": 0.7418,
+      "step": 28653
+    },
+    {
+      "epoch": 0.7869599846200324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2875243888010777e-06,
+      "loss": 0.7765,
+      "step": 28654
+    },
+    {
+      "epoch": 0.7869874488478756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2869582109639765e-06,
+      "loss": 0.8038,
+      "step": 28655
+    },
+    {
+      "epoch": 0.7870149130757189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2863920941554472e-06,
+      "loss": 0.8947,
+      "step": 28656
+    },
+    {
+      "epoch": 0.7870423773035621,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2858260383799714e-06,
+      "loss": 0.7528,
+      "step": 28657
+    },
+    {
+      "epoch": 0.7870698415314054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.285260043642028e-06,
+      "loss": 0.764,
+      "step": 28658
+    },
+    {
+      "epoch": 0.7870973057592486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2846941099460986e-06,
+      "loss": 0.7738,
+      "step": 28659
+    },
+    {
+      "epoch": 0.7871247699870918,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2841282372966544e-06,
+      "loss": 0.7823,
+      "step": 28660
+    },
+    {
+      "epoch": 0.7871522342149351,
+      "grad_norm": 0.0,
+      "learning_rate": 2.283562425698177e-06,
+      "loss": 0.8373,
+      "step": 28661
+    },
+    {
+      "epoch": 0.7871796984427782,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2829966751551437e-06,
+      "loss": 0.9133,
+      "step": 28662
+    },
+    {
+      "epoch": 0.7872071626706215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2824309856720317e-06,
+      "loss": 0.7791,
+      "step": 28663
+    },
+    {
+      "epoch": 0.7872346268984648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.281865357253312e-06,
+      "loss": 0.7966,
+      "step": 28664
+    },
+    {
+      "epoch": 0.787262091126308,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2812997899034674e-06,
+      "loss": 0.816,
+      "step": 28665
+    },
+    {
+      "epoch": 0.7872895553541512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2807342836269653e-06,
+      "loss": 0.7012,
+      "step": 28666
+    },
+    {
+      "epoch": 0.7873170195819944,
+      "grad_norm": 0.0,
+      "learning_rate": 2.280168838428284e-06,
+      "loss": 0.8815,
+      "step": 28667
+    },
+    {
+      "epoch": 0.7873444838098377,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2796034543118972e-06,
+      "loss": 0.9106,
+      "step": 28668
+    },
+    {
+      "epoch": 0.787371948037681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.279038131282277e-06,
+      "loss": 0.8284,
+      "step": 28669
+    },
+    {
+      "epoch": 0.7873994122655241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.278472869343903e-06,
+      "loss": 0.8173,
+      "step": 28670
+    },
+    {
+      "epoch": 0.7874268764933674,
+      "grad_norm": 0.0,
+      "learning_rate": 2.277907668501239e-06,
+      "loss": 0.785,
+      "step": 28671
+    },
+    {
+      "epoch": 0.7874543407212107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2773425287587635e-06,
+      "loss": 0.8446,
+      "step": 28672
+    },
+    {
+      "epoch": 0.7874818049490538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.276777450120943e-06,
+      "loss": 0.8612,
+      "step": 28673
+    },
+    {
+      "epoch": 0.7875092691768971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2762124325922498e-06,
+      "loss": 0.8593,
+      "step": 28674
+    },
+    {
+      "epoch": 0.7875367334047403,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2756474761771562e-06,
+      "loss": 0.8323,
+      "step": 28675
+    },
+    {
+      "epoch": 0.7875641976325836,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2750825808801324e-06,
+      "loss": 0.8371,
+      "step": 28676
+    },
+    {
+      "epoch": 0.7875916618604268,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2745177467056502e-06,
+      "loss": 0.8313,
+      "step": 28677
+    },
+    {
+      "epoch": 0.78761912608827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2739529736581735e-06,
+      "loss": 0.8514,
+      "step": 28678
+    },
+    {
+      "epoch": 0.7876465903161133,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2733882617421764e-06,
+      "loss": 0.7683,
+      "step": 28679
+    },
+    {
+      "epoch": 0.7876740545439564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2728236109621214e-06,
+      "loss": 0.7344,
+      "step": 28680
+    },
+    {
+      "epoch": 0.7877015187717997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.272259021322479e-06,
+      "loss": 0.9094,
+      "step": 28681
+    },
+    {
+      "epoch": 0.787728982999643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2716944928277173e-06,
+      "loss": 0.8271,
+      "step": 28682
+    },
+    {
+      "epoch": 0.7877564472274862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2711300254823045e-06,
+      "loss": 0.808,
+      "step": 28683
+    },
+    {
+      "epoch": 0.7877839114553294,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2705656192907024e-06,
+      "loss": 0.8307,
+      "step": 28684
+    },
+    {
+      "epoch": 0.7878113756831727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2700012742573795e-06,
+      "loss": 0.7919,
+      "step": 28685
+    },
+    {
+      "epoch": 0.7878388399110159,
+      "grad_norm": 0.0,
+      "learning_rate": 2.269436990386801e-06,
+      "loss": 0.6399,
+      "step": 28686
+    },
+    {
+      "epoch": 0.7878663041388592,
+      "grad_norm": 0.0,
+      "learning_rate": 2.268872767683433e-06,
+      "loss": 0.8547,
+      "step": 28687
+    },
+    {
+      "epoch": 0.7878937683667023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2683086061517367e-06,
+      "loss": 0.8215,
+      "step": 28688
+    },
+    {
+      "epoch": 0.7879212325945456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.26774450579618e-06,
+      "loss": 0.7193,
+      "step": 28689
+    },
+    {
+      "epoch": 0.7879486968223889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2671804666212217e-06,
+      "loss": 0.8488,
+      "step": 28690
+    },
+    {
+      "epoch": 0.787976161050232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.266616488631327e-06,
+      "loss": 0.8514,
+      "step": 28691
+    },
+    {
+      "epoch": 0.7880036252780753,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2660525718309577e-06,
+      "loss": 0.8257,
+      "step": 28692
+    },
+    {
+      "epoch": 0.7880310895059185,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2654887162245763e-06,
+      "loss": 0.8445,
+      "step": 28693
+    },
+    {
+      "epoch": 0.7880585537337618,
+      "grad_norm": 0.0,
+      "learning_rate": 2.264924921816648e-06,
+      "loss": 0.8235,
+      "step": 28694
+    },
+    {
+      "epoch": 0.788086017961605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.264361188611627e-06,
+      "loss": 0.8088,
+      "step": 28695
+    },
+    {
+      "epoch": 0.7881134821894482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.263797516613978e-06,
+      "loss": 0.8797,
+      "step": 28696
+    },
+    {
+      "epoch": 0.7881409464172915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2632339058281584e-06,
+      "loss": 0.7559,
+      "step": 28697
+    },
+    {
+      "epoch": 0.7881684106451348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2626703562586284e-06,
+      "loss": 0.739,
+      "step": 28698
+    },
+    {
+      "epoch": 0.7881958748729779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.262106867909848e-06,
+      "loss": 0.876,
+      "step": 28699
+    },
+    {
+      "epoch": 0.7882233391008212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.261543440786279e-06,
+      "loss": 0.6914,
+      "step": 28700
+    },
+    {
+      "epoch": 0.7882508033286644,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2609800748923716e-06,
+      "loss": 0.9662,
+      "step": 28701
+    },
+    {
+      "epoch": 0.7882782675565077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2604167702325885e-06,
+      "loss": 0.6567,
+      "step": 28702
+    },
+    {
+      "epoch": 0.7883057317843509,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2598535268113885e-06,
+      "loss": 0.7916,
+      "step": 28703
+    },
+    {
+      "epoch": 0.7883331960121941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.259290344633224e-06,
+      "loss": 0.8082,
+      "step": 28704
+    },
+    {
+      "epoch": 0.7883606602400374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2587272237025514e-06,
+      "loss": 0.7668,
+      "step": 28705
+    },
+    {
+      "epoch": 0.7883881244678805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2581641640238307e-06,
+      "loss": 0.8251,
+      "step": 28706
+    },
+    {
+      "epoch": 0.7884155886957238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2576011656015116e-06,
+      "loss": 0.7474,
+      "step": 28707
+    },
+    {
+      "epoch": 0.7884430529235671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2570382284400505e-06,
+      "loss": 0.8596,
+      "step": 28708
+    },
+    {
+      "epoch": 0.7884705171514103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.256475352543903e-06,
+      "loss": 0.8058,
+      "step": 28709
+    },
+    {
+      "epoch": 0.7884979813792535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2559125379175217e-06,
+      "loss": 0.8163,
+      "step": 28710
+    },
+    {
+      "epoch": 0.7885254456070968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2553497845653627e-06,
+      "loss": 0.8635,
+      "step": 28711
+    },
+    {
+      "epoch": 0.78855290983494,
+      "grad_norm": 0.0,
+      "learning_rate": 2.254787092491877e-06,
+      "loss": 0.8083,
+      "step": 28712
+    },
+    {
+      "epoch": 0.7885803740627833,
+      "grad_norm": 0.0,
+      "learning_rate": 2.254224461701513e-06,
+      "loss": 0.8652,
+      "step": 28713
+    },
+    {
+      "epoch": 0.7886078382906264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2536618921987253e-06,
+      "loss": 0.7395,
+      "step": 28714
+    },
+    {
+      "epoch": 0.7886353025184697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2530993839879657e-06,
+      "loss": 0.7966,
+      "step": 28715
+    },
+    {
+      "epoch": 0.788662766746313,
+      "grad_norm": 0.0,
+      "learning_rate": 2.252536937073684e-06,
+      "loss": 0.8428,
+      "step": 28716
+    },
+    {
+      "epoch": 0.7886902309741561,
+      "grad_norm": 0.0,
+      "learning_rate": 2.251974551460331e-06,
+      "loss": 0.7778,
+      "step": 28717
+    },
+    {
+      "epoch": 0.7887176952019994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2514122271523598e-06,
+      "loss": 0.6971,
+      "step": 28718
+    },
+    {
+      "epoch": 0.7887451594298426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2508499641542135e-06,
+      "loss": 0.7896,
+      "step": 28719
+    },
+    {
+      "epoch": 0.7887726236576859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.250287762470348e-06,
+      "loss": 0.8612,
+      "step": 28720
+    },
+    {
+      "epoch": 0.7888000878855291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2497256221052034e-06,
+      "loss": 0.8568,
+      "step": 28721
+    },
+    {
+      "epoch": 0.7888275521133723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.249163543063233e-06,
+      "loss": 0.9356,
+      "step": 28722
+    },
+    {
+      "epoch": 0.7888550163412156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2486015253488826e-06,
+      "loss": 0.8869,
+      "step": 28723
+    },
+    {
+      "epoch": 0.7888824805690589,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2480395689666025e-06,
+      "loss": 0.8729,
+      "step": 28724
+    },
+    {
+      "epoch": 0.788909944796902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2474776739208327e-06,
+      "loss": 0.791,
+      "step": 28725
+    },
+    {
+      "epoch": 0.7889374090247453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.246915840216023e-06,
+      "loss": 0.8772,
+      "step": 28726
+    },
+    {
+      "epoch": 0.7889648732525885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.246354067856621e-06,
+      "loss": 0.7351,
+      "step": 28727
+    },
+    {
+      "epoch": 0.7889923374804317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2457923568470664e-06,
+      "loss": 0.7814,
+      "step": 28728
+    },
+    {
+      "epoch": 0.789019801708275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2452307071918066e-06,
+      "loss": 0.707,
+      "step": 28729
+    },
+    {
+      "epoch": 0.7890472659361182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2446691188952883e-06,
+      "loss": 0.8313,
+      "step": 28730
+    },
+    {
+      "epoch": 0.7890747301639615,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2441075919619494e-06,
+      "loss": 0.8463,
+      "step": 28731
+    },
+    {
+      "epoch": 0.7891021943918046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2435461263962344e-06,
+      "loss": 0.7668,
+      "step": 28732
+    },
+    {
+      "epoch": 0.7891296586196479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.242984722202588e-06,
+      "loss": 0.7503,
+      "step": 28733
+    },
+    {
+      "epoch": 0.7891571228474912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.242423379385451e-06,
+      "loss": 0.8107,
+      "step": 28734
+    },
+    {
+      "epoch": 0.7891845870753343,
+      "grad_norm": 0.0,
+      "learning_rate": 2.241862097949268e-06,
+      "loss": 0.808,
+      "step": 28735
+    },
+    {
+      "epoch": 0.7892120513031776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.241300877898477e-06,
+      "loss": 0.7755,
+      "step": 28736
+    },
+    {
+      "epoch": 0.7892395155310209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2407397192375167e-06,
+      "loss": 0.7806,
+      "step": 28737
+    },
+    {
+      "epoch": 0.7892669797588641,
+      "grad_norm": 0.0,
+      "learning_rate": 2.240178621970829e-06,
+      "loss": 0.8604,
+      "step": 28738
+    },
+    {
+      "epoch": 0.7892944439867073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2396175861028545e-06,
+      "loss": 0.8974,
+      "step": 28739
+    },
+    {
+      "epoch": 0.7893219082145505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2390566116380307e-06,
+      "loss": 0.905,
+      "step": 28740
+    },
+    {
+      "epoch": 0.7893493724423938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2384956985807983e-06,
+      "loss": 0.762,
+      "step": 28741
+    },
+    {
+      "epoch": 0.7893768366702371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2379348469355968e-06,
+      "loss": 0.8374,
+      "step": 28742
+    },
+    {
+      "epoch": 0.7894043008980802,
+      "grad_norm": 0.0,
+      "learning_rate": 2.237374056706859e-06,
+      "loss": 0.7829,
+      "step": 28743
+    },
+    {
+      "epoch": 0.7894317651259235,
+      "grad_norm": 0.0,
+      "learning_rate": 2.236813327899027e-06,
+      "loss": 0.7892,
+      "step": 28744
+    },
+    {
+      "epoch": 0.7894592293537667,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2362526605165324e-06,
+      "loss": 0.8018,
+      "step": 28745
+    },
+    {
+      "epoch": 0.7894866935816099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2356920545638137e-06,
+      "loss": 0.815,
+      "step": 28746
+    },
+    {
+      "epoch": 0.7895141578094532,
+      "grad_norm": 0.0,
+      "learning_rate": 2.235131510045309e-06,
+      "loss": 0.8025,
+      "step": 28747
+    },
+    {
+      "epoch": 0.7895416220372964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.234571026965452e-06,
+      "loss": 0.8232,
+      "step": 28748
+    },
+    {
+      "epoch": 0.7895690862651397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.234010605328676e-06,
+      "loss": 0.8569,
+      "step": 28749
+    },
+    {
+      "epoch": 0.7895965504929829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2334502451394145e-06,
+      "loss": 0.7493,
+      "step": 28750
+    },
+    {
+      "epoch": 0.7896240147208261,
+      "grad_norm": 0.0,
+      "learning_rate": 2.232889946402107e-06,
+      "loss": 0.8012,
+      "step": 28751
+    },
+    {
+      "epoch": 0.7896514789486694,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2323297091211802e-06,
+      "loss": 0.7355,
+      "step": 28752
+    },
+    {
+      "epoch": 0.7896789431765125,
+      "grad_norm": 0.0,
+      "learning_rate": 2.23176953330107e-06,
+      "loss": 0.8923,
+      "step": 28753
+    },
+    {
+      "epoch": 0.7897064074043558,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2312094189462108e-06,
+      "loss": 0.8482,
+      "step": 28754
+    },
+    {
+      "epoch": 0.7897338716321991,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2306493660610296e-06,
+      "loss": 0.8782,
+      "step": 28755
+    },
+    {
+      "epoch": 0.7897613358600423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2300893746499595e-06,
+      "loss": 0.7933,
+      "step": 28756
+    },
+    {
+      "epoch": 0.7897888000878855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.229529444717432e-06,
+      "loss": 0.8642,
+      "step": 28757
+    },
+    {
+      "epoch": 0.7898162643157287,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2289695762678788e-06,
+      "loss": 0.77,
+      "step": 28758
+    },
+    {
+      "epoch": 0.789843728543572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2284097693057305e-06,
+      "loss": 0.9567,
+      "step": 28759
+    },
+    {
+      "epoch": 0.7898711927714153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2278500238354137e-06,
+      "loss": 0.8194,
+      "step": 28760
+    },
+    {
+      "epoch": 0.7898986569992584,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2272903398613567e-06,
+      "loss": 0.8266,
+      "step": 28761
+    },
+    {
+      "epoch": 0.7899261212271017,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2267307173879882e-06,
+      "loss": 0.8563,
+      "step": 28762
+    },
+    {
+      "epoch": 0.789953585454945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2261711564197386e-06,
+      "loss": 0.8751,
+      "step": 28763
+    },
+    {
+      "epoch": 0.7899810496827882,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2256116569610333e-06,
+      "loss": 0.8585,
+      "step": 28764
+    },
+    {
+      "epoch": 0.7900085139106314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2250522190163025e-06,
+      "loss": 0.7601,
+      "step": 28765
+    },
+    {
+      "epoch": 0.7900359781384746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2244928425899683e-06,
+      "loss": 0.7808,
+      "step": 28766
+    },
+    {
+      "epoch": 0.7900634423663179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.223933527686458e-06,
+      "loss": 0.8695,
+      "step": 28767
+    },
+    {
+      "epoch": 0.7900909065941611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2233742743102015e-06,
+      "loss": 0.8941,
+      "step": 28768
+    },
+    {
+      "epoch": 0.7901183708220043,
+      "grad_norm": 0.0,
+      "learning_rate": 2.222815082465617e-06,
+      "loss": 0.9246,
+      "step": 28769
+    },
+    {
+      "epoch": 0.7901458350498476,
+      "grad_norm": 0.0,
+      "learning_rate": 2.222255952157132e-06,
+      "loss": 0.745,
+      "step": 28770
+    },
+    {
+      "epoch": 0.7901732992776908,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2216968833891727e-06,
+      "loss": 0.8034,
+      "step": 28771
+    },
+    {
+      "epoch": 0.790200763505534,
+      "grad_norm": 0.0,
+      "learning_rate": 2.221137876166158e-06,
+      "loss": 0.7835,
+      "step": 28772
+    },
+    {
+      "epoch": 0.7902282277333773,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2205789304925153e-06,
+      "loss": 0.8763,
+      "step": 28773
+    },
+    {
+      "epoch": 0.7902556919612205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.220020046372664e-06,
+      "loss": 0.7648,
+      "step": 28774
+    },
+    {
+      "epoch": 0.7902831561890638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2194612238110312e-06,
+      "loss": 0.7819,
+      "step": 28775
+    },
+    {
+      "epoch": 0.790310620416907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2189024628120326e-06,
+      "loss": 0.7963,
+      "step": 28776
+    },
+    {
+      "epoch": 0.7903380846447502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2183437633800942e-06,
+      "loss": 0.8677,
+      "step": 28777
+    },
+    {
+      "epoch": 0.7903655488725935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2177851255196313e-06,
+      "loss": 0.9063,
+      "step": 28778
+    },
+    {
+      "epoch": 0.7903930131004366,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2172265492350675e-06,
+      "loss": 0.865,
+      "step": 28779
+    },
+    {
+      "epoch": 0.7904204773282799,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2166680345308213e-06,
+      "loss": 0.892,
+      "step": 28780
+    },
+    {
+      "epoch": 0.7904479415561232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.216109581411313e-06,
+      "loss": 0.8043,
+      "step": 28781
+    },
+    {
+      "epoch": 0.7904754057839664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.215551189880961e-06,
+      "loss": 0.7654,
+      "step": 28782
+    },
+    {
+      "epoch": 0.7905028700118096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.214992859944186e-06,
+      "loss": 0.7694,
+      "step": 28783
+    },
+    {
+      "epoch": 0.7905303342396528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2144345916054034e-06,
+      "loss": 0.7815,
+      "step": 28784
+    },
+    {
+      "epoch": 0.7905577984674961,
+      "grad_norm": 0.0,
+      "learning_rate": 2.213876384869027e-06,
+      "loss": 0.8329,
+      "step": 28785
+    },
+    {
+      "epoch": 0.7905852626953394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.213318239739477e-06,
+      "loss": 0.7984,
+      "step": 28786
+    },
+    {
+      "epoch": 0.7906127269231825,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2127601562211697e-06,
+      "loss": 0.8481,
+      "step": 28787
+    },
+    {
+      "epoch": 0.7906401911510258,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2122021343185196e-06,
+      "loss": 0.7739,
+      "step": 28788
+    },
+    {
+      "epoch": 0.7906676553788691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.211644174035946e-06,
+      "loss": 0.7054,
+      "step": 28789
+    },
+    {
+      "epoch": 0.7906951196067122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.211086275377858e-06,
+      "loss": 0.683,
+      "step": 28790
+    },
+    {
+      "epoch": 0.7907225838345555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.210528438348676e-06,
+      "loss": 0.8712,
+      "step": 28791
+    },
+    {
+      "epoch": 0.7907500480623987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2099706629528063e-06,
+      "loss": 0.8174,
+      "step": 28792
+    },
+    {
+      "epoch": 0.790777512290242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.209412949194667e-06,
+      "loss": 0.7784,
+      "step": 28793
+    },
+    {
+      "epoch": 0.7908049765180852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2088552970786704e-06,
+      "loss": 0.9177,
+      "step": 28794
+    },
+    {
+      "epoch": 0.7908324407459284,
+      "grad_norm": 0.0,
+      "learning_rate": 2.20829770660923e-06,
+      "loss": 0.7761,
+      "step": 28795
+    },
+    {
+      "epoch": 0.7908599049737717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.207740177790756e-06,
+      "loss": 0.8525,
+      "step": 28796
+    },
+    {
+      "epoch": 0.7908873692016148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.207182710627659e-06,
+      "loss": 0.7838,
+      "step": 28797
+    },
+    {
+      "epoch": 0.7909148334294581,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2066253051243503e-06,
+      "loss": 0.7161,
+      "step": 28798
+    },
+    {
+      "epoch": 0.7909422976573014,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2060679612852443e-06,
+      "loss": 0.8782,
+      "step": 28799
+    },
+    {
+      "epoch": 0.7909697618851446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.205510679114745e-06,
+      "loss": 0.8172,
+      "step": 28800
+    },
+    {
+      "epoch": 0.7909972261129878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2049534586172672e-06,
+      "loss": 0.7433,
+      "step": 28801
+    },
+    {
+      "epoch": 0.7910246903408311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2043962997972145e-06,
+      "loss": 0.8187,
+      "step": 28802
+    },
+    {
+      "epoch": 0.7910521545686743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2038392026589985e-06,
+      "loss": 0.6945,
+      "step": 28803
+    },
+    {
+      "epoch": 0.7910796187965176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.203282167207026e-06,
+      "loss": 0.8087,
+      "step": 28804
+    },
+    {
+      "epoch": 0.7911070830243607,
+      "grad_norm": 0.0,
+      "learning_rate": 2.202725193445705e-06,
+      "loss": 0.8393,
+      "step": 28805
+    },
+    {
+      "epoch": 0.791134547252204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.202168281379443e-06,
+      "loss": 0.7738,
+      "step": 28806
+    },
+    {
+      "epoch": 0.7911620114800473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.201611431012649e-06,
+      "loss": 0.7975,
+      "step": 28807
+    },
+    {
+      "epoch": 0.7911894757078904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.201054642349726e-06,
+      "loss": 0.7755,
+      "step": 28808
+    },
+    {
+      "epoch": 0.7912169399357337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2004979153950766e-06,
+      "loss": 0.7687,
+      "step": 28809
+    },
+    {
+      "epoch": 0.7912444041635769,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1999412501531103e-06,
+      "loss": 0.8771,
+      "step": 28810
+    },
+    {
+      "epoch": 0.7912718683914202,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1993846466282287e-06,
+      "loss": 0.8162,
+      "step": 28811
+    },
+    {
+      "epoch": 0.7912993326192634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1988281048248385e-06,
+      "loss": 0.7858,
+      "step": 28812
+    },
+    {
+      "epoch": 0.7913267968471066,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1982716247473445e-06,
+      "loss": 0.7748,
+      "step": 28813
+    },
+    {
+      "epoch": 0.7913542610749499,
+      "grad_norm": 0.0,
+      "learning_rate": 2.197715206400145e-06,
+      "loss": 0.6696,
+      "step": 28814
+    },
+    {
+      "epoch": 0.7913817253027932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.197158849787647e-06,
+      "loss": 0.7332,
+      "step": 28815
+    },
+    {
+      "epoch": 0.7914091895306363,
+      "grad_norm": 0.0,
+      "learning_rate": 2.196602554914249e-06,
+      "loss": 0.6839,
+      "step": 28816
+    },
+    {
+      "epoch": 0.7914366537584796,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1960463217843542e-06,
+      "loss": 0.847,
+      "step": 28817
+    },
+    {
+      "epoch": 0.7914641179863228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.195490150402366e-06,
+      "loss": 0.7584,
+      "step": 28818
+    },
+    {
+      "epoch": 0.791491582214166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1949340407726806e-06,
+      "loss": 0.8737,
+      "step": 28819
+    },
+    {
+      "epoch": 0.7915190464420093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1943779928997e-06,
+      "loss": 0.7974,
+      "step": 28820
+    },
+    {
+      "epoch": 0.7915465106698525,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1938220067878246e-06,
+      "loss": 0.8445,
+      "step": 28821
+    },
+    {
+      "epoch": 0.7915739748976958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.193266082441453e-06,
+      "loss": 0.8174,
+      "step": 28822
+    },
+    {
+      "epoch": 0.7916014391255389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1927102198649876e-06,
+      "loss": 0.8244,
+      "step": 28823
+    },
+    {
+      "epoch": 0.7916289033533822,
+      "grad_norm": 0.0,
+      "learning_rate": 2.192154419062823e-06,
+      "loss": 0.8184,
+      "step": 28824
+    },
+    {
+      "epoch": 0.7916563675812255,
+      "grad_norm": 0.0,
+      "learning_rate": 2.191598680039354e-06,
+      "loss": 0.8662,
+      "step": 28825
+    },
+    {
+      "epoch": 0.7916838318090686,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1910430027989814e-06,
+      "loss": 0.8673,
+      "step": 28826
+    },
+    {
+      "epoch": 0.7917112960369119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.190487387346102e-06,
+      "loss": 0.8839,
+      "step": 28827
+    },
+    {
+      "epoch": 0.7917387602647552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1899318336851105e-06,
+      "loss": 0.7428,
+      "step": 28828
+    },
+    {
+      "epoch": 0.7917662244925984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.189376341820404e-06,
+      "loss": 0.8056,
+      "step": 28829
+    },
+    {
+      "epoch": 0.7917936887204416,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1888209117563795e-06,
+      "loss": 0.8114,
+      "step": 28830
+    },
+    {
+      "epoch": 0.7918211529482848,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1882655434974276e-06,
+      "loss": 0.7441,
+      "step": 28831
+    },
+    {
+      "epoch": 0.7918486171761281,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1877102370479476e-06,
+      "loss": 0.7789,
+      "step": 28832
+    },
+    {
+      "epoch": 0.7918760814039714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1871549924123282e-06,
+      "loss": 0.8449,
+      "step": 28833
+    },
+    {
+      "epoch": 0.7919035456318145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1865998095949636e-06,
+      "loss": 0.7881,
+      "step": 28834
+    },
+    {
+      "epoch": 0.7919310098596578,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1860446886002484e-06,
+      "loss": 0.792,
+      "step": 28835
+    },
+    {
+      "epoch": 0.7919584740875011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1854896294325777e-06,
+      "loss": 0.8731,
+      "step": 28836
+    },
+    {
+      "epoch": 0.7919859383153443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.184934632096337e-06,
+      "loss": 0.7701,
+      "step": 28837
+    },
+    {
+      "epoch": 0.7920134025431875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.184379696595921e-06,
+      "loss": 0.8683,
+      "step": 28838
+    },
+    {
+      "epoch": 0.7920408667710307,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1838248229357217e-06,
+      "loss": 0.789,
+      "step": 28839
+    },
+    {
+      "epoch": 0.792068330998874,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1832700111201267e-06,
+      "loss": 0.8452,
+      "step": 28840
+    },
+    {
+      "epoch": 0.7920957952267172,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1827152611535273e-06,
+      "loss": 0.8076,
+      "step": 28841
+    },
+    {
+      "epoch": 0.7921232594545604,
+      "grad_norm": 0.0,
+      "learning_rate": 2.182160573040315e-06,
+      "loss": 0.751,
+      "step": 28842
+    },
+    {
+      "epoch": 0.7921507236824037,
+      "grad_norm": 0.0,
+      "learning_rate": 2.181605946784875e-06,
+      "loss": 0.8018,
+      "step": 28843
+    },
+    {
+      "epoch": 0.7921781879102469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.181051382391596e-06,
+      "loss": 0.8869,
+      "step": 28844
+    },
+    {
+      "epoch": 0.7922056521380901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.180496879864867e-06,
+      "loss": 0.8734,
+      "step": 28845
+    },
+    {
+      "epoch": 0.7922331163659334,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1799424392090774e-06,
+      "loss": 0.9113,
+      "step": 28846
+    },
+    {
+      "epoch": 0.7922605805937766,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1793880604286145e-06,
+      "loss": 0.8979,
+      "step": 28847
+    },
+    {
+      "epoch": 0.7922880448216199,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1788337435278615e-06,
+      "loss": 0.8625,
+      "step": 28848
+    },
+    {
+      "epoch": 0.7923155090494631,
+      "grad_norm": 0.0,
+      "learning_rate": 2.178279488511205e-06,
+      "loss": 0.7355,
+      "step": 28849
+    },
+    {
+      "epoch": 0.7923429732773063,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1777252953830286e-06,
+      "loss": 0.8759,
+      "step": 28850
+    },
+    {
+      "epoch": 0.7923704375051496,
+      "grad_norm": 0.0,
+      "learning_rate": 2.177171164147721e-06,
+      "loss": 0.6939,
+      "step": 28851
+    },
+    {
+      "epoch": 0.7923979017329927,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1766170948096653e-06,
+      "loss": 0.8511,
+      "step": 28852
+    },
+    {
+      "epoch": 0.792425365960836,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1760630873732457e-06,
+      "loss": 0.7894,
+      "step": 28853
+    },
+    {
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.0,
+      "learning_rate": 2.175509141842849e-06,
+      "loss": 0.8906,
+      "step": 28854
+    },
+    {
+      "epoch": 0.7924802944165225,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1749552582228506e-06,
+      "loss": 0.8813,
+      "step": 28855
+    },
+    {
+      "epoch": 0.7925077586443657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1744014365176413e-06,
+      "loss": 0.8148,
+      "step": 28856
+    },
+    {
+      "epoch": 0.7925352228722089,
+      "grad_norm": 0.0,
+      "learning_rate": 2.173847676731595e-06,
+      "loss": 0.8018,
+      "step": 28857
+    },
+    {
+      "epoch": 0.7925626871000522,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1732939788690986e-06,
+      "loss": 0.8985,
+      "step": 28858
+    },
+    {
+      "epoch": 0.7925901513278955,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1727403429345327e-06,
+      "loss": 0.7939,
+      "step": 28859
+    },
+    {
+      "epoch": 0.7926176155557386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1721867689322783e-06,
+      "loss": 0.8245,
+      "step": 28860
+    },
+    {
+      "epoch": 0.7926450797835819,
+      "grad_norm": 0.0,
+      "learning_rate": 2.171633256866713e-06,
+      "loss": 0.8272,
+      "step": 28861
+    },
+    {
+      "epoch": 0.7926725440114252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.171079806742217e-06,
+      "loss": 0.8487,
+      "step": 28862
+    },
+    {
+      "epoch": 0.7927000082392683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1705264185631737e-06,
+      "loss": 0.7252,
+      "step": 28863
+    },
+    {
+      "epoch": 0.7927274724671116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.169973092333955e-06,
+      "loss": 0.769,
+      "step": 28864
+    },
+    {
+      "epoch": 0.7927549366949548,
+      "grad_norm": 0.0,
+      "learning_rate": 2.169419828058942e-06,
+      "loss": 0.7389,
+      "step": 28865
+    },
+    {
+      "epoch": 0.7927824009227981,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1688666257425163e-06,
+      "loss": 0.8595,
+      "step": 28866
+    },
+    {
+      "epoch": 0.7928098651506413,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1683134853890475e-06,
+      "loss": 0.7427,
+      "step": 28867
+    },
+    {
+      "epoch": 0.7928373293784845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1677604070029166e-06,
+      "loss": 0.8649,
+      "step": 28868
+    },
+    {
+      "epoch": 0.7928647936063278,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1672073905884983e-06,
+      "loss": 0.8191,
+      "step": 28869
+    },
+    {
+      "epoch": 0.7928922578341709,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1666544361501707e-06,
+      "loss": 0.8398,
+      "step": 28870
+    },
+    {
+      "epoch": 0.7929197220620142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1661015436923093e-06,
+      "loss": 0.8449,
+      "step": 28871
+    },
+    {
+      "epoch": 0.7929471862898575,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1655487132192878e-06,
+      "loss": 0.8069,
+      "step": 28872
+    },
+    {
+      "epoch": 0.7929746505177007,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1649959447354763e-06,
+      "loss": 0.7952,
+      "step": 28873
+    },
+    {
+      "epoch": 0.7930021147455439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.164443238245253e-06,
+      "loss": 0.9033,
+      "step": 28874
+    },
+    {
+      "epoch": 0.7930295789733872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.163890593752989e-06,
+      "loss": 0.824,
+      "step": 28875
+    },
+    {
+      "epoch": 0.7930570432012304,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1633380112630575e-06,
+      "loss": 0.82,
+      "step": 28876
+    },
+    {
+      "epoch": 0.7930845074290737,
+      "grad_norm": 0.0,
+      "learning_rate": 2.162785490779833e-06,
+      "loss": 0.8965,
+      "step": 28877
+    },
+    {
+      "epoch": 0.7931119716569168,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1622330323076866e-06,
+      "loss": 0.7681,
+      "step": 28878
+    },
+    {
+      "epoch": 0.7931394358847601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1616806358509866e-06,
+      "loss": 0.8033,
+      "step": 28879
+    },
+    {
+      "epoch": 0.7931669001126034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.161128301414108e-06,
+      "loss": 0.7743,
+      "step": 28880
+    },
+    {
+      "epoch": 0.7931943643404465,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1605760290014165e-06,
+      "loss": 0.8151,
+      "step": 28881
+    },
+    {
+      "epoch": 0.7932218285682898,
+      "grad_norm": 0.0,
+      "learning_rate": 2.160023818617284e-06,
+      "loss": 0.7802,
+      "step": 28882
+    },
+    {
+      "epoch": 0.793249292796133,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1594716702660825e-06,
+      "loss": 0.7716,
+      "step": 28883
+    },
+    {
+      "epoch": 0.7932767570239763,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1589195839521758e-06,
+      "loss": 0.8711,
+      "step": 28884
+    },
+    {
+      "epoch": 0.7933042212518195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1583675596799346e-06,
+      "loss": 0.7468,
+      "step": 28885
+    },
+    {
+      "epoch": 0.7933316854796627,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1578155974537272e-06,
+      "loss": 0.7659,
+      "step": 28886
+    },
+    {
+      "epoch": 0.793359149707506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1572636972779225e-06,
+      "loss": 0.7944,
+      "step": 28887
+    },
+    {
+      "epoch": 0.7933866139353493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1567118591568836e-06,
+      "loss": 0.8446,
+      "step": 28888
+    },
+    {
+      "epoch": 0.7934140781631924,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1561600830949803e-06,
+      "loss": 0.8071,
+      "step": 28889
+    },
+    {
+      "epoch": 0.7934415423910357,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1556083690965757e-06,
+      "loss": 0.7606,
+      "step": 28890
+    },
+    {
+      "epoch": 0.7934690066188789,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1550567171660353e-06,
+      "loss": 0.8494,
+      "step": 28891
+    },
+    {
+      "epoch": 0.7934964708467221,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1545051273077245e-06,
+      "loss": 0.8053,
+      "step": 28892
+    },
+    {
+      "epoch": 0.7935239350745654,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1539535995260087e-06,
+      "loss": 0.8049,
+      "step": 28893
+    },
+    {
+      "epoch": 0.7935513993024086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.153402133825251e-06,
+      "loss": 0.7817,
+      "step": 28894
+    },
+    {
+      "epoch": 0.7935788635302519,
+      "grad_norm": 0.0,
+      "learning_rate": 2.152850730209819e-06,
+      "loss": 0.8335,
+      "step": 28895
+    },
+    {
+      "epoch": 0.793606327758095,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1522993886840706e-06,
+      "loss": 0.864,
+      "step": 28896
+    },
+    {
+      "epoch": 0.7936337919859383,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1517481092523663e-06,
+      "loss": 0.819,
+      "step": 28897
+    },
+    {
+      "epoch": 0.7936612562137816,
+      "grad_norm": 0.0,
+      "learning_rate": 2.151196891919072e-06,
+      "loss": 0.8092,
+      "step": 28898
+    },
+    {
+      "epoch": 0.7936887204416248,
+      "grad_norm": 0.0,
+      "learning_rate": 2.150645736688548e-06,
+      "loss": 0.8447,
+      "step": 28899
+    },
+    {
+      "epoch": 0.793716184669468,
+      "grad_norm": 0.0,
+      "learning_rate": 2.150094643565155e-06,
+      "loss": 0.797,
+      "step": 28900
+    },
+    {
+      "epoch": 0.7937436488973113,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1495436125532575e-06,
+      "loss": 0.8217,
+      "step": 28901
+    },
+    {
+      "epoch": 0.7937711131251545,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1489926436572084e-06,
+      "loss": 0.8238,
+      "step": 28902
+    },
+    {
+      "epoch": 0.7937985773529977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1484417368813715e-06,
+      "loss": 0.8446,
+      "step": 28903
+    },
+    {
+      "epoch": 0.7938260415808409,
+      "grad_norm": 0.0,
+      "learning_rate": 2.147890892230107e-06,
+      "loss": 0.8297,
+      "step": 28904
+    },
+    {
+      "epoch": 0.7938535058086842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1473401097077683e-06,
+      "loss": 0.9061,
+      "step": 28905
+    },
+    {
+      "epoch": 0.7938809700365275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1467893893187176e-06,
+      "loss": 0.8585,
+      "step": 28906
+    },
+    {
+      "epoch": 0.7939084342643706,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1462387310673127e-06,
+      "loss": 0.8044,
+      "step": 28907
+    },
+    {
+      "epoch": 0.7939358984922139,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1456881349579074e-06,
+      "loss": 0.7729,
+      "step": 28908
+    },
+    {
+      "epoch": 0.7939633627200571,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1451376009948597e-06,
+      "loss": 0.8996,
+      "step": 28909
+    },
+    {
+      "epoch": 0.7939908269479004,
+      "grad_norm": 0.0,
+      "learning_rate": 2.144587129182527e-06,
+      "loss": 0.9028,
+      "step": 28910
+    },
+    {
+      "epoch": 0.7940182911757436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1440367195252655e-06,
+      "loss": 0.8493,
+      "step": 28911
+    },
+    {
+      "epoch": 0.7940457554035868,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1434863720274257e-06,
+      "loss": 0.8525,
+      "step": 28912
+    },
+    {
+      "epoch": 0.7940732196314301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.142936086693368e-06,
+      "loss": 0.7166,
+      "step": 28913
+    },
+    {
+      "epoch": 0.7941006838592733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.142385863527442e-06,
+      "loss": 0.868,
+      "step": 28914
+    },
+    {
+      "epoch": 0.7941281480871165,
+      "grad_norm": 0.0,
+      "learning_rate": 2.141835702534001e-06,
+      "loss": 0.8381,
+      "step": 28915
+    },
+    {
+      "epoch": 0.7941556123149598,
+      "grad_norm": 0.0,
+      "learning_rate": 2.141285603717401e-06,
+      "loss": 0.8004,
+      "step": 28916
+    },
+    {
+      "epoch": 0.794183076542803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.140735567081993e-06,
+      "loss": 0.8596,
+      "step": 28917
+    },
+    {
+      "epoch": 0.7942105407706462,
+      "grad_norm": 0.0,
+      "learning_rate": 2.140185592632129e-06,
+      "loss": 0.7984,
+      "step": 28918
+    },
+    {
+      "epoch": 0.7942380049984895,
+      "grad_norm": 0.0,
+      "learning_rate": 2.139635680372165e-06,
+      "loss": 0.8843,
+      "step": 28919
+    },
+    {
+      "epoch": 0.7942654692263327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1390858303064477e-06,
+      "loss": 0.7653,
+      "step": 28920
+    },
+    {
+      "epoch": 0.794292933454176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1385360424393264e-06,
+      "loss": 0.7977,
+      "step": 28921
+    },
+    {
+      "epoch": 0.7943203976820191,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1379863167751525e-06,
+      "loss": 0.878,
+      "step": 28922
+    },
+    {
+      "epoch": 0.7943478619098624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.137436653318277e-06,
+      "loss": 0.8333,
+      "step": 28923
+    },
+    {
+      "epoch": 0.7943753261377057,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1368870520730467e-06,
+      "loss": 0.874,
+      "step": 28924
+    },
+    {
+      "epoch": 0.7944027903655488,
+      "grad_norm": 0.0,
+      "learning_rate": 2.136337513043816e-06,
+      "loss": 0.9091,
+      "step": 28925
+    },
+    {
+      "epoch": 0.7944302545933921,
+      "grad_norm": 0.0,
+      "learning_rate": 2.135788036234926e-06,
+      "loss": 0.7398,
+      "step": 28926
+    },
+    {
+      "epoch": 0.7944577188212354,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1352386216507302e-06,
+      "loss": 0.8301,
+      "step": 28927
+    },
+    {
+      "epoch": 0.7944851830490786,
+      "grad_norm": 0.0,
+      "learning_rate": 2.13468926929557e-06,
+      "loss": 0.7373,
+      "step": 28928
+    },
+    {
+      "epoch": 0.7945126472769218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.134139979173795e-06,
+      "loss": 0.7984,
+      "step": 28929
+    },
+    {
+      "epoch": 0.794540111504765,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1335907512897524e-06,
+      "loss": 0.7459,
+      "step": 28930
+    },
+    {
+      "epoch": 0.7945675757326083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.133041585647788e-06,
+      "loss": 0.8091,
+      "step": 28931
+    },
+    {
+      "epoch": 0.7945950399604516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1324924822522443e-06,
+      "loss": 0.9167,
+      "step": 28932
+    },
+    {
+      "epoch": 0.7946225041882947,
+      "grad_norm": 0.0,
+      "learning_rate": 2.131943441107468e-06,
+      "loss": 0.8655,
+      "step": 28933
+    },
+    {
+      "epoch": 0.794649968416138,
+      "grad_norm": 0.0,
+      "learning_rate": 2.131394462217802e-06,
+      "loss": 0.83,
+      "step": 28934
+    },
+    {
+      "epoch": 0.7946774326439812,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1308455455875943e-06,
+      "loss": 0.8602,
+      "step": 28935
+    },
+    {
+      "epoch": 0.7947048968718244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.130296691221182e-06,
+      "loss": 0.7478,
+      "step": 28936
+    },
+    {
+      "epoch": 0.7947323610996677,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1297478991229137e-06,
+      "loss": 0.7002,
+      "step": 28937
+    },
+    {
+      "epoch": 0.7947598253275109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.129199169297125e-06,
+      "loss": 0.831,
+      "step": 28938
+    },
+    {
+      "epoch": 0.7947872895553542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1286505017481617e-06,
+      "loss": 0.8601,
+      "step": 28939
+    },
+    {
+      "epoch": 0.7948147537831974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1281018964803656e-06,
+      "loss": 0.8449,
+      "step": 28940
+    },
+    {
+      "epoch": 0.7948422180110406,
+      "grad_norm": 0.0,
+      "learning_rate": 2.127553353498075e-06,
+      "loss": 0.791,
+      "step": 28941
+    },
+    {
+      "epoch": 0.7948696822388839,
+      "grad_norm": 0.0,
+      "learning_rate": 2.127004872805636e-06,
+      "loss": 0.7807,
+      "step": 28942
+    },
+    {
+      "epoch": 0.794897146466727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1264564544073806e-06,
+      "loss": 0.8041,
+      "step": 28943
+    },
+    {
+      "epoch": 0.7949246106945703,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1259080983076553e-06,
+      "loss": 0.8595,
+      "step": 28944
+    },
+    {
+      "epoch": 0.7949520749224136,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1253598045107917e-06,
+      "loss": 0.7883,
+      "step": 28945
+    },
+    {
+      "epoch": 0.7949795391502568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1248115730211317e-06,
+      "loss": 0.8252,
+      "step": 28946
+    },
+    {
+      "epoch": 0.7950070033781,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1242634038430134e-06,
+      "loss": 0.7879,
+      "step": 28947
+    },
+    {
+      "epoch": 0.7950344676059432,
+      "grad_norm": 0.0,
+      "learning_rate": 2.123715296980776e-06,
+      "loss": 0.6795,
+      "step": 28948
+    },
+    {
+      "epoch": 0.7950619318337865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1231672524387525e-06,
+      "loss": 0.9348,
+      "step": 28949
+    },
+    {
+      "epoch": 0.7950893960616298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.12261927022128e-06,
+      "loss": 0.8093,
+      "step": 28950
+    },
+    {
+      "epoch": 0.7951168602894729,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1220713503326985e-06,
+      "loss": 0.9398,
+      "step": 28951
+    },
+    {
+      "epoch": 0.7951443245173162,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1215234927773366e-06,
+      "loss": 0.835,
+      "step": 28952
+    },
+    {
+      "epoch": 0.7951717887451595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.120975697559533e-06,
+      "loss": 0.7754,
+      "step": 28953
+    },
+    {
+      "epoch": 0.7951992529730026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1204279646836245e-06,
+      "loss": 0.7154,
+      "step": 28954
+    },
+    {
+      "epoch": 0.7952267172008459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.11988029415394e-06,
+      "loss": 0.8596,
+      "step": 28955
+    },
+    {
+      "epoch": 0.7952541814286891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1193326859748143e-06,
+      "loss": 0.8073,
+      "step": 28956
+    },
+    {
+      "epoch": 0.7952816456565324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.118785140150581e-06,
+      "loss": 0.8062,
+      "step": 28957
+    },
+    {
+      "epoch": 0.7953091098843756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1182376566855733e-06,
+      "loss": 0.7559,
+      "step": 28958
+    },
+    {
+      "epoch": 0.7953365741122188,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1176902355841244e-06,
+      "loss": 0.7737,
+      "step": 28959
+    },
+    {
+      "epoch": 0.7953640383400621,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1171428768505644e-06,
+      "loss": 0.7984,
+      "step": 28960
+    },
+    {
+      "epoch": 0.7953915025679053,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1165955804892202e-06,
+      "loss": 0.8703,
+      "step": 28961
+    },
+    {
+      "epoch": 0.7954189667957485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.116048346504426e-06,
+      "loss": 0.8475,
+      "step": 28962
+    },
+    {
+      "epoch": 0.7954464310235918,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1155011749005105e-06,
+      "loss": 0.8897,
+      "step": 28963
+    },
+    {
+      "epoch": 0.795473895251435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1149540656818037e-06,
+      "loss": 0.9277,
+      "step": 28964
+    },
+    {
+      "epoch": 0.7955013594792782,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1144070188526355e-06,
+      "loss": 0.8304,
+      "step": 28965
+    },
+    {
+      "epoch": 0.7955288237071215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1138600344173364e-06,
+      "loss": 0.8235,
+      "step": 28966
+    },
+    {
+      "epoch": 0.7955562879349647,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1133131123802287e-06,
+      "loss": 0.7635,
+      "step": 28967
+    },
+    {
+      "epoch": 0.795583752162808,
+      "grad_norm": 0.0,
+      "learning_rate": 2.112766252745645e-06,
+      "loss": 0.8497,
+      "step": 28968
+    },
+    {
+      "epoch": 0.7956112163906511,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1122194555179067e-06,
+      "loss": 0.7133,
+      "step": 28969
+    },
+    {
+      "epoch": 0.7956386806184944,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1116727207013443e-06,
+      "loss": 0.8309,
+      "step": 28970
+    },
+    {
+      "epoch": 0.7956661448463377,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1111260483002825e-06,
+      "loss": 0.834,
+      "step": 28971
+    },
+    {
+      "epoch": 0.7956936090741809,
+      "grad_norm": 0.0,
+      "learning_rate": 2.11057943831905e-06,
+      "loss": 0.7645,
+      "step": 28972
+    },
+    {
+      "epoch": 0.7957210733020241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1100328907619673e-06,
+      "loss": 0.8476,
+      "step": 28973
+    },
+    {
+      "epoch": 0.7957485375298673,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1094864056333596e-06,
+      "loss": 0.8737,
+      "step": 28974
+    },
+    {
+      "epoch": 0.7957760017577106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1089399829375556e-06,
+      "loss": 0.7407,
+      "step": 28975
+    },
+    {
+      "epoch": 0.7958034659855538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1083936226788714e-06,
+      "loss": 0.794,
+      "step": 28976
+    },
+    {
+      "epoch": 0.795830930213397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1078473248616338e-06,
+      "loss": 0.8535,
+      "step": 28977
+    },
+    {
+      "epoch": 0.7958583944412403,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1073010894901692e-06,
+      "loss": 0.7689,
+      "step": 28978
+    },
+    {
+      "epoch": 0.7958858586690836,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1067549165687918e-06,
+      "loss": 0.8574,
+      "step": 28979
+    },
+    {
+      "epoch": 0.7959133228969267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1062088061018284e-06,
+      "loss": 0.8699,
+      "step": 28980
+    },
+    {
+      "epoch": 0.79594078712477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1056627580935974e-06,
+      "loss": 0.7962,
+      "step": 28981
+    },
+    {
+      "epoch": 0.7959682513526132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.105116772548421e-06,
+      "loss": 0.853,
+      "step": 28982
+    },
+    {
+      "epoch": 0.7959957155804565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.104570849470622e-06,
+      "loss": 0.7766,
+      "step": 28983
+    },
+    {
+      "epoch": 0.7960231798082997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.104024988864516e-06,
+      "loss": 0.8054,
+      "step": 28984
+    },
+    {
+      "epoch": 0.7960506440361429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.10347919073442e-06,
+      "loss": 0.8931,
+      "step": 28985
+    },
+    {
+      "epoch": 0.7960781082639862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.102933455084657e-06,
+      "loss": 0.7827,
+      "step": 28986
+    },
+    {
+      "epoch": 0.7961055724918293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.102387781919543e-06,
+      "loss": 0.8749,
+      "step": 28987
+    },
+    {
+      "epoch": 0.7961330367196726,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1018421712433946e-06,
+      "loss": 0.7297,
+      "step": 28988
+    },
+    {
+      "epoch": 0.7961605009475159,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1012966230605313e-06,
+      "loss": 0.8391,
+      "step": 28989
+    },
+    {
+      "epoch": 0.7961879651753591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.100751137375272e-06,
+      "loss": 0.8572,
+      "step": 28990
+    },
+    {
+      "epoch": 0.7962154294032023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1002057141919262e-06,
+      "loss": 0.8184,
+      "step": 28991
+    },
+    {
+      "epoch": 0.7962428936310456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0996603535148165e-06,
+      "loss": 0.8033,
+      "step": 28992
+    },
+    {
+      "epoch": 0.7962703578588888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.099115055348251e-06,
+      "loss": 0.7166,
+      "step": 28993
+    },
+    {
+      "epoch": 0.796297822086732,
+      "grad_norm": 0.0,
+      "learning_rate": 2.098569819696548e-06,
+      "loss": 0.7738,
+      "step": 28994
+    },
+    {
+      "epoch": 0.7963252863145752,
+      "grad_norm": 0.0,
+      "learning_rate": 2.098024646564021e-06,
+      "loss": 0.6955,
+      "step": 28995
+    },
+    {
+      "epoch": 0.7963527505424185,
+      "grad_norm": 0.0,
+      "learning_rate": 2.097479535954987e-06,
+      "loss": 0.8601,
+      "step": 28996
+    },
+    {
+      "epoch": 0.7963802147702618,
+      "grad_norm": 0.0,
+      "learning_rate": 2.096934487873753e-06,
+      "loss": 0.768,
+      "step": 28997
+    },
+    {
+      "epoch": 0.7964076789981049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.096389502324635e-06,
+      "loss": 0.8884,
+      "step": 28998
+    },
+    {
+      "epoch": 0.7964351432259482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0958445793119474e-06,
+      "loss": 0.8463,
+      "step": 28999
+    },
+    {
+      "epoch": 0.7964626074537914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.095299718839997e-06,
+      "loss": 0.8724,
+      "step": 29000
+    },
+    {
+      "epoch": 0.7964900716816347,
+      "grad_norm": 0.0,
+      "learning_rate": 2.094754920913099e-06,
+      "loss": 0.911,
+      "step": 29001
+    },
+    {
+      "epoch": 0.7965175359094779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.09421018553556e-06,
+      "loss": 0.7873,
+      "step": 29002
+    },
+    {
+      "epoch": 0.7965450001373211,
+      "grad_norm": 0.0,
+      "learning_rate": 2.093665512711692e-06,
+      "loss": 0.7901,
+      "step": 29003
+    },
+    {
+      "epoch": 0.7965724643651644,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0931209024458054e-06,
+      "loss": 0.7274,
+      "step": 29004
+    },
+    {
+      "epoch": 0.7965999285930077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0925763547422097e-06,
+      "loss": 0.7933,
+      "step": 29005
+    },
+    {
+      "epoch": 0.7966273928208508,
+      "grad_norm": 0.0,
+      "learning_rate": 2.092031869605212e-06,
+      "loss": 0.8905,
+      "step": 29006
+    },
+    {
+      "epoch": 0.7966548570486941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0914874470391246e-06,
+      "loss": 0.8046,
+      "step": 29007
+    },
+    {
+      "epoch": 0.7966823212765373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0909430870482515e-06,
+      "loss": 0.7729,
+      "step": 29008
+    },
+    {
+      "epoch": 0.7967097855043805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.090398789636898e-06,
+      "loss": 0.8532,
+      "step": 29009
+    },
+    {
+      "epoch": 0.7967372497322238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.089854554809374e-06,
+      "loss": 0.8116,
+      "step": 29010
+    },
+    {
+      "epoch": 0.796764713960067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0893103825699835e-06,
+      "loss": 0.797,
+      "step": 29011
+    },
+    {
+      "epoch": 0.7967921781879103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.088766272923034e-06,
+      "loss": 0.7615,
+      "step": 29012
+    },
+    {
+      "epoch": 0.7968196424157534,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0882222258728334e-06,
+      "loss": 0.8509,
+      "step": 29013
+    },
+    {
+      "epoch": 0.7968471066435967,
+      "grad_norm": 0.0,
+      "learning_rate": 2.087678241423681e-06,
+      "loss": 0.8361,
+      "step": 29014
+    },
+    {
+      "epoch": 0.79687457087144,
+      "grad_norm": 0.0,
+      "learning_rate": 2.087134319579882e-06,
+      "loss": 0.8548,
+      "step": 29015
+    },
+    {
+      "epoch": 0.7969020350992831,
+      "grad_norm": 0.0,
+      "learning_rate": 2.086590460345744e-06,
+      "loss": 0.8285,
+      "step": 29016
+    },
+    {
+      "epoch": 0.7969294993271264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.086046663725566e-06,
+      "loss": 0.8822,
+      "step": 29017
+    },
+    {
+      "epoch": 0.7969569635549697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.085502929723653e-06,
+      "loss": 0.7592,
+      "step": 29018
+    },
+    {
+      "epoch": 0.7969844277828129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.084959258344308e-06,
+      "loss": 0.7514,
+      "step": 29019
+    },
+    {
+      "epoch": 0.7970118920106561,
+      "grad_norm": 0.0,
+      "learning_rate": 2.084415649591829e-06,
+      "loss": 0.8812,
+      "step": 29020
+    },
+    {
+      "epoch": 0.7970393562384993,
+      "grad_norm": 0.0,
+      "learning_rate": 2.083872103470519e-06,
+      "loss": 0.8242,
+      "step": 29021
+    },
+    {
+      "epoch": 0.7970668204663426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0833286199846802e-06,
+      "loss": 0.8008,
+      "step": 29022
+    },
+    {
+      "epoch": 0.7970942846941859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.082785199138614e-06,
+      "loss": 0.7173,
+      "step": 29023
+    },
+    {
+      "epoch": 0.797121748922029,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0822418409366163e-06,
+      "loss": 0.8625,
+      "step": 29024
+    },
+    {
+      "epoch": 0.7971492131498723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0816985453829898e-06,
+      "loss": 0.8896,
+      "step": 29025
+    },
+    {
+      "epoch": 0.7971766773777155,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0811553124820293e-06,
+      "loss": 0.8325,
+      "step": 29026
+    },
+    {
+      "epoch": 0.7972041416055587,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0806121422380364e-06,
+      "loss": 0.8846,
+      "step": 29027
+    },
+    {
+      "epoch": 0.797231605833402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0800690346553065e-06,
+      "loss": 0.7365,
+      "step": 29028
+    },
+    {
+      "epoch": 0.7972590700612452,
+      "grad_norm": 0.0,
+      "learning_rate": 2.079525989738139e-06,
+      "loss": 0.8281,
+      "step": 29029
+    },
+    {
+      "epoch": 0.7972865342890885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.07898300749083e-06,
+      "loss": 0.7633,
+      "step": 29030
+    },
+    {
+      "epoch": 0.7973139985169317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.078440087917679e-06,
+      "loss": 0.846,
+      "step": 29031
+    },
+    {
+      "epoch": 0.7973414627447749,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0778972310229773e-06,
+      "loss": 0.8029,
+      "step": 29032
+    },
+    {
+      "epoch": 0.7973689269726182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0773544368110198e-06,
+      "loss": 0.7725,
+      "step": 29033
+    },
+    {
+      "epoch": 0.7973963912004614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0768117052861025e-06,
+      "loss": 0.7521,
+      "step": 29034
+    },
+    {
+      "epoch": 0.7974238554283046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0762690364525196e-06,
+      "loss": 0.7782,
+      "step": 29035
+    },
+    {
+      "epoch": 0.7974513196561479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0757264303145663e-06,
+      "loss": 0.8438,
+      "step": 29036
+    },
+    {
+      "epoch": 0.7974787838839911,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0751838868765375e-06,
+      "loss": 0.8031,
+      "step": 29037
+    },
+    {
+      "epoch": 0.7975062481118343,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0746414061427213e-06,
+      "loss": 0.7896,
+      "step": 29038
+    },
+    {
+      "epoch": 0.7975337123396776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0740989881174114e-06,
+      "loss": 0.7936,
+      "step": 29039
+    },
+    {
+      "epoch": 0.7975611765675208,
+      "grad_norm": 0.0,
+      "learning_rate": 2.073556632804904e-06,
+      "loss": 0.7947,
+      "step": 29040
+    },
+    {
+      "epoch": 0.7975886407953641,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0730143402094838e-06,
+      "loss": 0.8284,
+      "step": 29041
+    },
+    {
+      "epoch": 0.7976161050232072,
+      "grad_norm": 0.0,
+      "learning_rate": 2.072472110335444e-06,
+      "loss": 0.8222,
+      "step": 29042
+    },
+    {
+      "epoch": 0.7976435692510505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.07192994318708e-06,
+      "loss": 0.8304,
+      "step": 29043
+    },
+    {
+      "epoch": 0.7976710334788938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.071387838768674e-06,
+      "loss": 0.8204,
+      "step": 29044
+    },
+    {
+      "epoch": 0.797698497706737,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0708457970845175e-06,
+      "loss": 0.8802,
+      "step": 29045
+    },
+    {
+      "epoch": 0.7977259619345802,
+      "grad_norm": 0.0,
+      "learning_rate": 2.070303818138901e-06,
+      "loss": 0.8394,
+      "step": 29046
+    },
+    {
+      "epoch": 0.7977534261624234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.069761901936116e-06,
+      "loss": 0.8009,
+      "step": 29047
+    },
+    {
+      "epoch": 0.7977808903902667,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0692200484804438e-06,
+      "loss": 0.7792,
+      "step": 29048
+    },
+    {
+      "epoch": 0.79780835461811,
+      "grad_norm": 0.0,
+      "learning_rate": 2.068678257776178e-06,
+      "loss": 0.8943,
+      "step": 29049
+    },
+    {
+      "epoch": 0.7978358188459531,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0681365298275978e-06,
+      "loss": 0.669,
+      "step": 29050
+    },
+    {
+      "epoch": 0.7978632830737964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0675948646389944e-06,
+      "loss": 0.799,
+      "step": 29051
+    },
+    {
+      "epoch": 0.7978907473016397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.067053262214653e-06,
+      "loss": 0.8694,
+      "step": 29052
+    },
+    {
+      "epoch": 0.7979182115294828,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0665117225588605e-06,
+      "loss": 0.8031,
+      "step": 29053
+    },
+    {
+      "epoch": 0.7979456757573261,
+      "grad_norm": 0.0,
+      "learning_rate": 2.065970245675899e-06,
+      "loss": 0.8429,
+      "step": 29054
+    },
+    {
+      "epoch": 0.7979731399851693,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0654288315700577e-06,
+      "loss": 0.6923,
+      "step": 29055
+    },
+    {
+      "epoch": 0.7980006042130126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0648874802456165e-06,
+      "loss": 0.8353,
+      "step": 29056
+    },
+    {
+      "epoch": 0.7980280684408558,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0643461917068564e-06,
+      "loss": 0.8194,
+      "step": 29057
+    },
+    {
+      "epoch": 0.798055532668699,
+      "grad_norm": 0.0,
+      "learning_rate": 2.063804965958064e-06,
+      "loss": 0.8393,
+      "step": 29058
+    },
+    {
+      "epoch": 0.7980829968965423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.06326380300352e-06,
+      "loss": 0.832,
+      "step": 29059
+    },
+    {
+      "epoch": 0.7981104611243854,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0627227028475104e-06,
+      "loss": 0.8838,
+      "step": 29060
+    },
+    {
+      "epoch": 0.7981379253522287,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0621816654943117e-06,
+      "loss": 0.7604,
+      "step": 29061
+    },
+    {
+      "epoch": 0.798165389580072,
+      "grad_norm": 0.0,
+      "learning_rate": 2.061640690948206e-06,
+      "loss": 0.8983,
+      "step": 29062
+    },
+    {
+      "epoch": 0.7981928538079152,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0610997792134724e-06,
+      "loss": 0.7887,
+      "step": 29063
+    },
+    {
+      "epoch": 0.7982203180357584,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0605589302943974e-06,
+      "loss": 0.8687,
+      "step": 29064
+    },
+    {
+      "epoch": 0.7982477822636017,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0600181441952516e-06,
+      "loss": 0.7865,
+      "step": 29065
+    },
+    {
+      "epoch": 0.7982752464914449,
+      "grad_norm": 0.0,
+      "learning_rate": 2.059477420920322e-06,
+      "loss": 0.7633,
+      "step": 29066
+    },
+    {
+      "epoch": 0.7983027107192882,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0589367604738788e-06,
+      "loss": 0.8365,
+      "step": 29067
+    },
+    {
+      "epoch": 0.7983301749471313,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0583961628602046e-06,
+      "loss": 0.9144,
+      "step": 29068
+    },
+    {
+      "epoch": 0.7983576391749746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.057855628083576e-06,
+      "loss": 0.7894,
+      "step": 29069
+    },
+    {
+      "epoch": 0.7983851034028179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0573151561482694e-06,
+      "loss": 0.9166,
+      "step": 29070
+    },
+    {
+      "epoch": 0.798412567630661,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0567747470585663e-06,
+      "loss": 0.8625,
+      "step": 29071
+    },
+    {
+      "epoch": 0.7984400318585043,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0562344008187376e-06,
+      "loss": 0.8517,
+      "step": 29072
+    },
+    {
+      "epoch": 0.7984674960863475,
+      "grad_norm": 0.0,
+      "learning_rate": 2.055694117433057e-06,
+      "loss": 0.7172,
+      "step": 29073
+    },
+    {
+      "epoch": 0.7984949603141908,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0551538969058017e-06,
+      "loss": 0.765,
+      "step": 29074
+    },
+    {
+      "epoch": 0.798522424542034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0546137392412457e-06,
+      "loss": 0.8021,
+      "step": 29075
+    },
+    {
+      "epoch": 0.7985498887698772,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0540736444436636e-06,
+      "loss": 0.8571,
+      "step": 29076
+    },
+    {
+      "epoch": 0.7985773529977205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.05353361251733e-06,
+      "loss": 0.7231,
+      "step": 29077
+    },
+    {
+      "epoch": 0.7986048172255638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.052993643466518e-06,
+      "loss": 0.7805,
+      "step": 29078
+    },
+    {
+      "epoch": 0.7986322814534069,
+      "grad_norm": 0.0,
+      "learning_rate": 2.052453737295497e-06,
+      "loss": 0.7625,
+      "step": 29079
+    },
+    {
+      "epoch": 0.7986597456812502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0519138940085427e-06,
+      "loss": 0.8196,
+      "step": 29080
+    },
+    {
+      "epoch": 0.7986872099090934,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0513741136099217e-06,
+      "loss": 0.7846,
+      "step": 29081
+    },
+    {
+      "epoch": 0.7987146741369366,
+      "grad_norm": 0.0,
+      "learning_rate": 2.050834396103908e-06,
+      "loss": 0.7626,
+      "step": 29082
+    },
+    {
+      "epoch": 0.7987421383647799,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0502947414947714e-06,
+      "loss": 0.7649,
+      "step": 29083
+    },
+    {
+      "epoch": 0.7987696025926231,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0497551497867853e-06,
+      "loss": 0.8565,
+      "step": 29084
+    },
+    {
+      "epoch": 0.7987970668204664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.049215620984214e-06,
+      "loss": 0.764,
+      "step": 29085
+    },
+    {
+      "epoch": 0.7988245310483095,
+      "grad_norm": 0.0,
+      "learning_rate": 2.048676155091327e-06,
+      "loss": 0.7909,
+      "step": 29086
+    },
+    {
+      "epoch": 0.7988519952761528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0481367521123986e-06,
+      "loss": 0.7953,
+      "step": 29087
+    },
+    {
+      "epoch": 0.7988794595039961,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0475974120516885e-06,
+      "loss": 0.982,
+      "step": 29088
+    },
+    {
+      "epoch": 0.7989069237318392,
+      "grad_norm": 0.0,
+      "learning_rate": 2.047058134913469e-06,
+      "loss": 0.8812,
+      "step": 29089
+    },
+    {
+      "epoch": 0.7989343879596825,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0465189207020085e-06,
+      "loss": 0.7731,
+      "step": 29090
+    },
+    {
+      "epoch": 0.7989618521875258,
+      "grad_norm": 0.0,
+      "learning_rate": 2.045979769421569e-06,
+      "loss": 0.7376,
+      "step": 29091
+    },
+    {
+      "epoch": 0.798989316415369,
+      "grad_norm": 0.0,
+      "learning_rate": 2.045440681076418e-06,
+      "loss": 0.7241,
+      "step": 29092
+    },
+    {
+      "epoch": 0.7990167806432122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.044901655670821e-06,
+      "loss": 0.8577,
+      "step": 29093
+    },
+    {
+      "epoch": 0.7990442448710554,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0443626932090456e-06,
+      "loss": 0.8434,
+      "step": 29094
+    },
+    {
+      "epoch": 0.7990717090988987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0438237936953552e-06,
+      "loss": 0.8834,
+      "step": 29095
+    },
+    {
+      "epoch": 0.799099173326742,
+      "grad_norm": 0.0,
+      "learning_rate": 2.043284957134014e-06,
+      "loss": 0.8245,
+      "step": 29096
+    },
+    {
+      "epoch": 0.7991266375545851,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0427461835292807e-06,
+      "loss": 0.7247,
+      "step": 29097
+    },
+    {
+      "epoch": 0.7991541017824284,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0422074728854223e-06,
+      "loss": 0.8016,
+      "step": 29098
+    },
+    {
+      "epoch": 0.7991815660102716,
+      "grad_norm": 0.0,
+      "learning_rate": 2.041668825206702e-06,
+      "loss": 0.7455,
+      "step": 29099
+    },
+    {
+      "epoch": 0.7992090302381148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.041130240497379e-06,
+      "loss": 0.8748,
+      "step": 29100
+    },
+    {
+      "epoch": 0.7992364944659581,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0405917187617175e-06,
+      "loss": 0.8147,
+      "step": 29101
+    },
+    {
+      "epoch": 0.7992639586938013,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0400532600039802e-06,
+      "loss": 0.7497,
+      "step": 29102
+    },
+    {
+      "epoch": 0.7992914229216446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0395148642284213e-06,
+      "loss": 0.8016,
+      "step": 29103
+    },
+    {
+      "epoch": 0.7993188871494878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0389765314393085e-06,
+      "loss": 0.8875,
+      "step": 29104
+    },
+    {
+      "epoch": 0.799346351377331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.038438261640894e-06,
+      "loss": 0.7881,
+      "step": 29105
+    },
+    {
+      "epoch": 0.7993738156051743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0379000548374394e-06,
+      "loss": 0.7954,
+      "step": 29106
+    },
+    {
+      "epoch": 0.7994012798330175,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0373619110332043e-06,
+      "loss": 0.9612,
+      "step": 29107
+    },
+    {
+      "epoch": 0.7994287440608607,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0368238302324483e-06,
+      "loss": 0.8338,
+      "step": 29108
+    },
+    {
+      "epoch": 0.799456208288704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0362858124394247e-06,
+      "loss": 0.7575,
+      "step": 29109
+    },
+    {
+      "epoch": 0.7994836725165472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.035747857658392e-06,
+      "loss": 0.8148,
+      "step": 29110
+    },
+    {
+      "epoch": 0.7995111367443904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.035209965893611e-06,
+      "loss": 0.8092,
+      "step": 29111
+    },
+    {
+      "epoch": 0.7995386009722336,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0346721371493304e-06,
+      "loss": 0.8577,
+      "step": 29112
+    },
+    {
+      "epoch": 0.7995660652000769,
+      "grad_norm": 0.0,
+      "learning_rate": 2.03413437142981e-06,
+      "loss": 0.7935,
+      "step": 29113
+    },
+    {
+      "epoch": 0.7995935294279202,
+      "grad_norm": 0.0,
+      "learning_rate": 2.033596668739306e-06,
+      "loss": 0.8437,
+      "step": 29114
+    },
+    {
+      "epoch": 0.7996209936557633,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0330590290820694e-06,
+      "loss": 0.8312,
+      "step": 29115
+    },
+    {
+      "epoch": 0.7996484578836066,
+      "grad_norm": 0.0,
+      "learning_rate": 2.032521452462355e-06,
+      "loss": 0.8666,
+      "step": 29116
+    },
+    {
+      "epoch": 0.7996759221114499,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0319839388844178e-06,
+      "loss": 0.8997,
+      "step": 29117
+    },
+    {
+      "epoch": 0.799703386339293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.03144648835251e-06,
+      "loss": 0.8751,
+      "step": 29118
+    },
+    {
+      "epoch": 0.7997308505671363,
+      "grad_norm": 0.0,
+      "learning_rate": 2.030909100870887e-06,
+      "loss": 0.7332,
+      "step": 29119
+    },
+    {
+      "epoch": 0.7997583147949795,
+      "grad_norm": 0.0,
+      "learning_rate": 2.030371776443798e-06,
+      "loss": 0.9062,
+      "step": 29120
+    },
+    {
+      "epoch": 0.7997857790228228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.029834515075492e-06,
+      "loss": 0.7648,
+      "step": 29121
+    },
+    {
+      "epoch": 0.799813243250666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0292973167702225e-06,
+      "loss": 0.869,
+      "step": 29122
+    },
+    {
+      "epoch": 0.7998407074785092,
+      "grad_norm": 0.0,
+      "learning_rate": 2.02876018153224e-06,
+      "loss": 0.904,
+      "step": 29123
+    },
+    {
+      "epoch": 0.7998681717063525,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0282231093657943e-06,
+      "loss": 0.8571,
+      "step": 29124
+    },
+    {
+      "epoch": 0.7998956359341957,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0276861002751392e-06,
+      "loss": 0.8858,
+      "step": 29125
+    },
+    {
+      "epoch": 0.7999231001620389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.027149154264517e-06,
+      "loss": 0.8548,
+      "step": 29126
+    },
+    {
+      "epoch": 0.7999505643898822,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0266122713381787e-06,
+      "loss": 0.8122,
+      "step": 29127
+    },
+    {
+      "epoch": 0.7999780286177254,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0260754515003754e-06,
+      "loss": 0.7999,
+      "step": 29128
+    },
+    {
+      "epoch": 0.8000054928455687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.025538694755349e-06,
+      "loss": 0.8756,
+      "step": 29129
+    },
+    {
+      "epoch": 0.8000329570734119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.02500200110735e-06,
+      "loss": 0.8233,
+      "step": 29130
+    },
+    {
+      "epoch": 0.8000604213012551,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0244653705606263e-06,
+      "loss": 0.855,
+      "step": 29131
+    },
+    {
+      "epoch": 0.8000878855290984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0239288031194204e-06,
+      "loss": 0.7574,
+      "step": 29132
+    },
+    {
+      "epoch": 0.8001153497569415,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0233922987879797e-06,
+      "loss": 0.7955,
+      "step": 29133
+    },
+    {
+      "epoch": 0.8001428139847848,
+      "grad_norm": 0.0,
+      "learning_rate": 2.022855857570547e-06,
+      "loss": 0.7951,
+      "step": 29134
+    },
+    {
+      "epoch": 0.8001702782126281,
+      "grad_norm": 0.0,
+      "learning_rate": 2.022319479471373e-06,
+      "loss": 0.8576,
+      "step": 29135
+    },
+    {
+      "epoch": 0.8001977424404713,
+      "grad_norm": 0.0,
+      "learning_rate": 2.021783164494695e-06,
+      "loss": 0.7609,
+      "step": 29136
+    },
+    {
+      "epoch": 0.8002252066683145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0212469126447608e-06,
+      "loss": 0.8114,
+      "step": 29137
+    },
+    {
+      "epoch": 0.8002526708961577,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0207107239258096e-06,
+      "loss": 0.806,
+      "step": 29138
+    },
+    {
+      "epoch": 0.800280135124001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0201745983420862e-06,
+      "loss": 0.9049,
+      "step": 29139
+    },
+    {
+      "epoch": 0.8003075993518443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0196385358978323e-06,
+      "loss": 0.882,
+      "step": 29140
+    },
+    {
+      "epoch": 0.8003350635796874,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0191025365972893e-06,
+      "loss": 0.747,
+      "step": 29141
+    },
+    {
+      "epoch": 0.8003625278075307,
+      "grad_norm": 0.0,
+      "learning_rate": 2.018566600444699e-06,
+      "loss": 0.8249,
+      "step": 29142
+    },
+    {
+      "epoch": 0.800389992035374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0180307274443033e-06,
+      "loss": 0.781,
+      "step": 29143
+    },
+    {
+      "epoch": 0.8004174562632171,
+      "grad_norm": 0.0,
+      "learning_rate": 2.017494917600341e-06,
+      "loss": 0.7828,
+      "step": 29144
+    },
+    {
+      "epoch": 0.8004449204910604,
+      "grad_norm": 0.0,
+      "learning_rate": 2.016959170917049e-06,
+      "loss": 0.757,
+      "step": 29145
+    },
+    {
+      "epoch": 0.8004723847189036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0164234873986675e-06,
+      "loss": 0.8315,
+      "step": 29146
+    },
+    {
+      "epoch": 0.8004998489467469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0158878670494364e-06,
+      "loss": 0.8234,
+      "step": 29147
+    },
+    {
+      "epoch": 0.8005273131745901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.015352309873593e-06,
+      "loss": 0.6372,
+      "step": 29148
+    },
+    {
+      "epoch": 0.8005547774024333,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0148168158753767e-06,
+      "loss": 0.8134,
+      "step": 29149
+    },
+    {
+      "epoch": 0.8005822416302766,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0142813850590213e-06,
+      "loss": 0.8237,
+      "step": 29150
+    },
+    {
+      "epoch": 0.8006097058581197,
+      "grad_norm": 0.0,
+      "learning_rate": 2.013746017428765e-06,
+      "loss": 0.8004,
+      "step": 29151
+    },
+    {
+      "epoch": 0.800637170085963,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0132107129888465e-06,
+      "loss": 0.8624,
+      "step": 29152
+    },
+    {
+      "epoch": 0.8006646343138063,
+      "grad_norm": 0.0,
+      "learning_rate": 2.012675471743496e-06,
+      "loss": 0.7722,
+      "step": 29153
+    },
+    {
+      "epoch": 0.8006920985416495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0121402936969515e-06,
+      "loss": 0.7734,
+      "step": 29154
+    },
+    {
+      "epoch": 0.8007195627694927,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0116051788534487e-06,
+      "loss": 0.869,
+      "step": 29155
+    },
+    {
+      "epoch": 0.800747026997336,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0110701272172185e-06,
+      "loss": 0.8027,
+      "step": 29156
+    },
+    {
+      "epoch": 0.8007744912251792,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0105351387924953e-06,
+      "loss": 0.7727,
+      "step": 29157
+    },
+    {
+      "epoch": 0.8008019554530225,
+      "grad_norm": 0.0,
+      "learning_rate": 2.010000213583514e-06,
+      "loss": 0.7411,
+      "step": 29158
+    },
+    {
+      "epoch": 0.8008294196808656,
+      "grad_norm": 0.0,
+      "learning_rate": 2.009465351594507e-06,
+      "loss": 0.7193,
+      "step": 29159
+    },
+    {
+      "epoch": 0.8008568839087089,
+      "grad_norm": 0.0,
+      "learning_rate": 2.008930552829704e-06,
+      "loss": 0.8324,
+      "step": 29160
+    },
+    {
+      "epoch": 0.8008843481365522,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0083958172933403e-06,
+      "loss": 0.7889,
+      "step": 29161
+    },
+    {
+      "epoch": 0.8009118123643953,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0078611449896423e-06,
+      "loss": 0.8441,
+      "step": 29162
+    },
+    {
+      "epoch": 0.8009392765922386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.007326535922842e-06,
+      "loss": 0.8264,
+      "step": 29163
+    },
+    {
+      "epoch": 0.8009667408200818,
+      "grad_norm": 0.0,
+      "learning_rate": 2.006791990097171e-06,
+      "loss": 0.9952,
+      "step": 29164
+    },
+    {
+      "epoch": 0.8009942050479251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0062575075168577e-06,
+      "loss": 0.8159,
+      "step": 29165
+    },
+    {
+      "epoch": 0.8010216692757683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0057230881861313e-06,
+      "loss": 0.7822,
+      "step": 29166
+    },
+    {
+      "epoch": 0.8010491335036115,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0051887321092233e-06,
+      "loss": 0.7657,
+      "step": 29167
+    },
+    {
+      "epoch": 0.8010765977314548,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0046544392903587e-06,
+      "loss": 0.8445,
+      "step": 29168
+    },
+    {
+      "epoch": 0.8011040619592981,
+      "grad_norm": 0.0,
+      "learning_rate": 2.004120209733763e-06,
+      "loss": 0.8953,
+      "step": 29169
+    },
+    {
+      "epoch": 0.8011315261871412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.003586043443665e-06,
+      "loss": 0.8832,
+      "step": 29170
+    },
+    {
+      "epoch": 0.8011589904149845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.003051940424292e-06,
+      "loss": 0.8467,
+      "step": 29171
+    },
+    {
+      "epoch": 0.8011864546428277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.00251790067987e-06,
+      "loss": 0.7731,
+      "step": 29172
+    },
+    {
+      "epoch": 0.801213918870671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0019839242146265e-06,
+      "loss": 0.7189,
+      "step": 29173
+    },
+    {
+      "epoch": 0.8012413830985142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.001450011032783e-06,
+      "loss": 0.719,
+      "step": 29174
+    },
+    {
+      "epoch": 0.8012688473263574,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0009161611385643e-06,
+      "loss": 0.7864,
+      "step": 29175
+    },
+    {
+      "epoch": 0.8012963115542007,
+      "grad_norm": 0.0,
+      "learning_rate": 2.000382374536197e-06,
+      "loss": 0.8109,
+      "step": 29176
+    },
+    {
+      "epoch": 0.8013237757820438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999848651229902e-06,
+      "loss": 0.8239,
+      "step": 29177
+    },
+    {
+      "epoch": 0.8013512400098871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999314991223903e-06,
+      "loss": 0.8335,
+      "step": 29178
+    },
+    {
+      "epoch": 0.8013787042377304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987813945224265e-06,
+      "loss": 0.8062,
+      "step": 29179
+    },
+    {
+      "epoch": 0.8014061684655736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982478611296876e-06,
+      "loss": 0.8593,
+      "step": 29180
+    },
+    {
+      "epoch": 0.8014336326934168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977143910499107e-06,
+      "loss": 0.7245,
+      "step": 29181
+    },
+    {
+      "epoch": 0.8014610969212601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997180984287318e-06,
+      "loss": 0.9376,
+      "step": 29182
+    },
+    {
+      "epoch": 0.8014885611491033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996647640846131e-06,
+      "loss": 0.7347,
+      "step": 29183
+    },
+    {
+      "epoch": 0.8015160253769466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961143607305666e-06,
+      "loss": 0.7597,
+      "step": 29184
+    },
+    {
+      "epoch": 0.8015434896047897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995581143944848e-06,
+      "loss": 0.7721,
+      "step": 29185
+    },
+    {
+      "epoch": 0.801570953832633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950479904931898e-06,
+      "loss": 0.769,
+      "step": 29186
+    },
+    {
+      "epoch": 0.8015984180604763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945149003798112e-06,
+      "loss": 0.8698,
+      "step": 29187
+    },
+    {
+      "epoch": 0.8016258822883194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939818736089333e-06,
+      "loss": 0.7906,
+      "step": 29188
+    },
+    {
+      "epoch": 0.8016533465161627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993448910184771e-06,
+      "loss": 0.7784,
+      "step": 29189
+    },
+    {
+      "epoch": 0.8016808107440059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929160101115465e-06,
+      "loss": 0.8436,
+      "step": 29190
+    },
+    {
+      "epoch": 0.8017082749718492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923831733934684e-06,
+      "loss": 0.8262,
+      "step": 29191
+    },
+    {
+      "epoch": 0.8017357391996924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918504000347593e-06,
+      "loss": 0.8171,
+      "step": 29192
+    },
+    {
+      "epoch": 0.8017632034275356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991317690039629e-06,
+      "loss": 0.8546,
+      "step": 29193
+    },
+    {
+      "epoch": 0.8017906676553789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907850434122955e-06,
+      "loss": 0.8545,
+      "step": 29194
+    },
+    {
+      "epoch": 0.8018181318832222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990252460156974e-06,
+      "loss": 0.8705,
+      "step": 29195
+    },
+    {
+      "epoch": 0.8018455961110653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989719940277879e-06,
+      "loss": 0.8201,
+      "step": 29196
+    },
+    {
+      "epoch": 0.8018730603389086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989187483779221e-06,
+      "loss": 0.803,
+      "step": 29197
+    },
+    {
+      "epoch": 0.8019005245667518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988655090665215e-06,
+      "loss": 0.7704,
+      "step": 29198
+    },
+    {
+      "epoch": 0.801927988794595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881227609400724e-06,
+      "loss": 0.8108,
+      "step": 29199
+    },
+    {
+      "epoch": 0.8019554530224383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875904946080094e-06,
+      "loss": 0.7955,
+      "step": 29200
+    },
+    {
+      "epoch": 0.8019829172502815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870582916732307e-06,
+      "loss": 0.7865,
+      "step": 29201
+    },
+    {
+      "epoch": 0.8020103814781248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865261521399535e-06,
+      "loss": 0.7982,
+      "step": 29202
+    },
+    {
+      "epoch": 0.8020378457059679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985994076012383e-06,
+      "loss": 0.8537,
+      "step": 29203
+    },
+    {
+      "epoch": 0.8020653099338112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985462063294732e-06,
+      "loss": 0.8056,
+      "step": 29204
+    },
+    {
+      "epoch": 0.8020927741616545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849301139912104e-06,
+      "loss": 0.7891,
+      "step": 29205
+    },
+    {
+      "epoch": 0.8021202383894976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843982281060247e-06,
+      "loss": 0.7784,
+      "step": 29206
+    },
+    {
+      "epoch": 0.8021477026173409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838664056433886e-06,
+      "loss": 0.8975,
+      "step": 29207
+    },
+    {
+      "epoch": 0.8021751668451842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833346466075067e-06,
+      "loss": 0.817,
+      "step": 29208
+    },
+    {
+      "epoch": 0.8022026310730274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828029510025836e-06,
+      "loss": 0.7871,
+      "step": 29209
+    },
+    {
+      "epoch": 0.8022300953008706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822713188328303e-06,
+      "loss": 0.7891,
+      "step": 29210
+    },
+    {
+      "epoch": 0.8022575595287138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817397501024503e-06,
+      "loss": 0.789,
+      "step": 29211
+    },
+    {
+      "epoch": 0.8022850237565571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981208244815652e-06,
+      "loss": 0.7372,
+      "step": 29212
+    },
+    {
+      "epoch": 0.8023124879844004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980676802976641e-06,
+      "loss": 0.8224,
+      "step": 29213
+    },
+    {
+      "epoch": 0.8023399522122435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801454245896234e-06,
+      "loss": 0.8502,
+      "step": 29214
+    },
+    {
+      "epoch": 0.8023674164400868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796141096587997e-06,
+      "loss": 0.8123,
+      "step": 29215
+    },
+    {
+      "epoch": 0.80239488066793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979082858188378e-06,
+      "loss": 0.8295,
+      "step": 29216
+    },
+    {
+      "epoch": 0.8024223448957732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978551670182558e-06,
+      "loss": 0.7809,
+      "step": 29217
+    },
+    {
+      "epoch": 0.8024498091236165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9780205456455435e-06,
+      "loss": 0.8761,
+      "step": 29218
+    },
+    {
+      "epoch": 0.8024772733514597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774894845815386e-06,
+      "loss": 0.8149,
+      "step": 29219
+    },
+    {
+      "epoch": 0.802504737579303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976958486994747e-06,
+      "loss": 0.8239,
+      "step": 29220
+    },
+    {
+      "epoch": 0.8025322018071462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976427552889364e-06,
+      "loss": 0.7114,
+      "step": 29221
+    },
+    {
+      "epoch": 0.8025596660349894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758966822695945e-06,
+      "loss": 0.8343,
+      "step": 29222
+    },
+    {
+      "epoch": 0.8025871302628327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97536587513964e-06,
+      "loss": 0.714,
+      "step": 29223
+    },
+    {
+      "epoch": 0.8026145944906758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748351315037006e-06,
+      "loss": 0.8166,
+      "step": 29224
+    },
+    {
+      "epoch": 0.8026420587185191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743044513659725e-06,
+      "loss": 0.8106,
+      "step": 29225
+    },
+    {
+      "epoch": 0.8026695229463624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973773834730658e-06,
+      "loss": 0.7766,
+      "step": 29226
+    },
+    {
+      "epoch": 0.8026969871742056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973243281601952e-06,
+      "loss": 0.7925,
+      "step": 29227
+    },
+    {
+      "epoch": 0.8027244514020488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972712791984055e-06,
+      "loss": 0.7786,
+      "step": 29228
+    },
+    {
+      "epoch": 0.8027519156298921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721823658811623e-06,
+      "loss": 0.8524,
+      "step": 29229
+    },
+    {
+      "epoch": 0.8027793798577353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971652003297474e-06,
+      "loss": 0.8154,
+      "step": 29230
+    },
+    {
+      "epoch": 0.8028068440855786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711217042371866e-06,
+      "loss": 0.8157,
+      "step": 29231
+    },
+    {
+      "epoch": 0.8028343083134217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970591468704495e-06,
+      "loss": 0.822,
+      "step": 29232
+    },
+    {
+      "epoch": 0.802861772541265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700612967035902e-06,
+      "loss": 0.7282,
+      "step": 29233
+    },
+    {
+      "epoch": 0.8028892367691083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9695311882386727e-06,
+      "loss": 0.7062,
+      "step": 29234
+    },
+    {
+      "epoch": 0.8029167009969514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690011433139334e-06,
+      "loss": 0.8669,
+      "step": 29235
+    },
+    {
+      "epoch": 0.8029441652247947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684711619335694e-06,
+      "loss": 0.8195,
+      "step": 29236
+    },
+    {
+      "epoch": 0.8029716294526379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679412441017722e-06,
+      "loss": 0.9267,
+      "step": 29237
+    },
+    {
+      "epoch": 0.8029990936804812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674113898227376e-06,
+      "loss": 0.7069,
+      "step": 29238
+    },
+    {
+      "epoch": 0.8030265579083244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966881599100653e-06,
+      "loss": 0.7681,
+      "step": 29239
+    },
+    {
+      "epoch": 0.8030540221361676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663518719397156e-06,
+      "loss": 0.7878,
+      "step": 29240
+    },
+    {
+      "epoch": 0.8030814863640109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9658222083441113e-06,
+      "loss": 0.8762,
+      "step": 29241
+    },
+    {
+      "epoch": 0.8031089505918542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9652926083180337e-06,
+      "loss": 0.719,
+      "step": 29242
+    },
+    {
+      "epoch": 0.8031364148196973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9647630718656763e-06,
+      "loss": 0.8735,
+      "step": 29243
+    },
+    {
+      "epoch": 0.8031638790475406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642335989912232e-06,
+      "loss": 0.8127,
+      "step": 29244
+    },
+    {
+      "epoch": 0.8031913432753838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963704189698867e-06,
+      "loss": 0.7324,
+      "step": 29245
+    },
+    {
+      "epoch": 0.803218807503227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631748439927967e-06,
+      "loss": 0.8435,
+      "step": 29246
+    },
+    {
+      "epoch": 0.8032462717310703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962645561877201e-06,
+      "loss": 0.8376,
+      "step": 29247
+    },
+    {
+      "epoch": 0.8032737359589135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962116343356266e-06,
+      "loss": 0.8918,
+      "step": 29248
+    },
+    {
+      "epoch": 0.8033012001867568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961587188434182e-06,
+      "loss": 0.8134,
+      "step": 29249
+    },
+    {
+      "epoch": 0.8033286644145999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610580971151318e-06,
+      "loss": 0.7798,
+      "step": 29250
+    },
+    {
+      "epoch": 0.8033561286424432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960529069403304e-06,
+      "loss": 0.7195,
+      "step": 29251
+    },
+    {
+      "epoch": 0.8033835928702865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960000105302884e-06,
+      "loss": 0.8073,
+      "step": 29252
+    },
+    {
+      "epoch": 0.8034110570981297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9594712048180574e-06,
+      "loss": 0.8567,
+      "step": 29253
+    },
+    {
+      "epoch": 0.8034385213259729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9589423679530095e-06,
+      "loss": 0.7358,
+      "step": 29254
+    },
+    {
+      "epoch": 0.8034659855538162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9584135947119267e-06,
+      "loss": 0.7996,
+      "step": 29255
+    },
+    {
+      "epoch": 0.8034934497816594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578848850989907e-06,
+      "loss": 0.7836,
+      "step": 29256
+    },
+    {
+      "epoch": 0.8035209140095027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957356239118382e-06,
+      "loss": 0.78,
+      "step": 29257
+    },
+    {
+      "epoch": 0.8035483782373458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568276567742872e-06,
+      "loss": 0.8284,
+      "step": 29258
+    },
+    {
+      "epoch": 0.8035758424651891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562991380708863e-06,
+      "loss": 0.7458,
+      "step": 29259
+    },
+    {
+      "epoch": 0.8036033066930324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9557706830123625e-06,
+      "loss": 0.7798,
+      "step": 29260
+    },
+    {
+      "epoch": 0.8036307709208755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552422916029e-06,
+      "loss": 0.8462,
+      "step": 29261
+    },
+    {
+      "epoch": 0.8036582351487188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954713963846675e-06,
+      "loss": 0.9251,
+      "step": 29262
+    },
+    {
+      "epoch": 0.803685699376562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954185699747868e-06,
+      "loss": 0.7705,
+      "step": 29263
+    },
+    {
+      "epoch": 0.8037131636044053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953657499310665e-06,
+      "loss": 0.72,
+      "step": 29264
+    },
+    {
+      "epoch": 0.8037406278322485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953129362539237e-06,
+      "loss": 0.8442,
+      "step": 29265
+    },
+    {
+      "epoch": 0.8037680920600917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9526012894377653e-06,
+      "loss": 0.8642,
+      "step": 29266
+    },
+    {
+      "epoch": 0.803795556287935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9520732800104337e-06,
+      "loss": 0.7478,
+      "step": 29267
+    },
+    {
+      "epoch": 0.8038230205157783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951545334261413e-06,
+      "loss": 0.7942,
+      "step": 29268
+    },
+    {
+      "epoch": 0.8038504847436214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951017452194882e-06,
+      "loss": 0.6789,
+      "step": 29269
+    },
+    {
+      "epoch": 0.8038779489714647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504896338150204e-06,
+      "loss": 0.7595,
+      "step": 29270
+    },
+    {
+      "epoch": 0.8039054131993079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499618791260045e-06,
+      "loss": 0.7836,
+      "step": 29271
+    },
+    {
+      "epoch": 0.8039328774271511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9494341881320065e-06,
+      "loss": 0.9366,
+      "step": 29272
+    },
+    {
+      "epoch": 0.8039603416549944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948906560837206e-06,
+      "loss": 0.7382,
+      "step": 29273
+    },
+    {
+      "epoch": 0.8039878058828376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9483789972457736e-06,
+      "loss": 0.7284,
+      "step": 29274
+    },
+    {
+      "epoch": 0.8040152701106809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9478514973618844e-06,
+      "loss": 0.8847,
+      "step": 29275
+    },
+    {
+      "epoch": 0.804042734338524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9473240611897137e-06,
+      "loss": 0.8224,
+      "step": 29276
+    },
+    {
+      "epoch": 0.8040701985663673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467966887334344e-06,
+      "loss": 0.7887,
+      "step": 29277
+    },
+    {
+      "epoch": 0.8040976627942106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946269379997219e-06,
+      "loss": 0.7245,
+      "step": 29278
+    },
+    {
+      "epoch": 0.8041251270220537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457421349852433e-06,
+      "loss": 0.8088,
+      "step": 29279
+    },
+    {
+      "epoch": 0.804152591249897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9452149537016753e-06,
+      "loss": 0.767,
+      "step": 29280
+    },
+    {
+      "epoch": 0.8041800554777403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944687836150684e-06,
+      "loss": 0.8139,
+      "step": 29281
+    },
+    {
+      "epoch": 0.8042075197055835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944160782336443e-06,
+      "loss": 0.765,
+      "step": 29282
+    },
+    {
+      "epoch": 0.8042349839334267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943633792263122e-06,
+      "loss": 0.7408,
+      "step": 29283
+    },
+    {
+      "epoch": 0.8042624481612699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9431068659348905e-06,
+      "loss": 0.8564,
+      "step": 29284
+    },
+    {
+      "epoch": 0.8042899123891132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425800033559226e-06,
+      "loss": 0.7864,
+      "step": 29285
+    },
+    {
+      "epoch": 0.8043173766169565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9420532045303795e-06,
+      "loss": 0.8119,
+      "step": 29286
+    },
+    {
+      "epoch": 0.8043448408447996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9415264694624327e-06,
+      "loss": 0.9382,
+      "step": 29287
+    },
+    {
+      "epoch": 0.8043723050726429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9409997981562535e-06,
+      "loss": 0.7334,
+      "step": 29288
+    },
+    {
+      "epoch": 0.8043997693004861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9404731906160024e-06,
+      "loss": 0.8226,
+      "step": 29289
+    },
+    {
+      "epoch": 0.8044272335283293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399466468458495e-06,
+      "loss": 0.788,
+      "step": 29290
+    },
+    {
+      "epoch": 0.8044546977561726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394201668499634e-06,
+      "loss": 0.7308,
+      "step": 29291
+    },
+    {
+      "epoch": 0.8044821619840158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9388937506325047e-06,
+      "loss": 0.7663,
+      "step": 29292
+    },
+    {
+      "epoch": 0.8045096262118591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9383673981976404e-06,
+      "loss": 0.8807,
+      "step": 29293
+    },
+    {
+      "epoch": 0.8045370904397023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9378411095495374e-06,
+      "loss": 0.763,
+      "step": 29294
+    },
+    {
+      "epoch": 0.8045645546675455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373148846923608e-06,
+      "loss": 0.8013,
+      "step": 29295
+    },
+    {
+      "epoch": 0.8045920188953888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9367887236302684e-06,
+      "loss": 0.9513,
+      "step": 29296
+    },
+    {
+      "epoch": 0.804619483123232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936262626367431e-06,
+      "loss": 0.799,
+      "step": 29297
+    },
+    {
+      "epoch": 0.8046469473510752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935736592908004e-06,
+      "loss": 0.8151,
+      "step": 29298
+    },
+    {
+      "epoch": 0.8046744115789185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935210623256153e-06,
+      "loss": 0.7822,
+      "step": 29299
+    },
+    {
+      "epoch": 0.8047018758067617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9346847174160387e-06,
+      "loss": 0.8245,
+      "step": 29300
+    },
+    {
+      "epoch": 0.804729340034605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934158875391824e-06,
+      "loss": 0.7873,
+      "step": 29301
+    },
+    {
+      "epoch": 0.8047568042624481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336330971876717e-06,
+      "loss": 0.8076,
+      "step": 29302
+    },
+    {
+      "epoch": 0.8047842684902914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933107382807736e-06,
+      "loss": 0.7953,
+      "step": 29303
+    },
+    {
+      "epoch": 0.8048117327181347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325817322561815e-06,
+      "loss": 0.8837,
+      "step": 29304
+    },
+    {
+      "epoch": 0.8048391969459778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9320561455371635e-06,
+      "loss": 0.7689,
+      "step": 29305
+    },
+    {
+      "epoch": 0.8048666611738211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9315306226548424e-06,
+      "loss": 0.807,
+      "step": 29306
+    },
+    {
+      "epoch": 0.8048941254016644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9310051636133764e-06,
+      "loss": 0.763,
+      "step": 29307
+    },
+    {
+      "epoch": 0.8049215896295076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930479768416925e-06,
+      "loss": 0.7106,
+      "step": 29308
+    },
+    {
+      "epoch": 0.8049490538573508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929954437069642e-06,
+      "loss": 0.8159,
+      "step": 29309
+    },
+    {
+      "epoch": 0.804976518085194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929429169575685e-06,
+      "loss": 0.815,
+      "step": 29310
+    },
+    {
+      "epoch": 0.8050039823130373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9289039659392106e-06,
+      "loss": 0.8038,
+      "step": 29311
+    },
+    {
+      "epoch": 0.8050314465408805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928378826164378e-06,
+      "loss": 0.7756,
+      "step": 29312
+    },
+    {
+      "epoch": 0.8050589107687237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927853750255334e-06,
+      "loss": 0.8428,
+      "step": 29313
+    },
+    {
+      "epoch": 0.805086374996567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927328738216242e-06,
+      "loss": 0.8036,
+      "step": 29314
+    },
+    {
+      "epoch": 0.8051138392244102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926803790051249e-06,
+      "loss": 0.8354,
+      "step": 29315
+    },
+    {
+      "epoch": 0.8051413034522534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262789057645126e-06,
+      "loss": 0.7598,
+      "step": 29316
+    },
+    {
+      "epoch": 0.8051687676800967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257540853601843e-06,
+      "loss": 0.8315,
+      "step": 29317
+    },
+    {
+      "epoch": 0.8051962319079399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9252293288424186e-06,
+      "loss": 0.8097,
+      "step": 29318
+    },
+    {
+      "epoch": 0.8052236961357832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9247046362153677e-06,
+      "loss": 0.8378,
+      "step": 29319
+    },
+    {
+      "epoch": 0.8052511603636264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241800074831787e-06,
+      "loss": 0.8711,
+      "step": 29320
+    },
+    {
+      "epoch": 0.8052786245914696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9236554426500087e-06,
+      "loss": 0.7862,
+      "step": 29321
+    },
+    {
+      "epoch": 0.8053060888193129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231309417200027e-06,
+      "loss": 0.8596,
+      "step": 29322
+    },
+    {
+      "epoch": 0.805333553047156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9226065046973143e-06,
+      "loss": 0.7076,
+      "step": 29323
+    },
+    {
+      "epoch": 0.8053610172749993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9220821315860916e-06,
+      "loss": 0.8572,
+      "step": 29324
+    },
+    {
+      "epoch": 0.8053884815028426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921557822390484e-06,
+      "loss": 0.7961,
+      "step": 29325
+    },
+    {
+      "epoch": 0.8054159457306858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9210335771146426e-06,
+      "loss": 0.9391,
+      "step": 29326
+    },
+    {
+      "epoch": 0.805443409958529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9205093957627108e-06,
+      "loss": 0.8468,
+      "step": 29327
+    },
+    {
+      "epoch": 0.8054708741863722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9199852783388407e-06,
+      "loss": 0.9179,
+      "step": 29328
+    },
+    {
+      "epoch": 0.8054983384142155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194612248471744e-06,
+      "loss": 0.8607,
+      "step": 29329
+    },
+    {
+      "epoch": 0.8055258026420588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9189372352918613e-06,
+      "loss": 0.7726,
+      "step": 29330
+    },
+    {
+      "epoch": 0.8055532668699019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918413309677046e-06,
+      "loss": 0.8083,
+      "step": 29331
+    },
+    {
+      "epoch": 0.8055807310977452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917889448006879e-06,
+      "loss": 0.7646,
+      "step": 29332
+    },
+    {
+      "epoch": 0.8056081953255885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9173656502854986e-06,
+      "loss": 0.8583,
+      "step": 29333
+    },
+    {
+      "epoch": 0.8056356595534316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168419165170514e-06,
+      "loss": 0.7298,
+      "step": 29334
+    },
+    {
+      "epoch": 0.8056631237812749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9163182467056827e-06,
+      "loss": 0.7507,
+      "step": 29335
+    },
+    {
+      "epoch": 0.8056905880091181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915794640855537e-06,
+      "loss": 0.76,
+      "step": 29336
+    },
+    {
+      "epoch": 0.8057180522369614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9152710989707545e-06,
+      "loss": 0.8221,
+      "step": 29337
+    },
+    {
+      "epoch": 0.8057455164648046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147476210554806e-06,
+      "loss": 0.807,
+      "step": 29338
+    },
+    {
+      "epoch": 0.8057729806926478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9142242071138527e-06,
+      "loss": 0.8442,
+      "step": 29339
+    },
+    {
+      "epoch": 0.8058004449204911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913700857150016e-06,
+      "loss": 0.7798,
+      "step": 29340
+    },
+    {
+      "epoch": 0.8058279091483342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9131775711681087e-06,
+      "loss": 0.9026,
+      "step": 29341
+    },
+    {
+      "epoch": 0.8058553733761775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912654349172275e-06,
+      "loss": 0.8452,
+      "step": 29342
+    },
+    {
+      "epoch": 0.8058828376040208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912131191166654e-06,
+      "loss": 0.7873,
+      "step": 29343
+    },
+    {
+      "epoch": 0.805910301831864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911608097155382e-06,
+      "loss": 0.7997,
+      "step": 29344
+    },
+    {
+      "epoch": 0.8059377660597072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9110850671426017e-06,
+      "loss": 0.8474,
+      "step": 29345
+    },
+    {
+      "epoch": 0.8059652302875505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910562101132447e-06,
+      "loss": 0.7671,
+      "step": 29346
+    },
+    {
+      "epoch": 0.8059926945153937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9100391991290602e-06,
+      "loss": 0.8608,
+      "step": 29347
+    },
+    {
+      "epoch": 0.806020158743237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9095163611365752e-06,
+      "loss": 0.7978,
+      "step": 29348
+    },
+    {
+      "epoch": 0.8060476229710801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9089935871591312e-06,
+      "loss": 0.7684,
+      "step": 29349
+    },
+    {
+      "epoch": 0.8060750871989234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9084708772008677e-06,
+      "loss": 0.9174,
+      "step": 29350
+    },
+    {
+      "epoch": 0.8061025514267667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907948231265914e-06,
+      "loss": 0.9796,
+      "step": 29351
+    },
+    {
+      "epoch": 0.8061300156546098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9074256493584096e-06,
+      "loss": 0.8194,
+      "step": 29352
+    },
+    {
+      "epoch": 0.8061574798824531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906903131482487e-06,
+      "loss": 0.7868,
+      "step": 29353
+    },
+    {
+      "epoch": 0.8061849441102963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9063806776422822e-06,
+      "loss": 0.9313,
+      "step": 29354
+    },
+    {
+      "epoch": 0.8062124083381396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9058582878419275e-06,
+      "loss": 0.9286,
+      "step": 29355
+    },
+    {
+      "epoch": 0.8062398725659828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9053359620855605e-06,
+      "loss": 0.8342,
+      "step": 29356
+    },
+    {
+      "epoch": 0.806267336793826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9048137003773082e-06,
+      "loss": 0.8052,
+      "step": 29357
+    },
+    {
+      "epoch": 0.8062948010216693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904291502721306e-06,
+      "loss": 0.8245,
+      "step": 29358
+    },
+    {
+      "epoch": 0.8063222652495126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9037693691216852e-06,
+      "loss": 0.8281,
+      "step": 29359
+    },
+    {
+      "epoch": 0.8063497294773557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903247299582579e-06,
+      "loss": 0.7363,
+      "step": 29360
+    },
+    {
+      "epoch": 0.806377193705199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9027252941081175e-06,
+      "loss": 0.7455,
+      "step": 29361
+    },
+    {
+      "epoch": 0.8064046579330422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9022033527024275e-06,
+      "loss": 0.7789,
+      "step": 29362
+    },
+    {
+      "epoch": 0.8064321221608854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9016814753696411e-06,
+      "loss": 0.8369,
+      "step": 29363
+    },
+    {
+      "epoch": 0.8064595863887287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9011596621138883e-06,
+      "loss": 0.8746,
+      "step": 29364
+    },
+    {
+      "epoch": 0.8064870506165719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900637912939297e-06,
+      "loss": 0.7295,
+      "step": 29365
+    },
+    {
+      "epoch": 0.8065145148444152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9001162278499962e-06,
+      "loss": 0.7993,
+      "step": 29366
+    },
+    {
+      "epoch": 0.8065419790722583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8995946068501158e-06,
+      "loss": 0.7609,
+      "step": 29367
+    },
+    {
+      "epoch": 0.8065694433001016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8990730499437816e-06,
+      "loss": 0.7094,
+      "step": 29368
+    },
+    {
+      "epoch": 0.8065969075279449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8985515571351154e-06,
+      "loss": 0.7484,
+      "step": 29369
+    },
+    {
+      "epoch": 0.806624371755788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8980301284282477e-06,
+      "loss": 0.8802,
+      "step": 29370
+    },
+    {
+      "epoch": 0.8066518359836313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8975087638273049e-06,
+      "loss": 0.8777,
+      "step": 29371
+    },
+    {
+      "epoch": 0.8066793002114746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8969874633364105e-06,
+      "loss": 0.8645,
+      "step": 29372
+    },
+    {
+      "epoch": 0.8067067644393178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964662269596935e-06,
+      "loss": 0.7913,
+      "step": 29373
+    },
+    {
+      "epoch": 0.806734228667161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8959450547012714e-06,
+      "loss": 0.747,
+      "step": 29374
+    },
+    {
+      "epoch": 0.8067616928950042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8954239465652714e-06,
+      "loss": 0.8248,
+      "step": 29375
+    },
+    {
+      "epoch": 0.8067891571228475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8949029025558196e-06,
+      "loss": 0.7996,
+      "step": 29376
+    },
+    {
+      "epoch": 0.8068166213506908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894381922677032e-06,
+      "loss": 0.7596,
+      "step": 29377
+    },
+    {
+      "epoch": 0.8068440855785339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8938610069330354e-06,
+      "loss": 0.7023,
+      "step": 29378
+    },
+    {
+      "epoch": 0.8068715498063772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893340155327953e-06,
+      "loss": 0.8537,
+      "step": 29379
+    },
+    {
+      "epoch": 0.8068990140342204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8928193678659012e-06,
+      "loss": 0.7769,
+      "step": 29380
+    },
+    {
+      "epoch": 0.8069264782620637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8922986445510017e-06,
+      "loss": 0.8328,
+      "step": 29381
+    },
+    {
+      "epoch": 0.8069539424899069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8917779853873775e-06,
+      "loss": 0.7422,
+      "step": 29382
+    },
+    {
+      "epoch": 0.8069814067177501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8912573903791453e-06,
+      "loss": 0.6628,
+      "step": 29383
+    },
+    {
+      "epoch": 0.8070088709455934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8907368595304298e-06,
+      "loss": 0.7921,
+      "step": 29384
+    },
+    {
+      "epoch": 0.8070363351734366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8902163928453443e-06,
+      "loss": 0.9007,
+      "step": 29385
+    },
+    {
+      "epoch": 0.8070637994012798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8896959903280054e-06,
+      "loss": 0.8538,
+      "step": 29386
+    },
+    {
+      "epoch": 0.8070912636291231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8891756519825344e-06,
+      "loss": 0.7405,
+      "step": 29387
+    },
+    {
+      "epoch": 0.8071187278569663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8886553778130468e-06,
+      "loss": 0.7931,
+      "step": 29388
+    },
+    {
+      "epoch": 0.8071461920848095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8881351678236603e-06,
+      "loss": 0.8073,
+      "step": 29389
+    },
+    {
+      "epoch": 0.8071736563126528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8876150220184907e-06,
+      "loss": 0.8357,
+      "step": 29390
+    },
+    {
+      "epoch": 0.807201120540496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8870949404016558e-06,
+      "loss": 0.7911,
+      "step": 29391
+    },
+    {
+      "epoch": 0.8072285847683393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8865749229772678e-06,
+      "loss": 0.7538,
+      "step": 29392
+    },
+    {
+      "epoch": 0.8072560489961824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8860549697494402e-06,
+      "loss": 0.7556,
+      "step": 29393
+    },
+    {
+      "epoch": 0.8072835132240257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8855350807222883e-06,
+      "loss": 0.9075,
+      "step": 29394
+    },
+    {
+      "epoch": 0.807310977451869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8850152558999258e-06,
+      "loss": 0.8799,
+      "step": 29395
+    },
+    {
+      "epoch": 0.8073384416797121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884495495286467e-06,
+      "loss": 0.7379,
+      "step": 29396
+    },
+    {
+      "epoch": 0.8073659059075554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8839757988860252e-06,
+      "loss": 0.743,
+      "step": 29397
+    },
+    {
+      "epoch": 0.8073933701353987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8834561667027086e-06,
+      "loss": 0.788,
+      "step": 29398
+    },
+    {
+      "epoch": 0.8074208343632419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829365987406301e-06,
+      "loss": 0.8261,
+      "step": 29399
+    },
+    {
+      "epoch": 0.8074482985910851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8824170950039044e-06,
+      "loss": 0.8375,
+      "step": 29400
+    },
+    {
+      "epoch": 0.8074757628189283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8818976554966372e-06,
+      "loss": 0.8831,
+      "step": 29401
+    },
+    {
+      "epoch": 0.8075032270467716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813782802229396e-06,
+      "loss": 0.8777,
+      "step": 29402
+    },
+    {
+      "epoch": 0.8075306912746149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880858969186925e-06,
+      "loss": 0.9859,
+      "step": 29403
+    },
+    {
+      "epoch": 0.807558155502458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803397223926967e-06,
+      "loss": 0.8356,
+      "step": 29404
+    },
+    {
+      "epoch": 0.8075856197303013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879820539844366e-06,
+      "loss": 0.7168,
+      "step": 29405
+    },
+    {
+      "epoch": 0.8076130839581445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8793014215460393e-06,
+      "loss": 0.8005,
+      "step": 29406
+    },
+    {
+      "epoch": 0.8076405481859877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787823675018291e-06,
+      "loss": 0.9141,
+      "step": 29407
+    },
+    {
+      "epoch": 0.807668012413831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8782633777158355e-06,
+      "loss": 0.8007,
+      "step": 29408
+    },
+    {
+      "epoch": 0.8076954766416742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8777444521921718e-06,
+      "loss": 0.9556,
+      "step": 29409
+    },
+    {
+      "epoch": 0.8077229408695175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8772255909349368e-06,
+      "loss": 0.8105,
+      "step": 29410
+    },
+    {
+      "epoch": 0.8077504050973607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8767067939482398e-06,
+      "loss": 0.7001,
+      "step": 29411
+    },
+    {
+      "epoch": 0.8077778693252039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876188061236185e-06,
+      "loss": 0.8487,
+      "step": 29412
+    },
+    {
+      "epoch": 0.8078053335530472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875669392802877e-06,
+      "loss": 0.8499,
+      "step": 29413
+    },
+    {
+      "epoch": 0.8078327977808903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87515078865242e-06,
+      "loss": 0.7439,
+      "step": 29414
+    },
+    {
+      "epoch": 0.8078602620087336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8746322487889202e-06,
+      "loss": 0.8104,
+      "step": 29415
+    },
+    {
+      "epoch": 0.8078877262365769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741137732164771e-06,
+      "loss": 0.8198,
+      "step": 29416
+    },
+    {
+      "epoch": 0.8079151904644201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87359536193919e-06,
+      "loss": 0.8648,
+      "step": 29417
+    },
+    {
+      "epoch": 0.8079426546922633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8730770149611655e-06,
+      "loss": 0.8261,
+      "step": 29418
+    },
+    {
+      "epoch": 0.8079701189201065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725587322865035e-06,
+      "loss": 0.8252,
+      "step": 29419
+    },
+    {
+      "epoch": 0.8079975831479498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8720405139193077e-06,
+      "loss": 0.8364,
+      "step": 29420
+    },
+    {
+      "epoch": 0.8080250473757931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8715223598636734e-06,
+      "loss": 0.7795,
+      "step": 29421
+    },
+    {
+      "epoch": 0.8080525116036362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871004270123702e-06,
+      "loss": 0.7503,
+      "step": 29422
+    },
+    {
+      "epoch": 0.8080799758314795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8704862447034943e-06,
+      "loss": 0.7874,
+      "step": 29423
+    },
+    {
+      "epoch": 0.8081074400593228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8699682836071508e-06,
+      "loss": 0.8413,
+      "step": 29424
+    },
+    {
+      "epoch": 0.808134904287166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869450386838766e-06,
+      "loss": 0.8416,
+      "step": 29425
+    },
+    {
+      "epoch": 0.8081623685150092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8689325544024405e-06,
+      "loss": 0.7272,
+      "step": 29426
+    },
+    {
+      "epoch": 0.8081898327428524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8684147863022684e-06,
+      "loss": 0.8392,
+      "step": 29427
+    },
+    {
+      "epoch": 0.8082172969706957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8678970825423481e-06,
+      "loss": 0.7939,
+      "step": 29428
+    },
+    {
+      "epoch": 0.8082447611985389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8673794431267767e-06,
+      "loss": 0.7338,
+      "step": 29429
+    },
+    {
+      "epoch": 0.8082722254263821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668618680596485e-06,
+      "loss": 0.8068,
+      "step": 29430
+    },
+    {
+      "epoch": 0.8082996896542254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8663443573450635e-06,
+      "loss": 0.7682,
+      "step": 29431
+    },
+    {
+      "epoch": 0.8083271538820687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8658269109871085e-06,
+      "loss": 0.867,
+      "step": 29432
+    },
+    {
+      "epoch": 0.8083546181099118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865309528989886e-06,
+      "loss": 0.833,
+      "step": 29433
+    },
+    {
+      "epoch": 0.8083820823377551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8647922113574813e-06,
+      "loss": 0.8101,
+      "step": 29434
+    },
+    {
+      "epoch": 0.8084095465655983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8642749580939934e-06,
+      "loss": 0.786,
+      "step": 29435
+    },
+    {
+      "epoch": 0.8084370107934415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8637577692035124e-06,
+      "loss": 0.8167,
+      "step": 29436
+    },
+    {
+      "epoch": 0.8084644750212848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863240644690133e-06,
+      "loss": 0.8613,
+      "step": 29437
+    },
+    {
+      "epoch": 0.808491939249128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8627235845579471e-06,
+      "loss": 0.8089,
+      "step": 29438
+    },
+    {
+      "epoch": 0.8085194034769713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8622065888110418e-06,
+      "loss": 0.7809,
+      "step": 29439
+    },
+    {
+      "epoch": 0.8085468677048144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8616896574535126e-06,
+      "loss": 0.7769,
+      "step": 29440
+    },
+    {
+      "epoch": 0.8085743319326577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861172790489445e-06,
+      "loss": 0.8104,
+      "step": 29441
+    },
+    {
+      "epoch": 0.808601796160501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606559879229313e-06,
+      "loss": 0.7956,
+      "step": 29442
+    },
+    {
+      "epoch": 0.8086292603883442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8601392497580595e-06,
+      "loss": 0.8064,
+      "step": 29443
+    },
+    {
+      "epoch": 0.8086567246161874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8596225759989216e-06,
+      "loss": 0.7858,
+      "step": 29444
+    },
+    {
+      "epoch": 0.8086841888440307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8591059666496014e-06,
+      "loss": 0.7326,
+      "step": 29445
+    },
+    {
+      "epoch": 0.8087116530718739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8585894217141876e-06,
+      "loss": 0.8308,
+      "step": 29446
+    },
+    {
+      "epoch": 0.8087391172997171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858072941196769e-06,
+      "loss": 0.8901,
+      "step": 29447
+    },
+    {
+      "epoch": 0.8087665815275603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8575565251014328e-06,
+      "loss": 0.9275,
+      "step": 29448
+    },
+    {
+      "epoch": 0.8087940457554036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8570401734322607e-06,
+      "loss": 0.8496,
+      "step": 29449
+    },
+    {
+      "epoch": 0.8088215099832469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8565238861933443e-06,
+      "loss": 0.8022,
+      "step": 29450
+    },
+    {
+      "epoch": 0.80884897421109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8560076633887626e-06,
+      "loss": 0.8944,
+      "step": 29451
+    },
+    {
+      "epoch": 0.8088764384389333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855491505022603e-06,
+      "loss": 0.8357,
+      "step": 29452
+    },
+    {
+      "epoch": 0.8089039026667765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8549754110989492e-06,
+      "loss": 0.8295,
+      "step": 29453
+    },
+    {
+      "epoch": 0.8089313668946198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8544593816218847e-06,
+      "loss": 0.7257,
+      "step": 29454
+    },
+    {
+      "epoch": 0.808958831122463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8539434165954962e-06,
+      "loss": 0.8865,
+      "step": 29455
+    },
+    {
+      "epoch": 0.8089862953503062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8534275160238603e-06,
+      "loss": 0.8199,
+      "step": 29456
+    },
+    {
+      "epoch": 0.8090137595781495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8529116799110635e-06,
+      "loss": 0.7998,
+      "step": 29457
+    },
+    {
+      "epoch": 0.8090412238059927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852395908261183e-06,
+      "loss": 0.9227,
+      "step": 29458
+    },
+    {
+      "epoch": 0.8090686880338359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8518802010783032e-06,
+      "loss": 0.7759,
+      "step": 29459
+    },
+    {
+      "epoch": 0.8090961522616792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8513645583665018e-06,
+      "loss": 0.8155,
+      "step": 29460
+    },
+    {
+      "epoch": 0.8091236164895224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8508489801298624e-06,
+      "loss": 0.8113,
+      "step": 29461
+    },
+    {
+      "epoch": 0.8091510807173656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8503334663724636e-06,
+      "loss": 0.7621,
+      "step": 29462
+    },
+    {
+      "epoch": 0.8091785449452089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8498180170983815e-06,
+      "loss": 0.696,
+      "step": 29463
+    },
+    {
+      "epoch": 0.8092060091730521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.849302632311698e-06,
+      "loss": 0.666,
+      "step": 29464
+    },
+    {
+      "epoch": 0.8092334734008954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8487873120164878e-06,
+      "loss": 0.7619,
+      "step": 29465
+    },
+    {
+      "epoch": 0.8092609376287385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8482720562168288e-06,
+      "loss": 0.8358,
+      "step": 29466
+    },
+    {
+      "epoch": 0.8092884018565818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8477568649167998e-06,
+      "loss": 0.7696,
+      "step": 29467
+    },
+    {
+      "epoch": 0.8093158660844251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8472417381204789e-06,
+      "loss": 0.8247,
+      "step": 29468
+    },
+    {
+      "epoch": 0.8093433303122682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8467266758319368e-06,
+      "loss": 0.7985,
+      "step": 29469
+    },
+    {
+      "epoch": 0.8093707945401115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846211678055252e-06,
+      "loss": 0.8559,
+      "step": 29470
+    },
+    {
+      "epoch": 0.8093982587679548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8456967447944974e-06,
+      "loss": 0.824,
+      "step": 29471
+    },
+    {
+      "epoch": 0.809425722995798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8451818760537521e-06,
+      "loss": 0.7971,
+      "step": 29472
+    },
+    {
+      "epoch": 0.8094531872236412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8446670718370852e-06,
+      "loss": 0.7889,
+      "step": 29473
+    },
+    {
+      "epoch": 0.8094806514514844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8441523321485733e-06,
+      "loss": 0.8324,
+      "step": 29474
+    },
+    {
+      "epoch": 0.8095081156793277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8436376569922843e-06,
+      "loss": 0.9426,
+      "step": 29475
+    },
+    {
+      "epoch": 0.809535579907171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8431230463722938e-06,
+      "loss": 0.917,
+      "step": 29476
+    },
+    {
+      "epoch": 0.8095630441350141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842608500292674e-06,
+      "loss": 0.812,
+      "step": 29477
+    },
+    {
+      "epoch": 0.8095905083628574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8420940187574954e-06,
+      "loss": 0.7523,
+      "step": 29478
+    },
+    {
+      "epoch": 0.8096179725907006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841579601770831e-06,
+      "loss": 0.789,
+      "step": 29479
+    },
+    {
+      "epoch": 0.8096454368185438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8410652493367476e-06,
+      "loss": 0.8085,
+      "step": 29480
+    },
+    {
+      "epoch": 0.8096729010463871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8405509614593175e-06,
+      "loss": 0.8148,
+      "step": 29481
+    },
+    {
+      "epoch": 0.8097003652742303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8400367381426076e-06,
+      "loss": 0.8468,
+      "step": 29482
+    },
+    {
+      "epoch": 0.8097278295020736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839522579390688e-06,
+      "loss": 0.838,
+      "step": 29483
+    },
+    {
+      "epoch": 0.8097552937299168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8390084852076263e-06,
+      "loss": 0.743,
+      "step": 29484
+    },
+    {
+      "epoch": 0.80978275795776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8384944555974925e-06,
+      "loss": 0.7665,
+      "step": 29485
+    },
+    {
+      "epoch": 0.8098102221856033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8379804905643505e-06,
+      "loss": 0.7835,
+      "step": 29486
+    },
+    {
+      "epoch": 0.8098376864134464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837466590112268e-06,
+      "loss": 0.853,
+      "step": 29487
+    },
+    {
+      "epoch": 0.8098651506412897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8369527542453148e-06,
+      "loss": 0.7838,
+      "step": 29488
+    },
+    {
+      "epoch": 0.809892614869133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83643898296755e-06,
+      "loss": 0.8094,
+      "step": 29489
+    },
+    {
+      "epoch": 0.8099200790969762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8359252762830438e-06,
+      "loss": 0.7624,
+      "step": 29490
+    },
+    {
+      "epoch": 0.8099475433248194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8354116341958606e-06,
+      "loss": 0.8228,
+      "step": 29491
+    },
+    {
+      "epoch": 0.8099750075526626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8348980567100605e-06,
+      "loss": 0.8568,
+      "step": 29492
+    },
+    {
+      "epoch": 0.8100024717805059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8343845438297104e-06,
+      "loss": 0.7824,
+      "step": 29493
+    },
+    {
+      "epoch": 0.8100299360083492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8338710955588723e-06,
+      "loss": 0.8003,
+      "step": 29494
+    },
+    {
+      "epoch": 0.8100574002361923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83335771190161e-06,
+      "loss": 0.7651,
+      "step": 29495
+    },
+    {
+      "epoch": 0.8100848644640356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8328443928619877e-06,
+      "loss": 0.8268,
+      "step": 29496
+    },
+    {
+      "epoch": 0.8101123286918789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8323311384440635e-06,
+      "loss": 0.8103,
+      "step": 29497
+    },
+    {
+      "epoch": 0.810139792919722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8318179486518972e-06,
+      "loss": 0.791,
+      "step": 29498
+    },
+    {
+      "epoch": 0.8101672571475653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8313048234895503e-06,
+      "loss": 0.8638,
+      "step": 29499
+    },
+    {
+      "epoch": 0.8101947213754085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830791762961085e-06,
+      "loss": 0.8993,
+      "step": 29500
+    },
+    {
+      "epoch": 0.8102221856032518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8302787670705591e-06,
+      "loss": 0.8444,
+      "step": 29501
+    },
+    {
+      "epoch": 0.810249649831095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8297658358220326e-06,
+      "loss": 0.8105,
+      "step": 29502
+    },
+    {
+      "epoch": 0.8102771140589382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829252969219566e-06,
+      "loss": 0.7884,
+      "step": 29503
+    },
+    {
+      "epoch": 0.8103045782867815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8287401672672134e-06,
+      "loss": 0.7768,
+      "step": 29504
+    },
+    {
+      "epoch": 0.8103320425146247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828227429969035e-06,
+      "loss": 0.7941,
+      "step": 29505
+    },
+    {
+      "epoch": 0.8103595067424679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8277147573290843e-06,
+      "loss": 0.809,
+      "step": 29506
+    },
+    {
+      "epoch": 0.8103869709703112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8272021493514203e-06,
+      "loss": 0.8401,
+      "step": 29507
+    },
+    {
+      "epoch": 0.8104144351981544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8266896060400985e-06,
+      "loss": 0.8558,
+      "step": 29508
+    },
+    {
+      "epoch": 0.8104418994259976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826177127399178e-06,
+      "loss": 0.8211,
+      "step": 29509
+    },
+    {
+      "epoch": 0.8104693636538409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8256647134327065e-06,
+      "loss": 0.8557,
+      "step": 29510
+    },
+    {
+      "epoch": 0.8104968278816841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8251523641447432e-06,
+      "loss": 0.857,
+      "step": 29511
+    },
+    {
+      "epoch": 0.8105242921095274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8246400795393437e-06,
+      "loss": 0.8775,
+      "step": 29512
+    },
+    {
+      "epoch": 0.8105517563373705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8241278596205558e-06,
+      "loss": 0.8181,
+      "step": 29513
+    },
+    {
+      "epoch": 0.8105792205652138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8236157043924352e-06,
+      "loss": 0.6917,
+      "step": 29514
+    },
+    {
+      "epoch": 0.8106066847930571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8231036138590363e-06,
+      "loss": 0.8319,
+      "step": 29515
+    },
+    {
+      "epoch": 0.8106341490209003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8225915880244072e-06,
+      "loss": 0.8387,
+      "step": 29516
+    },
+    {
+      "epoch": 0.8106616132487435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822079626892601e-06,
+      "loss": 0.7376,
+      "step": 29517
+    },
+    {
+      "epoch": 0.8106890774765867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8215677304676692e-06,
+      "loss": 0.878,
+      "step": 29518
+    },
+    {
+      "epoch": 0.81071654170443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8210558987536608e-06,
+      "loss": 0.8507,
+      "step": 29519
+    },
+    {
+      "epoch": 0.8107440059322732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205441317546291e-06,
+      "loss": 0.9305,
+      "step": 29520
+    },
+    {
+      "epoch": 0.8107714701601164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8200324294746209e-06,
+      "loss": 0.8615,
+      "step": 29521
+    },
+    {
+      "epoch": 0.8107989343879597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8195207919176817e-06,
+      "loss": 0.9132,
+      "step": 29522
+    },
+    {
+      "epoch": 0.810826398615803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8190092190878627e-06,
+      "loss": 0.802,
+      "step": 29523
+    },
+    {
+      "epoch": 0.8108538628436461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818497710989212e-06,
+      "loss": 0.8586,
+      "step": 29524
+    },
+    {
+      "epoch": 0.8108813270714894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8179862676257766e-06,
+      "loss": 0.783,
+      "step": 29525
+    },
+    {
+      "epoch": 0.8109087912993326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8174748890016037e-06,
+      "loss": 0.8337,
+      "step": 29526
+    },
+    {
+      "epoch": 0.8109362555271759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8169635751207414e-06,
+      "loss": 0.8403,
+      "step": 29527
+    },
+    {
+      "epoch": 0.8109637197550191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8164523259872324e-06,
+      "loss": 0.7699,
+      "step": 29528
+    },
+    {
+      "epoch": 0.8109911839828623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8159411416051209e-06,
+      "loss": 0.8691,
+      "step": 29529
+    },
+    {
+      "epoch": 0.8110186482107056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8154300219784526e-06,
+      "loss": 0.8167,
+      "step": 29530
+    },
+    {
+      "epoch": 0.8110461124385487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8149189671112733e-06,
+      "loss": 0.7698,
+      "step": 29531
+    },
+    {
+      "epoch": 0.811073576666392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814407977007625e-06,
+      "loss": 0.8044,
+      "step": 29532
+    },
+    {
+      "epoch": 0.8111010408942353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8138970516715547e-06,
+      "loss": 0.8507,
+      "step": 29533
+    },
+    {
+      "epoch": 0.8111285051220785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8133861911070993e-06,
+      "loss": 0.864,
+      "step": 29534
+    },
+    {
+      "epoch": 0.8111559693499217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8128753953183032e-06,
+      "loss": 0.8518,
+      "step": 29535
+    },
+    {
+      "epoch": 0.811183433577765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8123646643092108e-06,
+      "loss": 0.8243,
+      "step": 29536
+    },
+    {
+      "epoch": 0.8112108978056082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811853998083858e-06,
+      "loss": 0.8362,
+      "step": 29537
+    },
+    {
+      "epoch": 0.8112383620334515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811343396646289e-06,
+      "loss": 0.845,
+      "step": 29538
+    },
+    {
+      "epoch": 0.8112658262612946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8108328600005454e-06,
+      "loss": 0.8966,
+      "step": 29539
+    },
+    {
+      "epoch": 0.8112932904891379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8103223881506615e-06,
+      "loss": 0.7447,
+      "step": 29540
+    },
+    {
+      "epoch": 0.8113207547169812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8098119811006787e-06,
+      "loss": 0.851,
+      "step": 29541
+    },
+    {
+      "epoch": 0.8113482189448243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809301638854637e-06,
+      "loss": 0.7609,
+      "step": 29542
+    },
+    {
+      "epoch": 0.8113756831726676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808791361416573e-06,
+      "loss": 0.8591,
+      "step": 29543
+    },
+    {
+      "epoch": 0.8114031474005108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808281148790527e-06,
+      "loss": 0.8798,
+      "step": 29544
+    },
+    {
+      "epoch": 0.8114306116283541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807771000980535e-06,
+      "loss": 0.8493,
+      "step": 29545
+    },
+    {
+      "epoch": 0.8114580758561973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8072609179906286e-06,
+      "loss": 0.7899,
+      "step": 29546
+    },
+    {
+      "epoch": 0.8114855400840405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8067508998248473e-06,
+      "loss": 0.8327,
+      "step": 29547
+    },
+    {
+      "epoch": 0.8115130043118838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8062409464872278e-06,
+      "loss": 0.784,
+      "step": 29548
+    },
+    {
+      "epoch": 0.8115404685397271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805731057981802e-06,
+      "loss": 0.8234,
+      "step": 29549
+    },
+    {
+      "epoch": 0.8115679327675702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8052212343126107e-06,
+      "loss": 0.7566,
+      "step": 29550
+    },
+    {
+      "epoch": 0.8115953969954135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8047114754836792e-06,
+      "loss": 0.8337,
+      "step": 29551
+    },
+    {
+      "epoch": 0.8116228612232567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042017814990488e-06,
+      "loss": 0.8457,
+      "step": 29552
+    },
+    {
+      "epoch": 0.8116503254510999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803692152362746e-06,
+      "loss": 0.8095,
+      "step": 29553
+    },
+    {
+      "epoch": 0.8116777896789432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8031825880788044e-06,
+      "loss": 0.8131,
+      "step": 29554
+    },
+    {
+      "epoch": 0.8117052539067864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802673088651259e-06,
+      "loss": 0.8198,
+      "step": 29555
+    },
+    {
+      "epoch": 0.8117327181346297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8021636540841415e-06,
+      "loss": 0.7566,
+      "step": 29556
+    },
+    {
+      "epoch": 0.8117601823624728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8016542843814776e-06,
+      "loss": 0.8497,
+      "step": 29557
+    },
+    {
+      "epoch": 0.8117876465903161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8011449795473012e-06,
+      "loss": 0.7692,
+      "step": 29558
+    },
+    {
+      "epoch": 0.8118151108181594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8006357395856423e-06,
+      "loss": 0.7352,
+      "step": 29559
+    },
+    {
+      "epoch": 0.8118425750460025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8001265645005306e-06,
+      "loss": 0.813,
+      "step": 29560
+    },
+    {
+      "epoch": 0.8118700392738458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7996174542959922e-06,
+      "loss": 0.7716,
+      "step": 29561
+    },
+    {
+      "epoch": 0.8118975035016891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7991084089760603e-06,
+      "loss": 0.8473,
+      "step": 29562
+    },
+    {
+      "epoch": 0.8119249677295323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7985994285447562e-06,
+      "loss": 0.7937,
+      "step": 29563
+    },
+    {
+      "epoch": 0.8119524319573755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7980905130061098e-06,
+      "loss": 0.8074,
+      "step": 29564
+    },
+    {
+      "epoch": 0.8119798961852187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7975816623641485e-06,
+      "loss": 0.8801,
+      "step": 29565
+    },
+    {
+      "epoch": 0.812007360413062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7970728766228985e-06,
+      "loss": 0.8618,
+      "step": 29566
+    },
+    {
+      "epoch": 0.8120348246409053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796564155786389e-06,
+      "loss": 0.8247,
+      "step": 29567
+    },
+    {
+      "epoch": 0.8120622888687484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7960554998586376e-06,
+      "loss": 0.9074,
+      "step": 29568
+    },
+    {
+      "epoch": 0.8120897530965917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795546908843676e-06,
+      "loss": 0.8771,
+      "step": 29569
+    },
+    {
+      "epoch": 0.8121172173244349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7950383827455231e-06,
+      "loss": 0.8114,
+      "step": 29570
+    },
+    {
+      "epoch": 0.8121446815522781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7945299215682045e-06,
+      "loss": 0.7044,
+      "step": 29571
+    },
+    {
+      "epoch": 0.8121721457801214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7940215253157444e-06,
+      "loss": 0.7852,
+      "step": 29572
+    },
+    {
+      "epoch": 0.8121996100079646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7935131939921635e-06,
+      "loss": 0.8568,
+      "step": 29573
+    },
+    {
+      "epoch": 0.8122270742358079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7930049276014883e-06,
+      "loss": 0.8478,
+      "step": 29574
+    },
+    {
+      "epoch": 0.8122545384636511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7924967261477344e-06,
+      "loss": 0.9055,
+      "step": 29575
+    },
+    {
+      "epoch": 0.8122820026914943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7919885896349287e-06,
+      "loss": 0.8181,
+      "step": 29576
+    },
+    {
+      "epoch": 0.8123094669193376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7914805180670858e-06,
+      "loss": 0.8267,
+      "step": 29577
+    },
+    {
+      "epoch": 0.8123369311471808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790972511448229e-06,
+      "loss": 0.8292,
+      "step": 29578
+    },
+    {
+      "epoch": 0.812364395375024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7904645697823774e-06,
+      "loss": 0.8481,
+      "step": 29579
+    },
+    {
+      "epoch": 0.8123918596028673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7899566930735535e-06,
+      "loss": 0.8207,
+      "step": 29580
+    },
+    {
+      "epoch": 0.8124193238307105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7894488813257682e-06,
+      "loss": 0.8573,
+      "step": 29581
+    },
+    {
+      "epoch": 0.8124467880585537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7889411345430452e-06,
+      "loss": 0.8228,
+      "step": 29582
+    },
+    {
+      "epoch": 0.8124742522863969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7884334527294e-06,
+      "loss": 0.8193,
+      "step": 29583
+    },
+    {
+      "epoch": 0.8125017165142402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787925835888853e-06,
+      "loss": 0.8408,
+      "step": 29584
+    },
+    {
+      "epoch": 0.8125291807420835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787418284025415e-06,
+      "loss": 0.8184,
+      "step": 29585
+    },
+    {
+      "epoch": 0.8125566449699266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7869107971431076e-06,
+      "loss": 0.8532,
+      "step": 29586
+    },
+    {
+      "epoch": 0.8125841091977699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786403375245942e-06,
+      "loss": 0.8927,
+      "step": 29587
+    },
+    {
+      "epoch": 0.8126115734256132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7858960183379326e-06,
+      "loss": 0.7521,
+      "step": 29588
+    },
+    {
+      "epoch": 0.8126390376534564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7853887264230973e-06,
+      "loss": 0.8491,
+      "step": 29589
+    },
+    {
+      "epoch": 0.8126665018812996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7848814995054476e-06,
+      "loss": 0.7726,
+      "step": 29590
+    },
+    {
+      "epoch": 0.8126939661091428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7843743375889998e-06,
+      "loss": 0.7876,
+      "step": 29591
+    },
+    {
+      "epoch": 0.8127214303369861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7838672406777625e-06,
+      "loss": 0.8332,
+      "step": 29592
+    },
+    {
+      "epoch": 0.8127488945648293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7833602087757528e-06,
+      "loss": 0.7769,
+      "step": 29593
+    },
+    {
+      "epoch": 0.8127763587926725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7828532418869782e-06,
+      "loss": 0.7777,
+      "step": 29594
+    },
+    {
+      "epoch": 0.8128038230205158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7823463400154505e-06,
+      "loss": 0.7935,
+      "step": 29595
+    },
+    {
+      "epoch": 0.812831287248359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7818395031651825e-06,
+      "loss": 0.8375,
+      "step": 29596
+    },
+    {
+      "epoch": 0.8128587514762022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781332731340184e-06,
+      "loss": 0.7506,
+      "step": 29597
+    },
+    {
+      "epoch": 0.8128862157040455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7808260245444663e-06,
+      "loss": 0.7729,
+      "step": 29598
+    },
+    {
+      "epoch": 0.8129136799318887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7803193827820354e-06,
+      "loss": 0.8297,
+      "step": 29599
+    },
+    {
+      "epoch": 0.812941144159732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779812806056903e-06,
+      "loss": 0.7652,
+      "step": 29600
+    },
+    {
+      "epoch": 0.8129686083875752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7793062943730744e-06,
+      "loss": 0.7305,
+      "step": 29601
+    },
+    {
+      "epoch": 0.8129960726154184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778799847734558e-06,
+      "loss": 0.8075,
+      "step": 29602
+    },
+    {
+      "epoch": 0.8130235368432617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7782934661453644e-06,
+      "loss": 0.8616,
+      "step": 29603
+    },
+    {
+      "epoch": 0.8130510010711048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7777871496094944e-06,
+      "loss": 0.7653,
+      "step": 29604
+    },
+    {
+      "epoch": 0.8130784652989481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7772808981309586e-06,
+      "loss": 0.8039,
+      "step": 29605
+    },
+    {
+      "epoch": 0.8131059295267914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7767747117137613e-06,
+      "loss": 0.744,
+      "step": 29606
+    },
+    {
+      "epoch": 0.8131333937546346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7762685903619082e-06,
+      "loss": 0.7649,
+      "step": 29607
+    },
+    {
+      "epoch": 0.8131608579824778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775762534079406e-06,
+      "loss": 0.8346,
+      "step": 29608
+    },
+    {
+      "epoch": 0.813188322210321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775256542870255e-06,
+      "loss": 0.7608,
+      "step": 29609
+    },
+    {
+      "epoch": 0.8132157864381643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7747506167384598e-06,
+      "loss": 0.7094,
+      "step": 29610
+    },
+    {
+      "epoch": 0.8132432506660076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7742447556880228e-06,
+      "loss": 0.8272,
+      "step": 29611
+    },
+    {
+      "epoch": 0.8132707148938507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773738959722947e-06,
+      "loss": 0.7518,
+      "step": 29612
+    },
+    {
+      "epoch": 0.813298179121694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7732332288472365e-06,
+      "loss": 0.7588,
+      "step": 29613
+    },
+    {
+      "epoch": 0.8133256433495373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7727275630648898e-06,
+      "loss": 0.9491,
+      "step": 29614
+    },
+    {
+      "epoch": 0.8133531075773804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7722219623799131e-06,
+      "loss": 0.811,
+      "step": 29615
+    },
+    {
+      "epoch": 0.8133805718052237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7717164267963005e-06,
+      "loss": 0.8341,
+      "step": 29616
+    },
+    {
+      "epoch": 0.8134080360330669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771210956318058e-06,
+      "loss": 0.743,
+      "step": 29617
+    },
+    {
+      "epoch": 0.8134355002609102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7707055509491788e-06,
+      "loss": 0.8487,
+      "step": 29618
+    },
+    {
+      "epoch": 0.8134629644887534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7702002106936655e-06,
+      "loss": 0.8472,
+      "step": 29619
+    },
+    {
+      "epoch": 0.8134904287165966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7696949355555159e-06,
+      "loss": 0.7793,
+      "step": 29620
+    },
+    {
+      "epoch": 0.8135178929444399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7691897255387304e-06,
+      "loss": 0.8452,
+      "step": 29621
+    },
+    {
+      "epoch": 0.8135453571722832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7686845806473018e-06,
+      "loss": 0.9148,
+      "step": 29622
+    },
+    {
+      "epoch": 0.8135728214001263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7681795008852287e-06,
+      "loss": 0.8349,
+      "step": 29623
+    },
+    {
+      "epoch": 0.8136002856279696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767674486256512e-06,
+      "loss": 0.747,
+      "step": 29624
+    },
+    {
+      "epoch": 0.8136277498558128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7671695367651397e-06,
+      "loss": 0.8325,
+      "step": 29625
+    },
+    {
+      "epoch": 0.813655214083656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766664652415112e-06,
+      "loss": 0.8853,
+      "step": 29626
+    },
+    {
+      "epoch": 0.8136826783114993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7661598332104246e-06,
+      "loss": 0.8747,
+      "step": 29627
+    },
+    {
+      "epoch": 0.8137101425393425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7656550791550685e-06,
+      "loss": 0.8291,
+      "step": 29628
+    },
+    {
+      "epoch": 0.8137376067671858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7651503902530387e-06,
+      "loss": 0.9596,
+      "step": 29629
+    },
+    {
+      "epoch": 0.8137650709950289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7646457665083284e-06,
+      "loss": 0.7042,
+      "step": 29630
+    },
+    {
+      "epoch": 0.8137925352228722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764141207924931e-06,
+      "loss": 0.9524,
+      "step": 29631
+    },
+    {
+      "epoch": 0.8138199994507155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7636367145068422e-06,
+      "loss": 0.8156,
+      "step": 29632
+    },
+    {
+      "epoch": 0.8138474636785586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7631322862580491e-06,
+      "loss": 0.8634,
+      "step": 29633
+    },
+    {
+      "epoch": 0.8138749279064019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7626279231825416e-06,
+      "loss": 0.7427,
+      "step": 29634
+    },
+    {
+      "epoch": 0.8139023921342452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7621236252843133e-06,
+      "loss": 0.7459,
+      "step": 29635
+    },
+    {
+      "epoch": 0.8139298563620884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761619392567353e-06,
+      "loss": 0.8296,
+      "step": 29636
+    },
+    {
+      "epoch": 0.8139573205899316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7611152250356523e-06,
+      "loss": 0.8411,
+      "step": 29637
+    },
+    {
+      "epoch": 0.8139847848177748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7606111226931989e-06,
+      "loss": 0.7723,
+      "step": 29638
+    },
+    {
+      "epoch": 0.8140122490456181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760107085543985e-06,
+      "loss": 0.775,
+      "step": 29639
+    },
+    {
+      "epoch": 0.8140397132734614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7596031135919922e-06,
+      "loss": 0.774,
+      "step": 29640
+    },
+    {
+      "epoch": 0.8140671775013045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759099206841215e-06,
+      "loss": 0.8433,
+      "step": 29641
+    },
+    {
+      "epoch": 0.8140946417291478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7585953652956344e-06,
+      "loss": 0.7898,
+      "step": 29642
+    },
+    {
+      "epoch": 0.814122105956991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7580915889592387e-06,
+      "loss": 0.7598,
+      "step": 29643
+    },
+    {
+      "epoch": 0.8141495701848342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7575878778360157e-06,
+      "loss": 0.8445,
+      "step": 29644
+    },
+    {
+      "epoch": 0.8141770344126775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757084231929953e-06,
+      "loss": 0.7294,
+      "step": 29645
+    },
+    {
+      "epoch": 0.8142044986405207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7565806512450302e-06,
+      "loss": 0.7599,
+      "step": 29646
+    },
+    {
+      "epoch": 0.814231962868364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7560771357852347e-06,
+      "loss": 0.7934,
+      "step": 29647
+    },
+    {
+      "epoch": 0.8142594270962072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7555736855545524e-06,
+      "loss": 0.7112,
+      "step": 29648
+    },
+    {
+      "epoch": 0.8142868913240504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7550703005569625e-06,
+      "loss": 0.8927,
+      "step": 29649
+    },
+    {
+      "epoch": 0.8143143555518937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7545669807964494e-06,
+      "loss": 0.7953,
+      "step": 29650
+    },
+    {
+      "epoch": 0.8143418197797369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7540637262769999e-06,
+      "loss": 0.8558,
+      "step": 29651
+    },
+    {
+      "epoch": 0.8143692840075801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7535605370025887e-06,
+      "loss": 0.8324,
+      "step": 29652
+    },
+    {
+      "epoch": 0.8143967482354234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7530574129772015e-06,
+      "loss": 0.7688,
+      "step": 29653
+    },
+    {
+      "epoch": 0.8144242124632666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7525543542048185e-06,
+      "loss": 0.7962,
+      "step": 29654
+    },
+    {
+      "epoch": 0.8144516766911098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7520513606894207e-06,
+      "loss": 0.7426,
+      "step": 29655
+    },
+    {
+      "epoch": 0.814479140918953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7515484324349895e-06,
+      "loss": 0.7764,
+      "step": 29656
+    },
+    {
+      "epoch": 0.8145066051467963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751045569445502e-06,
+      "loss": 0.7983,
+      "step": 29657
+    },
+    {
+      "epoch": 0.8145340693746396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7505427717249357e-06,
+      "loss": 0.7234,
+      "step": 29658
+    },
+    {
+      "epoch": 0.8145615336024827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75004003927727e-06,
+      "loss": 0.8511,
+      "step": 29659
+    },
+    {
+      "epoch": 0.814588997830326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7495373721064835e-06,
+      "loss": 0.7789,
+      "step": 29660
+    },
+    {
+      "epoch": 0.8146164620581693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7490347702165523e-06,
+      "loss": 0.8398,
+      "step": 29661
+    },
+    {
+      "epoch": 0.8146439262860125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7485322336114574e-06,
+      "loss": 0.7867,
+      "step": 29662
+    },
+    {
+      "epoch": 0.8146713905138557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7480297622951692e-06,
+      "loss": 0.8741,
+      "step": 29663
+    },
+    {
+      "epoch": 0.8146988547416989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7475273562716667e-06,
+      "loss": 0.7656,
+      "step": 29664
+    },
+    {
+      "epoch": 0.8147263189695422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7470250155449253e-06,
+      "loss": 0.7807,
+      "step": 29665
+    },
+    {
+      "epoch": 0.8147537831973855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7465227401189177e-06,
+      "loss": 0.8308,
+      "step": 29666
+    },
+    {
+      "epoch": 0.8147812474252286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7460205299976186e-06,
+      "loss": 0.8662,
+      "step": 29667
+    },
+    {
+      "epoch": 0.8148087116530719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7455183851850055e-06,
+      "loss": 0.8161,
+      "step": 29668
+    },
+    {
+      "epoch": 0.8148361758809151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7450163056850455e-06,
+      "loss": 0.8561,
+      "step": 29669
+    },
+    {
+      "epoch": 0.8148636401087583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445142915017144e-06,
+      "loss": 0.8602,
+      "step": 29670
+    },
+    {
+      "epoch": 0.8148911043366016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7440123426389833e-06,
+      "loss": 0.8414,
+      "step": 29671
+    },
+    {
+      "epoch": 0.8149185685644448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7435104591008279e-06,
+      "loss": 0.8049,
+      "step": 29672
+    },
+    {
+      "epoch": 0.814946032792288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7430086408912138e-06,
+      "loss": 0.89,
+      "step": 29673
+    },
+    {
+      "epoch": 0.8149734970201313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742506888014116e-06,
+      "loss": 0.7781,
+      "step": 29674
+    },
+    {
+      "epoch": 0.8150009612479745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7420052004734989e-06,
+      "loss": 0.8159,
+      "step": 29675
+    },
+    {
+      "epoch": 0.8150284254758178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741503578273337e-06,
+      "loss": 0.811,
+      "step": 29676
+    },
+    {
+      "epoch": 0.8150558897036609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7410020214175972e-06,
+      "loss": 0.8462,
+      "step": 29677
+    },
+    {
+      "epoch": 0.8150833539315042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7405005299102485e-06,
+      "loss": 0.7638,
+      "step": 29678
+    },
+    {
+      "epoch": 0.8151108181593475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7399991037552587e-06,
+      "loss": 0.87,
+      "step": 29679
+    },
+    {
+      "epoch": 0.8151382823871907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7394977429565985e-06,
+      "loss": 0.7958,
+      "step": 29680
+    },
+    {
+      "epoch": 0.8151657466150339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738996447518232e-06,
+      "loss": 0.8828,
+      "step": 29681
+    },
+    {
+      "epoch": 0.8151932108428771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7384952174441227e-06,
+      "loss": 0.7901,
+      "step": 29682
+    },
+    {
+      "epoch": 0.8152206750707204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.73799405273824e-06,
+      "loss": 0.7527,
+      "step": 29683
+    },
+    {
+      "epoch": 0.8152481392985637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737492953404548e-06,
+      "loss": 0.7718,
+      "step": 29684
+    },
+    {
+      "epoch": 0.8152756035264068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736991919447013e-06,
+      "loss": 0.8275,
+      "step": 29685
+    },
+    {
+      "epoch": 0.8153030677542501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7364909508696015e-06,
+      "loss": 0.8363,
+      "step": 29686
+    },
+    {
+      "epoch": 0.8153305319820934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7359900476762715e-06,
+      "loss": 0.7327,
+      "step": 29687
+    },
+    {
+      "epoch": 0.8153579962099365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735489209870992e-06,
+      "loss": 0.805,
+      "step": 29688
+    },
+    {
+      "epoch": 0.8153854604377798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734988437457721e-06,
+      "loss": 0.8072,
+      "step": 29689
+    },
+    {
+      "epoch": 0.815412924665623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7344877304404229e-06,
+      "loss": 0.9151,
+      "step": 29690
+    },
+    {
+      "epoch": 0.8154403888934663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7339870888230592e-06,
+      "loss": 0.896,
+      "step": 29691
+    },
+    {
+      "epoch": 0.8154678531213095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7334865126095946e-06,
+      "loss": 0.808,
+      "step": 29692
+    },
+    {
+      "epoch": 0.8154953173491527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7329860018039845e-06,
+      "loss": 0.864,
+      "step": 29693
+    },
+    {
+      "epoch": 0.815522781576996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7324855564101916e-06,
+      "loss": 0.8939,
+      "step": 29694
+    },
+    {
+      "epoch": 0.8155502458048391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731985176432175e-06,
+      "loss": 0.8779,
+      "step": 29695
+    },
+    {
+      "epoch": 0.8155777100326824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7314848618738967e-06,
+      "loss": 0.7897,
+      "step": 29696
+    },
+    {
+      "epoch": 0.8156051742605257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7309846127393114e-06,
+      "loss": 0.7779,
+      "step": 29697
+    },
+    {
+      "epoch": 0.8156326384883689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7304844290323807e-06,
+      "loss": 0.8437,
+      "step": 29698
+    },
+    {
+      "epoch": 0.8156601027162121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729984310757058e-06,
+      "loss": 0.7975,
+      "step": 29699
+    },
+    {
+      "epoch": 0.8156875669440554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7294842579173033e-06,
+      "loss": 0.8068,
+      "step": 29700
+    },
+    {
+      "epoch": 0.8157150311718986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7289842705170734e-06,
+      "loss": 0.8326,
+      "step": 29701
+    },
+    {
+      "epoch": 0.8157424953997419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7284843485603231e-06,
+      "loss": 0.8806,
+      "step": 29702
+    },
+    {
+      "epoch": 0.815769959627585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7279844920510093e-06,
+      "loss": 0.8246,
+      "step": 29703
+    },
+    {
+      "epoch": 0.8157974238554283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7274847009930872e-06,
+      "loss": 0.8508,
+      "step": 29704
+    },
+    {
+      "epoch": 0.8158248880832716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7269849753905122e-06,
+      "loss": 0.9174,
+      "step": 29705
+    },
+    {
+      "epoch": 0.8158523523111147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264853152472338e-06,
+      "loss": 0.8167,
+      "step": 29706
+    },
+    {
+      "epoch": 0.815879816538958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7259857205672092e-06,
+      "loss": 0.853,
+      "step": 29707
+    },
+    {
+      "epoch": 0.8159072807668012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7254861913543897e-06,
+      "loss": 0.7907,
+      "step": 29708
+    },
+    {
+      "epoch": 0.8159347449946445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7249867276127297e-06,
+      "loss": 0.804,
+      "step": 29709
+    },
+    {
+      "epoch": 0.8159622092224877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7244873293461818e-06,
+      "loss": 0.8222,
+      "step": 29710
+    },
+    {
+      "epoch": 0.8159896734503309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7239879965586926e-06,
+      "loss": 0.7725,
+      "step": 29711
+    },
+    {
+      "epoch": 0.8160171376781742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7234887292542202e-06,
+      "loss": 0.7392,
+      "step": 29712
+    },
+    {
+      "epoch": 0.8160446019060175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722989527436707e-06,
+      "loss": 0.7992,
+      "step": 29713
+    },
+    {
+      "epoch": 0.8160720661338606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7224903911101088e-06,
+      "loss": 0.7114,
+      "step": 29714
+    },
+    {
+      "epoch": 0.8160995303617039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721991320278371e-06,
+      "loss": 0.8701,
+      "step": 29715
+    },
+    {
+      "epoch": 0.8161269945895471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7214923149454476e-06,
+      "loss": 0.809,
+      "step": 29716
+    },
+    {
+      "epoch": 0.8161544588173903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720993375115282e-06,
+      "loss": 0.7425,
+      "step": 29717
+    },
+    {
+      "epoch": 0.8161819230452336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7204945007918238e-06,
+      "loss": 0.8788,
+      "step": 29718
+    },
+    {
+      "epoch": 0.8162093872730768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7199956919790195e-06,
+      "loss": 0.8622,
+      "step": 29719
+    },
+    {
+      "epoch": 0.8162368515009201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71949694868082e-06,
+      "loss": 0.8776,
+      "step": 29720
+    },
+    {
+      "epoch": 0.8162643157287632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7189982709011643e-06,
+      "loss": 0.8607,
+      "step": 29721
+    },
+    {
+      "epoch": 0.8162917799566065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718499658644006e-06,
+      "loss": 0.9027,
+      "step": 29722
+    },
+    {
+      "epoch": 0.8163192441844498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7180011119132834e-06,
+      "loss": 0.7904,
+      "step": 29723
+    },
+    {
+      "epoch": 0.816346708412293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717502630712944e-06,
+      "loss": 0.7719,
+      "step": 29724
+    },
+    {
+      "epoch": 0.8163741726401362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7170042150469313e-06,
+      "loss": 0.8473,
+      "step": 29725
+    },
+    {
+      "epoch": 0.8164016368679795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7165058649191911e-06,
+      "loss": 0.8425,
+      "step": 29726
+    },
+    {
+      "epoch": 0.8164291010958227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7160075803336674e-06,
+      "loss": 0.818,
+      "step": 29727
+    },
+    {
+      "epoch": 0.816456565323666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7155093612942975e-06,
+      "loss": 0.7347,
+      "step": 29728
+    },
+    {
+      "epoch": 0.8164840295515091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7150112078050296e-06,
+      "loss": 0.8777,
+      "step": 29729
+    },
+    {
+      "epoch": 0.8165114937793524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7145131198697996e-06,
+      "loss": 0.8687,
+      "step": 29730
+    },
+    {
+      "epoch": 0.8165389580071957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7140150974925518e-06,
+      "loss": 0.7745,
+      "step": 29731
+    },
+    {
+      "epoch": 0.8165664222350388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7135171406772267e-06,
+      "loss": 0.9035,
+      "step": 29732
+    },
+    {
+      "epoch": 0.8165938864628821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7130192494277654e-06,
+      "loss": 0.838,
+      "step": 29733
+    },
+    {
+      "epoch": 0.8166213506907253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7125214237481035e-06,
+      "loss": 0.8685,
+      "step": 29734
+    },
+    {
+      "epoch": 0.8166488149185686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7120236636421817e-06,
+      "loss": 0.7623,
+      "step": 29735
+    },
+    {
+      "epoch": 0.8166762791464118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711525969113943e-06,
+      "loss": 0.8136,
+      "step": 29736
+    },
+    {
+      "epoch": 0.816703743374255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7110283401673177e-06,
+      "loss": 0.8484,
+      "step": 29737
+    },
+    {
+      "epoch": 0.8167312076020983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7105307768062485e-06,
+      "loss": 0.6883,
+      "step": 29738
+    },
+    {
+      "epoch": 0.8167586718299416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7100332790346719e-06,
+      "loss": 0.7851,
+      "step": 29739
+    },
+    {
+      "epoch": 0.8167861360577847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7095358468565215e-06,
+      "loss": 0.8424,
+      "step": 29740
+    },
+    {
+      "epoch": 0.816813600285628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7090384802757343e-06,
+      "loss": 0.7355,
+      "step": 29741
+    },
+    {
+      "epoch": 0.8168410645134712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.708541179296247e-06,
+      "loss": 0.8198,
+      "step": 29742
+    },
+    {
+      "epoch": 0.8168685287413144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.708043943921993e-06,
+      "loss": 0.8202,
+      "step": 29743
+    },
+    {
+      "epoch": 0.8168959929691577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7075467741569107e-06,
+      "loss": 0.8415,
+      "step": 29744
+    },
+    {
+      "epoch": 0.8169234571970009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7070496700049298e-06,
+      "loss": 0.8577,
+      "step": 29745
+    },
+    {
+      "epoch": 0.8169509214248442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706552631469982e-06,
+      "loss": 0.8135,
+      "step": 29746
+    },
+    {
+      "epoch": 0.8169783856526873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7060556585560017e-06,
+      "loss": 0.8448,
+      "step": 29747
+    },
+    {
+      "epoch": 0.8170058498805306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7055587512669225e-06,
+      "loss": 0.7465,
+      "step": 29748
+    },
+    {
+      "epoch": 0.8170333141083739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7050619096066757e-06,
+      "loss": 0.8506,
+      "step": 29749
+    },
+    {
+      "epoch": 0.817060778336217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7045651335791914e-06,
+      "loss": 0.8503,
+      "step": 29750
+    },
+    {
+      "epoch": 0.8170882425640603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704068423188404e-06,
+      "loss": 0.7063,
+      "step": 29751
+    },
+    {
+      "epoch": 0.8171157067919036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7035717784382387e-06,
+      "loss": 0.7881,
+      "step": 29752
+    },
+    {
+      "epoch": 0.8171431710197468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703075199332629e-06,
+      "loss": 0.8174,
+      "step": 29753
+    },
+    {
+      "epoch": 0.81717063524759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7025786858755e-06,
+      "loss": 0.8616,
+      "step": 29754
+    },
+    {
+      "epoch": 0.8171980994754332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7020822380707835e-06,
+      "loss": 0.9045,
+      "step": 29755
+    },
+    {
+      "epoch": 0.8172255637032765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7015858559224053e-06,
+      "loss": 0.7647,
+      "step": 29756
+    },
+    {
+      "epoch": 0.8172530279311198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7010895394342975e-06,
+      "loss": 0.7941,
+      "step": 29757
+    },
+    {
+      "epoch": 0.8172804921589629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7005932886103815e-06,
+      "loss": 0.873,
+      "step": 29758
+    },
+    {
+      "epoch": 0.8173079563868062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7000971034545854e-06,
+      "loss": 0.7858,
+      "step": 29759
+    },
+    {
+      "epoch": 0.8173354206146494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6996009839708394e-06,
+      "loss": 0.8581,
+      "step": 29760
+    },
+    {
+      "epoch": 0.8173628848424926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6991049301630636e-06,
+      "loss": 0.7401,
+      "step": 29761
+    },
+    {
+      "epoch": 0.8173903490703359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6986089420351836e-06,
+      "loss": 0.8664,
+      "step": 29762
+    },
+    {
+      "epoch": 0.8174178132981791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6981130195911288e-06,
+      "loss": 0.8313,
+      "step": 29763
+    },
+    {
+      "epoch": 0.8174452775260224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697617162834817e-06,
+      "loss": 0.8435,
+      "step": 29764
+    },
+    {
+      "epoch": 0.8174727417538656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697121371770174e-06,
+      "loss": 0.7612,
+      "step": 29765
+    },
+    {
+      "epoch": 0.8175002059817088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6966256464011233e-06,
+      "loss": 0.8905,
+      "step": 29766
+    },
+    {
+      "epoch": 0.8175276702095521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6961299867315873e-06,
+      "loss": 0.8195,
+      "step": 29767
+    },
+    {
+      "epoch": 0.8175551344373952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6956343927654895e-06,
+      "loss": 0.8546,
+      "step": 29768
+    },
+    {
+      "epoch": 0.8175825986652385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695138864506749e-06,
+      "loss": 0.7734,
+      "step": 29769
+    },
+    {
+      "epoch": 0.8176100628930818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694643401959284e-06,
+      "loss": 0.8033,
+      "step": 29770
+    },
+    {
+      "epoch": 0.817637527120925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6941480051270176e-06,
+      "loss": 0.6577,
+      "step": 29771
+    },
+    {
+      "epoch": 0.8176649913487682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6936526740138704e-06,
+      "loss": 0.8513,
+      "step": 29772
+    },
+    {
+      "epoch": 0.8176924555766114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6931574086237602e-06,
+      "loss": 0.7514,
+      "step": 29773
+    },
+    {
+      "epoch": 0.8177199198044547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692662208960607e-06,
+      "loss": 0.8605,
+      "step": 29774
+    },
+    {
+      "epoch": 0.817747384032298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6921670750283303e-06,
+      "loss": 0.7988,
+      "step": 29775
+    },
+    {
+      "epoch": 0.8177748482601411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6916720068308435e-06,
+      "loss": 0.8586,
+      "step": 29776
+    },
+    {
+      "epoch": 0.8178023124879844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911770043720688e-06,
+      "loss": 0.8741,
+      "step": 29777
+    },
+    {
+      "epoch": 0.8178297767158277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6906820676559165e-06,
+      "loss": 0.7772,
+      "step": 29778
+    },
+    {
+      "epoch": 0.8178572409436708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690187196686308e-06,
+      "loss": 0.7983,
+      "step": 29779
+    },
+    {
+      "epoch": 0.8178847051715141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689692391467156e-06,
+      "loss": 0.7922,
+      "step": 29780
+    },
+    {
+      "epoch": 0.8179121693993573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68919765200238e-06,
+      "loss": 0.8541,
+      "step": 29781
+    },
+    {
+      "epoch": 0.8179396336272006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6887029782958896e-06,
+      "loss": 0.7986,
+      "step": 29782
+    },
+    {
+      "epoch": 0.8179670978550438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6882083703515994e-06,
+      "loss": 0.6662,
+      "step": 29783
+    },
+    {
+      "epoch": 0.817994562082887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6877138281734284e-06,
+      "loss": 0.8168,
+      "step": 29784
+    },
+    {
+      "epoch": 0.8180220263107303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6872193517652813e-06,
+      "loss": 0.8375,
+      "step": 29785
+    },
+    {
+      "epoch": 0.8180494905385735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6867249411310782e-06,
+      "loss": 0.7456,
+      "step": 29786
+    },
+    {
+      "epoch": 0.8180769547664167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686230596274725e-06,
+      "loss": 0.8555,
+      "step": 29787
+    },
+    {
+      "epoch": 0.81810441899426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6857363172001362e-06,
+      "loss": 0.6992,
+      "step": 29788
+    },
+    {
+      "epoch": 0.8181318832221032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685242103911221e-06,
+      "loss": 0.7232,
+      "step": 29789
+    },
+    {
+      "epoch": 0.8181593474499465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684747956411893e-06,
+      "loss": 0.7563,
+      "step": 29790
+    },
+    {
+      "epoch": 0.8181868116777897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6842538747060588e-06,
+      "loss": 0.831,
+      "step": 29791
+    },
+    {
+      "epoch": 0.8182142759056329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6837598587976312e-06,
+      "loss": 0.8966,
+      "step": 29792
+    },
+    {
+      "epoch": 0.8182417401334762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6832659086905178e-06,
+      "loss": 0.7551,
+      "step": 29793
+    },
+    {
+      "epoch": 0.8182692043613193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6827720243886235e-06,
+      "loss": 0.8992,
+      "step": 29794
+    },
+    {
+      "epoch": 0.8182966685891626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6822782058958587e-06,
+      "loss": 0.8054,
+      "step": 29795
+    },
+    {
+      "epoch": 0.8183241328170059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681784453216131e-06,
+      "loss": 0.7182,
+      "step": 29796
+    },
+    {
+      "epoch": 0.818351597044849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6812907663533473e-06,
+      "loss": 0.8597,
+      "step": 29797
+    },
+    {
+      "epoch": 0.8183790612726923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6807971453114146e-06,
+      "loss": 0.7473,
+      "step": 29798
+    },
+    {
+      "epoch": 0.8184065255005355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6803035900942355e-06,
+      "loss": 0.8501,
+      "step": 29799
+    },
+    {
+      "epoch": 0.8184339897283788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6798101007057178e-06,
+      "loss": 0.7506,
+      "step": 29800
+    },
+    {
+      "epoch": 0.818461453956222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679316677149767e-06,
+      "loss": 0.7466,
+      "step": 29801
+    },
+    {
+      "epoch": 0.8184889181840652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678823319430284e-06,
+      "loss": 0.7995,
+      "step": 29802
+    },
+    {
+      "epoch": 0.8185163824119085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6783300275511737e-06,
+      "loss": 0.7114,
+      "step": 29803
+    },
+    {
+      "epoch": 0.8185438466397518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6778368015163426e-06,
+      "loss": 0.8413,
+      "step": 29804
+    },
+    {
+      "epoch": 0.8185713108675949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6773436413296883e-06,
+      "loss": 0.8455,
+      "step": 29805
+    },
+    {
+      "epoch": 0.8185987750954382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6768505469951146e-06,
+      "loss": 0.8364,
+      "step": 29806
+    },
+    {
+      "epoch": 0.8186262393232814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6763575185165237e-06,
+      "loss": 0.7638,
+      "step": 29807
+    },
+    {
+      "epoch": 0.8186537035511247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675864555897818e-06,
+      "loss": 0.8552,
+      "step": 29808
+    },
+    {
+      "epoch": 0.8186811677789679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6753716591428948e-06,
+      "loss": 0.875,
+      "step": 29809
+    },
+    {
+      "epoch": 0.8187086320068111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674878828255657e-06,
+      "loss": 0.7984,
+      "step": 29810
+    },
+    {
+      "epoch": 0.8187360962346544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674386063240002e-06,
+      "loss": 0.7581,
+      "step": 29811
+    },
+    {
+      "epoch": 0.8187635604624975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673893364099829e-06,
+      "loss": 0.7636,
+      "step": 29812
+    },
+    {
+      "epoch": 0.8187910246903408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6734007308390355e-06,
+      "loss": 0.7653,
+      "step": 29813
+    },
+    {
+      "epoch": 0.8188184889181841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6729081634615218e-06,
+      "loss": 0.8024,
+      "step": 29814
+    },
+    {
+      "epoch": 0.8188459531460273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6724156619711851e-06,
+      "loss": 0.8016,
+      "step": 29815
+    },
+    {
+      "epoch": 0.8188734173738705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671923226371923e-06,
+      "loss": 0.7822,
+      "step": 29816
+    },
+    {
+      "epoch": 0.8189008816017138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6714308566676306e-06,
+      "loss": 0.7654,
+      "step": 29817
+    },
+    {
+      "epoch": 0.818928345829557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6709385528622012e-06,
+      "loss": 0.7094,
+      "step": 29818
+    },
+    {
+      "epoch": 0.8189558100574003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6704463149595318e-06,
+      "loss": 0.7403,
+      "step": 29819
+    },
+    {
+      "epoch": 0.8189832742852434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669954142963518e-06,
+      "loss": 0.7143,
+      "step": 29820
+    },
+    {
+      "epoch": 0.8190107385130867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6694620368780534e-06,
+      "loss": 0.7712,
+      "step": 29821
+    },
+    {
+      "epoch": 0.81903820274093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668969996707035e-06,
+      "loss": 0.8043,
+      "step": 29822
+    },
+    {
+      "epoch": 0.8190656669687731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6684780224543507e-06,
+      "loss": 0.8427,
+      "step": 29823
+    },
+    {
+      "epoch": 0.8190931311966164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6679861141238985e-06,
+      "loss": 0.9485,
+      "step": 29824
+    },
+    {
+      "epoch": 0.8191205954244597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6674942717195642e-06,
+      "loss": 0.7773,
+      "step": 29825
+    },
+    {
+      "epoch": 0.8191480596523029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6670024952452435e-06,
+      "loss": 0.7907,
+      "step": 29826
+    },
+    {
+      "epoch": 0.8191755238801461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6665107847048267e-06,
+      "loss": 0.8278,
+      "step": 29827
+    },
+    {
+      "epoch": 0.8192029881079893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6660191401022074e-06,
+      "loss": 0.8151,
+      "step": 29828
+    },
+    {
+      "epoch": 0.8192304523358326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6655275614412702e-06,
+      "loss": 0.7901,
+      "step": 29829
+    },
+    {
+      "epoch": 0.8192579165636759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6650360487259077e-06,
+      "loss": 0.8325,
+      "step": 29830
+    },
+    {
+      "epoch": 0.819285380791519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6645446019600087e-06,
+      "loss": 0.7827,
+      "step": 29831
+    },
+    {
+      "epoch": 0.8193128450193623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6640532211474637e-06,
+      "loss": 0.791,
+      "step": 29832
+    },
+    {
+      "epoch": 0.8193403092472055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663561906292156e-06,
+      "loss": 0.77,
+      "step": 29833
+    },
+    {
+      "epoch": 0.8193677734750487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6630706573979793e-06,
+      "loss": 0.7385,
+      "step": 29834
+    },
+    {
+      "epoch": 0.819395237702892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662579474468814e-06,
+      "loss": 0.8049,
+      "step": 29835
+    },
+    {
+      "epoch": 0.8194227019307352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662088357508549e-06,
+      "loss": 0.8051,
+      "step": 29836
+    },
+    {
+      "epoch": 0.8194501661585785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661597306521071e-06,
+      "loss": 0.7482,
+      "step": 29837
+    },
+    {
+      "epoch": 0.8194776303864217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6611063215102652e-06,
+      "loss": 0.8065,
+      "step": 29838
+    },
+    {
+      "epoch": 0.8195050946142649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660615402480016e-06,
+      "loss": 0.7735,
+      "step": 29839
+    },
+    {
+      "epoch": 0.8195325588421082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6601245494342111e-06,
+      "loss": 0.7753,
+      "step": 29840
+    },
+    {
+      "epoch": 0.8195600230699513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659633762376731e-06,
+      "loss": 0.8405,
+      "step": 29841
+    },
+    {
+      "epoch": 0.8195874872977946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6591430413114573e-06,
+      "loss": 0.8127,
+      "step": 29842
+    },
+    {
+      "epoch": 0.8196149515256379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6586523862422743e-06,
+      "loss": 0.8434,
+      "step": 29843
+    },
+    {
+      "epoch": 0.8196424157534811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6581617971730646e-06,
+      "loss": 0.8861,
+      "step": 29844
+    },
+    {
+      "epoch": 0.8196698799813243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6576712741077127e-06,
+      "loss": 0.8334,
+      "step": 29845
+    },
+    {
+      "epoch": 0.8196973442091675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6571808170500948e-06,
+      "loss": 0.79,
+      "step": 29846
+    },
+    {
+      "epoch": 0.8197248084370108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6566904260040939e-06,
+      "loss": 0.8675,
+      "step": 29847
+    },
+    {
+      "epoch": 0.8197522726648541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6562001009735929e-06,
+      "loss": 0.7548,
+      "step": 29848
+    },
+    {
+      "epoch": 0.8197797368926972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6557098419624663e-06,
+      "loss": 0.7305,
+      "step": 29849
+    },
+    {
+      "epoch": 0.8198072011205405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6552196489745952e-06,
+      "loss": 0.7379,
+      "step": 29850
+    },
+    {
+      "epoch": 0.8198346653483838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6547295220138627e-06,
+      "loss": 0.7006,
+      "step": 29851
+    },
+    {
+      "epoch": 0.819862129576227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6542394610841384e-06,
+      "loss": 0.7817,
+      "step": 29852
+    },
+    {
+      "epoch": 0.8198895938040702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6537494661893062e-06,
+      "loss": 0.7583,
+      "step": 29853
+    },
+    {
+      "epoch": 0.8199170580319134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6532595373332395e-06,
+      "loss": 0.8349,
+      "step": 29854
+    },
+    {
+      "epoch": 0.8199445222597567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6527696745198173e-06,
+      "loss": 0.8364,
+      "step": 29855
+    },
+    {
+      "epoch": 0.8199719864876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652279877752917e-06,
+      "loss": 0.8657,
+      "step": 29856
+    },
+    {
+      "epoch": 0.8199994507154431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6517901470364128e-06,
+      "loss": 0.7504,
+      "step": 29857
+    },
+    {
+      "epoch": 0.8200269149432864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6513004823741751e-06,
+      "loss": 0.8424,
+      "step": 29858
+    },
+    {
+      "epoch": 0.8200543791711296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650810883770083e-06,
+      "loss": 0.8242,
+      "step": 29859
+    },
+    {
+      "epoch": 0.8200818433989728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6503213512280091e-06,
+      "loss": 0.7553,
+      "step": 29860
+    },
+    {
+      "epoch": 0.8201093076268161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649831884751827e-06,
+      "loss": 0.766,
+      "step": 29861
+    },
+    {
+      "epoch": 0.8201367718546593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6493424843454099e-06,
+      "loss": 0.835,
+      "step": 29862
+    },
+    {
+      "epoch": 0.8201642360825026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6488531500126315e-06,
+      "loss": 0.7774,
+      "step": 29863
+    },
+    {
+      "epoch": 0.8201917003103458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64836388175736e-06,
+      "loss": 0.8315,
+      "step": 29864
+    },
+    {
+      "epoch": 0.820219164538189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6478746795834698e-06,
+      "loss": 0.8428,
+      "step": 29865
+    },
+    {
+      "epoch": 0.8202466287660323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6473855434948293e-06,
+      "loss": 0.8848,
+      "step": 29866
+    },
+    {
+      "epoch": 0.8202740929938754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6468964734953096e-06,
+      "loss": 0.7983,
+      "step": 29867
+    },
+    {
+      "epoch": 0.8203015572217187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646407469588781e-06,
+      "loss": 0.9188,
+      "step": 29868
+    },
+    {
+      "epoch": 0.820329021449562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6459185317791138e-06,
+      "loss": 0.8568,
+      "step": 29869
+    },
+    {
+      "epoch": 0.8203564856774052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6454296600701735e-06,
+      "loss": 0.7885,
+      "step": 29870
+    },
+    {
+      "epoch": 0.8203839499052484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6449408544658307e-06,
+      "loss": 0.7559,
+      "step": 29871
+    },
+    {
+      "epoch": 0.8204114141330916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6444521149699533e-06,
+      "loss": 0.791,
+      "step": 29872
+    },
+    {
+      "epoch": 0.8204388783609349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6439634415864059e-06,
+      "loss": 0.8431,
+      "step": 29873
+    },
+    {
+      "epoch": 0.8204663425887782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6434748343190566e-06,
+      "loss": 0.7466,
+      "step": 29874
+    },
+    {
+      "epoch": 0.8204938068166213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6429862931717733e-06,
+      "loss": 0.8063,
+      "step": 29875
+    },
+    {
+      "epoch": 0.8205212710444646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6424978181484174e-06,
+      "loss": 0.8316,
+      "step": 29876
+    },
+    {
+      "epoch": 0.8205487352723079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6420094092528561e-06,
+      "loss": 0.8228,
+      "step": 29877
+    },
+    {
+      "epoch": 0.820576199500151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6415210664889536e-06,
+      "loss": 0.7968,
+      "step": 29878
+    },
+    {
+      "epoch": 0.8206036637279943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641032789860575e-06,
+      "loss": 0.9252,
+      "step": 29879
+    },
+    {
+      "epoch": 0.8206311279558375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6405445793715847e-06,
+      "loss": 0.7508,
+      "step": 29880
+    },
+    {
+      "epoch": 0.8206585921836808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.640056435025844e-06,
+      "loss": 0.6655,
+      "step": 29881
+    },
+    {
+      "epoch": 0.820686056411524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6395683568272136e-06,
+      "loss": 0.7425,
+      "step": 29882
+    },
+    {
+      "epoch": 0.8207135206393672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6390803447795555e-06,
+      "loss": 0.8156,
+      "step": 29883
+    },
+    {
+      "epoch": 0.8207409848672105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6385923988867336e-06,
+      "loss": 0.7417,
+      "step": 29884
+    },
+    {
+      "epoch": 0.8207684490950536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6381045191526068e-06,
+      "loss": 0.8472,
+      "step": 29885
+    },
+    {
+      "epoch": 0.8207959133228969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6376167055810355e-06,
+      "loss": 0.7886,
+      "step": 29886
+    },
+    {
+      "epoch": 0.8208233775507402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6371289581758842e-06,
+      "loss": 0.7097,
+      "step": 29887
+    },
+    {
+      "epoch": 0.8208508417785834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6366412769410044e-06,
+      "loss": 0.7967,
+      "step": 29888
+    },
+    {
+      "epoch": 0.8208783060064266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6361536618802608e-06,
+      "loss": 0.8936,
+      "step": 29889
+    },
+    {
+      "epoch": 0.8209057702342699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635666112997507e-06,
+      "loss": 0.8263,
+      "step": 29890
+    },
+    {
+      "epoch": 0.8209332344621131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6351786302966032e-06,
+      "loss": 0.8591,
+      "step": 29891
+    },
+    {
+      "epoch": 0.8209606986899564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6346912137814063e-06,
+      "loss": 0.7905,
+      "step": 29892
+    },
+    {
+      "epoch": 0.8209881629177995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6342038634557756e-06,
+      "loss": 0.7967,
+      "step": 29893
+    },
+    {
+      "epoch": 0.8210156271456428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6337165793235622e-06,
+      "loss": 0.7942,
+      "step": 29894
+    },
+    {
+      "epoch": 0.8210430913734861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633229361388624e-06,
+      "loss": 0.7079,
+      "step": 29895
+    },
+    {
+      "epoch": 0.8210705556013292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6327422096548174e-06,
+      "loss": 0.9027,
+      "step": 29896
+    },
+    {
+      "epoch": 0.8210980198291725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322551241259943e-06,
+      "loss": 0.8146,
+      "step": 29897
+    },
+    {
+      "epoch": 0.8211254840570157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6317681048060096e-06,
+      "loss": 0.87,
+      "step": 29898
+    },
+    {
+      "epoch": 0.821152948284859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6312811516987203e-06,
+      "loss": 0.7692,
+      "step": 29899
+    },
+    {
+      "epoch": 0.8211804125127022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6307942648079734e-06,
+      "loss": 0.885,
+      "step": 29900
+    },
+    {
+      "epoch": 0.8212078767405454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6303074441376231e-06,
+      "loss": 0.9211,
+      "step": 29901
+    },
+    {
+      "epoch": 0.8212353409683887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6298206896915237e-06,
+      "loss": 0.8033,
+      "step": 29902
+    },
+    {
+      "epoch": 0.821262805196232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6293340014735249e-06,
+      "loss": 0.82,
+      "step": 29903
+    },
+    {
+      "epoch": 0.8212902694240751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6288473794874804e-06,
+      "loss": 0.8269,
+      "step": 29904
+    },
+    {
+      "epoch": 0.8213177336519184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6283608237372384e-06,
+      "loss": 0.7481,
+      "step": 29905
+    },
+    {
+      "epoch": 0.8213451978797616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6278743342266457e-06,
+      "loss": 0.8145,
+      "step": 29906
+    },
+    {
+      "epoch": 0.8213726621076048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6273879109595537e-06,
+      "loss": 0.8576,
+      "step": 29907
+    },
+    {
+      "epoch": 0.8214001263354481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6269015539398126e-06,
+      "loss": 0.8044,
+      "step": 29908
+    },
+    {
+      "epoch": 0.8214275905632913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6264152631712705e-06,
+      "loss": 0.7766,
+      "step": 29909
+    },
+    {
+      "epoch": 0.8214550547911346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6259290386577754e-06,
+      "loss": 0.7838,
+      "step": 29910
+    },
+    {
+      "epoch": 0.8214825190189777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6254428804031718e-06,
+      "loss": 0.9449,
+      "step": 29911
+    },
+    {
+      "epoch": 0.821509983246821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6249567884113082e-06,
+      "loss": 0.7997,
+      "step": 29912
+    },
+    {
+      "epoch": 0.8215374474746643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6244707626860335e-06,
+      "loss": 0.778,
+      "step": 29913
+    },
+    {
+      "epoch": 0.8215649117025074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6239848032311867e-06,
+      "loss": 0.8708,
+      "step": 29914
+    },
+    {
+      "epoch": 0.8215923759303507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6234989100506181e-06,
+      "loss": 0.8226,
+      "step": 29915
+    },
+    {
+      "epoch": 0.821619840158194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6230130831481728e-06,
+      "loss": 0.7883,
+      "step": 29916
+    },
+    {
+      "epoch": 0.8216473043860372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6225273225276904e-06,
+      "loss": 0.784,
+      "step": 29917
+    },
+    {
+      "epoch": 0.8216747686138804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6220416281930173e-06,
+      "loss": 0.8483,
+      "step": 29918
+    },
+    {
+      "epoch": 0.8217022328417236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6215560001479958e-06,
+      "loss": 0.8214,
+      "step": 29919
+    },
+    {
+      "epoch": 0.8217296970695669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6210704383964704e-06,
+      "loss": 0.7173,
+      "step": 29920
+    },
+    {
+      "epoch": 0.8217571612974102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6205849429422803e-06,
+      "loss": 0.8212,
+      "step": 29921
+    },
+    {
+      "epoch": 0.8217846255252533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6200995137892695e-06,
+      "loss": 0.8666,
+      "step": 29922
+    },
+    {
+      "epoch": 0.8218120897530966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6196141509412755e-06,
+      "loss": 0.786,
+      "step": 29923
+    },
+    {
+      "epoch": 0.8218395539809398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6191288544021411e-06,
+      "loss": 0.8884,
+      "step": 29924
+    },
+    {
+      "epoch": 0.821867018208783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6186436241757042e-06,
+      "loss": 0.8151,
+      "step": 29925
+    },
+    {
+      "epoch": 0.8218944824366263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6181584602658062e-06,
+      "loss": 0.8255,
+      "step": 29926
+    },
+    {
+      "epoch": 0.8219219466644695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6176733626762865e-06,
+      "loss": 0.8306,
+      "step": 29927
+    },
+    {
+      "epoch": 0.8219494108923128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6171883314109826e-06,
+      "loss": 0.9011,
+      "step": 29928
+    },
+    {
+      "epoch": 0.821976875120156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616703366473733e-06,
+      "loss": 0.7926,
+      "step": 29929
+    },
+    {
+      "epoch": 0.8220043393479992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6162184678683712e-06,
+      "loss": 0.8889,
+      "step": 29930
+    },
+    {
+      "epoch": 0.8220318035758425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6157336355987362e-06,
+      "loss": 0.781,
+      "step": 29931
+    },
+    {
+      "epoch": 0.8220592678036857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615248869668664e-06,
+      "loss": 0.9196,
+      "step": 29932
+    },
+    {
+      "epoch": 0.8220867320315289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6147641700819915e-06,
+      "loss": 0.6888,
+      "step": 29933
+    },
+    {
+      "epoch": 0.8221141962593722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6142795368425556e-06,
+      "loss": 0.8164,
+      "step": 29934
+    },
+    {
+      "epoch": 0.8221416604872154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6137949699541856e-06,
+      "loss": 0.7526,
+      "step": 29935
+    },
+    {
+      "epoch": 0.8221691247150587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613310469420718e-06,
+      "loss": 0.7705,
+      "step": 29936
+    },
+    {
+      "epoch": 0.8221965889429018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612826035245989e-06,
+      "loss": 0.8316,
+      "step": 29937
+    },
+    {
+      "epoch": 0.8222240531707451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6123416674338277e-06,
+      "loss": 0.727,
+      "step": 29938
+    },
+    {
+      "epoch": 0.8222515173985884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6118573659880675e-06,
+      "loss": 0.7516,
+      "step": 29939
+    },
+    {
+      "epoch": 0.8222789816264315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6113731309125436e-06,
+      "loss": 0.8303,
+      "step": 29940
+    },
+    {
+      "epoch": 0.8223064458542748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6108889622110824e-06,
+      "loss": 0.7147,
+      "step": 29941
+    },
+    {
+      "epoch": 0.8223339100821181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6104048598875178e-06,
+      "loss": 0.7799,
+      "step": 29942
+    },
+    {
+      "epoch": 0.8223613743099613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6099208239456787e-06,
+      "loss": 0.8357,
+      "step": 29943
+    },
+    {
+      "epoch": 0.8223888385378045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609436854389399e-06,
+      "loss": 0.8008,
+      "step": 29944
+    },
+    {
+      "epoch": 0.8224163027656477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6089529512225034e-06,
+      "loss": 0.7832,
+      "step": 29945
+    },
+    {
+      "epoch": 0.822443766993491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608469114448824e-06,
+      "loss": 0.7892,
+      "step": 29946
+    },
+    {
+      "epoch": 0.8224712312213343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607985344072186e-06,
+      "loss": 0.7948,
+      "step": 29947
+    },
+    {
+      "epoch": 0.8224986954491774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6075016400964183e-06,
+      "loss": 0.841,
+      "step": 29948
+    },
+    {
+      "epoch": 0.8225261596770207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607018002525348e-06,
+      "loss": 0.7659,
+      "step": 29949
+    },
+    {
+      "epoch": 0.8225536239048639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606534431362803e-06,
+      "loss": 0.7706,
+      "step": 29950
+    },
+    {
+      "epoch": 0.8225810881327071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606050926612609e-06,
+      "loss": 0.8136,
+      "step": 29951
+    },
+    {
+      "epoch": 0.8226085523605504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6055674882785933e-06,
+      "loss": 0.6844,
+      "step": 29952
+    },
+    {
+      "epoch": 0.8226360165883936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6050841163645802e-06,
+      "loss": 0.7879,
+      "step": 29953
+    },
+    {
+      "epoch": 0.8226634808162369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6046008108743905e-06,
+      "loss": 0.7717,
+      "step": 29954
+    },
+    {
+      "epoch": 0.8226909450440801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6041175718118507e-06,
+      "loss": 0.8178,
+      "step": 29955
+    },
+    {
+      "epoch": 0.8227184092719233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6036343991807856e-06,
+      "loss": 0.795,
+      "step": 29956
+    },
+    {
+      "epoch": 0.8227458734997666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6031512929850179e-06,
+      "loss": 0.7552,
+      "step": 29957
+    },
+    {
+      "epoch": 0.8227733377276097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6026682532283722e-06,
+      "loss": 0.825,
+      "step": 29958
+    },
+    {
+      "epoch": 0.822800801955453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6021852799146653e-06,
+      "loss": 0.7953,
+      "step": 29959
+    },
+    {
+      "epoch": 0.8228282661832963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.601702373047721e-06,
+      "loss": 0.8823,
+      "step": 29960
+    },
+    {
+      "epoch": 0.8228557304111395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6012195326313639e-06,
+      "loss": 0.827,
+      "step": 29961
+    },
+    {
+      "epoch": 0.8228831946389827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600736758669409e-06,
+      "loss": 0.7752,
+      "step": 29962
+    },
+    {
+      "epoch": 0.8229106588668259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6002540511656795e-06,
+      "loss": 0.7662,
+      "step": 29963
+    },
+    {
+      "epoch": 0.8229381230946692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5997714101239925e-06,
+      "loss": 0.8522,
+      "step": 29964
+    },
+    {
+      "epoch": 0.8229655873225125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5992888355481673e-06,
+      "loss": 0.8442,
+      "step": 29965
+    },
+    {
+      "epoch": 0.8229930515503556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598806327442024e-06,
+      "loss": 0.7602,
+      "step": 29966
+    },
+    {
+      "epoch": 0.8230205157781989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5983238858093787e-06,
+      "loss": 0.794,
+      "step": 29967
+    },
+    {
+      "epoch": 0.8230479800060422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5978415106540513e-06,
+      "loss": 0.8085,
+      "step": 29968
+    },
+    {
+      "epoch": 0.8230754442338853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5973592019798577e-06,
+      "loss": 0.8846,
+      "step": 29969
+    },
+    {
+      "epoch": 0.8231029084617286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5968769597906098e-06,
+      "loss": 0.823,
+      "step": 29970
+    },
+    {
+      "epoch": 0.8231303726895718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5963947840901262e-06,
+      "loss": 0.821,
+      "step": 29971
+    },
+    {
+      "epoch": 0.8231578369174151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595912674882223e-06,
+      "loss": 0.9268,
+      "step": 29972
+    },
+    {
+      "epoch": 0.8231853011452583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5954306321707135e-06,
+      "loss": 0.8093,
+      "step": 29973
+    },
+    {
+      "epoch": 0.8232127653731015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594948655959413e-06,
+      "loss": 0.7368,
+      "step": 29974
+    },
+    {
+      "epoch": 0.8232402296009448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5944667462521367e-06,
+      "loss": 0.8455,
+      "step": 29975
+    },
+    {
+      "epoch": 0.823267693828788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593984903052692e-06,
+      "loss": 0.8338,
+      "step": 29976
+    },
+    {
+      "epoch": 0.8232951580566312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5935031263648992e-06,
+      "loss": 0.8149,
+      "step": 29977
+    },
+    {
+      "epoch": 0.8233226222844745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5930214161925628e-06,
+      "loss": 0.7123,
+      "step": 29978
+    },
+    {
+      "epoch": 0.8233500865123177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5925397725394976e-06,
+      "loss": 0.8183,
+      "step": 29979
+    },
+    {
+      "epoch": 0.823377550740161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5920581954095138e-06,
+      "loss": 0.8262,
+      "step": 29980
+    },
+    {
+      "epoch": 0.8234050149680042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5915766848064262e-06,
+      "loss": 0.8258,
+      "step": 29981
+    },
+    {
+      "epoch": 0.8234324791958474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5910952407340375e-06,
+      "loss": 0.7293,
+      "step": 29982
+    },
+    {
+      "epoch": 0.8234599434236907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590613863196161e-06,
+      "loss": 0.8306,
+      "step": 29983
+    },
+    {
+      "epoch": 0.8234874076515338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590132552196607e-06,
+      "loss": 0.788,
+      "step": 29984
+    },
+    {
+      "epoch": 0.8235148718793771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5896513077391805e-06,
+      "loss": 0.8347,
+      "step": 29985
+    },
+    {
+      "epoch": 0.8235423361072204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5891701298276896e-06,
+      "loss": 0.8527,
+      "step": 29986
+    },
+    {
+      "epoch": 0.8235698003350636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5886890184659453e-06,
+      "loss": 0.8033,
+      "step": 29987
+    },
+    {
+      "epoch": 0.8235972645629068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5882079736577504e-06,
+      "loss": 0.889,
+      "step": 29988
+    },
+    {
+      "epoch": 0.82362472879075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587726995406912e-06,
+      "loss": 0.8133,
+      "step": 29989
+    },
+    {
+      "epoch": 0.8236521930185933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5872460837172355e-06,
+      "loss": 0.8692,
+      "step": 29990
+    },
+    {
+      "epoch": 0.8236796572464365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586765238592527e-06,
+      "loss": 0.8067,
+      "step": 29991
+    },
+    {
+      "epoch": 0.8237071214742797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5862844600365934e-06,
+      "loss": 0.7976,
+      "step": 29992
+    },
+    {
+      "epoch": 0.823734585702123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5858037480532362e-06,
+      "loss": 0.8789,
+      "step": 29993
+    },
+    {
+      "epoch": 0.8237620499299663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5853231026462567e-06,
+      "loss": 0.8156,
+      "step": 29994
+    },
+    {
+      "epoch": 0.8237895141578094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584842523819461e-06,
+      "loss": 0.8569,
+      "step": 29995
+    },
+    {
+      "epoch": 0.8238169783856527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58436201157665e-06,
+      "loss": 0.8343,
+      "step": 29996
+    },
+    {
+      "epoch": 0.8238444426134959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5838815659216279e-06,
+      "loss": 0.8235,
+      "step": 29997
+    },
+    {
+      "epoch": 0.8238719068413392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5834011868581933e-06,
+      "loss": 0.7849,
+      "step": 29998
+    },
+    {
+      "epoch": 0.8238993710691824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5829208743901525e-06,
+      "loss": 0.7043,
+      "step": 29999
+    },
+    {
+      "epoch": 0.8239268352970256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824406285212989e-06,
+      "loss": 0.7113,
+      "step": 30000
+    },
+    {
+      "epoch": 0.8239542995248689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5819604492554385e-06,
+      "loss": 0.8444,
+      "step": 30001
+    },
+    {
+      "epoch": 0.823981763752712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581480336596366e-06,
+      "loss": 0.743,
+      "step": 30002
+    },
+    {
+      "epoch": 0.8240092279805553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5810002905478816e-06,
+      "loss": 0.8029,
+      "step": 30003
+    },
+    {
+      "epoch": 0.8240366922083986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5805203111137845e-06,
+      "loss": 0.9018,
+      "step": 30004
+    },
+    {
+      "epoch": 0.8240641564362418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580040398297874e-06,
+      "loss": 0.7783,
+      "step": 30005
+    },
+    {
+      "epoch": 0.824091620664085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5795605521039436e-06,
+      "loss": 0.8972,
+      "step": 30006
+    },
+    {
+      "epoch": 0.8241190848919283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5790807725357915e-06,
+      "loss": 0.8202,
+      "step": 30007
+    },
+    {
+      "epoch": 0.8241465491197715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5786010595972156e-06,
+      "loss": 0.9152,
+      "step": 30008
+    },
+    {
+      "epoch": 0.8241740133476148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5781214132920098e-06,
+      "loss": 0.8893,
+      "step": 30009
+    },
+    {
+      "epoch": 0.8242014775754579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776418336239685e-06,
+      "loss": 0.7794,
+      "step": 30010
+    },
+    {
+      "epoch": 0.8242289418033012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57716232059689e-06,
+      "loss": 0.802,
+      "step": 30011
+    },
+    {
+      "epoch": 0.8242564060311445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5766828742145634e-06,
+      "loss": 0.7846,
+      "step": 30012
+    },
+    {
+      "epoch": 0.8242838702589876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5762034944807859e-06,
+      "loss": 0.8288,
+      "step": 30013
+    },
+    {
+      "epoch": 0.8243113344868309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5757241813993485e-06,
+      "loss": 0.768,
+      "step": 30014
+    },
+    {
+      "epoch": 0.8243387987146742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575244934974045e-06,
+      "loss": 0.801,
+      "step": 30015
+    },
+    {
+      "epoch": 0.8243662629425174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5747657552086692e-06,
+      "loss": 0.7503,
+      "step": 30016
+    },
+    {
+      "epoch": 0.8243937271703606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5742866421070103e-06,
+      "loss": 0.8952,
+      "step": 30017
+    },
+    {
+      "epoch": 0.8244211913982038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5738075956728571e-06,
+      "loss": 0.8009,
+      "step": 30018
+    },
+    {
+      "epoch": 0.8244486556260471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5733286159100013e-06,
+      "loss": 0.8684,
+      "step": 30019
+    },
+    {
+      "epoch": 0.8244761198538904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5728497028222345e-06,
+      "loss": 0.7719,
+      "step": 30020
+    },
+    {
+      "epoch": 0.8245035840817335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5723708564133454e-06,
+      "loss": 0.8329,
+      "step": 30021
+    },
+    {
+      "epoch": 0.8245310483095768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5718920766871226e-06,
+      "loss": 0.8837,
+      "step": 30022
+    },
+    {
+      "epoch": 0.82455851253742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5714133636473561e-06,
+      "loss": 0.7321,
+      "step": 30023
+    },
+    {
+      "epoch": 0.8245859767652632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5709347172978307e-06,
+      "loss": 0.8723,
+      "step": 30024
+    },
+    {
+      "epoch": 0.8246134409931065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5704561376423367e-06,
+      "loss": 0.8061,
+      "step": 30025
+    },
+    {
+      "epoch": 0.8246409052209497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5699776246846565e-06,
+      "loss": 0.8772,
+      "step": 30026
+    },
+    {
+      "epoch": 0.824668369448793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5694991784285785e-06,
+      "loss": 0.8148,
+      "step": 30027
+    },
+    {
+      "epoch": 0.8246958336766362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5690207988778916e-06,
+      "loss": 0.8003,
+      "step": 30028
+    },
+    {
+      "epoch": 0.8247232979044794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5685424860363763e-06,
+      "loss": 0.8282,
+      "step": 30029
+    },
+    {
+      "epoch": 0.8247507621323227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5680642399078171e-06,
+      "loss": 0.6966,
+      "step": 30030
+    },
+    {
+      "epoch": 0.8247782263601658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5675860604960014e-06,
+      "loss": 0.9113,
+      "step": 30031
+    },
+    {
+      "epoch": 0.8248056905880091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5671079478047123e-06,
+      "loss": 0.7262,
+      "step": 30032
+    },
+    {
+      "epoch": 0.8248331548158524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5666299018377307e-06,
+      "loss": 0.7319,
+      "step": 30033
+    },
+    {
+      "epoch": 0.8248606190436956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566151922598842e-06,
+      "loss": 0.8257,
+      "step": 30034
+    },
+    {
+      "epoch": 0.8248880832715388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5656740100918233e-06,
+      "loss": 0.848,
+      "step": 30035
+    },
+    {
+      "epoch": 0.824915547499382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5651961643204594e-06,
+      "loss": 0.795,
+      "step": 30036
+    },
+    {
+      "epoch": 0.8249430117272253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5647183852885307e-06,
+      "loss": 0.8539,
+      "step": 30037
+    },
+    {
+      "epoch": 0.8249704759550686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5642406729998184e-06,
+      "loss": 0.8627,
+      "step": 30038
+    },
+    {
+      "epoch": 0.8249979401829117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5637630274581006e-06,
+      "loss": 0.7753,
+      "step": 30039
+    },
+    {
+      "epoch": 0.825025404410755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56328544866716e-06,
+      "loss": 0.8043,
+      "step": 30040
+    },
+    {
+      "epoch": 0.8250528686385983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5628079366307737e-06,
+      "loss": 0.7341,
+      "step": 30041
+    },
+    {
+      "epoch": 0.8250803328664414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562330491352716e-06,
+      "loss": 0.7948,
+      "step": 30042
+    },
+    {
+      "epoch": 0.8251077970942847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5618531128367686e-06,
+      "loss": 0.777,
+      "step": 30043
+    },
+    {
+      "epoch": 0.8251352613221279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5613758010867076e-06,
+      "loss": 0.826,
+      "step": 30044
+    },
+    {
+      "epoch": 0.8251627255499712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608985561063105e-06,
+      "loss": 0.843,
+      "step": 30045
+    },
+    {
+      "epoch": 0.8251901897778144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5604213778993549e-06,
+      "loss": 0.8684,
+      "step": 30046
+    },
+    {
+      "epoch": 0.8252176540056576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559944266469613e-06,
+      "loss": 0.8368,
+      "step": 30047
+    },
+    {
+      "epoch": 0.8252451182335009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5594672218208605e-06,
+      "loss": 0.7172,
+      "step": 30048
+    },
+    {
+      "epoch": 0.825272582461344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558990243956875e-06,
+      "loss": 0.8794,
+      "step": 30049
+    },
+    {
+      "epoch": 0.8253000466891873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5585133328814272e-06,
+      "loss": 0.7146,
+      "step": 30050
+    },
+    {
+      "epoch": 0.8253275109170306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5580364885982913e-06,
+      "loss": 0.7963,
+      "step": 30051
+    },
+    {
+      "epoch": 0.8253549751448738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5575597111112427e-06,
+      "loss": 0.8019,
+      "step": 30052
+    },
+    {
+      "epoch": 0.825382439372717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55708300042405e-06,
+      "loss": 0.8454,
+      "step": 30053
+    },
+    {
+      "epoch": 0.8254099036005603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5566063565404877e-06,
+      "loss": 0.846,
+      "step": 30054
+    },
+    {
+      "epoch": 0.8254373678284035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556129779464326e-06,
+      "loss": 0.7934,
+      "step": 30055
+    },
+    {
+      "epoch": 0.8254648320562468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5556532691993386e-06,
+      "loss": 0.8074,
+      "step": 30056
+    },
+    {
+      "epoch": 0.8254922962840899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5551768257492905e-06,
+      "loss": 0.9427,
+      "step": 30057
+    },
+    {
+      "epoch": 0.8255197605119332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5547004491179584e-06,
+      "loss": 0.8042,
+      "step": 30058
+    },
+    {
+      "epoch": 0.8255472247397765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554224139309104e-06,
+      "loss": 0.7724,
+      "step": 30059
+    },
+    {
+      "epoch": 0.8255746889676197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5537478963265006e-06,
+      "loss": 0.7759,
+      "step": 30060
+    },
+    {
+      "epoch": 0.8256021531954629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5532717201739155e-06,
+      "loss": 0.8238,
+      "step": 30061
+    },
+    {
+      "epoch": 0.8256296174233061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5527956108551168e-06,
+      "loss": 0.73,
+      "step": 30062
+    },
+    {
+      "epoch": 0.8256570816511494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5523195683738712e-06,
+      "loss": 0.8273,
+      "step": 30063
+    },
+    {
+      "epoch": 0.8256845458789926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5518435927339471e-06,
+      "loss": 0.7159,
+      "step": 30064
+    },
+    {
+      "epoch": 0.8257120101068358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5513676839391102e-06,
+      "loss": 0.7423,
+      "step": 30065
+    },
+    {
+      "epoch": 0.8257394743346791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5508918419931218e-06,
+      "loss": 0.8702,
+      "step": 30066
+    },
+    {
+      "epoch": 0.8257669385625224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5504160668997504e-06,
+      "loss": 0.891,
+      "step": 30067
+    },
+    {
+      "epoch": 0.8257944027903655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5499403586627593e-06,
+      "loss": 0.822,
+      "step": 30068
+    },
+    {
+      "epoch": 0.8258218670182088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5494647172859145e-06,
+      "loss": 0.7877,
+      "step": 30069
+    },
+    {
+      "epoch": 0.825849331246052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5489891427729797e-06,
+      "loss": 0.8981,
+      "step": 30070
+    },
+    {
+      "epoch": 0.8258767954738953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548513635127714e-06,
+      "loss": 0.7927,
+      "step": 30071
+    },
+    {
+      "epoch": 0.8259042597017385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5480381943538824e-06,
+      "loss": 0.7343,
+      "step": 30072
+    },
+    {
+      "epoch": 0.8259317239295817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5475628204552495e-06,
+      "loss": 0.8205,
+      "step": 30073
+    },
+    {
+      "epoch": 0.825959188157425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5470875134355711e-06,
+      "loss": 0.8368,
+      "step": 30074
+    },
+    {
+      "epoch": 0.8259866523852681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54661227329861e-06,
+      "loss": 0.8972,
+      "step": 30075
+    },
+    {
+      "epoch": 0.8260141166131114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5461371000481308e-06,
+      "loss": 0.8716,
+      "step": 30076
+    },
+    {
+      "epoch": 0.8260415808409547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5456619936878858e-06,
+      "loss": 0.7725,
+      "step": 30077
+    },
+    {
+      "epoch": 0.8260690450687979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545186954221639e-06,
+      "loss": 0.7205,
+      "step": 30078
+    },
+    {
+      "epoch": 0.8260965092966411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5447119816531475e-06,
+      "loss": 0.8635,
+      "step": 30079
+    },
+    {
+      "epoch": 0.8261239735244844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5442370759861724e-06,
+      "loss": 0.8967,
+      "step": 30080
+    },
+    {
+      "epoch": 0.8261514377523276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5437622372244676e-06,
+      "loss": 0.783,
+      "step": 30081
+    },
+    {
+      "epoch": 0.8261789019801709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5432874653717933e-06,
+      "loss": 0.7607,
+      "step": 30082
+    },
+    {
+      "epoch": 0.826206366208014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542812760431901e-06,
+      "loss": 0.7971,
+      "step": 30083
+    },
+    {
+      "epoch": 0.8262338304358573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542338122408552e-06,
+      "loss": 0.7183,
+      "step": 30084
+    },
+    {
+      "epoch": 0.8262612946637006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5418635513054992e-06,
+      "loss": 0.904,
+      "step": 30085
+    },
+    {
+      "epoch": 0.8262887588915437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541389047126498e-06,
+      "loss": 0.8384,
+      "step": 30086
+    },
+    {
+      "epoch": 0.826316223119387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5409146098753057e-06,
+      "loss": 0.7408,
+      "step": 30087
+    },
+    {
+      "epoch": 0.8263436873472302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5404402395556717e-06,
+      "loss": 0.7965,
+      "step": 30088
+    },
+    {
+      "epoch": 0.8263711515750735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5399659361713537e-06,
+      "loss": 0.7901,
+      "step": 30089
+    },
+    {
+      "epoch": 0.8263986158029167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5394916997261e-06,
+      "loss": 0.7801,
+      "step": 30090
+    },
+    {
+      "epoch": 0.8264260800307599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5390175302236642e-06,
+      "loss": 0.8406,
+      "step": 30091
+    },
+    {
+      "epoch": 0.8264535442586032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5385434276678002e-06,
+      "loss": 0.8272,
+      "step": 30092
+    },
+    {
+      "epoch": 0.8264810084864465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5380693920622614e-06,
+      "loss": 0.8221,
+      "step": 30093
+    },
+    {
+      "epoch": 0.8265084727142896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5375954234107914e-06,
+      "loss": 0.7906,
+      "step": 30094
+    },
+    {
+      "epoch": 0.8265359369421329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5371215217171454e-06,
+      "loss": 0.8254,
+      "step": 30095
+    },
+    {
+      "epoch": 0.8265634011699761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5366476869850722e-06,
+      "loss": 0.8892,
+      "step": 30096
+    },
+    {
+      "epoch": 0.8265908653978193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5361739192183223e-06,
+      "loss": 0.7958,
+      "step": 30097
+    },
+    {
+      "epoch": 0.8266183296256626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5357002184206416e-06,
+      "loss": 0.82,
+      "step": 30098
+    },
+    {
+      "epoch": 0.8266457938535058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5352265845957814e-06,
+      "loss": 0.8086,
+      "step": 30099
+    },
+    {
+      "epoch": 0.8266732580813491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5347530177474857e-06,
+      "loss": 0.794,
+      "step": 30100
+    },
+    {
+      "epoch": 0.8267007223091922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5342795178795023e-06,
+      "loss": 0.7995,
+      "step": 30101
+    },
+    {
+      "epoch": 0.8267281865370355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5338060849955793e-06,
+      "loss": 0.787,
+      "step": 30102
+    },
+    {
+      "epoch": 0.8267556507648788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5333327190994618e-06,
+      "loss": 0.8794,
+      "step": 30103
+    },
+    {
+      "epoch": 0.826783114992722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5328594201948976e-06,
+      "loss": 0.8991,
+      "step": 30104
+    },
+    {
+      "epoch": 0.8268105792205652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5323861882856295e-06,
+      "loss": 0.8476,
+      "step": 30105
+    },
+    {
+      "epoch": 0.8268380434484085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5319130233753997e-06,
+      "loss": 0.7936,
+      "step": 30106
+    },
+    {
+      "epoch": 0.8268655076762517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5314399254679545e-06,
+      "loss": 0.8839,
+      "step": 30107
+    },
+    {
+      "epoch": 0.8268929719040949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5309668945670375e-06,
+      "loss": 0.8589,
+      "step": 30108
+    },
+    {
+      "epoch": 0.8269204361319381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53049393067639e-06,
+      "loss": 0.7823,
+      "step": 30109
+    },
+    {
+      "epoch": 0.8269479003597814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5300210337997557e-06,
+      "loss": 0.8636,
+      "step": 30110
+    },
+    {
+      "epoch": 0.8269753645876247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5295482039408793e-06,
+      "loss": 0.7908,
+      "step": 30111
+    },
+    {
+      "epoch": 0.8270028288154678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5290754411034948e-06,
+      "loss": 0.8306,
+      "step": 30112
+    },
+    {
+      "epoch": 0.8270302930433111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5286027452913499e-06,
+      "loss": 0.9219,
+      "step": 30113
+    },
+    {
+      "epoch": 0.8270577572711543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5281301165081785e-06,
+      "loss": 0.8199,
+      "step": 30114
+    },
+    {
+      "epoch": 0.8270852214989975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5276575547577243e-06,
+      "loss": 0.8194,
+      "step": 30115
+    },
+    {
+      "epoch": 0.8271126857268408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5271850600437245e-06,
+      "loss": 0.7189,
+      "step": 30116
+    },
+    {
+      "epoch": 0.827140149954684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5267126323699211e-06,
+      "loss": 0.7817,
+      "step": 30117
+    },
+    {
+      "epoch": 0.8271676141825273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5262402717400482e-06,
+      "loss": 0.8106,
+      "step": 30118
+    },
+    {
+      "epoch": 0.8271950784103705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.525767978157844e-06,
+      "loss": 0.7094,
+      "step": 30119
+    },
+    {
+      "epoch": 0.8272225426382137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5252957516270451e-06,
+      "loss": 0.8449,
+      "step": 30120
+    },
+    {
+      "epoch": 0.827250006866057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5248235921513932e-06,
+      "loss": 0.8543,
+      "step": 30121
+    },
+    {
+      "epoch": 0.8272774710939002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5243514997346165e-06,
+      "loss": 0.7641,
+      "step": 30122
+    },
+    {
+      "epoch": 0.8273049353217434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5238794743804563e-06,
+      "loss": 0.8604,
+      "step": 30123
+    },
+    {
+      "epoch": 0.8273323995495867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5234075160926432e-06,
+      "loss": 0.9404,
+      "step": 30124
+    },
+    {
+      "epoch": 0.8273598637774299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5229356248749127e-06,
+      "loss": 0.8238,
+      "step": 30125
+    },
+    {
+      "epoch": 0.8273873280052731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.522463800731e-06,
+      "loss": 0.8701,
+      "step": 30126
+    },
+    {
+      "epoch": 0.8274147922331163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5219920436646374e-06,
+      "loss": 0.7245,
+      "step": 30127
+    },
+    {
+      "epoch": 0.8274422564609596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5215203536795597e-06,
+      "loss": 0.7874,
+      "step": 30128
+    },
+    {
+      "epoch": 0.8274697206888029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5210487307794974e-06,
+      "loss": 0.8438,
+      "step": 30129
+    },
+    {
+      "epoch": 0.827497184916646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5205771749681797e-06,
+      "loss": 0.7697,
+      "step": 30130
+    },
+    {
+      "epoch": 0.8275246491444893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5201056862493402e-06,
+      "loss": 0.7975,
+      "step": 30131
+    },
+    {
+      "epoch": 0.8275521133723326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5196342646267093e-06,
+      "loss": 0.852,
+      "step": 30132
+    },
+    {
+      "epoch": 0.8275795776001758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5191629101040162e-06,
+      "loss": 0.8208,
+      "step": 30133
+    },
+    {
+      "epoch": 0.827607041828019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5186916226849924e-06,
+      "loss": 0.7426,
+      "step": 30134
+    },
+    {
+      "epoch": 0.8276345060558622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.518220402373367e-06,
+      "loss": 0.8497,
+      "step": 30135
+    },
+    {
+      "epoch": 0.8276619702837055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5177492491728662e-06,
+      "loss": 0.8587,
+      "step": 30136
+    },
+    {
+      "epoch": 0.8276894345115487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5172781630872213e-06,
+      "loss": 0.8534,
+      "step": 30137
+    },
+    {
+      "epoch": 0.8277168987393919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.516807144120155e-06,
+      "loss": 0.7923,
+      "step": 30138
+    },
+    {
+      "epoch": 0.8277443629672352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.516336192275396e-06,
+      "loss": 0.7416,
+      "step": 30139
+    },
+    {
+      "epoch": 0.8277718271950784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515865307556671e-06,
+      "loss": 0.865,
+      "step": 30140
+    },
+    {
+      "epoch": 0.8277992914229216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5153944899677097e-06,
+      "loss": 0.7833,
+      "step": 30141
+    },
+    {
+      "epoch": 0.8278267556507649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5149237395122306e-06,
+      "loss": 0.7956,
+      "step": 30142
+    },
+    {
+      "epoch": 0.8278542198786081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514453056193962e-06,
+      "loss": 0.8593,
+      "step": 30143
+    },
+    {
+      "epoch": 0.8278816841064514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5139824400166303e-06,
+      "loss": 0.8964,
+      "step": 30144
+    },
+    {
+      "epoch": 0.8279091483342946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513511890983954e-06,
+      "loss": 0.7339,
+      "step": 30145
+    },
+    {
+      "epoch": 0.8279366125621378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5130414090996614e-06,
+      "loss": 0.8473,
+      "step": 30146
+    },
+    {
+      "epoch": 0.8279640767899811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5125709943674716e-06,
+      "loss": 0.8655,
+      "step": 30147
+    },
+    {
+      "epoch": 0.8279915410178242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5121006467911058e-06,
+      "loss": 0.672,
+      "step": 30148
+    },
+    {
+      "epoch": 0.8280190052456675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5116303663742893e-06,
+      "loss": 0.8134,
+      "step": 30149
+    },
+    {
+      "epoch": 0.8280464694735108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.51116015312074e-06,
+      "loss": 0.8412,
+      "step": 30150
+    },
+    {
+      "epoch": 0.828073933701354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5106900070341802e-06,
+      "loss": 0.7057,
+      "step": 30151
+    },
+    {
+      "epoch": 0.8281013979291972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5102199281183327e-06,
+      "loss": 0.8885,
+      "step": 30152
+    },
+    {
+      "epoch": 0.8281288621570404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5097499163769124e-06,
+      "loss": 0.7302,
+      "step": 30153
+    },
+    {
+      "epoch": 0.8281563263848837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5092799718136386e-06,
+      "loss": 0.81,
+      "step": 30154
+    },
+    {
+      "epoch": 0.828183790612727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5088100944322304e-06,
+      "loss": 0.8643,
+      "step": 30155
+    },
+    {
+      "epoch": 0.8282112548405701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5083402842364048e-06,
+      "loss": 0.7665,
+      "step": 30156
+    },
+    {
+      "epoch": 0.8282387190684134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5078705412298811e-06,
+      "loss": 0.7821,
+      "step": 30157
+    },
+    {
+      "epoch": 0.8282661832962567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5074008654163775e-06,
+      "loss": 0.8188,
+      "step": 30158
+    },
+    {
+      "epoch": 0.8282936475240998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5069312567996053e-06,
+      "loss": 0.8464,
+      "step": 30159
+    },
+    {
+      "epoch": 0.8283211117519431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.506461715383284e-06,
+      "loss": 0.7405,
+      "step": 30160
+    },
+    {
+      "epoch": 0.8283485759797863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5059922411711292e-06,
+      "loss": 0.7887,
+      "step": 30161
+    },
+    {
+      "epoch": 0.8283760402076296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5055228341668516e-06,
+      "loss": 0.7179,
+      "step": 30162
+    },
+    {
+      "epoch": 0.8284035044354728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.505053494374168e-06,
+      "loss": 0.8883,
+      "step": 30163
+    },
+    {
+      "epoch": 0.828430968663316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.504584221796793e-06,
+      "loss": 0.863,
+      "step": 30164
+    },
+    {
+      "epoch": 0.8284584328911593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5041150164384378e-06,
+      "loss": 0.803,
+      "step": 30165
+    },
+    {
+      "epoch": 0.8284858971190024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5036458783028141e-06,
+      "loss": 0.846,
+      "step": 30166
+    },
+    {
+      "epoch": 0.8285133613468457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503176807393637e-06,
+      "loss": 0.8259,
+      "step": 30167
+    },
+    {
+      "epoch": 0.828540825574689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502707803714617e-06,
+      "loss": 0.8846,
+      "step": 30168
+    },
+    {
+      "epoch": 0.8285682898025322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5022388672694633e-06,
+      "loss": 0.8605,
+      "step": 30169
+    },
+    {
+      "epoch": 0.8285957540303754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5017699980618895e-06,
+      "loss": 0.7425,
+      "step": 30170
+    },
+    {
+      "epoch": 0.8286232182582187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5013011960956004e-06,
+      "loss": 0.7892,
+      "step": 30171
+    },
+    {
+      "epoch": 0.8286506824860619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5008324613743098e-06,
+      "loss": 0.8373,
+      "step": 30172
+    },
+    {
+      "epoch": 0.8286781467139052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5003637939017234e-06,
+      "loss": 0.8213,
+      "step": 30173
+    },
+    {
+      "epoch": 0.8287056109417483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.499895193681552e-06,
+      "loss": 0.7262,
+      "step": 30174
+    },
+    {
+      "epoch": 0.8287330751695916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4994266607175023e-06,
+      "loss": 0.9109,
+      "step": 30175
+    },
+    {
+      "epoch": 0.8287605393974349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4989581950132836e-06,
+      "loss": 0.9255,
+      "step": 30176
+    },
+    {
+      "epoch": 0.828788003625278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4984897965726008e-06,
+      "loss": 0.8112,
+      "step": 30177
+    },
+    {
+      "epoch": 0.8288154678531213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4980214653991576e-06,
+      "loss": 0.8712,
+      "step": 30178
+    },
+    {
+      "epoch": 0.8288429320809645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.497553201496662e-06,
+      "loss": 0.7754,
+      "step": 30179
+    },
+    {
+      "epoch": 0.8288703963088078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4970850048688191e-06,
+      "loss": 0.8872,
+      "step": 30180
+    },
+    {
+      "epoch": 0.828897860536651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4966168755193334e-06,
+      "loss": 0.8708,
+      "step": 30181
+    },
+    {
+      "epoch": 0.8289253247644942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4961488134519108e-06,
+      "loss": 0.8492,
+      "step": 30182
+    },
+    {
+      "epoch": 0.8289527889923375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.495680818670251e-06,
+      "loss": 0.7754,
+      "step": 30183
+    },
+    {
+      "epoch": 0.8289802532201808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4952128911780584e-06,
+      "loss": 0.7712,
+      "step": 30184
+    },
+    {
+      "epoch": 0.8290077174480239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494745030979039e-06,
+      "loss": 0.9445,
+      "step": 30185
+    },
+    {
+      "epoch": 0.8290351816758672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4942772380768888e-06,
+      "loss": 0.812,
+      "step": 30186
+    },
+    {
+      "epoch": 0.8290626459037104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4938095124753117e-06,
+      "loss": 0.7917,
+      "step": 30187
+    },
+    {
+      "epoch": 0.8290901101315536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.493341854178011e-06,
+      "loss": 0.9054,
+      "step": 30188
+    },
+    {
+      "epoch": 0.8291175743593969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4928742631886828e-06,
+      "loss": 0.7916,
+      "step": 30189
+    },
+    {
+      "epoch": 0.8291450385872401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4924067395110276e-06,
+      "loss": 0.936,
+      "step": 30190
+    },
+    {
+      "epoch": 0.8291725028150834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4919392831487468e-06,
+      "loss": 0.8271,
+      "step": 30191
+    },
+    {
+      "epoch": 0.8291999670429265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4914718941055407e-06,
+      "loss": 0.7356,
+      "step": 30192
+    },
+    {
+      "epoch": 0.8292274312707698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491004572385102e-06,
+      "loss": 0.8726,
+      "step": 30193
+    },
+    {
+      "epoch": 0.8292548954986131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490537317991133e-06,
+      "loss": 0.7488,
+      "step": 30194
+    },
+    {
+      "epoch": 0.8292823597264563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490070130927328e-06,
+      "loss": 0.8679,
+      "step": 30195
+    },
+    {
+      "epoch": 0.8293098239542995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4896030111973836e-06,
+      "loss": 0.7666,
+      "step": 30196
+    },
+    {
+      "epoch": 0.8293372881821428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4891359588049969e-06,
+      "loss": 0.891,
+      "step": 30197
+    },
+    {
+      "epoch": 0.829364752409986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.488668973753863e-06,
+      "loss": 0.7265,
+      "step": 30198
+    },
+    {
+      "epoch": 0.8293922166378292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4882020560476785e-06,
+      "loss": 0.7892,
+      "step": 30199
+    },
+    {
+      "epoch": 0.8294196808656724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4877352056901373e-06,
+      "loss": 0.7916,
+      "step": 30200
+    },
+    {
+      "epoch": 0.8294471450935157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4872684226849332e-06,
+      "loss": 0.861,
+      "step": 30201
+    },
+    {
+      "epoch": 0.829474609321359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4868017070357565e-06,
+      "loss": 0.737,
+      "step": 30202
+    },
+    {
+      "epoch": 0.8295020735492021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486335058746302e-06,
+      "loss": 0.8172,
+      "step": 30203
+    },
+    {
+      "epoch": 0.8295295377770454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4858684778202626e-06,
+      "loss": 0.799,
+      "step": 30204
+    },
+    {
+      "epoch": 0.8295570020048886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4854019642613327e-06,
+      "loss": 0.8167,
+      "step": 30205
+    },
+    {
+      "epoch": 0.8295844662327319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4849355180731973e-06,
+      "loss": 0.8356,
+      "step": 30206
+    },
+    {
+      "epoch": 0.8296119304605751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4844691392595511e-06,
+      "loss": 0.7937,
+      "step": 30207
+    },
+    {
+      "epoch": 0.8296393946884183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4840028278240837e-06,
+      "loss": 0.7258,
+      "step": 30208
+    },
+    {
+      "epoch": 0.8296668589162616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4835365837704873e-06,
+      "loss": 0.7674,
+      "step": 30209
+    },
+    {
+      "epoch": 0.8296943231441049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.483070407102446e-06,
+      "loss": 0.8561,
+      "step": 30210
+    },
+    {
+      "epoch": 0.829721787371948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4826042978236532e-06,
+      "loss": 0.7542,
+      "step": 30211
+    },
+    {
+      "epoch": 0.8297492515997913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4821382559377916e-06,
+      "loss": 0.7813,
+      "step": 30212
+    },
+    {
+      "epoch": 0.8297767158276345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4816722814485519e-06,
+      "loss": 0.7849,
+      "step": 30213
+    },
+    {
+      "epoch": 0.8298041800554777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812063743596205e-06,
+      "loss": 0.8644,
+      "step": 30214
+    },
+    {
+      "epoch": 0.829831644283321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480740534674685e-06,
+      "loss": 0.8204,
+      "step": 30215
+    },
+    {
+      "epoch": 0.8298591085111642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4802747623974323e-06,
+      "loss": 0.7547,
+      "step": 30216
+    },
+    {
+      "epoch": 0.8298865727390075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4798090575315438e-06,
+      "loss": 0.6997,
+      "step": 30217
+    },
+    {
+      "epoch": 0.8299140369668507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4793434200807078e-06,
+      "loss": 0.7047,
+      "step": 30218
+    },
+    {
+      "epoch": 0.8299415011946939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4788778500486056e-06,
+      "loss": 0.7229,
+      "step": 30219
+    },
+    {
+      "epoch": 0.8299689654225372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4784123474389233e-06,
+      "loss": 0.8867,
+      "step": 30220
+    },
+    {
+      "epoch": 0.8299964296503803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4779469122553425e-06,
+      "loss": 0.737,
+      "step": 30221
+    },
+    {
+      "epoch": 0.8300238938782236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4774815445015466e-06,
+      "loss": 0.7883,
+      "step": 30222
+    },
+    {
+      "epoch": 0.8300513581060669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.477016244181222e-06,
+      "loss": 0.8825,
+      "step": 30223
+    },
+    {
+      "epoch": 0.8300788223339101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476551011298043e-06,
+      "loss": 0.7797,
+      "step": 30224
+    },
+    {
+      "epoch": 0.8301062865617533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476085845855696e-06,
+      "loss": 0.6815,
+      "step": 30225
+    },
+    {
+      "epoch": 0.8301337507895965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4756207478578577e-06,
+      "loss": 0.7764,
+      "step": 30226
+    },
+    {
+      "epoch": 0.8301612150174398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4751557173082097e-06,
+      "loss": 0.8666,
+      "step": 30227
+    },
+    {
+      "epoch": 0.8301886792452831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4746907542104328e-06,
+      "loss": 0.825,
+      "step": 30228
+    },
+    {
+      "epoch": 0.8302161434731262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4742258585682056e-06,
+      "loss": 0.7485,
+      "step": 30229
+    },
+    {
+      "epoch": 0.8302436077009695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4737610303852057e-06,
+      "loss": 0.8647,
+      "step": 30230
+    },
+    {
+      "epoch": 0.8302710719288128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4732962696651098e-06,
+      "loss": 0.8462,
+      "step": 30231
+    },
+    {
+      "epoch": 0.8302985361566559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4728315764115975e-06,
+      "loss": 0.8083,
+      "step": 30232
+    },
+    {
+      "epoch": 0.8303260003844992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4723669506283467e-06,
+      "loss": 0.7369,
+      "step": 30233
+    },
+    {
+      "epoch": 0.8303534646123424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4719023923190301e-06,
+      "loss": 0.9099,
+      "step": 30234
+    },
+    {
+      "epoch": 0.8303809288401857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471437901487328e-06,
+      "loss": 0.9111,
+      "step": 30235
+    },
+    {
+      "epoch": 0.8304083930680289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.470973478136909e-06,
+      "loss": 0.7971,
+      "step": 30236
+    },
+    {
+      "epoch": 0.8304358572958721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4705091222714529e-06,
+      "loss": 0.8395,
+      "step": 30237
+    },
+    {
+      "epoch": 0.8304633215237154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4700448338946326e-06,
+      "loss": 0.799,
+      "step": 30238
+    },
+    {
+      "epoch": 0.8304907857515585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.469580613010121e-06,
+      "loss": 0.8521,
+      "step": 30239
+    },
+    {
+      "epoch": 0.8305182499794018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4691164596215946e-06,
+      "loss": 0.7364,
+      "step": 30240
+    },
+    {
+      "epoch": 0.8305457142072451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4686523737327218e-06,
+      "loss": 0.7917,
+      "step": 30241
+    },
+    {
+      "epoch": 0.8305731784350883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4681883553471776e-06,
+      "loss": 0.8322,
+      "step": 30242
+    },
+    {
+      "epoch": 0.8306006426629315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46772440446863e-06,
+      "loss": 0.8244,
+      "step": 30243
+    },
+    {
+      "epoch": 0.8306281068907748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4672605211007518e-06,
+      "loss": 0.7822,
+      "step": 30244
+    },
+    {
+      "epoch": 0.830655571118618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4667967052472143e-06,
+      "loss": 0.7102,
+      "step": 30245
+    },
+    {
+      "epoch": 0.8306830353464613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4663329569116857e-06,
+      "loss": 0.8203,
+      "step": 30246
+    },
+    {
+      "epoch": 0.8307104995743044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4658692760978388e-06,
+      "loss": 0.797,
+      "step": 30247
+    },
+    {
+      "epoch": 0.8307379638021477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4654056628093383e-06,
+      "loss": 0.7654,
+      "step": 30248
+    },
+    {
+      "epoch": 0.830765428029991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4649421170498567e-06,
+      "loss": 0.8484,
+      "step": 30249
+    },
+    {
+      "epoch": 0.8307928922578341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464478638823056e-06,
+      "loss": 0.707,
+      "step": 30250
+    },
+    {
+      "epoch": 0.8308203564856774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464015228132607e-06,
+      "loss": 0.812,
+      "step": 30251
+    },
+    {
+      "epoch": 0.8308478207135206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4635518849821762e-06,
+      "loss": 0.7647,
+      "step": 30252
+    },
+    {
+      "epoch": 0.8308752849413639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4630886093754316e-06,
+      "loss": 0.8651,
+      "step": 30253
+    },
+    {
+      "epoch": 0.8309027491692071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4626254013160347e-06,
+      "loss": 0.7924,
+      "step": 30254
+    },
+    {
+      "epoch": 0.8309302133970503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4621622608076536e-06,
+      "loss": 0.8105,
+      "step": 30255
+    },
+    {
+      "epoch": 0.8309576776248936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4616991878539511e-06,
+      "loss": 0.8312,
+      "step": 30256
+    },
+    {
+      "epoch": 0.8309851418527369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.461236182458594e-06,
+      "loss": 0.7891,
+      "step": 30257
+    },
+    {
+      "epoch": 0.83101260608058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460773244625242e-06,
+      "loss": 0.765,
+      "step": 30258
+    },
+    {
+      "epoch": 0.8310400703084233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460310374357562e-06,
+      "loss": 0.8299,
+      "step": 30259
+    },
+    {
+      "epoch": 0.8310675345362665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.459847571659212e-06,
+      "loss": 0.8833,
+      "step": 30260
+    },
+    {
+      "epoch": 0.8310949987641097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4593848365338559e-06,
+      "loss": 0.8265,
+      "step": 30261
+    },
+    {
+      "epoch": 0.831122462991953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.458922168985155e-06,
+      "loss": 0.7561,
+      "step": 30262
+    },
+    {
+      "epoch": 0.8311499272197962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4584595690167713e-06,
+      "loss": 0.8254,
+      "step": 30263
+    },
+    {
+      "epoch": 0.8311773914476395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457997036632366e-06,
+      "loss": 0.6868,
+      "step": 30264
+    },
+    {
+      "epoch": 0.8312048556754826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4575345718355971e-06,
+      "loss": 0.7526,
+      "step": 30265
+    },
+    {
+      "epoch": 0.8312323199033259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457072174630122e-06,
+      "loss": 0.8011,
+      "step": 30266
+    },
+    {
+      "epoch": 0.8312597841311692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4566098450195997e-06,
+      "loss": 0.8289,
+      "step": 30267
+    },
+    {
+      "epoch": 0.8312872483590124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4561475830076909e-06,
+      "loss": 0.7752,
+      "step": 30268
+    },
+    {
+      "epoch": 0.8313147125868556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4556853885980515e-06,
+      "loss": 0.8303,
+      "step": 30269
+    },
+    {
+      "epoch": 0.8313421768146989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4552232617943408e-06,
+      "loss": 0.7675,
+      "step": 30270
+    },
+    {
+      "epoch": 0.8313696410425421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4547612026002123e-06,
+      "loss": 0.7587,
+      "step": 30271
+    },
+    {
+      "epoch": 0.8313971052703854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4542992110193234e-06,
+      "loss": 0.7859,
+      "step": 30272
+    },
+    {
+      "epoch": 0.8314245694982285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4538372870553307e-06,
+      "loss": 0.8178,
+      "step": 30273
+    },
+    {
+      "epoch": 0.8314520337260718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4533754307118865e-06,
+      "loss": 0.9069,
+      "step": 30274
+    },
+    {
+      "epoch": 0.8314794979539151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4529136419926459e-06,
+      "loss": 0.8493,
+      "step": 30275
+    },
+    {
+      "epoch": 0.8315069621817582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4524519209012654e-06,
+      "loss": 0.8822,
+      "step": 30276
+    },
+    {
+      "epoch": 0.8315344264096015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4519902674413944e-06,
+      "loss": 0.713,
+      "step": 30277
+    },
+    {
+      "epoch": 0.8315618906374447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4515286816166884e-06,
+      "loss": 0.8746,
+      "step": 30278
+    },
+    {
+      "epoch": 0.831589354865288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4510671634307971e-06,
+      "loss": 0.694,
+      "step": 30279
+    },
+    {
+      "epoch": 0.8316168190931312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4506057128873751e-06,
+      "loss": 0.869,
+      "step": 30280
+    },
+    {
+      "epoch": 0.8316442833209744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4501443299900753e-06,
+      "loss": 0.7989,
+      "step": 30281
+    },
+    {
+      "epoch": 0.8316717475488177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4496830147425445e-06,
+      "loss": 0.775,
+      "step": 30282
+    },
+    {
+      "epoch": 0.831699211776661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449221767148431e-06,
+      "loss": 0.8297,
+      "step": 30283
+    },
+    {
+      "epoch": 0.8317266760045041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4487605872113886e-06,
+      "loss": 0.7578,
+      "step": 30284
+    },
+    {
+      "epoch": 0.8317541402323474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4482994749350642e-06,
+      "loss": 0.7157,
+      "step": 30285
+    },
+    {
+      "epoch": 0.8317816044601906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4478384303231064e-06,
+      "loss": 0.6827,
+      "step": 30286
+    },
+    {
+      "epoch": 0.8318090686880338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447377453379164e-06,
+      "loss": 0.8605,
+      "step": 30287
+    },
+    {
+      "epoch": 0.8318365329158771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4469165441068866e-06,
+      "loss": 0.8702,
+      "step": 30288
+    },
+    {
+      "epoch": 0.8318639971437203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.446455702509919e-06,
+      "loss": 0.8612,
+      "step": 30289
+    },
+    {
+      "epoch": 0.8318914613715636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445994928591905e-06,
+      "loss": 0.8574,
+      "step": 30290
+    },
+    {
+      "epoch": 0.8319189255994067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455342223564916e-06,
+      "loss": 0.7907,
+      "step": 30291
+    },
+    {
+      "epoch": 0.83194638982725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445073583807326e-06,
+      "loss": 0.7513,
+      "step": 30292
+    },
+    {
+      "epoch": 0.8319738540550933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4446130129480519e-06,
+      "loss": 0.7931,
+      "step": 30293
+    },
+    {
+      "epoch": 0.8320013182829364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444152509782316e-06,
+      "loss": 0.8172,
+      "step": 30294
+    },
+    {
+      "epoch": 0.8320287825107797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4436920743137572e-06,
+      "loss": 0.7521,
+      "step": 30295
+    },
+    {
+      "epoch": 0.832056246738623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.443231706546021e-06,
+      "loss": 0.7343,
+      "step": 30296
+    },
+    {
+      "epoch": 0.8320837109664662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4427714064827525e-06,
+      "loss": 0.8504,
+      "step": 30297
+    },
+    {
+      "epoch": 0.8321111751943094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4423111741275897e-06,
+      "loss": 0.6905,
+      "step": 30298
+    },
+    {
+      "epoch": 0.8321386394221526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4418510094841742e-06,
+      "loss": 0.8442,
+      "step": 30299
+    },
+    {
+      "epoch": 0.8321661036499959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.441390912556152e-06,
+      "loss": 0.7559,
+      "step": 30300
+    },
+    {
+      "epoch": 0.8321935678778392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440930883347158e-06,
+      "loss": 0.8209,
+      "step": 30301
+    },
+    {
+      "epoch": 0.8322210321056823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4404709218608348e-06,
+      "loss": 0.7973,
+      "step": 30302
+    },
+    {
+      "epoch": 0.8322484963335256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4400110281008206e-06,
+      "loss": 0.8378,
+      "step": 30303
+    },
+    {
+      "epoch": 0.8322759605613688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4395512020707569e-06,
+      "loss": 0.8192,
+      "step": 30304
+    },
+    {
+      "epoch": 0.832303424789212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4390914437742786e-06,
+      "loss": 0.7397,
+      "step": 30305
+    },
+    {
+      "epoch": 0.8323308890170553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4386317532150273e-06,
+      "loss": 0.7495,
+      "step": 30306
+    },
+    {
+      "epoch": 0.8323583532448985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4381721303966346e-06,
+      "loss": 0.8373,
+      "step": 30307
+    },
+    {
+      "epoch": 0.8323858174727418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4377125753227416e-06,
+      "loss": 0.7752,
+      "step": 30308
+    },
+    {
+      "epoch": 0.832413281700585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4372530879969825e-06,
+      "loss": 0.8296,
+      "step": 30309
+    },
+    {
+      "epoch": 0.8324407459284282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4367936684229954e-06,
+      "loss": 0.8273,
+      "step": 30310
+    },
+    {
+      "epoch": 0.8324682101562715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4363343166044119e-06,
+      "loss": 0.9475,
+      "step": 30311
+    },
+    {
+      "epoch": 0.8324956743841146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4358750325448723e-06,
+      "loss": 0.8375,
+      "step": 30312
+    },
+    {
+      "epoch": 0.8325231386119579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4354158162480058e-06,
+      "loss": 0.7737,
+      "step": 30313
+    },
+    {
+      "epoch": 0.8325506028398012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4349566677174454e-06,
+      "loss": 0.8137,
+      "step": 30314
+    },
+    {
+      "epoch": 0.8325780670676444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4344975869568257e-06,
+      "loss": 0.908,
+      "step": 30315
+    },
+    {
+      "epoch": 0.8326055312954876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4340385739697793e-06,
+      "loss": 0.7731,
+      "step": 30316
+    },
+    {
+      "epoch": 0.8326329955233308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433579628759937e-06,
+      "loss": 0.7732,
+      "step": 30317
+    },
+    {
+      "epoch": 0.8326604597511741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4331207513309331e-06,
+      "loss": 0.765,
+      "step": 30318
+    },
+    {
+      "epoch": 0.8326879239790174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4326619416863952e-06,
+      "loss": 0.8614,
+      "step": 30319
+    },
+    {
+      "epoch": 0.8327153882068605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4322031998299535e-06,
+      "loss": 0.84,
+      "step": 30320
+    },
+    {
+      "epoch": 0.8327428524347038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4317445257652418e-06,
+      "loss": 0.8555,
+      "step": 30321
+    },
+    {
+      "epoch": 0.8327703166625471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4312859194958838e-06,
+      "loss": 0.7332,
+      "step": 30322
+    },
+    {
+      "epoch": 0.8327977808903902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4308273810255114e-06,
+      "loss": 0.8721,
+      "step": 30323
+    },
+    {
+      "epoch": 0.8328252451182335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4303689103577545e-06,
+      "loss": 0.785,
+      "step": 30324
+    },
+    {
+      "epoch": 0.8328527093460767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4299105074962361e-06,
+      "loss": 0.8793,
+      "step": 30325
+    },
+    {
+      "epoch": 0.83288017357392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4294521724445843e-06,
+      "loss": 0.8198,
+      "step": 30326
+    },
+    {
+      "epoch": 0.8329076378017632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4289939052064283e-06,
+      "loss": 0.8665,
+      "step": 30327
+    },
+    {
+      "epoch": 0.8329351020296064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4285357057853933e-06,
+      "loss": 0.8314,
+      "step": 30328
+    },
+    {
+      "epoch": 0.8329625662574497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4280775741851026e-06,
+      "loss": 0.8811,
+      "step": 30329
+    },
+    {
+      "epoch": 0.8329900304852929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427619510409185e-06,
+      "loss": 0.9102,
+      "step": 30330
+    },
+    {
+      "epoch": 0.8330174947131361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4271615144612605e-06,
+      "loss": 0.7865,
+      "step": 30331
+    },
+    {
+      "epoch": 0.8330449589409794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.426703586344954e-06,
+      "loss": 0.8538,
+      "step": 30332
+    },
+    {
+      "epoch": 0.8330724231688226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4262457260638907e-06,
+      "loss": 0.8486,
+      "step": 30333
+    },
+    {
+      "epoch": 0.8330998873966659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4257879336216918e-06,
+      "loss": 0.7916,
+      "step": 30334
+    },
+    {
+      "epoch": 0.8331273516245091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4253302090219823e-06,
+      "loss": 0.8126,
+      "step": 30335
+    },
+    {
+      "epoch": 0.8331548158523523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4248725522683803e-06,
+      "loss": 0.7888,
+      "step": 30336
+    },
+    {
+      "epoch": 0.8331822800801956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4244149633645099e-06,
+      "loss": 0.8546,
+      "step": 30337
+    },
+    {
+      "epoch": 0.8332097443080387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.423957442313988e-06,
+      "loss": 0.7961,
+      "step": 30338
+    },
+    {
+      "epoch": 0.833237208535882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4234999891204359e-06,
+      "loss": 0.9014,
+      "step": 30339
+    },
+    {
+      "epoch": 0.8332646727637253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4230426037874756e-06,
+      "loss": 0.7219,
+      "step": 30340
+    },
+    {
+      "epoch": 0.8332921369915685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4225852863187261e-06,
+      "loss": 0.9136,
+      "step": 30341
+    },
+    {
+      "epoch": 0.8333196012194117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4221280367178026e-06,
+      "loss": 0.8619,
+      "step": 30342
+    },
+    {
+      "epoch": 0.8333470654472549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.421670854988324e-06,
+      "loss": 0.8543,
+      "step": 30343
+    },
+    {
+      "epoch": 0.8333745296750982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.421213741133909e-06,
+      "loss": 0.8424,
+      "step": 30344
+    },
+    {
+      "epoch": 0.8334019939029415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4207566951581764e-06,
+      "loss": 0.775,
+      "step": 30345
+    },
+    {
+      "epoch": 0.8334294581307846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4202997170647382e-06,
+      "loss": 0.8723,
+      "step": 30346
+    },
+    {
+      "epoch": 0.8334569223586279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4198428068572134e-06,
+      "loss": 0.8333,
+      "step": 30347
+    },
+    {
+      "epoch": 0.8334843865864712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4193859645392138e-06,
+      "loss": 0.815,
+      "step": 30348
+    },
+    {
+      "epoch": 0.8335118508143143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4189291901143576e-06,
+      "loss": 0.7757,
+      "step": 30349
+    },
+    {
+      "epoch": 0.8335393150421576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4184724835862561e-06,
+      "loss": 0.9114,
+      "step": 30350
+    },
+    {
+      "epoch": 0.8335667792700008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4180158449585246e-06,
+      "loss": 0.7988,
+      "step": 30351
+    },
+    {
+      "epoch": 0.8335942434978441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4175592742347787e-06,
+      "loss": 0.8555,
+      "step": 30352
+    },
+    {
+      "epoch": 0.8336217077256873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.417102771418627e-06,
+      "loss": 0.7706,
+      "step": 30353
+    },
+    {
+      "epoch": 0.8336491719535305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416646336513684e-06,
+      "loss": 0.6889,
+      "step": 30354
+    },
+    {
+      "epoch": 0.8336766361813738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416189969523558e-06,
+      "loss": 0.9009,
+      "step": 30355
+    },
+    {
+      "epoch": 0.8337041004092169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415733670451862e-06,
+      "loss": 0.7887,
+      "step": 30356
+    },
+    {
+      "epoch": 0.8337315646370602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4152774393022061e-06,
+      "loss": 0.8169,
+      "step": 30357
+    },
+    {
+      "epoch": 0.8337590288649035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414821276078201e-06,
+      "loss": 0.8052,
+      "step": 30358
+    },
+    {
+      "epoch": 0.8337864930927467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414365180783458e-06,
+      "loss": 0.8989,
+      "step": 30359
+    },
+    {
+      "epoch": 0.8338139573205899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.413909153421581e-06,
+      "loss": 0.8704,
+      "step": 30360
+    },
+    {
+      "epoch": 0.8338414215484332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4134531939961826e-06,
+      "loss": 0.7942,
+      "step": 30361
+    },
+    {
+      "epoch": 0.8338688857762764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4129973025108668e-06,
+      "loss": 0.8711,
+      "step": 30362
+    },
+    {
+      "epoch": 0.8338963500041197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4125414789692426e-06,
+      "loss": 0.8143,
+      "step": 30363
+    },
+    {
+      "epoch": 0.8339238142319628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4120857233749162e-06,
+      "loss": 0.8431,
+      "step": 30364
+    },
+    {
+      "epoch": 0.8339512784598061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4116300357314972e-06,
+      "loss": 0.7298,
+      "step": 30365
+    },
+    {
+      "epoch": 0.8339787426876494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4111744160425855e-06,
+      "loss": 0.8038,
+      "step": 30366
+    },
+    {
+      "epoch": 0.8340062069154925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4107188643117885e-06,
+      "loss": 0.8456,
+      "step": 30367
+    },
+    {
+      "epoch": 0.8340336711433358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4102633805427124e-06,
+      "loss": 0.8192,
+      "step": 30368
+    },
+    {
+      "epoch": 0.834061135371179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4098079647389606e-06,
+      "loss": 0.848,
+      "step": 30369
+    },
+    {
+      "epoch": 0.8340885995990223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4093526169041338e-06,
+      "loss": 0.8069,
+      "step": 30370
+    },
+    {
+      "epoch": 0.8341160638268655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4088973370418391e-06,
+      "loss": 0.7847,
+      "step": 30371
+    },
+    {
+      "epoch": 0.8341435280547087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4084421251556747e-06,
+      "loss": 0.7653,
+      "step": 30372
+    },
+    {
+      "epoch": 0.834170992282552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.407986981249243e-06,
+      "loss": 0.8322,
+      "step": 30373
+    },
+    {
+      "epoch": 0.8341984565103953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4075319053261482e-06,
+      "loss": 0.8362,
+      "step": 30374
+    },
+    {
+      "epoch": 0.8342259207382384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4070768973899884e-06,
+      "loss": 0.7857,
+      "step": 30375
+    },
+    {
+      "epoch": 0.8342533849660817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4066219574443674e-06,
+      "loss": 0.6893,
+      "step": 30376
+    },
+    {
+      "epoch": 0.8342808491939249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4061670854928788e-06,
+      "loss": 0.8473,
+      "step": 30377
+    },
+    {
+      "epoch": 0.8343083134217681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4057122815391288e-06,
+      "loss": 0.8747,
+      "step": 30378
+    },
+    {
+      "epoch": 0.8343357776496114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40525754558671e-06,
+      "loss": 0.828,
+      "step": 30379
+    },
+    {
+      "epoch": 0.8343632418774546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.404802877639222e-06,
+      "loss": 0.8304,
+      "step": 30380
+    },
+    {
+      "epoch": 0.8343907061052979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4043482777002637e-06,
+      "loss": 0.8331,
+      "step": 30381
+    },
+    {
+      "epoch": 0.834418170333141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4038937457734314e-06,
+      "loss": 0.7025,
+      "step": 30382
+    },
+    {
+      "epoch": 0.8344456345609843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4034392818623232e-06,
+      "loss": 0.8662,
+      "step": 30383
+    },
+    {
+      "epoch": 0.8344730987888276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402984885970532e-06,
+      "loss": 0.8367,
+      "step": 30384
+    },
+    {
+      "epoch": 0.8345005630166707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4025305581016569e-06,
+      "loss": 0.8273,
+      "step": 30385
+    },
+    {
+      "epoch": 0.834528027244514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4020762982592873e-06,
+      "loss": 0.8403,
+      "step": 30386
+    },
+    {
+      "epoch": 0.8345554914723573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4016221064470214e-06,
+      "loss": 0.8723,
+      "step": 30387
+    },
+    {
+      "epoch": 0.8345829557002005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4011679826684531e-06,
+      "loss": 0.8129,
+      "step": 30388
+    },
+    {
+      "epoch": 0.8346104199280437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400713926927173e-06,
+      "loss": 0.8531,
+      "step": 30389
+    },
+    {
+      "epoch": 0.8346378841558869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4002599392267757e-06,
+      "loss": 0.7759,
+      "step": 30390
+    },
+    {
+      "epoch": 0.8346653483837302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.399806019570853e-06,
+      "loss": 0.7365,
+      "step": 30391
+    },
+    {
+      "epoch": 0.8346928126115735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3993521679629961e-06,
+      "loss": 0.8447,
+      "step": 30392
+    },
+    {
+      "epoch": 0.8347202768394166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3988983844067983e-06,
+      "loss": 0.9037,
+      "step": 30393
+    },
+    {
+      "epoch": 0.8347477410672599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3984446689058495e-06,
+      "loss": 0.9141,
+      "step": 30394
+    },
+    {
+      "epoch": 0.8347752052951031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3979910214637348e-06,
+      "loss": 0.845,
+      "step": 30395
+    },
+    {
+      "epoch": 0.8348026695229464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3975374420840482e-06,
+      "loss": 0.8695,
+      "step": 30396
+    },
+    {
+      "epoch": 0.8348301337507896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3970839307703776e-06,
+      "loss": 0.7598,
+      "step": 30397
+    },
+    {
+      "epoch": 0.8348575979786328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3966304875263103e-06,
+      "loss": 0.8228,
+      "step": 30398
+    },
+    {
+      "epoch": 0.8348850622064761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3961771123554358e-06,
+      "loss": 0.7906,
+      "step": 30399
+    },
+    {
+      "epoch": 0.8349125264343193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.395723805261343e-06,
+      "loss": 0.8344,
+      "step": 30400
+    },
+    {
+      "epoch": 0.8349399906621625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3952705662476152e-06,
+      "loss": 0.7866,
+      "step": 30401
+    },
+    {
+      "epoch": 0.8349674548900058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3948173953178411e-06,
+      "loss": 0.834,
+      "step": 30402
+    },
+    {
+      "epoch": 0.834994919117849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3943642924756029e-06,
+      "loss": 0.7207,
+      "step": 30403
+    },
+    {
+      "epoch": 0.8350223833456922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3939112577244884e-06,
+      "loss": 0.8351,
+      "step": 30404
+    },
+    {
+      "epoch": 0.8350498475735355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3934582910680806e-06,
+      "loss": 0.8331,
+      "step": 30405
+    },
+    {
+      "epoch": 0.8350773118013787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3930053925099673e-06,
+      "loss": 0.7945,
+      "step": 30406
+    },
+    {
+      "epoch": 0.835104776029222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3925525620537285e-06,
+      "loss": 0.8548,
+      "step": 30407
+    },
+    {
+      "epoch": 0.8351322402570651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3920997997029463e-06,
+      "loss": 0.7776,
+      "step": 30408
+    },
+    {
+      "epoch": 0.8351597044849084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.391647105461208e-06,
+      "loss": 0.8167,
+      "step": 30409
+    },
+    {
+      "epoch": 0.8351871687127517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3911944793320897e-06,
+      "loss": 0.7437,
+      "step": 30410
+    },
+    {
+      "epoch": 0.8352146329405948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3907419213191754e-06,
+      "loss": 0.7775,
+      "step": 30411
+    },
+    {
+      "epoch": 0.8352420971684381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3902894314260484e-06,
+      "loss": 0.7876,
+      "step": 30412
+    },
+    {
+      "epoch": 0.8352695613962814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3898370096562853e-06,
+      "loss": 0.785,
+      "step": 30413
+    },
+    {
+      "epoch": 0.8352970256241246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3893846560134649e-06,
+      "loss": 0.841,
+      "step": 30414
+    },
+    {
+      "epoch": 0.8353244898519678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3889323705011704e-06,
+      "loss": 0.8389,
+      "step": 30415
+    },
+    {
+      "epoch": 0.835351954079811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3884801531229774e-06,
+      "loss": 0.9141,
+      "step": 30416
+    },
+    {
+      "epoch": 0.8353794183076543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3880280038824679e-06,
+      "loss": 0.7947,
+      "step": 30417
+    },
+    {
+      "epoch": 0.8354068825354976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3875759227832164e-06,
+      "loss": 0.7973,
+      "step": 30418
+    },
+    {
+      "epoch": 0.8354343467633407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3871239098287991e-06,
+      "loss": 0.7745,
+      "step": 30419
+    },
+    {
+      "epoch": 0.835461810991184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3866719650227933e-06,
+      "loss": 0.7735,
+      "step": 30420
+    },
+    {
+      "epoch": 0.8354892752190273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3862200883687748e-06,
+      "loss": 0.9046,
+      "step": 30421
+    },
+    {
+      "epoch": 0.8355167394468704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3857682798703197e-06,
+      "loss": 0.7784,
+      "step": 30422
+    },
+    {
+      "epoch": 0.8355442036747137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3853165395310031e-06,
+      "loss": 0.8267,
+      "step": 30423
+    },
+    {
+      "epoch": 0.8355716679025569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3848648673544018e-06,
+      "loss": 0.7787,
+      "step": 30424
+    },
+    {
+      "epoch": 0.8355991321304002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3844132633440866e-06,
+      "loss": 0.8602,
+      "step": 30425
+    },
+    {
+      "epoch": 0.8356265963582434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3839617275036278e-06,
+      "loss": 0.7048,
+      "step": 30426
+    },
+    {
+      "epoch": 0.8356540605860866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3835102598366024e-06,
+      "loss": 0.805,
+      "step": 30427
+    },
+    {
+      "epoch": 0.8356815248139299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.383058860346581e-06,
+      "loss": 0.8008,
+      "step": 30428
+    },
+    {
+      "epoch": 0.835708989041773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3826075290371367e-06,
+      "loss": 0.7021,
+      "step": 30429
+    },
+    {
+      "epoch": 0.8357364532696163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3821562659118404e-06,
+      "loss": 0.8363,
+      "step": 30430
+    },
+    {
+      "epoch": 0.8357639174974596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3817050709742607e-06,
+      "loss": 0.8854,
+      "step": 30431
+    },
+    {
+      "epoch": 0.8357913817253028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3812539442279681e-06,
+      "loss": 0.7083,
+      "step": 30432
+    },
+    {
+      "epoch": 0.835818845953146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3808028856765366e-06,
+      "loss": 0.7787,
+      "step": 30433
+    },
+    {
+      "epoch": 0.8358463101809893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3803518953235285e-06,
+      "loss": 0.8085,
+      "step": 30434
+    },
+    {
+      "epoch": 0.8358737744088325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3799009731725143e-06,
+      "loss": 0.7595,
+      "step": 30435
+    },
+    {
+      "epoch": 0.8359012386366758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.379450119227066e-06,
+      "loss": 0.7306,
+      "step": 30436
+    },
+    {
+      "epoch": 0.8359287028645189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3789993334907447e-06,
+      "loss": 0.7557,
+      "step": 30437
+    },
+    {
+      "epoch": 0.8359561670923622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3785486159671203e-06,
+      "loss": 0.8253,
+      "step": 30438
+    },
+    {
+      "epoch": 0.8359836313202055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3780979666597593e-06,
+      "loss": 0.8506,
+      "step": 30439
+    },
+    {
+      "epoch": 0.8360110955480486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3776473855722273e-06,
+      "loss": 0.8208,
+      "step": 30440
+    },
+    {
+      "epoch": 0.8360385597758919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3771968727080897e-06,
+      "loss": 0.7802,
+      "step": 30441
+    },
+    {
+      "epoch": 0.8360660240037351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.376746428070912e-06,
+      "loss": 0.8086,
+      "step": 30442
+    },
+    {
+      "epoch": 0.8360934882315784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3762960516642543e-06,
+      "loss": 0.8409,
+      "step": 30443
+    },
+    {
+      "epoch": 0.8361209524594216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3758457434916828e-06,
+      "loss": 0.7959,
+      "step": 30444
+    },
+    {
+      "epoch": 0.8361484166872648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3753955035567612e-06,
+      "loss": 0.8434,
+      "step": 30445
+    },
+    {
+      "epoch": 0.8361758809151081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.37494533186305e-06,
+      "loss": 0.8911,
+      "step": 30446
+    },
+    {
+      "epoch": 0.8362033451429514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3744952284141154e-06,
+      "loss": 0.6945,
+      "step": 30447
+    },
+    {
+      "epoch": 0.8362308093707945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3740451932135125e-06,
+      "loss": 0.7858,
+      "step": 30448
+    },
+    {
+      "epoch": 0.8362582735986378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3735952262648089e-06,
+      "loss": 0.7467,
+      "step": 30449
+    },
+    {
+      "epoch": 0.836285737826481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.373145327571559e-06,
+      "loss": 0.8522,
+      "step": 30450
+    },
+    {
+      "epoch": 0.8363132020543242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372695497137324e-06,
+      "loss": 0.7865,
+      "step": 30451
+    },
+    {
+      "epoch": 0.8363406662821675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3722457349656648e-06,
+      "loss": 0.8811,
+      "step": 30452
+    },
+    {
+      "epoch": 0.8363681305100107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3717960410601415e-06,
+      "loss": 0.7793,
+      "step": 30453
+    },
+    {
+      "epoch": 0.836395594737854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3713464154243083e-06,
+      "loss": 0.778,
+      "step": 30454
+    },
+    {
+      "epoch": 0.8364230589656971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3708968580617243e-06,
+      "loss": 0.7745,
+      "step": 30455
+    },
+    {
+      "epoch": 0.8364505231935404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3704473689759468e-06,
+      "loss": 0.824,
+      "step": 30456
+    },
+    {
+      "epoch": 0.8364779874213837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.369997948170535e-06,
+      "loss": 0.7809,
+      "step": 30457
+    },
+    {
+      "epoch": 0.8365054516492268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3695485956490407e-06,
+      "loss": 0.8896,
+      "step": 30458
+    },
+    {
+      "epoch": 0.8365329158770701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3690993114150219e-06,
+      "loss": 0.8625,
+      "step": 30459
+    },
+    {
+      "epoch": 0.8365603801049134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3686500954720316e-06,
+      "loss": 0.8042,
+      "step": 30460
+    },
+    {
+      "epoch": 0.8365878443327566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3682009478236246e-06,
+      "loss": 0.7732,
+      "step": 30461
+    },
+    {
+      "epoch": 0.8366153085605998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.367751868473356e-06,
+      "loss": 0.7486,
+      "step": 30462
+    },
+    {
+      "epoch": 0.836642772788443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3673028574247782e-06,
+      "loss": 0.7891,
+      "step": 30463
+    },
+    {
+      "epoch": 0.8366702370162863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3668539146814464e-06,
+      "loss": 0.8417,
+      "step": 30464
+    },
+    {
+      "epoch": 0.8366977012441296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3664050402469076e-06,
+      "loss": 0.9293,
+      "step": 30465
+    },
+    {
+      "epoch": 0.8367251654719727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.365956234124719e-06,
+      "loss": 0.8344,
+      "step": 30466
+    },
+    {
+      "epoch": 0.836752629699816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3655074963184278e-06,
+      "loss": 0.7766,
+      "step": 30467
+    },
+    {
+      "epoch": 0.8367800939276592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3650588268315845e-06,
+      "loss": 0.8961,
+      "step": 30468
+    },
+    {
+      "epoch": 0.8368075581555025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3646102256677419e-06,
+      "loss": 0.7762,
+      "step": 30469
+    },
+    {
+      "epoch": 0.8368350223833457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364161692830448e-06,
+      "loss": 0.8435,
+      "step": 30470
+    },
+    {
+      "epoch": 0.8368624866111889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3637132283232535e-06,
+      "loss": 0.7458,
+      "step": 30471
+    },
+    {
+      "epoch": 0.8368899508390322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3632648321497044e-06,
+      "loss": 0.7342,
+      "step": 30472
+    },
+    {
+      "epoch": 0.8369174150668754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3628165043133501e-06,
+      "loss": 0.7633,
+      "step": 30473
+    },
+    {
+      "epoch": 0.8369448792947186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3623682448177367e-06,
+      "loss": 0.8443,
+      "step": 30474
+    },
+    {
+      "epoch": 0.8369723435225619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3619200536664102e-06,
+      "loss": 0.7522,
+      "step": 30475
+    },
+    {
+      "epoch": 0.8369998077504051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.36147193086292e-06,
+      "loss": 0.9061,
+      "step": 30476
+    },
+    {
+      "epoch": 0.8370272719782483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3610238764108107e-06,
+      "loss": 0.7335,
+      "step": 30477
+    },
+    {
+      "epoch": 0.8370547362060916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3605758903136257e-06,
+      "loss": 0.841,
+      "step": 30478
+    },
+    {
+      "epoch": 0.8370822004339348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3601279725749106e-06,
+      "loss": 0.8332,
+      "step": 30479
+    },
+    {
+      "epoch": 0.837109664661778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3596801231982094e-06,
+      "loss": 0.8103,
+      "step": 30480
+    },
+    {
+      "epoch": 0.8371371288896212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.359232342187068e-06,
+      "loss": 0.8359,
+      "step": 30481
+    },
+    {
+      "epoch": 0.8371645931174645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3587846295450257e-06,
+      "loss": 0.8278,
+      "step": 30482
+    },
+    {
+      "epoch": 0.8371920573453078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358336985275629e-06,
+      "loss": 0.6936,
+      "step": 30483
+    },
+    {
+      "epoch": 0.8372195215731509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3578894093824159e-06,
+      "loss": 0.7582,
+      "step": 30484
+    },
+    {
+      "epoch": 0.8372469858009942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3574419018689279e-06,
+      "loss": 0.81,
+      "step": 30485
+    },
+    {
+      "epoch": 0.8372744500288375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3569944627387088e-06,
+      "loss": 0.7097,
+      "step": 30486
+    },
+    {
+      "epoch": 0.8373019142566807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3565470919952962e-06,
+      "loss": 0.8641,
+      "step": 30487
+    },
+    {
+      "epoch": 0.8373293784845239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3560997896422346e-06,
+      "loss": 0.8117,
+      "step": 30488
+    },
+    {
+      "epoch": 0.8373568427123671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3556525556830568e-06,
+      "loss": 0.8214,
+      "step": 30489
+    },
+    {
+      "epoch": 0.8373843069402104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.355205390121307e-06,
+      "loss": 0.8093,
+      "step": 30490
+    },
+    {
+      "epoch": 0.8374117711680537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3547582929605185e-06,
+      "loss": 0.8248,
+      "step": 30491
+    },
+    {
+      "epoch": 0.8374392353958968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3543112642042312e-06,
+      "loss": 0.8228,
+      "step": 30492
+    },
+    {
+      "epoch": 0.8374666996237401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.353864303855983e-06,
+      "loss": 0.8446,
+      "step": 30493
+    },
+    {
+      "epoch": 0.8374941638515833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3534174119193088e-06,
+      "loss": 0.8778,
+      "step": 30494
+    },
+    {
+      "epoch": 0.8375216280794265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3529705883977472e-06,
+      "loss": 0.8601,
+      "step": 30495
+    },
+    {
+      "epoch": 0.8375490923072698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3525238332948299e-06,
+      "loss": 0.7513,
+      "step": 30496
+    },
+    {
+      "epoch": 0.837576556535113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.352077146614097e-06,
+      "loss": 0.9033,
+      "step": 30497
+    },
+    {
+      "epoch": 0.8376040207629563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3516305283590758e-06,
+      "loss": 0.8152,
+      "step": 30498
+    },
+    {
+      "epoch": 0.8376314849907995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3511839785333048e-06,
+      "loss": 0.7241,
+      "step": 30499
+    },
+    {
+      "epoch": 0.8376589492186427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3507374971403165e-06,
+      "loss": 0.8678,
+      "step": 30500
+    },
+    {
+      "epoch": 0.837686413446486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350291084183646e-06,
+      "loss": 0.8354,
+      "step": 30501
+    },
+    {
+      "epoch": 0.8377138776743291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34984473966682e-06,
+      "loss": 0.8729,
+      "step": 30502
+    },
+    {
+      "epoch": 0.8377413419021724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3493984635933732e-06,
+      "loss": 0.817,
+      "step": 30503
+    },
+    {
+      "epoch": 0.8377688061300157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3489522559668378e-06,
+      "loss": 0.6948,
+      "step": 30504
+    },
+    {
+      "epoch": 0.8377962703578589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3485061167907443e-06,
+      "loss": 0.8281,
+      "step": 30505
+    },
+    {
+      "epoch": 0.8378237345857021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3480600460686221e-06,
+      "loss": 0.8228,
+      "step": 30506
+    },
+    {
+      "epoch": 0.8378511988135453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3476140438039976e-06,
+      "loss": 0.8295,
+      "step": 30507
+    },
+    {
+      "epoch": 0.8378786630413886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3471681100004031e-06,
+      "loss": 0.8103,
+      "step": 30508
+    },
+    {
+      "epoch": 0.8379061272692319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3467222446613658e-06,
+      "loss": 0.7746,
+      "step": 30509
+    },
+    {
+      "epoch": 0.837933591497075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3462764477904144e-06,
+      "loss": 0.7395,
+      "step": 30510
+    },
+    {
+      "epoch": 0.8379610557249183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3458307193910758e-06,
+      "loss": 0.8641,
+      "step": 30511
+    },
+    {
+      "epoch": 0.8379885199527616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3453850594668782e-06,
+      "loss": 0.8216,
+      "step": 30512
+    },
+    {
+      "epoch": 0.8380159841806047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3449394680213457e-06,
+      "loss": 0.8258,
+      "step": 30513
+    },
+    {
+      "epoch": 0.838043448408448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3444939450580063e-06,
+      "loss": 0.9105,
+      "step": 30514
+    },
+    {
+      "epoch": 0.8380709126362912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3440484905803807e-06,
+      "loss": 0.7953,
+      "step": 30515
+    },
+    {
+      "epoch": 0.8380983768641345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3436031045919972e-06,
+      "loss": 0.8126,
+      "step": 30516
+    },
+    {
+      "epoch": 0.8381258410919777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3431577870963796e-06,
+      "loss": 0.7981,
+      "step": 30517
+    },
+    {
+      "epoch": 0.8381533053198209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3427125380970529e-06,
+      "loss": 0.7257,
+      "step": 30518
+    },
+    {
+      "epoch": 0.8381807695476642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3422673575975353e-06,
+      "loss": 0.8601,
+      "step": 30519
+    },
+    {
+      "epoch": 0.8382082337755073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341822245601353e-06,
+      "loss": 0.7295,
+      "step": 30520
+    },
+    {
+      "epoch": 0.8382356980033506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3413772021120287e-06,
+      "loss": 0.749,
+      "step": 30521
+    },
+    {
+      "epoch": 0.8382631622311939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3409322271330793e-06,
+      "loss": 0.797,
+      "step": 30522
+    },
+    {
+      "epoch": 0.8382906264590371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340487320668028e-06,
+      "loss": 0.7955,
+      "step": 30523
+    },
+    {
+      "epoch": 0.8383180906868803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3400424827203984e-06,
+      "loss": 0.8451,
+      "step": 30524
+    },
+    {
+      "epoch": 0.8383455549147236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3395977132937054e-06,
+      "loss": 0.8578,
+      "step": 30525
+    },
+    {
+      "epoch": 0.8383730191425668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3391530123914699e-06,
+      "loss": 0.7945,
+      "step": 30526
+    },
+    {
+      "epoch": 0.8384004833704101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3387083800172097e-06,
+      "loss": 0.883,
+      "step": 30527
+    },
+    {
+      "epoch": 0.8384279475982532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3382638161744443e-06,
+      "loss": 0.7747,
+      "step": 30528
+    },
+    {
+      "epoch": 0.8384554118260965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3378193208666933e-06,
+      "loss": 0.8226,
+      "step": 30529
+    },
+    {
+      "epoch": 0.8384828760539398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3373748940974707e-06,
+      "loss": 0.7731,
+      "step": 30530
+    },
+    {
+      "epoch": 0.838510340281783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3369305358702922e-06,
+      "loss": 0.8611,
+      "step": 30531
+    },
+    {
+      "epoch": 0.8385378045096262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3364862461886742e-06,
+      "loss": 0.9247,
+      "step": 30532
+    },
+    {
+      "epoch": 0.8385652687374694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3360420250561334e-06,
+      "loss": 0.8472,
+      "step": 30533
+    },
+    {
+      "epoch": 0.8385927329653127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3355978724761842e-06,
+      "loss": 0.8037,
+      "step": 30534
+    },
+    {
+      "epoch": 0.838620197193156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3351537884523402e-06,
+      "loss": 0.7434,
+      "step": 30535
+    },
+    {
+      "epoch": 0.8386476614209991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3347097729881198e-06,
+      "loss": 0.7918,
+      "step": 30536
+    },
+    {
+      "epoch": 0.8386751256488424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334265826087029e-06,
+      "loss": 0.7354,
+      "step": 30537
+    },
+    {
+      "epoch": 0.8387025898766857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3338219477525849e-06,
+      "loss": 0.7489,
+      "step": 30538
+    },
+    {
+      "epoch": 0.8387300541045288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3333781379882981e-06,
+      "loss": 0.7859,
+      "step": 30539
+    },
+    {
+      "epoch": 0.8387575183323721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3329343967976793e-06,
+      "loss": 0.8055,
+      "step": 30540
+    },
+    {
+      "epoch": 0.8387849825602153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332490724184241e-06,
+      "loss": 0.9341,
+      "step": 30541
+    },
+    {
+      "epoch": 0.8388124467880586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3320471201514962e-06,
+      "loss": 0.7699,
+      "step": 30542
+    },
+    {
+      "epoch": 0.8388399110159018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3316035847029507e-06,
+      "loss": 0.8168,
+      "step": 30543
+    },
+    {
+      "epoch": 0.838867375243745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3311601178421141e-06,
+      "loss": 0.8009,
+      "step": 30544
+    },
+    {
+      "epoch": 0.8388948394715883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3307167195725e-06,
+      "loss": 0.875,
+      "step": 30545
+    },
+    {
+      "epoch": 0.8389223036994314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3302733898976105e-06,
+      "loss": 0.8495,
+      "step": 30546
+    },
+    {
+      "epoch": 0.8389497679272747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.329830128820956e-06,
+      "loss": 0.6872,
+      "step": 30547
+    },
+    {
+      "epoch": 0.838977232155118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3293869363460466e-06,
+      "loss": 0.7529,
+      "step": 30548
+    },
+    {
+      "epoch": 0.8390046963829612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3289438124763842e-06,
+      "loss": 0.8694,
+      "step": 30549
+    },
+    {
+      "epoch": 0.8390321606108044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3285007572154773e-06,
+      "loss": 0.7698,
+      "step": 30550
+    },
+    {
+      "epoch": 0.8390596248386477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3280577705668319e-06,
+      "loss": 0.8435,
+      "step": 30551
+    },
+    {
+      "epoch": 0.8390870890664909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3276148525339517e-06,
+      "loss": 0.8459,
+      "step": 30552
+    },
+    {
+      "epoch": 0.8391145532943342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.327172003120345e-06,
+      "loss": 0.7819,
+      "step": 30553
+    },
+    {
+      "epoch": 0.8391420175221773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3267292223295126e-06,
+      "loss": 0.9619,
+      "step": 30554
+    },
+    {
+      "epoch": 0.8391694817500206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3262865101649557e-06,
+      "loss": 0.786,
+      "step": 30555
+    },
+    {
+      "epoch": 0.8391969459778639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.325843866630181e-06,
+      "loss": 0.7404,
+      "step": 30556
+    },
+    {
+      "epoch": 0.839224410205707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3254012917286884e-06,
+      "loss": 0.8374,
+      "step": 30557
+    },
+    {
+      "epoch": 0.8392518744335503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3249587854639823e-06,
+      "loss": 0.7091,
+      "step": 30558
+    },
+    {
+      "epoch": 0.8392793386613935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3245163478395618e-06,
+      "loss": 0.9192,
+      "step": 30559
+    },
+    {
+      "epoch": 0.8393068028892368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3240739788589308e-06,
+      "loss": 0.8159,
+      "step": 30560
+    },
+    {
+      "epoch": 0.83933426711708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3236316785255855e-06,
+      "loss": 0.9098,
+      "step": 30561
+    },
+    {
+      "epoch": 0.8393617313449232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3231894468430283e-06,
+      "loss": 0.8181,
+      "step": 30562
+    },
+    {
+      "epoch": 0.8393891955727665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3227472838147559e-06,
+      "loss": 0.8416,
+      "step": 30563
+    },
+    {
+      "epoch": 0.8394166598006098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3223051894442685e-06,
+      "loss": 0.8295,
+      "step": 30564
+    },
+    {
+      "epoch": 0.8394441240284529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3218631637350655e-06,
+      "loss": 0.7741,
+      "step": 30565
+    },
+    {
+      "epoch": 0.8394715882562962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3214212066906406e-06,
+      "loss": 0.7855,
+      "step": 30566
+    },
+    {
+      "epoch": 0.8394990524841394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3209793183144926e-06,
+      "loss": 0.7889,
+      "step": 30567
+    },
+    {
+      "epoch": 0.8395265167119826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3205374986101193e-06,
+      "loss": 0.6947,
+      "step": 30568
+    },
+    {
+      "epoch": 0.8395539809398259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3200957475810161e-06,
+      "loss": 0.9315,
+      "step": 30569
+    },
+    {
+      "epoch": 0.8395814451676691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3196540652306767e-06,
+      "loss": 0.822,
+      "step": 30570
+    },
+    {
+      "epoch": 0.8396089093955124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3192124515625982e-06,
+      "loss": 0.8809,
+      "step": 30571
+    },
+    {
+      "epoch": 0.8396363736233555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3187709065802723e-06,
+      "loss": 0.8985,
+      "step": 30572
+    },
+    {
+      "epoch": 0.8396638378511988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.318329430287193e-06,
+      "loss": 0.8325,
+      "step": 30573
+    },
+    {
+      "epoch": 0.8396913020790421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.317888022686854e-06,
+      "loss": 0.749,
+      "step": 30574
+    },
+    {
+      "epoch": 0.8397187663068852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3174466837827481e-06,
+      "loss": 0.8824,
+      "step": 30575
+    },
+    {
+      "epoch": 0.8397462305347285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3170054135783671e-06,
+      "loss": 0.8655,
+      "step": 30576
+    },
+    {
+      "epoch": 0.8397736947625718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.316564212077206e-06,
+      "loss": 0.716,
+      "step": 30577
+    },
+    {
+      "epoch": 0.839801158990415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3161230792827518e-06,
+      "loss": 0.7972,
+      "step": 30578
+    },
+    {
+      "epoch": 0.8398286232182582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.315682015198493e-06,
+      "loss": 0.6426,
+      "step": 30579
+    },
+    {
+      "epoch": 0.8398560874461014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.315241019827922e-06,
+      "loss": 0.8422,
+      "step": 30580
+    },
+    {
+      "epoch": 0.8398835516739447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3148000931745287e-06,
+      "loss": 0.7525,
+      "step": 30581
+    },
+    {
+      "epoch": 0.839911015901788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3143592352418022e-06,
+      "loss": 0.8929,
+      "step": 30582
+    },
+    {
+      "epoch": 0.8399384801296311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313918446033231e-06,
+      "loss": 0.8152,
+      "step": 30583
+    },
+    {
+      "epoch": 0.8399659443574744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3134777255523002e-06,
+      "loss": 0.8351,
+      "step": 30584
+    },
+    {
+      "epoch": 0.8399934085853176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3130370738025e-06,
+      "loss": 0.8011,
+      "step": 30585
+    },
+    {
+      "epoch": 0.8400208728131608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3125964907873134e-06,
+      "loss": 0.7774,
+      "step": 30586
+    },
+    {
+      "epoch": 0.8400483370410041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3121559765102288e-06,
+      "loss": 0.7798,
+      "step": 30587
+    },
+    {
+      "epoch": 0.8400758012688473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.311715530974732e-06,
+      "loss": 0.8517,
+      "step": 30588
+    },
+    {
+      "epoch": 0.8401032654966906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3112751541843084e-06,
+      "loss": 0.7982,
+      "step": 30589
+    },
+    {
+      "epoch": 0.8401307297245338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3108348461424402e-06,
+      "loss": 0.921,
+      "step": 30590
+    },
+    {
+      "epoch": 0.840158193952377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3103946068526118e-06,
+      "loss": 0.7863,
+      "step": 30591
+    },
+    {
+      "epoch": 0.8401856581802203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.309954436318308e-06,
+      "loss": 0.8039,
+      "step": 30592
+    },
+    {
+      "epoch": 0.8402131224080635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3095143345430127e-06,
+      "loss": 0.7962,
+      "step": 30593
+    },
+    {
+      "epoch": 0.8402405866359067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090743015302042e-06,
+      "loss": 0.9345,
+      "step": 30594
+    },
+    {
+      "epoch": 0.84026805086375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3086343372833687e-06,
+      "loss": 0.9586,
+      "step": 30595
+    },
+    {
+      "epoch": 0.8402955150915932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3081944418059833e-06,
+      "loss": 0.9342,
+      "step": 30596
+    },
+    {
+      "epoch": 0.8403229793194364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3077546151015296e-06,
+      "loss": 0.8199,
+      "step": 30597
+    },
+    {
+      "epoch": 0.8403504435472796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3073148571734895e-06,
+      "loss": 0.8965,
+      "step": 30598
+    },
+    {
+      "epoch": 0.8403779077751229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30687516802534e-06,
+      "loss": 0.8398,
+      "step": 30599
+    },
+    {
+      "epoch": 0.8404053720029662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3064355476605629e-06,
+      "loss": 0.8712,
+      "step": 30600
+    },
+    {
+      "epoch": 0.8404328362308093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3059959960826363e-06,
+      "loss": 0.8448,
+      "step": 30601
+    },
+    {
+      "epoch": 0.8404603004586526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3055565132950376e-06,
+      "loss": 0.7492,
+      "step": 30602
+    },
+    {
+      "epoch": 0.8404877646864959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3051170993012419e-06,
+      "loss": 0.7398,
+      "step": 30603
+    },
+    {
+      "epoch": 0.840515228914339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3046777541047273e-06,
+      "loss": 0.7234,
+      "step": 30604
+    },
+    {
+      "epoch": 0.8405426931421823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.304238477708971e-06,
+      "loss": 0.8214,
+      "step": 30605
+    },
+    {
+      "epoch": 0.8405701573700255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3037992701174485e-06,
+      "loss": 0.9032,
+      "step": 30606
+    },
+    {
+      "epoch": 0.8405976215978688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3033601313336363e-06,
+      "loss": 0.8224,
+      "step": 30607
+    },
+    {
+      "epoch": 0.840625085825712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3029210613610056e-06,
+      "loss": 0.8443,
+      "step": 30608
+    },
+    {
+      "epoch": 0.8406525500535552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3024820602030342e-06,
+      "loss": 0.7622,
+      "step": 30609
+    },
+    {
+      "epoch": 0.8406800142813985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3020431278631918e-06,
+      "loss": 0.7908,
+      "step": 30610
+    },
+    {
+      "epoch": 0.8407074785092418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3016042643449533e-06,
+      "loss": 0.694,
+      "step": 30611
+    },
+    {
+      "epoch": 0.8407349427370849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3011654696517918e-06,
+      "loss": 0.7766,
+      "step": 30612
+    },
+    {
+      "epoch": 0.8407624069649282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3007267437871808e-06,
+      "loss": 0.8362,
+      "step": 30613
+    },
+    {
+      "epoch": 0.8407898711927714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3002880867545863e-06,
+      "loss": 0.8738,
+      "step": 30614
+    },
+    {
+      "epoch": 0.8408173354206147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2998494985574839e-06,
+      "loss": 0.7739,
+      "step": 30615
+    },
+    {
+      "epoch": 0.8408447996484579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2994109791993415e-06,
+      "loss": 0.7862,
+      "step": 30616
+    },
+    {
+      "epoch": 0.8408722638763011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298972528683633e-06,
+      "loss": 0.7525,
+      "step": 30617
+    },
+    {
+      "epoch": 0.8408997281041444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2985341470138213e-06,
+      "loss": 0.7768,
+      "step": 30618
+    },
+    {
+      "epoch": 0.8409271923319875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2980958341933802e-06,
+      "loss": 0.778,
+      "step": 30619
+    },
+    {
+      "epoch": 0.8409546565598308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2976575902257726e-06,
+      "loss": 0.8481,
+      "step": 30620
+    },
+    {
+      "epoch": 0.8409821207876741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2972194151144702e-06,
+      "loss": 0.8029,
+      "step": 30621
+    },
+    {
+      "epoch": 0.8410095850155173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296781308862939e-06,
+      "loss": 0.7606,
+      "step": 30622
+    },
+    {
+      "epoch": 0.8410370492433605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2963432714746448e-06,
+      "loss": 0.7977,
+      "step": 30623
+    },
+    {
+      "epoch": 0.8410645134712038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2959053029530566e-06,
+      "loss": 0.7932,
+      "step": 30624
+    },
+    {
+      "epoch": 0.841091977699047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2954674033016356e-06,
+      "loss": 0.7425,
+      "step": 30625
+    },
+    {
+      "epoch": 0.8411194419268903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2950295725238505e-06,
+      "loss": 0.8059,
+      "step": 30626
+    },
+    {
+      "epoch": 0.8411469061547334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2945918106231602e-06,
+      "loss": 0.8077,
+      "step": 30627
+    },
+    {
+      "epoch": 0.8411743703825767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2941541176030325e-06,
+      "loss": 0.8122,
+      "step": 30628
+    },
+    {
+      "epoch": 0.84120183461042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2937164934669299e-06,
+      "loss": 0.8501,
+      "step": 30629
+    },
+    {
+      "epoch": 0.8412292988382631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2932789382183164e-06,
+      "loss": 0.8288,
+      "step": 30630
+    },
+    {
+      "epoch": 0.8412567630661064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2928414518606503e-06,
+      "loss": 0.9142,
+      "step": 30631
+    },
+    {
+      "epoch": 0.8412842272939496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2924040343973964e-06,
+      "loss": 0.8989,
+      "step": 30632
+    },
+    {
+      "epoch": 0.8413116915217929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2919666858320157e-06,
+      "loss": 0.7208,
+      "step": 30633
+    },
+    {
+      "epoch": 0.8413391557496361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2915294061679662e-06,
+      "loss": 0.8216,
+      "step": 30634
+    },
+    {
+      "epoch": 0.8413666199774793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2910921954087097e-06,
+      "loss": 0.7425,
+      "step": 30635
+    },
+    {
+      "epoch": 0.8413940842053226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2906550535577078e-06,
+      "loss": 0.7943,
+      "step": 30636
+    },
+    {
+      "epoch": 0.8414215484331659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2902179806184134e-06,
+      "loss": 0.8175,
+      "step": 30637
+    },
+    {
+      "epoch": 0.841449012661009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2897809765942893e-06,
+      "loss": 0.7596,
+      "step": 30638
+    },
+    {
+      "epoch": 0.8414764768888523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2893440414887915e-06,
+      "loss": 0.8315,
+      "step": 30639
+    },
+    {
+      "epoch": 0.8415039411166955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2889071753053773e-06,
+      "loss": 0.853,
+      "step": 30640
+    },
+    {
+      "epoch": 0.8415314053445387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2884703780475073e-06,
+      "loss": 0.7746,
+      "step": 30641
+    },
+    {
+      "epoch": 0.841558869572382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2880336497186329e-06,
+      "loss": 0.7984,
+      "step": 30642
+    },
+    {
+      "epoch": 0.8415863338002252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2875969903222097e-06,
+      "loss": 0.8702,
+      "step": 30643
+    },
+    {
+      "epoch": 0.8416137980280685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2871603998616944e-06,
+      "loss": 0.7698,
+      "step": 30644
+    },
+    {
+      "epoch": 0.8416412622559116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28672387834054e-06,
+      "loss": 0.7809,
+      "step": 30645
+    },
+    {
+      "epoch": 0.8416687264837549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2862874257622016e-06,
+      "loss": 0.8313,
+      "step": 30646
+    },
+    {
+      "epoch": 0.8416961907115982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.285851042130133e-06,
+      "loss": 0.8275,
+      "step": 30647
+    },
+    {
+      "epoch": 0.8417236549394413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.285414727447788e-06,
+      "loss": 0.8277,
+      "step": 30648
+    },
+    {
+      "epoch": 0.8417511191672846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2849784817186161e-06,
+      "loss": 0.8017,
+      "step": 30649
+    },
+    {
+      "epoch": 0.8417785833951279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2845423049460715e-06,
+      "loss": 0.8289,
+      "step": 30650
+    },
+    {
+      "epoch": 0.8418060476229711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2841061971336032e-06,
+      "loss": 0.8768,
+      "step": 30651
+    },
+    {
+      "epoch": 0.8418335118508143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2836701582846622e-06,
+      "loss": 0.7917,
+      "step": 30652
+    },
+    {
+      "epoch": 0.8418609760786575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283234188402699e-06,
+      "loss": 0.7932,
+      "step": 30653
+    },
+    {
+      "epoch": 0.8418884403065008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2827982874911659e-06,
+      "loss": 0.7073,
+      "step": 30654
+    },
+    {
+      "epoch": 0.8419159045343441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2823624555535075e-06,
+      "loss": 0.8834,
+      "step": 30655
+    },
+    {
+      "epoch": 0.8419433687621872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.281926692593174e-06,
+      "loss": 0.8853,
+      "step": 30656
+    },
+    {
+      "epoch": 0.8419708329900305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.281490998613616e-06,
+      "loss": 0.7353,
+      "step": 30657
+    },
+    {
+      "epoch": 0.8419982972178737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2810553736182764e-06,
+      "loss": 0.8943,
+      "step": 30658
+    },
+    {
+      "epoch": 0.842025761445717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2806198176106032e-06,
+      "loss": 0.8054,
+      "step": 30659
+    },
+    {
+      "epoch": 0.8420532256735602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2801843305940464e-06,
+      "loss": 0.8602,
+      "step": 30660
+    },
+    {
+      "epoch": 0.8420806899014034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2797489125720464e-06,
+      "loss": 0.7821,
+      "step": 30661
+    },
+    {
+      "epoch": 0.8421081541292467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2793135635480514e-06,
+      "loss": 0.8397,
+      "step": 30662
+    },
+    {
+      "epoch": 0.8421356183570899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2788782835255054e-06,
+      "loss": 0.7467,
+      "step": 30663
+    },
+    {
+      "epoch": 0.8421630825849331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2784430725078522e-06,
+      "loss": 0.7884,
+      "step": 30664
+    },
+    {
+      "epoch": 0.8421905468127764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2780079304985393e-06,
+      "loss": 0.7494,
+      "step": 30665
+    },
+    {
+      "epoch": 0.8422180110406196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277572857501005e-06,
+      "loss": 0.8144,
+      "step": 30666
+    },
+    {
+      "epoch": 0.8422454752684628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2771378535186906e-06,
+      "loss": 0.7796,
+      "step": 30667
+    },
+    {
+      "epoch": 0.8422729394963061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2767029185550416e-06,
+      "loss": 0.844,
+      "step": 30668
+    },
+    {
+      "epoch": 0.8423004037241493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2762680526134974e-06,
+      "loss": 0.8499,
+      "step": 30669
+    },
+    {
+      "epoch": 0.8423278679519925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2758332556974994e-06,
+      "loss": 0.7926,
+      "step": 30670
+    },
+    {
+      "epoch": 0.8423553321798357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2753985278104885e-06,
+      "loss": 0.6702,
+      "step": 30671
+    },
+    {
+      "epoch": 0.842382796407679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2749638689559062e-06,
+      "loss": 0.8689,
+      "step": 30672
+    },
+    {
+      "epoch": 0.8424102606355223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2745292791371876e-06,
+      "loss": 0.8239,
+      "step": 30673
+    },
+    {
+      "epoch": 0.8424377248633654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2740947583577755e-06,
+      "loss": 0.8522,
+      "step": 30674
+    },
+    {
+      "epoch": 0.8424651890912087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2736603066211028e-06,
+      "loss": 0.8032,
+      "step": 30675
+    },
+    {
+      "epoch": 0.842492653319052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.273225923930611e-06,
+      "loss": 0.8314,
+      "step": 30676
+    },
+    {
+      "epoch": 0.8425201175468952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2727916102897352e-06,
+      "loss": 0.8856,
+      "step": 30677
+    },
+    {
+      "epoch": 0.8425475817747384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.272357365701915e-06,
+      "loss": 0.8346,
+      "step": 30678
+    },
+    {
+      "epoch": 0.8425750460025816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271923190170583e-06,
+      "loss": 0.9022,
+      "step": 30679
+    },
+    {
+      "epoch": 0.8426025102304249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2714890836991744e-06,
+      "loss": 0.7737,
+      "step": 30680
+    },
+    {
+      "epoch": 0.8426299744582681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2710550462911286e-06,
+      "loss": 0.7784,
+      "step": 30681
+    },
+    {
+      "epoch": 0.8426574386861113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.270621077949873e-06,
+      "loss": 0.7914,
+      "step": 30682
+    },
+    {
+      "epoch": 0.8426849029139546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2701871786788455e-06,
+      "loss": 0.8229,
+      "step": 30683
+    },
+    {
+      "epoch": 0.8427123671417978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2697533484814818e-06,
+      "loss": 0.8054,
+      "step": 30684
+    },
+    {
+      "epoch": 0.842739831369641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2693195873612073e-06,
+      "loss": 0.7442,
+      "step": 30685
+    },
+    {
+      "epoch": 0.8427672955974843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.26888589532146e-06,
+      "loss": 0.7583,
+      "step": 30686
+    },
+    {
+      "epoch": 0.8427947598253275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2684522723656689e-06,
+      "loss": 0.7684,
+      "step": 30687
+    },
+    {
+      "epoch": 0.8428222240531708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2680187184972648e-06,
+      "loss": 0.8493,
+      "step": 30688
+    },
+    {
+      "epoch": 0.842849688281014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2675852337196827e-06,
+      "loss": 0.7706,
+      "step": 30689
+    },
+    {
+      "epoch": 0.8428771525088572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2671518180363473e-06,
+      "loss": 0.8533,
+      "step": 30690
+    },
+    {
+      "epoch": 0.8429046167367005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2667184714506885e-06,
+      "loss": 0.8437,
+      "step": 30691
+    },
+    {
+      "epoch": 0.8429320809645436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2662851939661358e-06,
+      "loss": 0.8015,
+      "step": 30692
+    },
+    {
+      "epoch": 0.8429595451923869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2658519855861184e-06,
+      "loss": 0.9031,
+      "step": 30693
+    },
+    {
+      "epoch": 0.8429870094202302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2654188463140626e-06,
+      "loss": 0.7812,
+      "step": 30694
+    },
+    {
+      "epoch": 0.8430144736480734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.264985776153398e-06,
+      "loss": 0.78,
+      "step": 30695
+    },
+    {
+      "epoch": 0.8430419378759166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2645527751075482e-06,
+      "loss": 0.7839,
+      "step": 30696
+    },
+    {
+      "epoch": 0.8430694021037598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2641198431799407e-06,
+      "loss": 0.7583,
+      "step": 30697
+    },
+    {
+      "epoch": 0.8430968663316031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2636869803740026e-06,
+      "loss": 0.8958,
+      "step": 30698
+    },
+    {
+      "epoch": 0.8431243305594464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2632541866931557e-06,
+      "loss": 0.7724,
+      "step": 30699
+    },
+    {
+      "epoch": 0.8431517947872895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2628214621408253e-06,
+      "loss": 0.8886,
+      "step": 30700
+    },
+    {
+      "epoch": 0.8431792590151328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2623888067204382e-06,
+      "loss": 0.7794,
+      "step": 30701
+    },
+    {
+      "epoch": 0.8432067232429761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261956220435414e-06,
+      "loss": 0.7047,
+      "step": 30702
+    },
+    {
+      "epoch": 0.8432341874708192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2615237032891769e-06,
+      "loss": 0.7995,
+      "step": 30703
+    },
+    {
+      "epoch": 0.8432616516986625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2610912552851484e-06,
+      "loss": 0.7527,
+      "step": 30704
+    },
+    {
+      "epoch": 0.8432891159265057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606588764267536e-06,
+      "loss": 0.7721,
+      "step": 30705
+    },
+    {
+      "epoch": 0.843316580154349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2602265667174096e-06,
+      "loss": 0.8168,
+      "step": 30706
+    },
+    {
+      "epoch": 0.8433440443821922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2597943261605394e-06,
+      "loss": 0.8426,
+      "step": 30707
+    },
+    {
+      "epoch": 0.8433715086100354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2593621547595602e-06,
+      "loss": 0.9324,
+      "step": 30708
+    },
+    {
+      "epoch": 0.8433989728378787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2589300525178938e-06,
+      "loss": 0.9303,
+      "step": 30709
+    },
+    {
+      "epoch": 0.8434264370657218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2584980194389596e-06,
+      "loss": 0.8211,
+      "step": 30710
+    },
+    {
+      "epoch": 0.8434539012935651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2580660555261738e-06,
+      "loss": 0.7834,
+      "step": 30711
+    },
+    {
+      "epoch": 0.8434813655214084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2576341607829567e-06,
+      "loss": 0.8838,
+      "step": 30712
+    },
+    {
+      "epoch": 0.8435088297492516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.257202335212726e-06,
+      "loss": 0.7635,
+      "step": 30713
+    },
+    {
+      "epoch": 0.8435362939770948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2567705788188977e-06,
+      "loss": 0.7743,
+      "step": 30714
+    },
+    {
+      "epoch": 0.8435637582049381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2563388916048847e-06,
+      "loss": 0.8506,
+      "step": 30715
+    },
+    {
+      "epoch": 0.8435912224327813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255907273574105e-06,
+      "loss": 0.8844,
+      "step": 30716
+    },
+    {
+      "epoch": 0.8436186866606246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2554757247299754e-06,
+      "loss": 0.8475,
+      "step": 30717
+    },
+    {
+      "epoch": 0.8436461508884677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255044245075908e-06,
+      "loss": 0.839,
+      "step": 30718
+    },
+    {
+      "epoch": 0.843673615116311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2546128346153207e-06,
+      "loss": 0.7552,
+      "step": 30719
+    },
+    {
+      "epoch": 0.8437010793441543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2541814933516217e-06,
+      "loss": 0.8227,
+      "step": 30720
+    },
+    {
+      "epoch": 0.8437285435719974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2537502212882291e-06,
+      "loss": 0.8339,
+      "step": 30721
+    },
+    {
+      "epoch": 0.8437560077998407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2533190184285505e-06,
+      "loss": 0.9027,
+      "step": 30722
+    },
+    {
+      "epoch": 0.8437834720276839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2528878847759995e-06,
+      "loss": 0.9619,
+      "step": 30723
+    },
+    {
+      "epoch": 0.8438109362555272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2524568203339882e-06,
+      "loss": 0.8315,
+      "step": 30724
+    },
+    {
+      "epoch": 0.8438384004833704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2520258251059291e-06,
+      "loss": 0.8931,
+      "step": 30725
+    },
+    {
+      "epoch": 0.8438658647112136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2515948990952288e-06,
+      "loss": 0.8315,
+      "step": 30726
+    },
+    {
+      "epoch": 0.8438933289390569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2511640423052973e-06,
+      "loss": 0.7027,
+      "step": 30727
+    },
+    {
+      "epoch": 0.8439207931669002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2507332547395456e-06,
+      "loss": 0.7539,
+      "step": 30728
+    },
+    {
+      "epoch": 0.8439482573947433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2503025364013844e-06,
+      "loss": 0.8076,
+      "step": 30729
+    },
+    {
+      "epoch": 0.8439757216225866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.249871887294215e-06,
+      "loss": 0.7409,
+      "step": 30730
+    },
+    {
+      "epoch": 0.8440031858504298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.249441307421453e-06,
+      "loss": 0.8676,
+      "step": 30731
+    },
+    {
+      "epoch": 0.844030650078273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2490107967864972e-06,
+      "loss": 0.7622,
+      "step": 30732
+    },
+    {
+      "epoch": 0.8440581143061163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2485803553927577e-06,
+      "loss": 0.8777,
+      "step": 30733
+    },
+    {
+      "epoch": 0.8440855785339595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2481499832436417e-06,
+      "loss": 0.797,
+      "step": 30734
+    },
+    {
+      "epoch": 0.8441130427618028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2477196803425517e-06,
+      "loss": 0.9583,
+      "step": 30735
+    },
+    {
+      "epoch": 0.8441405069896459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2472894466928953e-06,
+      "loss": 0.8074,
+      "step": 30736
+    },
+    {
+      "epoch": 0.8441679712174892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2468592822980764e-06,
+      "loss": 0.8455,
+      "step": 30737
+    },
+    {
+      "epoch": 0.8441954354453325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2464291871614976e-06,
+      "loss": 0.8004,
+      "step": 30738
+    },
+    {
+      "epoch": 0.8442228996731757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2459991612865595e-06,
+      "loss": 0.8443,
+      "step": 30739
+    },
+    {
+      "epoch": 0.8442503639010189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2455692046766676e-06,
+      "loss": 0.8279,
+      "step": 30740
+    },
+    {
+      "epoch": 0.8442778281288622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2451393173352234e-06,
+      "loss": 0.7966,
+      "step": 30741
+    },
+    {
+      "epoch": 0.8443052923567054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2447094992656284e-06,
+      "loss": 0.8013,
+      "step": 30742
+    },
+    {
+      "epoch": 0.8443327565845486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2442797504712844e-06,
+      "loss": 0.9246,
+      "step": 30743
+    },
+    {
+      "epoch": 0.8443602208123918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.243850070955589e-06,
+      "loss": 0.833,
+      "step": 30744
+    },
+    {
+      "epoch": 0.8443876850402351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2434204607219469e-06,
+      "loss": 0.8018,
+      "step": 30745
+    },
+    {
+      "epoch": 0.8444151492680784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.24299091977375e-06,
+      "loss": 0.8225,
+      "step": 30746
+    },
+    {
+      "epoch": 0.8444426134959215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242561448114402e-06,
+      "loss": 0.7778,
+      "step": 30747
+    },
+    {
+      "epoch": 0.8444700777237648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2421320457473029e-06,
+      "loss": 0.8079,
+      "step": 30748
+    },
+    {
+      "epoch": 0.844497541951608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2417027126758453e-06,
+      "loss": 0.8263,
+      "step": 30749
+    },
+    {
+      "epoch": 0.8445250061794513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2412734489034272e-06,
+      "loss": 0.8643,
+      "step": 30750
+    },
+    {
+      "epoch": 0.8445524704072945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2408442544334476e-06,
+      "loss": 0.7713,
+      "step": 30751
+    },
+    {
+      "epoch": 0.8445799346351377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2404151292693012e-06,
+      "loss": 0.7192,
+      "step": 30752
+    },
+    {
+      "epoch": 0.844607398862981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2399860734143842e-06,
+      "loss": 0.8354,
+      "step": 30753
+    },
+    {
+      "epoch": 0.8446348630908243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.239557086872092e-06,
+      "loss": 0.7742,
+      "step": 30754
+    },
+    {
+      "epoch": 0.8446623273186674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2391281696458158e-06,
+      "loss": 0.7997,
+      "step": 30755
+    },
+    {
+      "epoch": 0.8446897915465107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2386993217389497e-06,
+      "loss": 0.9386,
+      "step": 30756
+    },
+    {
+      "epoch": 0.8447172557743539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2382705431548891e-06,
+      "loss": 0.7378,
+      "step": 30757
+    },
+    {
+      "epoch": 0.8447447200021971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2378418338970266e-06,
+      "loss": 0.8286,
+      "step": 30758
+    },
+    {
+      "epoch": 0.8447721842300404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2374131939687527e-06,
+      "loss": 0.706,
+      "step": 30759
+    },
+    {
+      "epoch": 0.8447996484578836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2369846233734627e-06,
+      "loss": 0.8427,
+      "step": 30760
+    },
+    {
+      "epoch": 0.8448271126857269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2365561221145417e-06,
+      "loss": 0.8687,
+      "step": 30761
+    },
+    {
+      "epoch": 0.84485457691357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2361276901953868e-06,
+      "loss": 0.8665,
+      "step": 30762
+    },
+    {
+      "epoch": 0.8448820411414133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.235699327619382e-06,
+      "loss": 0.7751,
+      "step": 30763
+    },
+    {
+      "epoch": 0.8449095053692566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2352710343899177e-06,
+      "loss": 0.8217,
+      "step": 30764
+    },
+    {
+      "epoch": 0.8449369695970997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.234842810510386e-06,
+      "loss": 0.8907,
+      "step": 30765
+    },
+    {
+      "epoch": 0.844964433824943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.234414655984174e-06,
+      "loss": 0.7184,
+      "step": 30766
+    },
+    {
+      "epoch": 0.8449918980527863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233986570814667e-06,
+      "loss": 0.8191,
+      "step": 30767
+    },
+    {
+      "epoch": 0.8450193622806295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2335585550052542e-06,
+      "loss": 0.805,
+      "step": 30768
+    },
+    {
+      "epoch": 0.8450468265084727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2331306085593242e-06,
+      "loss": 0.8374,
+      "step": 30769
+    },
+    {
+      "epoch": 0.8450742907363159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2327027314802574e-06,
+      "loss": 0.7513,
+      "step": 30770
+    },
+    {
+      "epoch": 0.8451017549641592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2322749237714438e-06,
+      "loss": 0.8705,
+      "step": 30771
+    },
+    {
+      "epoch": 0.8451292191920025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231847185436269e-06,
+      "loss": 0.8088,
+      "step": 30772
+    },
+    {
+      "epoch": 0.8451566834198456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231419516478114e-06,
+      "loss": 0.8203,
+      "step": 30773
+    },
+    {
+      "epoch": 0.8451841476476889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2309919169003647e-06,
+      "loss": 0.8668,
+      "step": 30774
+    },
+    {
+      "epoch": 0.8452116118755321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2305643867064031e-06,
+      "loss": 0.845,
+      "step": 30775
+    },
+    {
+      "epoch": 0.8452390761033753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2301369258996133e-06,
+      "loss": 0.8335,
+      "step": 30776
+    },
+    {
+      "epoch": 0.8452665403312186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2297095344833798e-06,
+      "loss": 0.9178,
+      "step": 30777
+    },
+    {
+      "epoch": 0.8452940045590618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2292822124610814e-06,
+      "loss": 0.8414,
+      "step": 30778
+    },
+    {
+      "epoch": 0.8453214687869051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2288549598360977e-06,
+      "loss": 0.8081,
+      "step": 30779
+    },
+    {
+      "epoch": 0.8453489330147483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2284277766118114e-06,
+      "loss": 0.8221,
+      "step": 30780
+    },
+    {
+      "epoch": 0.8453763972425915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2280006627916018e-06,
+      "loss": 0.8231,
+      "step": 30781
+    },
+    {
+      "epoch": 0.8454038614704348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22757361837885e-06,
+      "loss": 0.8152,
+      "step": 30782
+    },
+    {
+      "epoch": 0.845431325698278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.227146643376933e-06,
+      "loss": 0.7653,
+      "step": 30783
+    },
+    {
+      "epoch": 0.8454587899261212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2267197377892314e-06,
+      "loss": 0.7086,
+      "step": 30784
+    },
+    {
+      "epoch": 0.8454862541539645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2262929016191205e-06,
+      "loss": 0.7242,
+      "step": 30785
+    },
+    {
+      "epoch": 0.8455137183818077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2258661348699807e-06,
+      "loss": 0.8548,
+      "step": 30786
+    },
+    {
+      "epoch": 0.8455411826096509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.225439437545185e-06,
+      "loss": 0.7903,
+      "step": 30787
+    },
+    {
+      "epoch": 0.8455686468374941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2250128096481105e-06,
+      "loss": 0.7525,
+      "step": 30788
+    },
+    {
+      "epoch": 0.8455961110653374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2245862511821348e-06,
+      "loss": 0.805,
+      "step": 30789
+    },
+    {
+      "epoch": 0.8456235752931807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.224159762150634e-06,
+      "loss": 0.7586,
+      "step": 30790
+    },
+    {
+      "epoch": 0.8456510395210238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2237333425569776e-06,
+      "loss": 0.7748,
+      "step": 30791
+    },
+    {
+      "epoch": 0.8456785037488671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2233069924045437e-06,
+      "loss": 0.7867,
+      "step": 30792
+    },
+    {
+      "epoch": 0.8457059679767104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2228807116967056e-06,
+      "loss": 0.8147,
+      "step": 30793
+    },
+    {
+      "epoch": 0.8457334322045535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2224545004368338e-06,
+      "loss": 0.832,
+      "step": 30794
+    },
+    {
+      "epoch": 0.8457608964323968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2220283586283022e-06,
+      "loss": 0.9324,
+      "step": 30795
+    },
+    {
+      "epoch": 0.84578836066024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2216022862744847e-06,
+      "loss": 0.6624,
+      "step": 30796
+    },
+    {
+      "epoch": 0.8458158248880833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2211762833787477e-06,
+      "loss": 0.8919,
+      "step": 30797
+    },
+    {
+      "epoch": 0.8458432891159265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220750349944465e-06,
+      "loss": 0.9333,
+      "step": 30798
+    },
+    {
+      "epoch": 0.8458707533437697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2203244859750052e-06,
+      "loss": 0.8427,
+      "step": 30799
+    },
+    {
+      "epoch": 0.845898217571613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2198986914737398e-06,
+      "loss": 0.8838,
+      "step": 30800
+    },
+    {
+      "epoch": 0.8459256817994562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2194729664440398e-06,
+      "loss": 0.7827,
+      "step": 30801
+    },
+    {
+      "epoch": 0.8459531460272994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2190473108892698e-06,
+      "loss": 0.774,
+      "step": 30802
+    },
+    {
+      "epoch": 0.8459806102551427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2186217248127975e-06,
+      "loss": 0.7433,
+      "step": 30803
+    },
+    {
+      "epoch": 0.8460080744829859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.218196208217992e-06,
+      "loss": 0.814,
+      "step": 30804
+    },
+    {
+      "epoch": 0.8460355387108291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217770761108219e-06,
+      "loss": 0.8543,
+      "step": 30805
+    },
+    {
+      "epoch": 0.8460630029386724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2173453834868455e-06,
+      "loss": 0.7804,
+      "step": 30806
+    },
+    {
+      "epoch": 0.8460904671665156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2169200753572408e-06,
+      "loss": 0.7551,
+      "step": 30807
+    },
+    {
+      "epoch": 0.8461179313943589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2164948367227635e-06,
+      "loss": 0.7551,
+      "step": 30808
+    },
+    {
+      "epoch": 0.846145395622202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216069667586781e-06,
+      "loss": 0.8203,
+      "step": 30809
+    },
+    {
+      "epoch": 0.8461728598500453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2156445679526618e-06,
+      "loss": 0.8613,
+      "step": 30810
+    },
+    {
+      "epoch": 0.8462003240778886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2152195378237618e-06,
+      "loss": 0.8976,
+      "step": 30811
+    },
+    {
+      "epoch": 0.8462277883057318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2147945772034485e-06,
+      "loss": 0.8014,
+      "step": 30812
+    },
+    {
+      "epoch": 0.846255252533575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2143696860950849e-06,
+      "loss": 0.8031,
+      "step": 30813
+    },
+    {
+      "epoch": 0.8462827167614183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2139448645020301e-06,
+      "loss": 0.8636,
+      "step": 30814
+    },
+    {
+      "epoch": 0.8463101809892615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2135201124276475e-06,
+      "loss": 0.8618,
+      "step": 30815
+    },
+    {
+      "epoch": 0.8463376452171048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2130954298752962e-06,
+      "loss": 0.8957,
+      "step": 30816
+    },
+    {
+      "epoch": 0.8463651094449479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2126708168483404e-06,
+      "loss": 0.7424,
+      "step": 30817
+    },
+    {
+      "epoch": 0.8463925736727912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2122462733501338e-06,
+      "loss": 0.9009,
+      "step": 30818
+    },
+    {
+      "epoch": 0.8464200379006345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211821799384042e-06,
+      "loss": 0.7481,
+      "step": 30819
+    },
+    {
+      "epoch": 0.8464475021284776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2113973949534176e-06,
+      "loss": 0.7131,
+      "step": 30820
+    },
+    {
+      "epoch": 0.8464749663563209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210973060061622e-06,
+      "loss": 0.7791,
+      "step": 30821
+    },
+    {
+      "epoch": 0.8465024305841641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210548794712012e-06,
+      "loss": 0.8019,
+      "step": 30822
+    },
+    {
+      "epoch": 0.8465298948120074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2101245989079434e-06,
+      "loss": 0.7703,
+      "step": 30823
+    },
+    {
+      "epoch": 0.8465573590398506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.209700472652775e-06,
+      "loss": 0.7885,
+      "step": 30824
+    },
+    {
+      "epoch": 0.8465848232676938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2092764159498637e-06,
+      "loss": 0.8282,
+      "step": 30825
+    },
+    {
+      "epoch": 0.8466122874955371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2088524288025627e-06,
+      "loss": 0.7278,
+      "step": 30826
+    },
+    {
+      "epoch": 0.8466397517233804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2084285112142237e-06,
+      "loss": 0.7382,
+      "step": 30827
+    },
+    {
+      "epoch": 0.8466672159512235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2080046631882048e-06,
+      "loss": 0.8274,
+      "step": 30828
+    },
+    {
+      "epoch": 0.8466946801790668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2075808847278582e-06,
+      "loss": 0.8693,
+      "step": 30829
+    },
+    {
+      "epoch": 0.84672214440691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2071571758365374e-06,
+      "loss": 0.7254,
+      "step": 30830
+    },
+    {
+      "epoch": 0.8467496086347532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206733536517597e-06,
+      "loss": 0.7595,
+      "step": 30831
+    },
+    {
+      "epoch": 0.8467770728625965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206309966774385e-06,
+      "loss": 0.8705,
+      "step": 30832
+    },
+    {
+      "epoch": 0.8468045370904397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2058864666102555e-06,
+      "loss": 0.7563,
+      "step": 30833
+    },
+    {
+      "epoch": 0.846832001318283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2054630360285601e-06,
+      "loss": 0.7364,
+      "step": 30834
+    },
+    {
+      "epoch": 0.8468594655461261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2050396750326465e-06,
+      "loss": 0.7293,
+      "step": 30835
+    },
+    {
+      "epoch": 0.8468869297739694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2046163836258651e-06,
+      "loss": 0.8254,
+      "step": 30836
+    },
+    {
+      "epoch": 0.8469143940018127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2041931618115689e-06,
+      "loss": 0.8306,
+      "step": 30837
+    },
+    {
+      "epoch": 0.8469418582296558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2037700095931005e-06,
+      "loss": 0.8012,
+      "step": 30838
+    },
+    {
+      "epoch": 0.8469693224574991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203346926973812e-06,
+      "loss": 0.7975,
+      "step": 30839
+    },
+    {
+      "epoch": 0.8469967866853424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2029239139570492e-06,
+      "loss": 0.8162,
+      "step": 30840
+    },
+    {
+      "epoch": 0.8470242509131856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2025009705461633e-06,
+      "loss": 0.9511,
+      "step": 30841
+    },
+    {
+      "epoch": 0.8470517151410288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2020780967444946e-06,
+      "loss": 0.7127,
+      "step": 30842
+    },
+    {
+      "epoch": 0.847079179368872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2016552925553937e-06,
+      "loss": 0.848,
+      "step": 30843
+    },
+    {
+      "epoch": 0.8471066435967153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2012325579822025e-06,
+      "loss": 0.8309,
+      "step": 30844
+    },
+    {
+      "epoch": 0.8471341078245586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2008098930282685e-06,
+      "loss": 0.78,
+      "step": 30845
+    },
+    {
+      "epoch": 0.8471615720524017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2003872976969344e-06,
+      "loss": 0.8094,
+      "step": 30846
+    },
+    {
+      "epoch": 0.847189036280245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1999647719915441e-06,
+      "loss": 0.7694,
+      "step": 30847
+    },
+    {
+      "epoch": 0.8472165005080882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199542315915443e-06,
+      "loss": 0.9073,
+      "step": 30848
+    },
+    {
+      "epoch": 0.8472439647359314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1991199294719723e-06,
+      "loss": 0.8491,
+      "step": 30849
+    },
+    {
+      "epoch": 0.8472714289637747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1986976126644746e-06,
+      "loss": 0.7122,
+      "step": 30850
+    },
+    {
+      "epoch": 0.8472988931916179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1982753654962886e-06,
+      "loss": 0.7977,
+      "step": 30851
+    },
+    {
+      "epoch": 0.8473263574194612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1978531879707577e-06,
+      "loss": 0.8702,
+      "step": 30852
+    },
+    {
+      "epoch": 0.8473538216473044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1974310800912215e-06,
+      "loss": 0.7733,
+      "step": 30853
+    },
+    {
+      "epoch": 0.8473812858751476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1970090418610202e-06,
+      "loss": 0.8285,
+      "step": 30854
+    },
+    {
+      "epoch": 0.8474087501029909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1965870732834961e-06,
+      "loss": 0.8864,
+      "step": 30855
+    },
+    {
+      "epoch": 0.847436214330834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1961651743619829e-06,
+      "loss": 0.7292,
+      "step": 30856
+    },
+    {
+      "epoch": 0.8474636785586773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1957433450998212e-06,
+      "loss": 0.8647,
+      "step": 30857
+    },
+    {
+      "epoch": 0.8474911427865206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.195321585500351e-06,
+      "loss": 0.7304,
+      "step": 30858
+    },
+    {
+      "epoch": 0.8475186070143638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1948998955669034e-06,
+      "loss": 0.7548,
+      "step": 30859
+    },
+    {
+      "epoch": 0.847546071242207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1944782753028194e-06,
+      "loss": 0.8593,
+      "step": 30860
+    },
+    {
+      "epoch": 0.8475735354700502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1940567247114377e-06,
+      "loss": 0.7748,
+      "step": 30861
+    },
+    {
+      "epoch": 0.8476009996978935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1936352437960864e-06,
+      "loss": 0.8773,
+      "step": 30862
+    },
+    {
+      "epoch": 0.8476284639257368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.193213832560105e-06,
+      "loss": 0.7802,
+      "step": 30863
+    },
+    {
+      "epoch": 0.8476559281535799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1927924910068267e-06,
+      "loss": 0.8568,
+      "step": 30864
+    },
+    {
+      "epoch": 0.8476833923814232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1923712191395886e-06,
+      "loss": 0.7997,
+      "step": 30865
+    },
+    {
+      "epoch": 0.8477108566092665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1919500169617192e-06,
+      "loss": 0.8175,
+      "step": 30866
+    },
+    {
+      "epoch": 0.8477383208371096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1915288844765548e-06,
+      "loss": 0.8145,
+      "step": 30867
+    },
+    {
+      "epoch": 0.8477657850649529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1911078216874238e-06,
+      "loss": 0.9293,
+      "step": 30868
+    },
+    {
+      "epoch": 0.8477932492927961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1906868285976603e-06,
+      "loss": 0.7659,
+      "step": 30869
+    },
+    {
+      "epoch": 0.8478207135206394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1902659052105948e-06,
+      "loss": 0.8222,
+      "step": 30870
+    },
+    {
+      "epoch": 0.8478481777484826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1898450515295579e-06,
+      "loss": 0.7729,
+      "step": 30871
+    },
+    {
+      "epoch": 0.8478756419763258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1894242675578816e-06,
+      "loss": 0.8623,
+      "step": 30872
+    },
+    {
+      "epoch": 0.8479031062041691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189003553298892e-06,
+      "loss": 0.8511,
+      "step": 30873
+    },
+    {
+      "epoch": 0.8479305704320123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1885829087559208e-06,
+      "loss": 0.8502,
+      "step": 30874
+    },
+    {
+      "epoch": 0.8479580346598555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1881623339322923e-06,
+      "loss": 0.8577,
+      "step": 30875
+    },
+    {
+      "epoch": 0.8479854988876988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.187741828831338e-06,
+      "loss": 0.7885,
+      "step": 30876
+    },
+    {
+      "epoch": 0.848012963115542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1873213934563833e-06,
+      "loss": 0.6711,
+      "step": 30877
+    },
+    {
+      "epoch": 0.8480404273433853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1869010278107573e-06,
+      "loss": 0.7593,
+      "step": 30878
+    },
+    {
+      "epoch": 0.8480678915712285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1864807318977822e-06,
+      "loss": 0.7535,
+      "step": 30879
+    },
+    {
+      "epoch": 0.8480953557990717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1860605057207864e-06,
+      "loss": 0.7611,
+      "step": 30880
+    },
+    {
+      "epoch": 0.848122820026915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1856403492830948e-06,
+      "loss": 0.831,
+      "step": 30881
+    },
+    {
+      "epoch": 0.8481502842547581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1852202625880305e-06,
+      "loss": 0.7971,
+      "step": 30882
+    },
+    {
+      "epoch": 0.8481777484826014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1848002456389163e-06,
+      "loss": 0.9251,
+      "step": 30883
+    },
+    {
+      "epoch": 0.8482052127104447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1843802984390806e-06,
+      "loss": 0.9106,
+      "step": 30884
+    },
+    {
+      "epoch": 0.8482326769382879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1839604209918409e-06,
+      "loss": 0.9163,
+      "step": 30885
+    },
+    {
+      "epoch": 0.8482601411661311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1835406133005212e-06,
+      "loss": 0.6989,
+      "step": 30886
+    },
+    {
+      "epoch": 0.8482876053939743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.183120875368442e-06,
+      "loss": 0.8424,
+      "step": 30887
+    },
+    {
+      "epoch": 0.8483150696218176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1827012071989275e-06,
+      "loss": 0.7244,
+      "step": 30888
+    },
+    {
+      "epoch": 0.8483425338496609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1822816087952982e-06,
+      "loss": 0.7475,
+      "step": 30889
+    },
+    {
+      "epoch": 0.848369998077504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1818620801608716e-06,
+      "loss": 0.8253,
+      "step": 30890
+    },
+    {
+      "epoch": 0.8483974623053473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.181442621298966e-06,
+      "loss": 0.7662,
+      "step": 30891
+    },
+    {
+      "epoch": 0.8484249265331906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1810232322129034e-06,
+      "loss": 0.8437,
+      "step": 30892
+    },
+    {
+      "epoch": 0.8484523907610337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.180603912906001e-06,
+      "loss": 0.825,
+      "step": 30893
+    },
+    {
+      "epoch": 0.848479854988877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1801846633815762e-06,
+      "loss": 0.8162,
+      "step": 30894
+    },
+    {
+      "epoch": 0.8485073192167202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1797654836429473e-06,
+      "loss": 0.7593,
+      "step": 30895
+    },
+    {
+      "epoch": 0.8485347834445635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1793463736934318e-06,
+      "loss": 0.8191,
+      "step": 30896
+    },
+    {
+      "epoch": 0.8485622476724067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1789273335363427e-06,
+      "loss": 0.8012,
+      "step": 30897
+    },
+    {
+      "epoch": 0.8485897119002499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1785083631750006e-06,
+      "loss": 0.7482,
+      "step": 30898
+    },
+    {
+      "epoch": 0.8486171761280932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.178089462612715e-06,
+      "loss": 0.8067,
+      "step": 30899
+    },
+    {
+      "epoch": 0.8486446403559363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.177670631852802e-06,
+      "loss": 0.831,
+      "step": 30900
+    },
+    {
+      "epoch": 0.8486721045837796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1772518708985769e-06,
+      "loss": 0.7345,
+      "step": 30901
+    },
+    {
+      "epoch": 0.8486995688116229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.176833179753355e-06,
+      "loss": 0.8255,
+      "step": 30902
+    },
+    {
+      "epoch": 0.8487270330394661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1764145584204444e-06,
+      "loss": 0.795,
+      "step": 30903
+    },
+    {
+      "epoch": 0.8487544972673093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1759960069031594e-06,
+      "loss": 0.935,
+      "step": 30904
+    },
+    {
+      "epoch": 0.8487819614951526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.175577525204814e-06,
+      "loss": 0.7573,
+      "step": 30905
+    },
+    {
+      "epoch": 0.8488094257229958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1751591133287166e-06,
+      "loss": 0.7311,
+      "step": 30906
+    },
+    {
+      "epoch": 0.8488368899508391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1747407712781778e-06,
+      "loss": 0.8326,
+      "step": 30907
+    },
+    {
+      "epoch": 0.8488643541786822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1743224990565106e-06,
+      "loss": 0.792,
+      "step": 30908
+    },
+    {
+      "epoch": 0.8488918184065255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173904296667021e-06,
+      "loss": 0.8034,
+      "step": 30909
+    },
+    {
+      "epoch": 0.8489192826343688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173486164113019e-06,
+      "loss": 0.7475,
+      "step": 30910
+    },
+    {
+      "epoch": 0.8489467468622119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173068101397814e-06,
+      "loss": 0.8226,
+      "step": 30911
+    },
+    {
+      "epoch": 0.8489742110900552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1726501085247133e-06,
+      "loss": 0.7531,
+      "step": 30912
+    },
+    {
+      "epoch": 0.8490016753178984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1722321854970253e-06,
+      "loss": 0.8108,
+      "step": 30913
+    },
+    {
+      "epoch": 0.8490291395457417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1718143323180575e-06,
+      "loss": 0.7967,
+      "step": 30914
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1713965489911117e-06,
+      "loss": 0.7925,
+      "step": 30915
+    },
+    {
+      "epoch": 0.8490840680014281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1709788355194962e-06,
+      "loss": 0.8609,
+      "step": 30916
+    },
+    {
+      "epoch": 0.8491115322292714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1705611919065163e-06,
+      "loss": 0.8074,
+      "step": 30917
+    },
+    {
+      "epoch": 0.8491389964571147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170143618155476e-06,
+      "loss": 0.7643,
+      "step": 30918
+    },
+    {
+      "epoch": 0.8491664606849578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1697261142696803e-06,
+      "loss": 0.799,
+      "step": 30919
+    },
+    {
+      "epoch": 0.8491939249128011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1693086802524333e-06,
+      "loss": 0.7784,
+      "step": 30920
+    },
+    {
+      "epoch": 0.8492213891406443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168891316107036e-06,
+      "loss": 0.8105,
+      "step": 30921
+    },
+    {
+      "epoch": 0.8492488533684875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168474021836792e-06,
+      "loss": 0.7515,
+      "step": 30922
+    },
+    {
+      "epoch": 0.8492763175963308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1680567974450019e-06,
+      "loss": 0.7568,
+      "step": 30923
+    },
+    {
+      "epoch": 0.849303781824174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1676396429349667e-06,
+      "loss": 0.7377,
+      "step": 30924
+    },
+    {
+      "epoch": 0.8493312460520173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1672225583099883e-06,
+      "loss": 0.7824,
+      "step": 30925
+    },
+    {
+      "epoch": 0.8493587102798604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1668055435733683e-06,
+      "loss": 0.7454,
+      "step": 30926
+    },
+    {
+      "epoch": 0.8493861745077037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.166388598728403e-06,
+      "loss": 0.8163,
+      "step": 30927
+    },
+    {
+      "epoch": 0.849413638735547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1659717237783919e-06,
+      "loss": 0.8426,
+      "step": 30928
+    },
+    {
+      "epoch": 0.8494411029633901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1655549187266379e-06,
+      "loss": 0.8706,
+      "step": 30929
+    },
+    {
+      "epoch": 0.8494685671912334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165138183576433e-06,
+      "loss": 0.8185,
+      "step": 30930
+    },
+    {
+      "epoch": 0.8494960314190767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1647215183310789e-06,
+      "loss": 0.9085,
+      "step": 30931
+    },
+    {
+      "epoch": 0.8495234956469199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1643049229938685e-06,
+      "loss": 0.8101,
+      "step": 30932
+    },
+    {
+      "epoch": 0.8495509598747631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1638883975681003e-06,
+      "loss": 0.8236,
+      "step": 30933
+    },
+    {
+      "epoch": 0.8495784241026063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1634719420570695e-06,
+      "loss": 0.7979,
+      "step": 30934
+    },
+    {
+      "epoch": 0.8496058883304496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1630555564640723e-06,
+      "loss": 0.8399,
+      "step": 30935
+    },
+    {
+      "epoch": 0.8496333525582929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1626392407924014e-06,
+      "loss": 0.8514,
+      "step": 30936
+    },
+    {
+      "epoch": 0.849660816786136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1622229950453556e-06,
+      "loss": 0.8686,
+      "step": 30937
+    },
+    {
+      "epoch": 0.8496882810139793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1618068192262244e-06,
+      "loss": 0.8863,
+      "step": 30938
+    },
+    {
+      "epoch": 0.8497157452418225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1613907133382985e-06,
+      "loss": 0.817,
+      "step": 30939
+    },
+    {
+      "epoch": 0.8497432094696658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1609746773848741e-06,
+      "loss": 0.8287,
+      "step": 30940
+    },
+    {
+      "epoch": 0.849770673697509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1605587113692418e-06,
+      "loss": 0.7666,
+      "step": 30941
+    },
+    {
+      "epoch": 0.8497981379253522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1601428152946926e-06,
+      "loss": 0.8297,
+      "step": 30942
+    },
+    {
+      "epoch": 0.8498256021531955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1597269891645201e-06,
+      "loss": 0.8892,
+      "step": 30943
+    },
+    {
+      "epoch": 0.8498530663810387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15931123298201e-06,
+      "loss": 0.7379,
+      "step": 30944
+    },
+    {
+      "epoch": 0.8498805306088819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1588955467504536e-06,
+      "loss": 0.8391,
+      "step": 30945
+    },
+    {
+      "epoch": 0.8499079948367252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.158479930473142e-06,
+      "loss": 0.8088,
+      "step": 30946
+    },
+    {
+      "epoch": 0.8499354590645684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1580643841533612e-06,
+      "loss": 0.8498,
+      "step": 30947
+    },
+    {
+      "epoch": 0.8499629232924116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1576489077943987e-06,
+      "loss": 0.7515,
+      "step": 30948
+    },
+    {
+      "epoch": 0.8499903875202549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157233501399546e-06,
+      "loss": 0.7213,
+      "step": 30949
+    },
+    {
+      "epoch": 0.8500178517480981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1568181649720856e-06,
+      "loss": 0.7913,
+      "step": 30950
+    },
+    {
+      "epoch": 0.8500453159759414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1564028985153054e-06,
+      "loss": 0.8439,
+      "step": 30951
+    },
+    {
+      "epoch": 0.8500727802037845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1559877020324906e-06,
+      "loss": 0.8397,
+      "step": 30952
+    },
+    {
+      "epoch": 0.8501002444316278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1555725755269288e-06,
+      "loss": 0.7541,
+      "step": 30953
+    },
+    {
+      "epoch": 0.8501277086594711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1551575190019015e-06,
+      "loss": 0.8657,
+      "step": 30954
+    },
+    {
+      "epoch": 0.8501551728873142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1547425324606953e-06,
+      "loss": 0.7815,
+      "step": 30955
+    },
+    {
+      "epoch": 0.8501826371151575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1543276159065908e-06,
+      "loss": 0.8613,
+      "step": 30956
+    },
+    {
+      "epoch": 0.8502101013430008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1539127693428731e-06,
+      "loss": 0.8041,
+      "step": 30957
+    },
+    {
+      "epoch": 0.850237565570844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1534979927728229e-06,
+      "loss": 0.7719,
+      "step": 30958
+    },
+    {
+      "epoch": 0.8502650297986872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1530832861997244e-06,
+      "loss": 0.7881,
+      "step": 30959
+    },
+    {
+      "epoch": 0.8502924940265304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.152668649626858e-06,
+      "loss": 0.8405,
+      "step": 30960
+    },
+    {
+      "epoch": 0.8503199582543737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.152254083057507e-06,
+      "loss": 0.7737,
+      "step": 30961
+    },
+    {
+      "epoch": 0.850347422482217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1518395864949471e-06,
+      "loss": 0.7763,
+      "step": 30962
+    },
+    {
+      "epoch": 0.8503748867100601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1514251599424597e-06,
+      "loss": 0.8302,
+      "step": 30963
+    },
+    {
+      "epoch": 0.8504023509379034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.151010803403323e-06,
+      "loss": 0.8685,
+      "step": 30964
+    },
+    {
+      "epoch": 0.8504298151657466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1505965168808163e-06,
+      "loss": 0.7284,
+      "step": 30965
+    },
+    {
+      "epoch": 0.8504572793935898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1501823003782186e-06,
+      "loss": 0.8072,
+      "step": 30966
+    },
+    {
+      "epoch": 0.8504847436214331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1497681538988093e-06,
+      "loss": 0.8387,
+      "step": 30967
+    },
+    {
+      "epoch": 0.8505122078492763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.14935407744586e-06,
+      "loss": 0.8008,
+      "step": 30968
+    },
+    {
+      "epoch": 0.8505396720771196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1489400710226496e-06,
+      "loss": 0.955,
+      "step": 30969
+    },
+    {
+      "epoch": 0.8505671363049628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.148526134632455e-06,
+      "loss": 0.7999,
+      "step": 30970
+    },
+    {
+      "epoch": 0.850594600532806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1481122682785494e-06,
+      "loss": 0.856,
+      "step": 30971
+    },
+    {
+      "epoch": 0.8506220647606493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1476984719642082e-06,
+      "loss": 0.8236,
+      "step": 30972
+    },
+    {
+      "epoch": 0.8506495289884924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1472847456927072e-06,
+      "loss": 0.7834,
+      "step": 30973
+    },
+    {
+      "epoch": 0.8506769932163357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1468710894673174e-06,
+      "loss": 0.8096,
+      "step": 30974
+    },
+    {
+      "epoch": 0.850704457444179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1464575032913127e-06,
+      "loss": 0.794,
+      "step": 30975
+    },
+    {
+      "epoch": 0.8507319216720222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.146043987167965e-06,
+      "loss": 0.8678,
+      "step": 30976
+    },
+    {
+      "epoch": 0.8507593858998654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1456305411005498e-06,
+      "loss": 0.7742,
+      "step": 30977
+    },
+    {
+      "epoch": 0.8507868501277086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1452171650923338e-06,
+      "loss": 0.7476,
+      "step": 30978
+    },
+    {
+      "epoch": 0.8508143143555519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1448038591465915e-06,
+      "loss": 0.7034,
+      "step": 30979
+    },
+    {
+      "epoch": 0.8508417785833952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.144390623266588e-06,
+      "loss": 0.7758,
+      "step": 30980
+    },
+    {
+      "epoch": 0.8508692428112383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1439774574555973e-06,
+      "loss": 0.8015,
+      "step": 30981
+    },
+    {
+      "epoch": 0.8508967070390816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.143564361716888e-06,
+      "loss": 0.7472,
+      "step": 30982
+    },
+    {
+      "epoch": 0.8509241712669249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1431513360537273e-06,
+      "loss": 0.8411,
+      "step": 30983
+    },
+    {
+      "epoch": 0.850951635494768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1427383804693837e-06,
+      "loss": 0.7003,
+      "step": 30984
+    },
+    {
+      "epoch": 0.8509790997226113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.142325494967128e-06,
+      "loss": 0.8103,
+      "step": 30985
+    },
+    {
+      "epoch": 0.8510065639504545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1419126795502232e-06,
+      "loss": 0.8053,
+      "step": 30986
+    },
+    {
+      "epoch": 0.8510340281782978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1414999342219357e-06,
+      "loss": 0.7417,
+      "step": 30987
+    },
+    {
+      "epoch": 0.851061492406141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1410872589855314e-06,
+      "loss": 0.7411,
+      "step": 30988
+    },
+    {
+      "epoch": 0.8510889566339842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1406746538442759e-06,
+      "loss": 0.7683,
+      "step": 30989
+    },
+    {
+      "epoch": 0.8511164208618275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1402621188014373e-06,
+      "loss": 0.7797,
+      "step": 30990
+    },
+    {
+      "epoch": 0.8511438850896706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1398496538602743e-06,
+      "loss": 0.8428,
+      "step": 30991
+    },
+    {
+      "epoch": 0.8511713493175139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1394372590240521e-06,
+      "loss": 0.8997,
+      "step": 30992
+    },
+    {
+      "epoch": 0.8511988135453572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1390249342960357e-06,
+      "loss": 0.7567,
+      "step": 30993
+    },
+    {
+      "epoch": 0.8512262777732004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1386126796794883e-06,
+      "loss": 0.8094,
+      "step": 30994
+    },
+    {
+      "epoch": 0.8512537420010436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.138200495177667e-06,
+      "loss": 0.7729,
+      "step": 30995
+    },
+    {
+      "epoch": 0.8512812062288869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1377883807938394e-06,
+      "loss": 0.856,
+      "step": 30996
+    },
+    {
+      "epoch": 0.8513086704567301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1373763365312606e-06,
+      "loss": 0.7143,
+      "step": 30997
+    },
+    {
+      "epoch": 0.8513361346845734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1369643623931925e-06,
+      "loss": 0.8133,
+      "step": 30998
+    },
+    {
+      "epoch": 0.8513635989124165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1365524583828968e-06,
+      "loss": 0.7499,
+      "step": 30999
+    },
+    {
+      "epoch": 0.8513910631402598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136140624503631e-06,
+      "loss": 0.737,
+      "step": 31000
+    },
+    {
+      "epoch": 0.8514185273681031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135728860758656e-06,
+      "loss": 0.79,
+      "step": 31001
+    },
+    {
+      "epoch": 0.8514459915959462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1353171671512274e-06,
+      "loss": 0.8917,
+      "step": 31002
+    },
+    {
+      "epoch": 0.8514734558237895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1349055436846023e-06,
+      "loss": 0.7438,
+      "step": 31003
+    },
+    {
+      "epoch": 0.8515009200516328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1344939903620379e-06,
+      "loss": 0.7766,
+      "step": 31004
+    },
+    {
+      "epoch": 0.851528384279476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1340825071867911e-06,
+      "loss": 0.7966,
+      "step": 31005
+    },
+    {
+      "epoch": 0.8515558485073192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1336710941621177e-06,
+      "loss": 0.7772,
+      "step": 31006
+    },
+    {
+      "epoch": 0.8515833127351624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1332597512912735e-06,
+      "loss": 0.8405,
+      "step": 31007
+    },
+    {
+      "epoch": 0.8516107769630057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1328484785775152e-06,
+      "loss": 0.8284,
+      "step": 31008
+    },
+    {
+      "epoch": 0.851638241190849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.132437276024092e-06,
+      "loss": 0.8237,
+      "step": 31009
+    },
+    {
+      "epoch": 0.8516657054186921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1320261436342616e-06,
+      "loss": 0.807,
+      "step": 31010
+    },
+    {
+      "epoch": 0.8516931696465354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1316150814112735e-06,
+      "loss": 0.8358,
+      "step": 31011
+    },
+    {
+      "epoch": 0.8517206338743786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.131204089358383e-06,
+      "loss": 0.778,
+      "step": 31012
+    },
+    {
+      "epoch": 0.8517480981022219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1307931674788408e-06,
+      "loss": 0.7632,
+      "step": 31013
+    },
+    {
+      "epoch": 0.8517755623300651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.130382315775901e-06,
+      "loss": 0.7462,
+      "step": 31014
+    },
+    {
+      "epoch": 0.8518030265579083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1299715342528095e-06,
+      "loss": 0.7942,
+      "step": 31015
+    },
+    {
+      "epoch": 0.8518304907857516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1295608229128196e-06,
+      "loss": 0.7856,
+      "step": 31016
+    },
+    {
+      "epoch": 0.8518579550135948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.12915018175918e-06,
+      "loss": 0.7484,
+      "step": 31017
+    },
+    {
+      "epoch": 0.851885419241438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128739610795143e-06,
+      "loss": 0.8264,
+      "step": 31018
+    },
+    {
+      "epoch": 0.8519128834692813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1283291100239524e-06,
+      "loss": 0.8539,
+      "step": 31019
+    },
+    {
+      "epoch": 0.8519403476971245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1279186794488605e-06,
+      "loss": 0.8787,
+      "step": 31020
+    },
+    {
+      "epoch": 0.8519678119249677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1275083190731106e-06,
+      "loss": 0.8599,
+      "step": 31021
+    },
+    {
+      "epoch": 0.851995276152811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1270980288999511e-06,
+      "loss": 0.7929,
+      "step": 31022
+    },
+    {
+      "epoch": 0.8520227403806542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1266878089326294e-06,
+      "loss": 0.8439,
+      "step": 31023
+    },
+    {
+      "epoch": 0.8520502046084975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1262776591743919e-06,
+      "loss": 0.7196,
+      "step": 31024
+    },
+    {
+      "epoch": 0.8520776688363406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1258675796284846e-06,
+      "loss": 0.7838,
+      "step": 31025
+    },
+    {
+      "epoch": 0.8521051330641839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1254575702981496e-06,
+      "loss": 0.7388,
+      "step": 31026
+    },
+    {
+      "epoch": 0.8521325972920272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.125047631186631e-06,
+      "loss": 0.7319,
+      "step": 31027
+    },
+    {
+      "epoch": 0.8521600615198703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1246377622971728e-06,
+      "loss": 0.8105,
+      "step": 31028
+    },
+    {
+      "epoch": 0.8521875257477136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1242279636330179e-06,
+      "loss": 0.8061,
+      "step": 31029
+    },
+    {
+      "epoch": 0.8522149899755569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1238182351974103e-06,
+      "loss": 0.7871,
+      "step": 31030
+    },
+    {
+      "epoch": 0.8522424542034001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123408576993592e-06,
+      "loss": 0.8082,
+      "step": 31031
+    },
+    {
+      "epoch": 0.8522699184312433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1229989890248038e-06,
+      "loss": 0.6993,
+      "step": 31032
+    },
+    {
+      "epoch": 0.8522973826590865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1225894712942854e-06,
+      "loss": 0.8177,
+      "step": 31033
+    },
+    {
+      "epoch": 0.8523248468869298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1221800238052794e-06,
+      "loss": 0.8672,
+      "step": 31034
+    },
+    {
+      "epoch": 0.852352311114773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1217706465610224e-06,
+      "loss": 0.7315,
+      "step": 31035
+    },
+    {
+      "epoch": 0.8523797753426162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.121361339564755e-06,
+      "loss": 0.8094,
+      "step": 31036
+    },
+    {
+      "epoch": 0.8524072395704595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1209521028197158e-06,
+      "loss": 0.7566,
+      "step": 31037
+    },
+    {
+      "epoch": 0.8524347037983027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1205429363291443e-06,
+      "loss": 0.8815,
+      "step": 31038
+    },
+    {
+      "epoch": 0.8524621680261459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1201338400962757e-06,
+      "loss": 0.8376,
+      "step": 31039
+    },
+    {
+      "epoch": 0.8524896322539892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1197248141243477e-06,
+      "loss": 0.7543,
+      "step": 31040
+    },
+    {
+      "epoch": 0.8525170964818324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1193158584165987e-06,
+      "loss": 0.8082,
+      "step": 31041
+    },
+    {
+      "epoch": 0.8525445607096757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1189069729762602e-06,
+      "loss": 0.8178,
+      "step": 31042
+    },
+    {
+      "epoch": 0.8525720249375189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1184981578065712e-06,
+      "loss": 0.8102,
+      "step": 31043
+    },
+    {
+      "epoch": 0.8525994891653621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1180894129107655e-06,
+      "loss": 0.7232,
+      "step": 31044
+    },
+    {
+      "epoch": 0.8526269533932054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1176807382920761e-06,
+      "loss": 0.7459,
+      "step": 31045
+    },
+    {
+      "epoch": 0.8526544176210485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1172721339537374e-06,
+      "loss": 0.8248,
+      "step": 31046
+    },
+    {
+      "epoch": 0.8526818818488918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.116863599898982e-06,
+      "loss": 0.7981,
+      "step": 31047
+    },
+    {
+      "epoch": 0.8527093460767351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1164551361310427e-06,
+      "loss": 0.8385,
+      "step": 31048
+    },
+    {
+      "epoch": 0.8527368103045783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.116046742653154e-06,
+      "loss": 0.9305,
+      "step": 31049
+    },
+    {
+      "epoch": 0.8527642745324215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1156384194685444e-06,
+      "loss": 0.9388,
+      "step": 31050
+    },
+    {
+      "epoch": 0.8527917387602647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1152301665804421e-06,
+      "loss": 0.7706,
+      "step": 31051
+    },
+    {
+      "epoch": 0.852819202988108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1148219839920815e-06,
+      "loss": 0.7626,
+      "step": 31052
+    },
+    {
+      "epoch": 0.8528466672159513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114413871706691e-06,
+      "loss": 0.8074,
+      "step": 31053
+    },
+    {
+      "epoch": 0.8528741314437944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1140058297275002e-06,
+      "loss": 0.8345,
+      "step": 31054
+    },
+    {
+      "epoch": 0.8529015956716377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.113597858057739e-06,
+      "loss": 0.8552,
+      "step": 31055
+    },
+    {
+      "epoch": 0.852929059899481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1131899567006322e-06,
+      "loss": 0.8366,
+      "step": 31056
+    },
+    {
+      "epoch": 0.8529565241273241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1127821256594084e-06,
+      "loss": 0.8111,
+      "step": 31057
+    },
+    {
+      "epoch": 0.8529839883551674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1123743649372975e-06,
+      "loss": 0.7592,
+      "step": 31058
+    },
+    {
+      "epoch": 0.8530114525830106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1119666745375213e-06,
+      "loss": 0.8213,
+      "step": 31059
+    },
+    {
+      "epoch": 0.8530389168108539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1115590544633071e-06,
+      "loss": 0.8183,
+      "step": 31060
+    },
+    {
+      "epoch": 0.8530663810386971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1111515047178834e-06,
+      "loss": 0.8269,
+      "step": 31061
+    },
+    {
+      "epoch": 0.8530938452665403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.11074402530447e-06,
+      "loss": 0.8107,
+      "step": 31062
+    },
+    {
+      "epoch": 0.8531213094943836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.110336616226293e-06,
+      "loss": 0.7918,
+      "step": 31063
+    },
+    {
+      "epoch": 0.8531487737222267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1099292774865755e-06,
+      "loss": 0.7963,
+      "step": 31064
+    },
+    {
+      "epoch": 0.85317623795007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109522009088544e-06,
+      "loss": 0.9036,
+      "step": 31065
+    },
+    {
+      "epoch": 0.8532037021779133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1091148110354167e-06,
+      "loss": 0.8079,
+      "step": 31066
+    },
+    {
+      "epoch": 0.8532311664057565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.108707683330419e-06,
+      "loss": 0.88,
+      "step": 31067
+    },
+    {
+      "epoch": 0.8532586306335997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.108300625976767e-06,
+      "loss": 0.8139,
+      "step": 31068
+    },
+    {
+      "epoch": 0.853286094861443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107893638977686e-06,
+      "loss": 0.7818,
+      "step": 31069
+    },
+    {
+      "epoch": 0.8533135590892862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107486722336395e-06,
+      "loss": 0.8427,
+      "step": 31070
+    },
+    {
+      "epoch": 0.8533410233171295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107079876056113e-06,
+      "loss": 0.8029,
+      "step": 31071
+    },
+    {
+      "epoch": 0.8533684875449726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1066731001400598e-06,
+      "loss": 0.8452,
+      "step": 31072
+    },
+    {
+      "epoch": 0.8533959517728159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1062663945914564e-06,
+      "loss": 0.8375,
+      "step": 31073
+    },
+    {
+      "epoch": 0.8534234160006592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1058597594135178e-06,
+      "loss": 0.8168,
+      "step": 31074
+    },
+    {
+      "epoch": 0.8534508802285024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1054531946094615e-06,
+      "loss": 0.7636,
+      "step": 31075
+    },
+    {
+      "epoch": 0.8534783444563456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1050467001825038e-06,
+      "loss": 0.7824,
+      "step": 31076
+    },
+    {
+      "epoch": 0.8535058086841888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1046402761358621e-06,
+      "loss": 0.8127,
+      "step": 31077
+    },
+    {
+      "epoch": 0.8535332729120321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1042339224727528e-06,
+      "loss": 0.7666,
+      "step": 31078
+    },
+    {
+      "epoch": 0.8535607371398753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1038276391963921e-06,
+      "loss": 0.8438,
+      "step": 31079
+    },
+    {
+      "epoch": 0.8535882013677185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1034214263099907e-06,
+      "loss": 0.8316,
+      "step": 31080
+    },
+    {
+      "epoch": 0.8536156655955618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1030152838167663e-06,
+      "loss": 0.7154,
+      "step": 31081
+    },
+    {
+      "epoch": 0.8536431298234051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1026092117199328e-06,
+      "loss": 0.8754,
+      "step": 31082
+    },
+    {
+      "epoch": 0.8536705940512482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1022032100226999e-06,
+      "loss": 0.8204,
+      "step": 31083
+    },
+    {
+      "epoch": 0.8536980582790915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1017972787282815e-06,
+      "loss": 0.8377,
+      "step": 31084
+    },
+    {
+      "epoch": 0.8537255225069347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.101391417839892e-06,
+      "loss": 0.908,
+      "step": 31085
+    },
+    {
+      "epoch": 0.853752986734778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1009856273607388e-06,
+      "loss": 0.7977,
+      "step": 31086
+    },
+    {
+      "epoch": 0.8537804509626212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100579907294035e-06,
+      "loss": 0.8181,
+      "step": 31087
+    },
+    {
+      "epoch": 0.8538079151904644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1001742576429897e-06,
+      "loss": 0.8407,
+      "step": 31088
+    },
+    {
+      "epoch": 0.8538353794183077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0997686784108153e-06,
+      "loss": 0.8625,
+      "step": 31089
+    },
+    {
+      "epoch": 0.8538628436461508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0993631696007157e-06,
+      "loss": 0.8851,
+      "step": 31090
+    },
+    {
+      "epoch": 0.8538903078739941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.098957731215906e-06,
+      "loss": 0.7705,
+      "step": 31091
+    },
+    {
+      "epoch": 0.8539177721018374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0985523632595873e-06,
+      "loss": 0.7838,
+      "step": 31092
+    },
+    {
+      "epoch": 0.8539452363296806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0981470657349714e-06,
+      "loss": 0.7794,
+      "step": 31093
+    },
+    {
+      "epoch": 0.8539727005575238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0977418386452642e-06,
+      "loss": 0.7854,
+      "step": 31094
+    },
+    {
+      "epoch": 0.8540001647853671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0973366819936715e-06,
+      "loss": 0.7071,
+      "step": 31095
+    },
+    {
+      "epoch": 0.8540276290132103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0969315957833992e-06,
+      "loss": 0.8428,
+      "step": 31096
+    },
+    {
+      "epoch": 0.8540550932410536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.096526580017656e-06,
+      "loss": 0.86,
+      "step": 31097
+    },
+    {
+      "epoch": 0.8540825574688967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0961216346996428e-06,
+      "loss": 0.785,
+      "step": 31098
+    },
+    {
+      "epoch": 0.85411002169674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0957167598325612e-06,
+      "loss": 0.8674,
+      "step": 31099
+    },
+    {
+      "epoch": 0.8541374859245833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.095311955419619e-06,
+      "loss": 0.7525,
+      "step": 31100
+    },
+    {
+      "epoch": 0.8541649501524264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0949072214640177e-06,
+      "loss": 0.8524,
+      "step": 31101
+    },
+    {
+      "epoch": 0.8541924143802697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0945025579689605e-06,
+      "loss": 0.6493,
+      "step": 31102
+    },
+    {
+      "epoch": 0.8542198786081129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0940979649376505e-06,
+      "loss": 0.8914,
+      "step": 31103
+    },
+    {
+      "epoch": 0.8542473428359562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0936934423732848e-06,
+      "loss": 0.7855,
+      "step": 31104
+    },
+    {
+      "epoch": 0.8542748070637994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0932889902790654e-06,
+      "loss": 0.7676,
+      "step": 31105
+    },
+    {
+      "epoch": 0.8543022712916426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092884608658198e-06,
+      "loss": 0.964,
+      "step": 31106
+    },
+    {
+      "epoch": 0.8543297355194859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092480297513875e-06,
+      "loss": 0.8211,
+      "step": 31107
+    },
+    {
+      "epoch": 0.8543571997473292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0920760568492995e-06,
+      "loss": 0.7357,
+      "step": 31108
+    },
+    {
+      "epoch": 0.8543846639751723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0916718866676667e-06,
+      "loss": 0.8895,
+      "step": 31109
+    },
+    {
+      "epoch": 0.8544121282030156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0912677869721777e-06,
+      "loss": 0.9507,
+      "step": 31110
+    },
+    {
+      "epoch": 0.8544395924308588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0908637577660286e-06,
+      "loss": 0.8076,
+      "step": 31111
+    },
+    {
+      "epoch": 0.854467056658702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0904597990524169e-06,
+      "loss": 0.8339,
+      "step": 31112
+    },
+    {
+      "epoch": 0.8544945208865453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.09005591083454e-06,
+      "loss": 0.7954,
+      "step": 31113
+    },
+    {
+      "epoch": 0.8545219851143885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0896520931155907e-06,
+      "loss": 0.855,
+      "step": 31114
+    },
+    {
+      "epoch": 0.8545494493422318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0892483458987668e-06,
+      "loss": 0.8248,
+      "step": 31115
+    },
+    {
+      "epoch": 0.8545769135700749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0888446691872601e-06,
+      "loss": 0.7169,
+      "step": 31116
+    },
+    {
+      "epoch": 0.8546043777979182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0884410629842657e-06,
+      "loss": 0.8157,
+      "step": 31117
+    },
+    {
+      "epoch": 0.8546318420257615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0880375272929778e-06,
+      "loss": 0.8017,
+      "step": 31118
+    },
+    {
+      "epoch": 0.8546593062536046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0876340621165893e-06,
+      "loss": 0.7854,
+      "step": 31119
+    },
+    {
+      "epoch": 0.8546867704814479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0872306674582945e-06,
+      "loss": 0.9086,
+      "step": 31120
+    },
+    {
+      "epoch": 0.8547142347092912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0868273433212816e-06,
+      "loss": 0.8476,
+      "step": 31121
+    },
+    {
+      "epoch": 0.8547416989371344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0864240897087452e-06,
+      "loss": 0.8587,
+      "step": 31122
+    },
+    {
+      "epoch": 0.8547691631649776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0860209066238714e-06,
+      "loss": 0.9688,
+      "step": 31123
+    },
+    {
+      "epoch": 0.8547966273928208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0856177940698542e-06,
+      "loss": 0.8688,
+      "step": 31124
+    },
+    {
+      "epoch": 0.8548240916206641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0852147520498813e-06,
+      "loss": 0.7795,
+      "step": 31125
+    },
+    {
+      "epoch": 0.8548515558485074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0848117805671444e-06,
+      "loss": 0.7264,
+      "step": 31126
+    },
+    {
+      "epoch": 0.8548790200763505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0844088796248287e-06,
+      "loss": 0.6678,
+      "step": 31127
+    },
+    {
+      "epoch": 0.8549064843041938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0840060492261228e-06,
+      "loss": 0.7805,
+      "step": 31128
+    },
+    {
+      "epoch": 0.854933948532037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836032893742154e-06,
+      "loss": 0.759,
+      "step": 31129
+    },
+    {
+      "epoch": 0.8549614127598802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083200600072295e-06,
+      "loss": 0.7046,
+      "step": 31130
+    },
+    {
+      "epoch": 0.8549888769877235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0827979813235433e-06,
+      "loss": 0.9317,
+      "step": 31131
+    },
+    {
+      "epoch": 0.8550163412155667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0823954331311493e-06,
+      "loss": 0.8104,
+      "step": 31132
+    },
+    {
+      "epoch": 0.85504380544341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0819929554982954e-06,
+      "loss": 0.817,
+      "step": 31133
+    },
+    {
+      "epoch": 0.8550712696712532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0815905484281686e-06,
+      "loss": 0.7875,
+      "step": 31134
+    },
+    {
+      "epoch": 0.8550987338990964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0811882119239514e-06,
+      "loss": 0.8535,
+      "step": 31135
+    },
+    {
+      "epoch": 0.8551261981269397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0807859459888282e-06,
+      "loss": 0.8518,
+      "step": 31136
+    },
+    {
+      "epoch": 0.8551536623547829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080383750625983e-06,
+      "loss": 0.7627,
+      "step": 31137
+    },
+    {
+      "epoch": 0.8551811265826261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0799816258385941e-06,
+      "loss": 0.8022,
+      "step": 31138
+    },
+    {
+      "epoch": 0.8552085908104694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0795795716298496e-06,
+      "loss": 0.8738,
+      "step": 31139
+    },
+    {
+      "epoch": 0.8552360550383126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079177588002923e-06,
+      "loss": 0.7953,
+      "step": 31140
+    },
+    {
+      "epoch": 0.8552635192661558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0787756749610002e-06,
+      "loss": 0.8228,
+      "step": 31141
+    },
+    {
+      "epoch": 0.855290983493999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0783738325072602e-06,
+      "loss": 0.8066,
+      "step": 31142
+    },
+    {
+      "epoch": 0.8553184477218423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0779720606448817e-06,
+      "loss": 0.7644,
+      "step": 31143
+    },
+    {
+      "epoch": 0.8553459119496856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077570359377047e-06,
+      "loss": 0.8093,
+      "step": 31144
+    },
+    {
+      "epoch": 0.8553733761775287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0771687287069287e-06,
+      "loss": 0.7623,
+      "step": 31145
+    },
+    {
+      "epoch": 0.855400840405372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0767671686377113e-06,
+      "loss": 0.8042,
+      "step": 31146
+    },
+    {
+      "epoch": 0.8554283046332153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0763656791725653e-06,
+      "loss": 0.7669,
+      "step": 31147
+    },
+    {
+      "epoch": 0.8554557688610585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0759642603146714e-06,
+      "loss": 0.8257,
+      "step": 31148
+    },
+    {
+      "epoch": 0.8554832330889017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0755629120672051e-06,
+      "loss": 0.8376,
+      "step": 31149
+    },
+    {
+      "epoch": 0.8555106973167449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0751616344333438e-06,
+      "loss": 0.8893,
+      "step": 31150
+    },
+    {
+      "epoch": 0.8555381615445882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0747604274162582e-06,
+      "loss": 0.8444,
+      "step": 31151
+    },
+    {
+      "epoch": 0.8555656257724314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0743592910191259e-06,
+      "loss": 0.7918,
+      "step": 31152
+    },
+    {
+      "epoch": 0.8555930900002746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0739582252451207e-06,
+      "loss": 0.7754,
+      "step": 31153
+    },
+    {
+      "epoch": 0.8556205542281179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0735572300974173e-06,
+      "loss": 0.7777,
+      "step": 31154
+    },
+    {
+      "epoch": 0.8556480184559611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.073156305579185e-06,
+      "loss": 0.7367,
+      "step": 31155
+    },
+    {
+      "epoch": 0.8556754826838043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0727554516936001e-06,
+      "loss": 0.7701,
+      "step": 31156
+    },
+    {
+      "epoch": 0.8557029469116476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0723546684438302e-06,
+      "loss": 0.8242,
+      "step": 31157
+    },
+    {
+      "epoch": 0.8557304111394908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0719539558330482e-06,
+      "loss": 0.8021,
+      "step": 31158
+    },
+    {
+      "epoch": 0.855757875367334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0715533138644253e-06,
+      "loss": 0.8004,
+      "step": 31159
+    },
+    {
+      "epoch": 0.8557853395951773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0711527425411305e-06,
+      "loss": 0.845,
+      "step": 31160
+    },
+    {
+      "epoch": 0.8558128038230205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0707522418663352e-06,
+      "loss": 0.8352,
+      "step": 31161
+    },
+    {
+      "epoch": 0.8558402680508638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0703518118432088e-06,
+      "loss": 0.724,
+      "step": 31162
+    },
+    {
+      "epoch": 0.8558677322787069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0699514524749144e-06,
+      "loss": 0.7924,
+      "step": 31163
+    },
+    {
+      "epoch": 0.8558951965065502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.069551163764624e-06,
+      "loss": 0.7543,
+      "step": 31164
+    },
+    {
+      "epoch": 0.8559226607343935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0691509457155047e-06,
+      "loss": 0.6907,
+      "step": 31165
+    },
+    {
+      "epoch": 0.8559501249622367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0687507983307211e-06,
+      "loss": 0.8263,
+      "step": 31166
+    },
+    {
+      "epoch": 0.8559775891900799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0683507216134415e-06,
+      "loss": 0.7257,
+      "step": 31167
+    },
+    {
+      "epoch": 0.8560050534179231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0679507155668334e-06,
+      "loss": 0.8035,
+      "step": 31168
+    },
+    {
+      "epoch": 0.8560325176457664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0675507801940566e-06,
+      "loss": 0.8297,
+      "step": 31169
+    },
+    {
+      "epoch": 0.8560599818736097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0671509154982807e-06,
+      "loss": 0.8278,
+      "step": 31170
+    },
+    {
+      "epoch": 0.8560874461014528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0667511214826643e-06,
+      "loss": 0.788,
+      "step": 31171
+    },
+    {
+      "epoch": 0.8561149103292961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0663513981503737e-06,
+      "loss": 0.8278,
+      "step": 31172
+    },
+    {
+      "epoch": 0.8561423745571394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065951745504573e-06,
+      "loss": 0.7474,
+      "step": 31173
+    },
+    {
+      "epoch": 0.8561698387849825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0655521635484212e-06,
+      "loss": 0.7024,
+      "step": 31174
+    },
+    {
+      "epoch": 0.8561973030128258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0651526522850818e-06,
+      "loss": 0.782,
+      "step": 31175
+    },
+    {
+      "epoch": 0.856224767240669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0647532117177151e-06,
+      "loss": 0.72,
+      "step": 31176
+    },
+    {
+      "epoch": 0.8562522314685123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0643538418494824e-06,
+      "loss": 0.8699,
+      "step": 31177
+    },
+    {
+      "epoch": 0.8562796956963555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0639545426835462e-06,
+      "loss": 0.6311,
+      "step": 31178
+    },
+    {
+      "epoch": 0.8563071599241987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0635553142230625e-06,
+      "loss": 0.7939,
+      "step": 31179
+    },
+    {
+      "epoch": 0.856334624152042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.063156156471188e-06,
+      "loss": 0.7559,
+      "step": 31180
+    },
+    {
+      "epoch": 0.8563620883798851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0627570694310852e-06,
+      "loss": 0.7682,
+      "step": 31181
+    },
+    {
+      "epoch": 0.8563895526077284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0623580531059096e-06,
+      "loss": 0.7581,
+      "step": 31182
+    },
+    {
+      "epoch": 0.8564170168355717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061959107498819e-06,
+      "loss": 0.8086,
+      "step": 31183
+    },
+    {
+      "epoch": 0.8564444810634149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0615602326129703e-06,
+      "loss": 0.7968,
+      "step": 31184
+    },
+    {
+      "epoch": 0.8564719452912581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0611614284515225e-06,
+      "loss": 0.7727,
+      "step": 31185
+    },
+    {
+      "epoch": 0.8564994095191014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0607626950176265e-06,
+      "loss": 0.9147,
+      "step": 31186
+    },
+    {
+      "epoch": 0.8565268737469446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0603640323144381e-06,
+      "loss": 0.8444,
+      "step": 31187
+    },
+    {
+      "epoch": 0.8565543379747879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0599654403451121e-06,
+      "loss": 0.8126,
+      "step": 31188
+    },
+    {
+      "epoch": 0.856581802202631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0595669191128023e-06,
+      "loss": 0.8136,
+      "step": 31189
+    },
+    {
+      "epoch": 0.8566092664304743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0591684686206616e-06,
+      "loss": 0.8347,
+      "step": 31190
+    },
+    {
+      "epoch": 0.8566367306583176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0587700888718466e-06,
+      "loss": 0.7818,
+      "step": 31191
+    },
+    {
+      "epoch": 0.8566641948861607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0583717798695036e-06,
+      "loss": 0.7815,
+      "step": 31192
+    },
+    {
+      "epoch": 0.856691659114004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0579735416167858e-06,
+      "loss": 0.8041,
+      "step": 31193
+    },
+    {
+      "epoch": 0.8567191233418472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0575753741168483e-06,
+      "loss": 0.8792,
+      "step": 31194
+    },
+    {
+      "epoch": 0.8567465875696905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0571772773728362e-06,
+      "loss": 0.8843,
+      "step": 31195
+    },
+    {
+      "epoch": 0.8567740517975337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0567792513879017e-06,
+      "loss": 0.7959,
+      "step": 31196
+    },
+    {
+      "epoch": 0.8568015160253769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0563812961651954e-06,
+      "loss": 0.8231,
+      "step": 31197
+    },
+    {
+      "epoch": 0.8568289802532202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0559834117078637e-06,
+      "loss": 0.7901,
+      "step": 31198
+    },
+    {
+      "epoch": 0.8568564444810635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0555855980190554e-06,
+      "loss": 0.7794,
+      "step": 31199
+    },
+    {
+      "epoch": 0.8568839087089066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.055187855101918e-06,
+      "loss": 0.7976,
+      "step": 31200
+    },
+    {
+      "epoch": 0.8569113729367499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.054790182959602e-06,
+      "loss": 0.8024,
+      "step": 31201
+    },
+    {
+      "epoch": 0.8569388371645931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0543925815952483e-06,
+      "loss": 0.803,
+      "step": 31202
+    },
+    {
+      "epoch": 0.8569663013924363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0539950510120079e-06,
+      "loss": 0.7791,
+      "step": 31203
+    },
+    {
+      "epoch": 0.8569937656202796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0535975912130224e-06,
+      "loss": 0.8773,
+      "step": 31204
+    },
+    {
+      "epoch": 0.8570212298481228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0532002022014386e-06,
+      "loss": 0.7319,
+      "step": 31205
+    },
+    {
+      "epoch": 0.8570486940759661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0528028839803994e-06,
+      "loss": 0.7762,
+      "step": 31206
+    },
+    {
+      "epoch": 0.8570761583038093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.05240563655305e-06,
+      "loss": 0.7939,
+      "step": 31207
+    },
+    {
+      "epoch": 0.8571036225316525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0520084599225334e-06,
+      "loss": 0.8144,
+      "step": 31208
+    },
+    {
+      "epoch": 0.8571310867594958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0516113540919947e-06,
+      "loss": 0.8394,
+      "step": 31209
+    },
+    {
+      "epoch": 0.857158550987339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0512143190645718e-06,
+      "loss": 0.8199,
+      "step": 31210
+    },
+    {
+      "epoch": 0.8571860152151822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0508173548434064e-06,
+      "loss": 0.722,
+      "step": 31211
+    },
+    {
+      "epoch": 0.8572134794430255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0504204614316405e-06,
+      "loss": 0.7868,
+      "step": 31212
+    },
+    {
+      "epoch": 0.8572409436708687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0500236388324147e-06,
+      "loss": 0.7575,
+      "step": 31213
+    },
+    {
+      "epoch": 0.857268407898712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049626887048869e-06,
+      "loss": 0.7598,
+      "step": 31214
+    },
+    {
+      "epoch": 0.8572958721265551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0492302060841442e-06,
+      "loss": 0.8707,
+      "step": 31215
+    },
+    {
+      "epoch": 0.8573233363543984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0488335959413753e-06,
+      "loss": 0.7987,
+      "step": 31216
+    },
+    {
+      "epoch": 0.8573508005822417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0484370566237023e-06,
+      "loss": 0.7751,
+      "step": 31217
+    },
+    {
+      "epoch": 0.8573782648100848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0480405881342647e-06,
+      "loss": 0.8594,
+      "step": 31218
+    },
+    {
+      "epoch": 0.8574057290379281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0476441904761957e-06,
+      "loss": 0.8835,
+      "step": 31219
+    },
+    {
+      "epoch": 0.8574331932657714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.047247863652634e-06,
+      "loss": 0.6743,
+      "step": 31220
+    },
+    {
+      "epoch": 0.8574606574936146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0468516076667168e-06,
+      "loss": 0.7802,
+      "step": 31221
+    },
+    {
+      "epoch": 0.8574881217214578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.046455422521575e-06,
+      "loss": 0.8391,
+      "step": 31222
+    },
+    {
+      "epoch": 0.857515585949301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0460593082203474e-06,
+      "loss": 0.7704,
+      "step": 31223
+    },
+    {
+      "epoch": 0.8575430501771443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0456632647661669e-06,
+      "loss": 0.8053,
+      "step": 31224
+    },
+    {
+      "epoch": 0.8575705144049875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0452672921621688e-06,
+      "loss": 0.8113,
+      "step": 31225
+    },
+    {
+      "epoch": 0.8575979786328307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0448713904114826e-06,
+      "loss": 0.8012,
+      "step": 31226
+    },
+    {
+      "epoch": 0.857625442860674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0444755595172452e-06,
+      "loss": 0.7915,
+      "step": 31227
+    },
+    {
+      "epoch": 0.8576529070885172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0440797994825834e-06,
+      "loss": 0.8188,
+      "step": 31228
+    },
+    {
+      "epoch": 0.8576803713163604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0436841103106322e-06,
+      "loss": 0.7582,
+      "step": 31229
+    },
+    {
+      "epoch": 0.8577078355442037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0432884920045217e-06,
+      "loss": 0.7994,
+      "step": 31230
+    },
+    {
+      "epoch": 0.8577352997720469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0428929445673819e-06,
+      "loss": 0.7923,
+      "step": 31231
+    },
+    {
+      "epoch": 0.8577627639998902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0424974680023447e-06,
+      "loss": 0.789,
+      "step": 31232
+    },
+    {
+      "epoch": 0.8577902282277334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0421020623125355e-06,
+      "loss": 0.9395,
+      "step": 31233
+    },
+    {
+      "epoch": 0.8578176924555766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0417067275010861e-06,
+      "loss": 0.7682,
+      "step": 31234
+    },
+    {
+      "epoch": 0.8578451566834199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.041311463571122e-06,
+      "loss": 0.7747,
+      "step": 31235
+    },
+    {
+      "epoch": 0.857872620911263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0409162705257714e-06,
+      "loss": 0.7601,
+      "step": 31236
+    },
+    {
+      "epoch": 0.8579000851391063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0405211483681621e-06,
+      "loss": 0.7434,
+      "step": 31237
+    },
+    {
+      "epoch": 0.8579275493669496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0401260971014215e-06,
+      "loss": 0.8012,
+      "step": 31238
+    },
+    {
+      "epoch": 0.8579550135947928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0397311167286728e-06,
+      "loss": 0.8468,
+      "step": 31239
+    },
+    {
+      "epoch": 0.857982477822636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.039336207253041e-06,
+      "loss": 0.7069,
+      "step": 31240
+    },
+    {
+      "epoch": 0.8580099420504792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.038941368677654e-06,
+      "loss": 0.8142,
+      "step": 31241
+    },
+    {
+      "epoch": 0.8580374062783225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0385466010056345e-06,
+      "loss": 0.7476,
+      "step": 31242
+    },
+    {
+      "epoch": 0.8580648705061658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0381519042401045e-06,
+      "loss": 0.7733,
+      "step": 31243
+    },
+    {
+      "epoch": 0.8580923347340089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0377572783841894e-06,
+      "loss": 0.7361,
+      "step": 31244
+    },
+    {
+      "epoch": 0.8581197989618522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0373627234410088e-06,
+      "loss": 0.7685,
+      "step": 31245
+    },
+    {
+      "epoch": 0.8581472631896955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0369682394136859e-06,
+      "loss": 0.8486,
+      "step": 31246
+    },
+    {
+      "epoch": 0.8581747274175386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0365738263053425e-06,
+      "loss": 0.7497,
+      "step": 31247
+    },
+    {
+      "epoch": 0.8582021916453819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0361794841190996e-06,
+      "loss": 0.7888,
+      "step": 31248
+    },
+    {
+      "epoch": 0.8582296558732251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035785212858078e-06,
+      "loss": 0.7613,
+      "step": 31249
+    },
+    {
+      "epoch": 0.8582571201010684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035391012525394e-06,
+      "loss": 0.7206,
+      "step": 31250
+    },
+    {
+      "epoch": 0.8582845843289116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.034996883124172e-06,
+      "loss": 0.8394,
+      "step": 31251
+    },
+    {
+      "epoch": 0.8583120485567548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0346028246575247e-06,
+      "loss": 0.7579,
+      "step": 31252
+    },
+    {
+      "epoch": 0.8583395127845981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.034208837128573e-06,
+      "loss": 0.6824,
+      "step": 31253
+    },
+    {
+      "epoch": 0.8583669770124412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0338149205404336e-06,
+      "loss": 0.8358,
+      "step": 31254
+    },
+    {
+      "epoch": 0.8583944412402845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0334210748962237e-06,
+      "loss": 0.8042,
+      "step": 31255
+    },
+    {
+      "epoch": 0.8584219054681278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.033027300199062e-06,
+      "loss": 0.8427,
+      "step": 31256
+    },
+    {
+      "epoch": 0.858449369695971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0326335964520595e-06,
+      "loss": 0.7499,
+      "step": 31257
+    },
+    {
+      "epoch": 0.8584768339238142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0322399636583359e-06,
+      "loss": 0.7401,
+      "step": 31258
+    },
+    {
+      "epoch": 0.8585042981516575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0318464018210005e-06,
+      "loss": 0.7708,
+      "step": 31259
+    },
+    {
+      "epoch": 0.8585317623795007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0314529109431715e-06,
+      "loss": 0.8169,
+      "step": 31260
+    },
+    {
+      "epoch": 0.858559226607344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0310594910279604e-06,
+      "loss": 0.8058,
+      "step": 31261
+    },
+    {
+      "epoch": 0.8585866908351871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0306661420784824e-06,
+      "loss": 0.8289,
+      "step": 31262
+    },
+    {
+      "epoch": 0.8586141550630304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0302728640978477e-06,
+      "loss": 0.8488,
+      "step": 31263
+    },
+    {
+      "epoch": 0.8586416192908737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0298796570891679e-06,
+      "loss": 0.8265,
+      "step": 31264
+    },
+    {
+      "epoch": 0.8586690835187168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029486521055555e-06,
+      "loss": 0.8624,
+      "step": 31265
+    },
+    {
+      "epoch": 0.8586965477465601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029093456000121e-06,
+      "loss": 0.7843,
+      "step": 31266
+    },
+    {
+      "epoch": 0.8587240119744033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0287004619259733e-06,
+      "loss": 0.8379,
+      "step": 31267
+    },
+    {
+      "epoch": 0.8587514762022466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0283075388362252e-06,
+      "loss": 0.6828,
+      "step": 31268
+    },
+    {
+      "epoch": 0.8587789404300898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0279146867339795e-06,
+      "loss": 0.8523,
+      "step": 31269
+    },
+    {
+      "epoch": 0.858806404657933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0275219056223495e-06,
+      "loss": 0.7984,
+      "step": 31270
+    },
+    {
+      "epoch": 0.8588338688857763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0271291955044416e-06,
+      "loss": 0.7939,
+      "step": 31271
+    },
+    {
+      "epoch": 0.8588613331136196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0267365563833632e-06,
+      "loss": 0.8121,
+      "step": 31272
+    },
+    {
+      "epoch": 0.8588887973414627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.026343988262223e-06,
+      "loss": 0.7301,
+      "step": 31273
+    },
+    {
+      "epoch": 0.858916261569306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0259514911441238e-06,
+      "loss": 0.8245,
+      "step": 31274
+    },
+    {
+      "epoch": 0.8589437257971492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0255590650321733e-06,
+      "loss": 0.8028,
+      "step": 31275
+    },
+    {
+      "epoch": 0.8589711900249924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0251667099294749e-06,
+      "loss": 0.7748,
+      "step": 31276
+    },
+    {
+      "epoch": 0.8589986542528357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247744258391335e-06,
+      "loss": 0.9658,
+      "step": 31277
+    },
+    {
+      "epoch": 0.8590261184806789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0243822127642545e-06,
+      "loss": 0.688,
+      "step": 31278
+    },
+    {
+      "epoch": 0.8590535827085222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0239900707079387e-06,
+      "loss": 0.6582,
+      "step": 31279
+    },
+    {
+      "epoch": 0.8590810469363653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0235979996732937e-06,
+      "loss": 0.7784,
+      "step": 31280
+    },
+    {
+      "epoch": 0.8591085111642086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0232059996634157e-06,
+      "loss": 0.8383,
+      "step": 31281
+    },
+    {
+      "epoch": 0.8591359753920519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0228140706814115e-06,
+      "loss": 0.899,
+      "step": 31282
+    },
+    {
+      "epoch": 0.859163439619895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0224222127303773e-06,
+      "loss": 0.7509,
+      "step": 31283
+    },
+    {
+      "epoch": 0.8591909038477383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0220304258134172e-06,
+      "loss": 0.8072,
+      "step": 31284
+    },
+    {
+      "epoch": 0.8592183680755816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.021638709933629e-06,
+      "loss": 0.9058,
+      "step": 31285
+    },
+    {
+      "epoch": 0.8592458323034248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0212470650941164e-06,
+      "loss": 0.8656,
+      "step": 31286
+    },
+    {
+      "epoch": 0.859273296531268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0208554912979718e-06,
+      "loss": 0.8318,
+      "step": 31287
+    },
+    {
+      "epoch": 0.8593007607591112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.020463988548297e-06,
+      "loss": 0.7999,
+      "step": 31288
+    },
+    {
+      "epoch": 0.8593282249869545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0200725568481906e-06,
+      "loss": 0.7943,
+      "step": 31289
+    },
+    {
+      "epoch": 0.8593556892147978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019681196200749e-06,
+      "loss": 0.8168,
+      "step": 31290
+    },
+    {
+      "epoch": 0.8593831534426409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019289906609069e-06,
+      "loss": 0.7944,
+      "step": 31291
+    },
+    {
+      "epoch": 0.8594106176704842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0188986880762442e-06,
+      "loss": 0.7869,
+      "step": 31292
+    },
+    {
+      "epoch": 0.8594380818983274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0185075406053724e-06,
+      "loss": 0.8351,
+      "step": 31293
+    },
+    {
+      "epoch": 0.8594655461261707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0181164641995479e-06,
+      "loss": 0.9153,
+      "step": 31294
+    },
+    {
+      "epoch": 0.8594930103540139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0177254588618645e-06,
+      "loss": 0.8386,
+      "step": 31295
+    },
+    {
+      "epoch": 0.8595204745818571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.017334524595417e-06,
+      "loss": 0.795,
+      "step": 31296
+    },
+    {
+      "epoch": 0.8595479388097004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0169436614033002e-06,
+      "loss": 0.8555,
+      "step": 31297
+    },
+    {
+      "epoch": 0.8595754030375437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.016552869288603e-06,
+      "loss": 0.8254,
+      "step": 31298
+    },
+    {
+      "epoch": 0.8596028672653868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0161621482544214e-06,
+      "loss": 0.8915,
+      "step": 31299
+    },
+    {
+      "epoch": 0.8596303314932301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0157714983038424e-06,
+      "loss": 0.8344,
+      "step": 31300
+    },
+    {
+      "epoch": 0.8596577957210733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.01538091943996e-06,
+      "loss": 0.8574,
+      "step": 31301
+    },
+    {
+      "epoch": 0.8596852599489165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0149904116658637e-06,
+      "loss": 0.7543,
+      "step": 31302
+    },
+    {
+      "epoch": 0.8597127241767598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014599974984647e-06,
+      "loss": 0.8644,
+      "step": 31303
+    },
+    {
+      "epoch": 0.859740188404603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0142096093993925e-06,
+      "loss": 0.8422,
+      "step": 31304
+    },
+    {
+      "epoch": 0.8597676526324463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0138193149131925e-06,
+      "loss": 0.7471,
+      "step": 31305
+    },
+    {
+      "epoch": 0.8597951168602894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0134290915291378e-06,
+      "loss": 0.772,
+      "step": 31306
+    },
+    {
+      "epoch": 0.8598225810881327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0130389392503103e-06,
+      "loss": 0.8928,
+      "step": 31307
+    },
+    {
+      "epoch": 0.859850045315976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0126488580797999e-06,
+      "loss": 0.7789,
+      "step": 31308
+    },
+    {
+      "epoch": 0.8598775095438191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0122588480206952e-06,
+      "loss": 0.7719,
+      "step": 31309
+    },
+    {
+      "epoch": 0.8599049737716624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0118689090760792e-06,
+      "loss": 0.834,
+      "step": 31310
+    },
+    {
+      "epoch": 0.8599324379995057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.011479041249036e-06,
+      "loss": 0.6914,
+      "step": 31311
+    },
+    {
+      "epoch": 0.8599599022273489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0110892445426545e-06,
+      "loss": 0.9334,
+      "step": 31312
+    },
+    {
+      "epoch": 0.8599873664551921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0106995189600156e-06,
+      "loss": 0.8045,
+      "step": 31313
+    },
+    {
+      "epoch": 0.8600148306830353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0103098645042064e-06,
+      "loss": 0.7472,
+      "step": 31314
+    },
+    {
+      "epoch": 0.8600422949108786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.009920281178308e-06,
+      "loss": 0.7851,
+      "step": 31315
+    },
+    {
+      "epoch": 0.8600697591387219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0095307689854005e-06,
+      "loss": 0.7202,
+      "step": 31316
+    },
+    {
+      "epoch": 0.860097223366565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0091413279285689e-06,
+      "loss": 0.8195,
+      "step": 31317
+    },
+    {
+      "epoch": 0.8601246875944083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0087519580108928e-06,
+      "loss": 0.8099,
+      "step": 31318
+    },
+    {
+      "epoch": 0.8601521518222515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0083626592354545e-06,
+      "loss": 0.8098,
+      "step": 31319
+    },
+    {
+      "epoch": 0.8601796160500947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0079734316053346e-06,
+      "loss": 0.7509,
+      "step": 31320
+    },
+    {
+      "epoch": 0.860207080277938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0075842751236132e-06,
+      "loss": 0.7657,
+      "step": 31321
+    },
+    {
+      "epoch": 0.8602345445057812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0071951897933684e-06,
+      "loss": 0.7949,
+      "step": 31322
+    },
+    {
+      "epoch": 0.8602620087336245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0068061756176773e-06,
+      "loss": 0.7441,
+      "step": 31323
+    },
+    {
+      "epoch": 0.8602894729614677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064172325996191e-06,
+      "loss": 0.752,
+      "step": 31324
+    },
+    {
+      "epoch": 0.8603169371893109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0060283607422706e-06,
+      "loss": 0.7607,
+      "step": 31325
+    },
+    {
+      "epoch": 0.8603444014171542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0056395600487113e-06,
+      "loss": 0.8501,
+      "step": 31326
+    },
+    {
+      "epoch": 0.8603718656449973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0052508305220165e-06,
+      "loss": 0.7741,
+      "step": 31327
+    },
+    {
+      "epoch": 0.8603993298728406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0048621721652596e-06,
+      "loss": 0.8383,
+      "step": 31328
+    },
+    {
+      "epoch": 0.8604267941006839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0044735849815178e-06,
+      "loss": 0.8472,
+      "step": 31329
+    },
+    {
+      "epoch": 0.8604542583285271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.004085068973868e-06,
+      "loss": 0.8721,
+      "step": 31330
+    },
+    {
+      "epoch": 0.8604817225563703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0036966241453793e-06,
+      "loss": 0.8613,
+      "step": 31331
+    },
+    {
+      "epoch": 0.8605091867842135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0033082504991286e-06,
+      "loss": 0.7657,
+      "step": 31332
+    },
+    {
+      "epoch": 0.8605366510120568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.002919948038189e-06,
+      "loss": 0.7672,
+      "step": 31333
+    },
+    {
+      "epoch": 0.8605641152399001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0025317167656301e-06,
+      "loss": 0.8125,
+      "step": 31334
+    },
+    {
+      "epoch": 0.8605915794677432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0021435566845262e-06,
+      "loss": 0.8941,
+      "step": 31335
+    },
+    {
+      "epoch": 0.8606190436955865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0017554677979468e-06,
+      "loss": 0.8353,
+      "step": 31336
+    },
+    {
+      "epoch": 0.8606465079234298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0013674501089644e-06,
+      "loss": 0.7966,
+      "step": 31337
+    },
+    {
+      "epoch": 0.860673972151273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0009795036206504e-06,
+      "loss": 0.8114,
+      "step": 31338
+    },
+    {
+      "epoch": 0.8607014363791162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0005916283360717e-06,
+      "loss": 0.8561,
+      "step": 31339
+    },
+    {
+      "epoch": 0.8607289006069594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0002038242582968e-06,
+      "loss": 0.7799,
+      "step": 31340
+    },
+    {
+      "epoch": 0.8607563648348027,
+      "grad_norm": 0.0,
+      "learning_rate": 9.998160913903953e-07,
+      "loss": 0.7793,
+      "step": 31341
+    },
+    {
+      "epoch": 0.8607838290626459,
+      "grad_norm": 0.0,
+      "learning_rate": 9.994284297354362e-07,
+      "loss": 0.8299,
+      "step": 31342
+    },
+    {
+      "epoch": 0.8608112932904891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.990408392964845e-07,
+      "loss": 0.7728,
+      "step": 31343
+    },
+    {
+      "epoch": 0.8608387575183324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.986533200766092e-07,
+      "loss": 0.7257,
+      "step": 31344
+    },
+    {
+      "epoch": 0.8608662217461756,
+      "grad_norm": 0.0,
+      "learning_rate": 9.982658720788773e-07,
+      "loss": 0.8586,
+      "step": 31345
+    },
+    {
+      "epoch": 0.8608936859740188,
+      "grad_norm": 0.0,
+      "learning_rate": 9.978784953063526e-07,
+      "loss": 0.8022,
+      "step": 31346
+    },
+    {
+      "epoch": 0.8609211502018621,
+      "grad_norm": 0.0,
+      "learning_rate": 9.974911897620987e-07,
+      "loss": 0.7896,
+      "step": 31347
+    },
+    {
+      "epoch": 0.8609486144297053,
+      "grad_norm": 0.0,
+      "learning_rate": 9.971039554491814e-07,
+      "loss": 0.8566,
+      "step": 31348
+    },
+    {
+      "epoch": 0.8609760786575485,
+      "grad_norm": 0.0,
+      "learning_rate": 9.967167923706644e-07,
+      "loss": 0.8864,
+      "step": 31349
+    },
+    {
+      "epoch": 0.8610035428853918,
+      "grad_norm": 0.0,
+      "learning_rate": 9.963297005296146e-07,
+      "loss": 0.8164,
+      "step": 31350
+    },
+    {
+      "epoch": 0.861031007113235,
+      "grad_norm": 0.0,
+      "learning_rate": 9.959426799290895e-07,
+      "loss": 0.773,
+      "step": 31351
+    },
+    {
+      "epoch": 0.8610584713410783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.955557305721542e-07,
+      "loss": 0.8413,
+      "step": 31352
+    },
+    {
+      "epoch": 0.8610859355689214,
+      "grad_norm": 0.0,
+      "learning_rate": 9.951688524618696e-07,
+      "loss": 0.7244,
+      "step": 31353
+    },
+    {
+      "epoch": 0.8611133997967647,
+      "grad_norm": 0.0,
+      "learning_rate": 9.947820456012991e-07,
+      "loss": 0.866,
+      "step": 31354
+    },
+    {
+      "epoch": 0.861140864024608,
+      "grad_norm": 0.0,
+      "learning_rate": 9.94395309993499e-07,
+      "loss": 0.9059,
+      "step": 31355
+    },
+    {
+      "epoch": 0.8611683282524512,
+      "grad_norm": 0.0,
+      "learning_rate": 9.940086456415343e-07,
+      "loss": 0.8298,
+      "step": 31356
+    },
+    {
+      "epoch": 0.8611957924802944,
+      "grad_norm": 0.0,
+      "learning_rate": 9.936220525484596e-07,
+      "loss": 0.7751,
+      "step": 31357
+    },
+    {
+      "epoch": 0.8612232567081376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.93235530717336e-07,
+      "loss": 0.8233,
+      "step": 31358
+    },
+    {
+      "epoch": 0.8612507209359809,
+      "grad_norm": 0.0,
+      "learning_rate": 9.928490801512214e-07,
+      "loss": 0.7942,
+      "step": 31359
+    },
+    {
+      "epoch": 0.8612781851638242,
+      "grad_norm": 0.0,
+      "learning_rate": 9.92462700853175e-07,
+      "loss": 0.7888,
+      "step": 31360
+    },
+    {
+      "epoch": 0.8613056493916673,
+      "grad_norm": 0.0,
+      "learning_rate": 9.920763928262534e-07,
+      "loss": 0.6928,
+      "step": 31361
+    },
+    {
+      "epoch": 0.8613331136195106,
+      "grad_norm": 0.0,
+      "learning_rate": 9.916901560735115e-07,
+      "loss": 0.8235,
+      "step": 31362
+    },
+    {
+      "epoch": 0.8613605778473539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.913039905980082e-07,
+      "loss": 0.8532,
+      "step": 31363
+    },
+    {
+      "epoch": 0.861388042075197,
+      "grad_norm": 0.0,
+      "learning_rate": 9.909178964027966e-07,
+      "loss": 0.8813,
+      "step": 31364
+    },
+    {
+      "epoch": 0.8614155063030403,
+      "grad_norm": 0.0,
+      "learning_rate": 9.905318734909307e-07,
+      "loss": 0.8589,
+      "step": 31365
+    },
+    {
+      "epoch": 0.8614429705308835,
+      "grad_norm": 0.0,
+      "learning_rate": 9.901459218654674e-07,
+      "loss": 0.8126,
+      "step": 31366
+    },
+    {
+      "epoch": 0.8614704347587268,
+      "grad_norm": 0.0,
+      "learning_rate": 9.897600415294596e-07,
+      "loss": 0.8825,
+      "step": 31367
+    },
+    {
+      "epoch": 0.86149789898657,
+      "grad_norm": 0.0,
+      "learning_rate": 9.893742324859624e-07,
+      "loss": 0.7789,
+      "step": 31368
+    },
+    {
+      "epoch": 0.8615253632144132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.889884947380257e-07,
+      "loss": 0.8551,
+      "step": 31369
+    },
+    {
+      "epoch": 0.8615528274422565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.886028282887027e-07,
+      "loss": 0.9073,
+      "step": 31370
+    },
+    {
+      "epoch": 0.8615802916700996,
+      "grad_norm": 0.0,
+      "learning_rate": 9.882172331410444e-07,
+      "loss": 0.8832,
+      "step": 31371
+    },
+    {
+      "epoch": 0.8616077558979429,
+      "grad_norm": 0.0,
+      "learning_rate": 9.878317092981016e-07,
+      "loss": 0.7761,
+      "step": 31372
+    },
+    {
+      "epoch": 0.8616352201257862,
+      "grad_norm": 0.0,
+      "learning_rate": 9.87446256762925e-07,
+      "loss": 0.7488,
+      "step": 31373
+    },
+    {
+      "epoch": 0.8616626843536294,
+      "grad_norm": 0.0,
+      "learning_rate": 9.87060875538567e-07,
+      "loss": 0.7206,
+      "step": 31374
+    },
+    {
+      "epoch": 0.8616901485814726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.866755656280724e-07,
+      "loss": 0.8237,
+      "step": 31375
+    },
+    {
+      "epoch": 0.8617176128093159,
+      "grad_norm": 0.0,
+      "learning_rate": 9.862903270344926e-07,
+      "loss": 0.7864,
+      "step": 31376
+    },
+    {
+      "epoch": 0.8617450770371591,
+      "grad_norm": 0.0,
+      "learning_rate": 9.859051597608748e-07,
+      "loss": 0.7958,
+      "step": 31377
+    },
+    {
+      "epoch": 0.8617725412650024,
+      "grad_norm": 0.0,
+      "learning_rate": 9.855200638102691e-07,
+      "loss": 0.7632,
+      "step": 31378
+    },
+    {
+      "epoch": 0.8618000054928455,
+      "grad_norm": 0.0,
+      "learning_rate": 9.851350391857194e-07,
+      "loss": 0.7881,
+      "step": 31379
+    },
+    {
+      "epoch": 0.8618274697206888,
+      "grad_norm": 0.0,
+      "learning_rate": 9.847500858902748e-07,
+      "loss": 0.7925,
+      "step": 31380
+    },
+    {
+      "epoch": 0.8618549339485321,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84365203926977e-07,
+      "loss": 0.7961,
+      "step": 31381
+    },
+    {
+      "epoch": 0.8618823981763752,
+      "grad_norm": 0.0,
+      "learning_rate": 9.839803932988745e-07,
+      "loss": 0.8382,
+      "step": 31382
+    },
+    {
+      "epoch": 0.8619098624042185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.83595654009012e-07,
+      "loss": 0.7814,
+      "step": 31383
+    },
+    {
+      "epoch": 0.8619373266320617,
+      "grad_norm": 0.0,
+      "learning_rate": 9.832109860604333e-07,
+      "loss": 0.7959,
+      "step": 31384
+    },
+    {
+      "epoch": 0.861964790859905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.82826389456184e-07,
+      "loss": 0.8304,
+      "step": 31385
+    },
+    {
+      "epoch": 0.8619922550877482,
+      "grad_norm": 0.0,
+      "learning_rate": 9.824418641993028e-07,
+      "loss": 0.8234,
+      "step": 31386
+    },
+    {
+      "epoch": 0.8620197193155914,
+      "grad_norm": 0.0,
+      "learning_rate": 9.82057410292837e-07,
+      "loss": 0.8057,
+      "step": 31387
+    },
+    {
+      "epoch": 0.8620471835434347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.816730277398235e-07,
+      "loss": 0.7949,
+      "step": 31388
+    },
+    {
+      "epoch": 0.862074647771278,
+      "grad_norm": 0.0,
+      "learning_rate": 9.812887165433072e-07,
+      "loss": 0.897,
+      "step": 31389
+    },
+    {
+      "epoch": 0.8621021119991211,
+      "grad_norm": 0.0,
+      "learning_rate": 9.80904476706328e-07,
+      "loss": 0.9092,
+      "step": 31390
+    },
+    {
+      "epoch": 0.8621295762269644,
+      "grad_norm": 0.0,
+      "learning_rate": 9.805203082319258e-07,
+      "loss": 0.8901,
+      "step": 31391
+    },
+    {
+      "epoch": 0.8621570404548076,
+      "grad_norm": 0.0,
+      "learning_rate": 9.801362111231428e-07,
+      "loss": 0.7599,
+      "step": 31392
+    },
+    {
+      "epoch": 0.8621845046826508,
+      "grad_norm": 0.0,
+      "learning_rate": 9.79752185383015e-07,
+      "loss": 0.8505,
+      "step": 31393
+    },
+    {
+      "epoch": 0.8622119689104941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.793682310145825e-07,
+      "loss": 0.7699,
+      "step": 31394
+    },
+    {
+      "epoch": 0.8622394331383373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.789843480208815e-07,
+      "loss": 0.7669,
+      "step": 31395
+    },
+    {
+      "epoch": 0.8622668973661806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78600536404951e-07,
+      "loss": 0.7642,
+      "step": 31396
+    },
+    {
+      "epoch": 0.8622943615940238,
+      "grad_norm": 0.0,
+      "learning_rate": 9.782167961698275e-07,
+      "loss": 0.9125,
+      "step": 31397
+    },
+    {
+      "epoch": 0.862321825821867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.778331273185494e-07,
+      "loss": 0.88,
+      "step": 31398
+    },
+    {
+      "epoch": 0.8623492900497103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.774495298541486e-07,
+      "loss": 0.7474,
+      "step": 31399
+    },
+    {
+      "epoch": 0.8623767542775534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77066003779662e-07,
+      "loss": 0.8509,
+      "step": 31400
+    },
+    {
+      "epoch": 0.8624042185053967,
+      "grad_norm": 0.0,
+      "learning_rate": 9.766825490981235e-07,
+      "loss": 0.7373,
+      "step": 31401
+    },
+    {
+      "epoch": 0.86243168273324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76299165812571e-07,
+      "loss": 0.7815,
+      "step": 31402
+    },
+    {
+      "epoch": 0.8624591469610832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.759158539260328e-07,
+      "loss": 0.7516,
+      "step": 31403
+    },
+    {
+      "epoch": 0.8624866111889264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.755326134415454e-07,
+      "loss": 0.7893,
+      "step": 31404
+    },
+    {
+      "epoch": 0.8625140754167696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.75149444362139e-07,
+      "loss": 0.7751,
+      "step": 31405
+    },
+    {
+      "epoch": 0.8625415396446129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.747663466908452e-07,
+      "loss": 0.8643,
+      "step": 31406
+    },
+    {
+      "epoch": 0.8625690038724562,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74383320430695e-07,
+      "loss": 0.6578,
+      "step": 31407
+    },
+    {
+      "epoch": 0.8625964681002993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74000365584722e-07,
+      "loss": 0.8522,
+      "step": 31408
+    },
+    {
+      "epoch": 0.8626239323281426,
+      "grad_norm": 0.0,
+      "learning_rate": 9.736174821559553e-07,
+      "loss": 0.8184,
+      "step": 31409
+    },
+    {
+      "epoch": 0.8626513965559859,
+      "grad_norm": 0.0,
+      "learning_rate": 9.732346701474227e-07,
+      "loss": 0.8581,
+      "step": 31410
+    },
+    {
+      "epoch": 0.862678860783829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.728519295621553e-07,
+      "loss": 0.7791,
+      "step": 31411
+    },
+    {
+      "epoch": 0.8627063250116723,
+      "grad_norm": 0.0,
+      "learning_rate": 9.724692604031782e-07,
+      "loss": 0.7379,
+      "step": 31412
+    },
+    {
+      "epoch": 0.8627337892395155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.720866626735215e-07,
+      "loss": 0.7784,
+      "step": 31413
+    },
+    {
+      "epoch": 0.8627612534673588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.717041363762115e-07,
+      "loss": 0.8564,
+      "step": 31414
+    },
+    {
+      "epoch": 0.862788717695202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.713216815142779e-07,
+      "loss": 0.7527,
+      "step": 31415
+    },
+    {
+      "epoch": 0.8628161819230452,
+      "grad_norm": 0.0,
+      "learning_rate": 9.709392980907428e-07,
+      "loss": 0.8275,
+      "step": 31416
+    },
+    {
+      "epoch": 0.8628436461508885,
+      "grad_norm": 0.0,
+      "learning_rate": 9.70556986108634e-07,
+      "loss": 0.8,
+      "step": 31417
+    },
+    {
+      "epoch": 0.8628711103787317,
+      "grad_norm": 0.0,
+      "learning_rate": 9.701747455709765e-07,
+      "loss": 0.8115,
+      "step": 31418
+    },
+    {
+      "epoch": 0.8628985746065749,
+      "grad_norm": 0.0,
+      "learning_rate": 9.697925764807937e-07,
+      "loss": 0.8157,
+      "step": 31419
+    },
+    {
+      "epoch": 0.8629260388344182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.694104788411095e-07,
+      "loss": 0.7982,
+      "step": 31420
+    },
+    {
+      "epoch": 0.8629535030622614,
+      "grad_norm": 0.0,
+      "learning_rate": 9.690284526549498e-07,
+      "loss": 0.7858,
+      "step": 31421
+    },
+    {
+      "epoch": 0.8629809672901047,
+      "grad_norm": 0.0,
+      "learning_rate": 9.68646497925333e-07,
+      "loss": 0.7099,
+      "step": 31422
+    },
+    {
+      "epoch": 0.8630084315179479,
+      "grad_norm": 0.0,
+      "learning_rate": 9.682646146552832e-07,
+      "loss": 0.6999,
+      "step": 31423
+    },
+    {
+      "epoch": 0.8630358957457911,
+      "grad_norm": 0.0,
+      "learning_rate": 9.678828028478228e-07,
+      "loss": 0.8312,
+      "step": 31424
+    },
+    {
+      "epoch": 0.8630633599736344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.675010625059723e-07,
+      "loss": 0.7787,
+      "step": 31425
+    },
+    {
+      "epoch": 0.8630908242014775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.67119393632754e-07,
+      "loss": 0.7555,
+      "step": 31426
+    },
+    {
+      "epoch": 0.8631182884293208,
+      "grad_norm": 0.0,
+      "learning_rate": 9.667377962311863e-07,
+      "loss": 0.7999,
+      "step": 31427
+    },
+    {
+      "epoch": 0.8631457526571641,
+      "grad_norm": 0.0,
+      "learning_rate": 9.663562703042862e-07,
+      "loss": 0.8452,
+      "step": 31428
+    },
+    {
+      "epoch": 0.8631732168850073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.659748158550753e-07,
+      "loss": 0.8633,
+      "step": 31429
+    },
+    {
+      "epoch": 0.8632006811128505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.655934328865701e-07,
+      "loss": 0.7661,
+      "step": 31430
+    },
+    {
+      "epoch": 0.8632281453406937,
+      "grad_norm": 0.0,
+      "learning_rate": 9.652121214017896e-07,
+      "loss": 0.7898,
+      "step": 31431
+    },
+    {
+      "epoch": 0.863255609568537,
+      "grad_norm": 0.0,
+      "learning_rate": 9.648308814037511e-07,
+      "loss": 0.8857,
+      "step": 31432
+    },
+    {
+      "epoch": 0.8632830737963803,
+      "grad_norm": 0.0,
+      "learning_rate": 9.644497128954722e-07,
+      "loss": 0.7698,
+      "step": 31433
+    },
+    {
+      "epoch": 0.8633105380242234,
+      "grad_norm": 0.0,
+      "learning_rate": 9.64068615879965e-07,
+      "loss": 0.8262,
+      "step": 31434
+    },
+    {
+      "epoch": 0.8633380022520667,
+      "grad_norm": 0.0,
+      "learning_rate": 9.636875903602493e-07,
+      "loss": 0.7233,
+      "step": 31435
+    },
+    {
+      "epoch": 0.86336546647991,
+      "grad_norm": 0.0,
+      "learning_rate": 9.63306636339335e-07,
+      "loss": 0.7398,
+      "step": 31436
+    },
+    {
+      "epoch": 0.8633929307077531,
+      "grad_norm": 0.0,
+      "learning_rate": 9.629257538202397e-07,
+      "loss": 0.8076,
+      "step": 31437
+    },
+    {
+      "epoch": 0.8634203949355964,
+      "grad_norm": 0.0,
+      "learning_rate": 9.625449428059763e-07,
+      "loss": 0.8832,
+      "step": 31438
+    },
+    {
+      "epoch": 0.8634478591634396,
+      "grad_norm": 0.0,
+      "learning_rate": 9.621642032995603e-07,
+      "loss": 0.7734,
+      "step": 31439
+    },
+    {
+      "epoch": 0.8634753233912829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.617835353040007e-07,
+      "loss": 0.8537,
+      "step": 31440
+    },
+    {
+      "epoch": 0.8635027876191261,
+      "grad_norm": 0.0,
+      "learning_rate": 9.6140293882231e-07,
+      "loss": 0.8027,
+      "step": 31441
+    },
+    {
+      "epoch": 0.8635302518469693,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61022413857502e-07,
+      "loss": 0.8586,
+      "step": 31442
+    },
+    {
+      "epoch": 0.8635577160748126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.60641960412585e-07,
+      "loss": 0.9403,
+      "step": 31443
+    },
+    {
+      "epoch": 0.8635851803026557,
+      "grad_norm": 0.0,
+      "learning_rate": 9.6026157849057e-07,
+      "loss": 0.8938,
+      "step": 31444
+    },
+    {
+      "epoch": 0.863612644530499,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59881268094469e-07,
+      "loss": 0.7963,
+      "step": 31445
+    },
+    {
+      "epoch": 0.8636401087583423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.595010292272866e-07,
+      "loss": 0.8248,
+      "step": 31446
+    },
+    {
+      "epoch": 0.8636675729861855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.591208618920346e-07,
+      "loss": 0.7666,
+      "step": 31447
+    },
+    {
+      "epoch": 0.8636950372140287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.587407660917202e-07,
+      "loss": 0.8882,
+      "step": 31448
+    },
+    {
+      "epoch": 0.863722501441872,
+      "grad_norm": 0.0,
+      "learning_rate": 9.583607418293518e-07,
+      "loss": 0.8657,
+      "step": 31449
+    },
+    {
+      "epoch": 0.8637499656697152,
+      "grad_norm": 0.0,
+      "learning_rate": 9.579807891079373e-07,
+      "loss": 0.8722,
+      "step": 31450
+    },
+    {
+      "epoch": 0.8637774298975585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.576009079304805e-07,
+      "loss": 0.8616,
+      "step": 31451
+    },
+    {
+      "epoch": 0.8638048941254016,
+      "grad_norm": 0.0,
+      "learning_rate": 9.572210982999875e-07,
+      "loss": 0.8882,
+      "step": 31452
+    },
+    {
+      "epoch": 0.8638323583532449,
+      "grad_norm": 0.0,
+      "learning_rate": 9.568413602194626e-07,
+      "loss": 0.8698,
+      "step": 31453
+    },
+    {
+      "epoch": 0.8638598225810882,
+      "grad_norm": 0.0,
+      "learning_rate": 9.564616936919123e-07,
+      "loss": 0.8266,
+      "step": 31454
+    },
+    {
+      "epoch": 0.8638872868089313,
+      "grad_norm": 0.0,
+      "learning_rate": 9.560820987203413e-07,
+      "loss": 0.7208,
+      "step": 31455
+    },
+    {
+      "epoch": 0.8639147510367746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.55702575307751e-07,
+      "loss": 0.8542,
+      "step": 31456
+    },
+    {
+      "epoch": 0.8639422152646178,
+      "grad_norm": 0.0,
+      "learning_rate": 9.55323123457147e-07,
+      "loss": 0.8142,
+      "step": 31457
+    },
+    {
+      "epoch": 0.8639696794924611,
+      "grad_norm": 0.0,
+      "learning_rate": 9.549437431715281e-07,
+      "loss": 0.7969,
+      "step": 31458
+    },
+    {
+      "epoch": 0.8639971437203043,
+      "grad_norm": 0.0,
+      "learning_rate": 9.545644344539007e-07,
+      "loss": 0.8054,
+      "step": 31459
+    },
+    {
+      "epoch": 0.8640246079481475,
+      "grad_norm": 0.0,
+      "learning_rate": 9.541851973072613e-07,
+      "loss": 0.6985,
+      "step": 31460
+    },
+    {
+      "epoch": 0.8640520721759908,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53806031734612e-07,
+      "loss": 0.7002,
+      "step": 31461
+    },
+    {
+      "epoch": 0.8640795364038341,
+      "grad_norm": 0.0,
+      "learning_rate": 9.534269377389538e-07,
+      "loss": 0.7775,
+      "step": 31462
+    },
+    {
+      "epoch": 0.8641070006316772,
+      "grad_norm": 0.0,
+      "learning_rate": 9.530479153232885e-07,
+      "loss": 0.7724,
+      "step": 31463
+    },
+    {
+      "epoch": 0.8641344648595205,
+      "grad_norm": 0.0,
+      "learning_rate": 9.526689644906095e-07,
+      "loss": 0.8536,
+      "step": 31464
+    },
+    {
+      "epoch": 0.8641619290873637,
+      "grad_norm": 0.0,
+      "learning_rate": 9.522900852439187e-07,
+      "loss": 0.7788,
+      "step": 31465
+    },
+    {
+      "epoch": 0.8641893933152069,
+      "grad_norm": 0.0,
+      "learning_rate": 9.51911277586216e-07,
+      "loss": 0.9103,
+      "step": 31466
+    },
+    {
+      "epoch": 0.8642168575430502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.515325415204946e-07,
+      "loss": 0.8131,
+      "step": 31467
+    },
+    {
+      "epoch": 0.8642443217708934,
+      "grad_norm": 0.0,
+      "learning_rate": 9.511538770497519e-07,
+      "loss": 0.8479,
+      "step": 31468
+    },
+    {
+      "epoch": 0.8642717859987367,
+      "grad_norm": 0.0,
+      "learning_rate": 9.507752841769869e-07,
+      "loss": 0.8269,
+      "step": 31469
+    },
+    {
+      "epoch": 0.8642992502265798,
+      "grad_norm": 0.0,
+      "learning_rate": 9.503967629051913e-07,
+      "loss": 0.8848,
+      "step": 31470
+    },
+    {
+      "epoch": 0.8643267144544231,
+      "grad_norm": 0.0,
+      "learning_rate": 9.50018313237362e-07,
+      "loss": 0.7744,
+      "step": 31471
+    },
+    {
+      "epoch": 0.8643541786822664,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496399351764929e-07,
+      "loss": 0.8307,
+      "step": 31472
+    },
+    {
+      "epoch": 0.8643816429101095,
+      "grad_norm": 0.0,
+      "learning_rate": 9.492616287255795e-07,
+      "loss": 0.7734,
+      "step": 31473
+    },
+    {
+      "epoch": 0.8644091071379528,
+      "grad_norm": 0.0,
+      "learning_rate": 9.488833938876152e-07,
+      "loss": 0.7602,
+      "step": 31474
+    },
+    {
+      "epoch": 0.8644365713657961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.485052306655917e-07,
+      "loss": 0.7346,
+      "step": 31475
+    },
+    {
+      "epoch": 0.8644640355936393,
+      "grad_norm": 0.0,
+      "learning_rate": 9.481271390624991e-07,
+      "loss": 0.8166,
+      "step": 31476
+    },
+    {
+      "epoch": 0.8644914998214825,
+      "grad_norm": 0.0,
+      "learning_rate": 9.477491190813315e-07,
+      "loss": 0.7968,
+      "step": 31477
+    },
+    {
+      "epoch": 0.8645189640493257,
+      "grad_norm": 0.0,
+      "learning_rate": 9.473711707250788e-07,
+      "loss": 0.8663,
+      "step": 31478
+    },
+    {
+      "epoch": 0.864546428277169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.469932939967319e-07,
+      "loss": 0.8288,
+      "step": 31479
+    },
+    {
+      "epoch": 0.8645738925050123,
+      "grad_norm": 0.0,
+      "learning_rate": 9.46615488899283e-07,
+      "loss": 0.8385,
+      "step": 31480
+    },
+    {
+      "epoch": 0.8646013567328554,
+      "grad_norm": 0.0,
+      "learning_rate": 9.462377554357171e-07,
+      "loss": 0.8558,
+      "step": 31481
+    },
+    {
+      "epoch": 0.8646288209606987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.458600936090279e-07,
+      "loss": 0.8814,
+      "step": 31482
+    },
+    {
+      "epoch": 0.8646562851885419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.454825034221981e-07,
+      "loss": 0.728,
+      "step": 31483
+    },
+    {
+      "epoch": 0.8646837494163852,
+      "grad_norm": 0.0,
+      "learning_rate": 9.451049848782179e-07,
+      "loss": 0.8694,
+      "step": 31484
+    },
+    {
+      "epoch": 0.8647112136442284,
+      "grad_norm": 0.0,
+      "learning_rate": 9.447275379800758e-07,
+      "loss": 0.7649,
+      "step": 31485
+    },
+    {
+      "epoch": 0.8647386778720716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.443501627307583e-07,
+      "loss": 0.7764,
+      "step": 31486
+    },
+    {
+      "epoch": 0.8647661420999149,
+      "grad_norm": 0.0,
+      "learning_rate": 9.439728591332475e-07,
+      "loss": 0.761,
+      "step": 31487
+    },
+    {
+      "epoch": 0.8647936063277581,
+      "grad_norm": 0.0,
+      "learning_rate": 9.435956271905322e-07,
+      "loss": 0.8118,
+      "step": 31488
+    },
+    {
+      "epoch": 0.8648210705556013,
+      "grad_norm": 0.0,
+      "learning_rate": 9.432184669055966e-07,
+      "loss": 0.9008,
+      "step": 31489
+    },
+    {
+      "epoch": 0.8648485347834446,
+      "grad_norm": 0.0,
+      "learning_rate": 9.428413782814272e-07,
+      "loss": 0.8482,
+      "step": 31490
+    },
+    {
+      "epoch": 0.8648759990112878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.424643613210016e-07,
+      "loss": 0.8383,
+      "step": 31491
+    },
+    {
+      "epoch": 0.864903463239131,
+      "grad_norm": 0.0,
+      "learning_rate": 9.420874160273096e-07,
+      "loss": 0.8586,
+      "step": 31492
+    },
+    {
+      "epoch": 0.8649309274669743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.417105424033291e-07,
+      "loss": 0.7921,
+      "step": 31493
+    },
+    {
+      "epoch": 0.8649583916948175,
+      "grad_norm": 0.0,
+      "learning_rate": 9.413337404520428e-07,
+      "loss": 0.7963,
+      "step": 31494
+    },
+    {
+      "epoch": 0.8649858559226608,
+      "grad_norm": 0.0,
+      "learning_rate": 9.409570101764331e-07,
+      "loss": 0.7643,
+      "step": 31495
+    },
+    {
+      "epoch": 0.8650133201505039,
+      "grad_norm": 0.0,
+      "learning_rate": 9.405803515794809e-07,
+      "loss": 0.849,
+      "step": 31496
+    },
+    {
+      "epoch": 0.8650407843783472,
+      "grad_norm": 0.0,
+      "learning_rate": 9.40203764664166e-07,
+      "loss": 0.8297,
+      "step": 31497
+    },
+    {
+      "epoch": 0.8650682486061905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.398272494334703e-07,
+      "loss": 0.7856,
+      "step": 31498
+    },
+    {
+      "epoch": 0.8650957128340336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.394508058903717e-07,
+      "loss": 0.7993,
+      "step": 31499
+    },
+    {
+      "epoch": 0.8651231770618769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.390744340378455e-07,
+      "loss": 0.8694,
+      "step": 31500
+    },
+    {
+      "epoch": 0.8651506412897202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.386981338788714e-07,
+      "loss": 0.8628,
+      "step": 31501
+    },
+    {
+      "epoch": 0.8651781055175634,
+      "grad_norm": 0.0,
+      "learning_rate": 9.383219054164294e-07,
+      "loss": 0.8718,
+      "step": 31502
+    },
+    {
+      "epoch": 0.8652055697454066,
+      "grad_norm": 0.0,
+      "learning_rate": 9.379457486534937e-07,
+      "loss": 0.7903,
+      "step": 31503
+    },
+    {
+      "epoch": 0.8652330339732498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.375696635930431e-07,
+      "loss": 0.8079,
+      "step": 31504
+    },
+    {
+      "epoch": 0.8652604982010931,
+      "grad_norm": 0.0,
+      "learning_rate": 9.371936502380507e-07,
+      "loss": 0.787,
+      "step": 31505
+    },
+    {
+      "epoch": 0.8652879624289364,
+      "grad_norm": 0.0,
+      "learning_rate": 9.368177085914943e-07,
+      "loss": 0.7588,
+      "step": 31506
+    },
+    {
+      "epoch": 0.8653154266567795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.364418386563445e-07,
+      "loss": 0.8467,
+      "step": 31507
+    },
+    {
+      "epoch": 0.8653428908846228,
+      "grad_norm": 0.0,
+      "learning_rate": 9.360660404355781e-07,
+      "loss": 0.85,
+      "step": 31508
+    },
+    {
+      "epoch": 0.865370355112466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.356903139321682e-07,
+      "loss": 0.8215,
+      "step": 31509
+    },
+    {
+      "epoch": 0.8653978193403092,
+      "grad_norm": 0.0,
+      "learning_rate": 9.353146591490903e-07,
+      "loss": 0.7635,
+      "step": 31510
+    },
+    {
+      "epoch": 0.8654252835681525,
+      "grad_norm": 0.0,
+      "learning_rate": 9.349390760893118e-07,
+      "loss": 0.7979,
+      "step": 31511
+    },
+    {
+      "epoch": 0.8654527477959957,
+      "grad_norm": 0.0,
+      "learning_rate": 9.34563564755806e-07,
+      "loss": 0.8591,
+      "step": 31512
+    },
+    {
+      "epoch": 0.865480212023839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.341881251515462e-07,
+      "loss": 0.8064,
+      "step": 31513
+    },
+    {
+      "epoch": 0.8655076762516822,
+      "grad_norm": 0.0,
+      "learning_rate": 9.33812757279503e-07,
+      "loss": 0.8066,
+      "step": 31514
+    },
+    {
+      "epoch": 0.8655351404795254,
+      "grad_norm": 0.0,
+      "learning_rate": 9.334374611426433e-07,
+      "loss": 0.7559,
+      "step": 31515
+    },
+    {
+      "epoch": 0.8655626047073687,
+      "grad_norm": 0.0,
+      "learning_rate": 9.330622367439402e-07,
+      "loss": 0.6959,
+      "step": 31516
+    },
+    {
+      "epoch": 0.8655900689352118,
+      "grad_norm": 0.0,
+      "learning_rate": 9.326870840863589e-07,
+      "loss": 0.9093,
+      "step": 31517
+    },
+    {
+      "epoch": 0.8656175331630551,
+      "grad_norm": 0.0,
+      "learning_rate": 9.323120031728694e-07,
+      "loss": 0.7462,
+      "step": 31518
+    },
+    {
+      "epoch": 0.8656449973908984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.319369940064405e-07,
+      "loss": 0.7799,
+      "step": 31519
+    },
+    {
+      "epoch": 0.8656724616187416,
+      "grad_norm": 0.0,
+      "learning_rate": 9.315620565900385e-07,
+      "loss": 0.7743,
+      "step": 31520
+    },
+    {
+      "epoch": 0.8656999258465848,
+      "grad_norm": 0.0,
+      "learning_rate": 9.311871909266323e-07,
+      "loss": 0.7383,
+      "step": 31521
+    },
+    {
+      "epoch": 0.865727390074428,
+      "grad_norm": 0.0,
+      "learning_rate": 9.30812397019184e-07,
+      "loss": 0.8274,
+      "step": 31522
+    },
+    {
+      "epoch": 0.8657548543022713,
+      "grad_norm": 0.0,
+      "learning_rate": 9.304376748706634e-07,
+      "loss": 0.8195,
+      "step": 31523
+    },
+    {
+      "epoch": 0.8657823185301146,
+      "grad_norm": 0.0,
+      "learning_rate": 9.300630244840314e-07,
+      "loss": 0.8497,
+      "step": 31524
+    },
+    {
+      "epoch": 0.8658097827579577,
+      "grad_norm": 0.0,
+      "learning_rate": 9.296884458622535e-07,
+      "loss": 0.7302,
+      "step": 31525
+    },
+    {
+      "epoch": 0.865837246985801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.293139390082939e-07,
+      "loss": 0.868,
+      "step": 31526
+    },
+    {
+      "epoch": 0.8658647112136443,
+      "grad_norm": 0.0,
+      "learning_rate": 9.289395039251159e-07,
+      "loss": 0.8016,
+      "step": 31527
+    },
+    {
+      "epoch": 0.8658921754414874,
+      "grad_norm": 0.0,
+      "learning_rate": 9.285651406156848e-07,
+      "loss": 0.8321,
+      "step": 31528
+    },
+    {
+      "epoch": 0.8659196396693307,
+      "grad_norm": 0.0,
+      "learning_rate": 9.281908490829583e-07,
+      "loss": 0.7689,
+      "step": 31529
+    },
+    {
+      "epoch": 0.8659471038971739,
+      "grad_norm": 0.0,
+      "learning_rate": 9.278166293299018e-07,
+      "loss": 0.7763,
+      "step": 31530
+    },
+    {
+      "epoch": 0.8659745681250172,
+      "grad_norm": 0.0,
+      "learning_rate": 9.274424813594718e-07,
+      "loss": 0.947,
+      "step": 31531
+    },
+    {
+      "epoch": 0.8660020323528604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.270684051746326e-07,
+      "loss": 0.8563,
+      "step": 31532
+    },
+    {
+      "epoch": 0.8660294965807036,
+      "grad_norm": 0.0,
+      "learning_rate": 9.26694400778343e-07,
+      "loss": 0.7853,
+      "step": 31533
+    },
+    {
+      "epoch": 0.8660569608085469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.263204681735616e-07,
+      "loss": 0.691,
+      "step": 31534
+    },
+    {
+      "epoch": 0.86608442503639,
+      "grad_norm": 0.0,
+      "learning_rate": 9.259466073632461e-07,
+      "loss": 0.8252,
+      "step": 31535
+    },
+    {
+      "epoch": 0.8661118892642333,
+      "grad_norm": 0.0,
+      "learning_rate": 9.255728183503565e-07,
+      "loss": 0.7614,
+      "step": 31536
+    },
+    {
+      "epoch": 0.8661393534920766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.251991011378502e-07,
+      "loss": 0.8661,
+      "step": 31537
+    },
+    {
+      "epoch": 0.8661668177199198,
+      "grad_norm": 0.0,
+      "learning_rate": 9.248254557286862e-07,
+      "loss": 0.7967,
+      "step": 31538
+    },
+    {
+      "epoch": 0.866194281947763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.244518821258175e-07,
+      "loss": 0.8119,
+      "step": 31539
+    },
+    {
+      "epoch": 0.8662217461756063,
+      "grad_norm": 0.0,
+      "learning_rate": 9.240783803321995e-07,
+      "loss": 0.7137,
+      "step": 31540
+    },
+    {
+      "epoch": 0.8662492104034495,
+      "grad_norm": 0.0,
+      "learning_rate": 9.237049503507889e-07,
+      "loss": 0.7987,
+      "step": 31541
+    },
+    {
+      "epoch": 0.8662766746312928,
+      "grad_norm": 0.0,
+      "learning_rate": 9.233315921845409e-07,
+      "loss": 0.8064,
+      "step": 31542
+    },
+    {
+      "epoch": 0.8663041388591359,
+      "grad_norm": 0.0,
+      "learning_rate": 9.229583058364089e-07,
+      "loss": 0.8912,
+      "step": 31543
+    },
+    {
+      "epoch": 0.8663316030869792,
+      "grad_norm": 0.0,
+      "learning_rate": 9.225850913093482e-07,
+      "loss": 0.8932,
+      "step": 31544
+    },
+    {
+      "epoch": 0.8663590673148225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.222119486063108e-07,
+      "loss": 0.7549,
+      "step": 31545
+    },
+    {
+      "epoch": 0.8663865315426656,
+      "grad_norm": 0.0,
+      "learning_rate": 9.218388777302479e-07,
+      "loss": 0.7921,
+      "step": 31546
+    },
+    {
+      "epoch": 0.8664139957705089,
+      "grad_norm": 0.0,
+      "learning_rate": 9.214658786841136e-07,
+      "loss": 0.8361,
+      "step": 31547
+    },
+    {
+      "epoch": 0.8664414599983521,
+      "grad_norm": 0.0,
+      "learning_rate": 9.210929514708566e-07,
+      "loss": 0.756,
+      "step": 31548
+    },
+    {
+      "epoch": 0.8664689242261954,
+      "grad_norm": 0.0,
+      "learning_rate": 9.207200960934281e-07,
+      "loss": 0.8186,
+      "step": 31549
+    },
+    {
+      "epoch": 0.8664963884540386,
+      "grad_norm": 0.0,
+      "learning_rate": 9.203473125547801e-07,
+      "loss": 0.7896,
+      "step": 31550
+    },
+    {
+      "epoch": 0.8665238526818818,
+      "grad_norm": 0.0,
+      "learning_rate": 9.199746008578625e-07,
+      "loss": 0.7667,
+      "step": 31551
+    },
+    {
+      "epoch": 0.8665513169097251,
+      "grad_norm": 0.0,
+      "learning_rate": 9.196019610056217e-07,
+      "loss": 0.7799,
+      "step": 31552
+    },
+    {
+      "epoch": 0.8665787811375684,
+      "grad_norm": 0.0,
+      "learning_rate": 9.192293930010077e-07,
+      "loss": 0.8592,
+      "step": 31553
+    },
+    {
+      "epoch": 0.8666062453654115,
+      "grad_norm": 0.0,
+      "learning_rate": 9.188568968469702e-07,
+      "loss": 0.8158,
+      "step": 31554
+    },
+    {
+      "epoch": 0.8666337095932548,
+      "grad_norm": 0.0,
+      "learning_rate": 9.184844725464526e-07,
+      "loss": 0.8537,
+      "step": 31555
+    },
+    {
+      "epoch": 0.866661173821098,
+      "grad_norm": 0.0,
+      "learning_rate": 9.181121201024035e-07,
+      "loss": 0.7643,
+      "step": 31556
+    },
+    {
+      "epoch": 0.8666886380489413,
+      "grad_norm": 0.0,
+      "learning_rate": 9.177398395177717e-07,
+      "loss": 0.887,
+      "step": 31557
+    },
+    {
+      "epoch": 0.8667161022767845,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17367630795497e-07,
+      "loss": 0.7548,
+      "step": 31558
+    },
+    {
+      "epoch": 0.8667435665046277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.169954939385295e-07,
+      "loss": 0.8509,
+      "step": 31559
+    },
+    {
+      "epoch": 0.866771030732471,
+      "grad_norm": 0.0,
+      "learning_rate": 9.166234289498121e-07,
+      "loss": 0.8366,
+      "step": 31560
+    },
+    {
+      "epoch": 0.8667984949603141,
+      "grad_norm": 0.0,
+      "learning_rate": 9.162514358322883e-07,
+      "loss": 0.7791,
+      "step": 31561
+    },
+    {
+      "epoch": 0.8668259591881574,
+      "grad_norm": 0.0,
+      "learning_rate": 9.158795145889033e-07,
+      "loss": 0.8109,
+      "step": 31562
+    },
+    {
+      "epoch": 0.8668534234160007,
+      "grad_norm": 0.0,
+      "learning_rate": 9.155076652225981e-07,
+      "loss": 0.8709,
+      "step": 31563
+    },
+    {
+      "epoch": 0.8668808876438439,
+      "grad_norm": 0.0,
+      "learning_rate": 9.151358877363148e-07,
+      "loss": 0.8676,
+      "step": 31564
+    },
+    {
+      "epoch": 0.8669083518716871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.147641821329944e-07,
+      "loss": 0.7848,
+      "step": 31565
+    },
+    {
+      "epoch": 0.8669358160995304,
+      "grad_norm": 0.0,
+      "learning_rate": 9.143925484155791e-07,
+      "loss": 0.8031,
+      "step": 31566
+    },
+    {
+      "epoch": 0.8669632803273736,
+      "grad_norm": 0.0,
+      "learning_rate": 9.140209865870098e-07,
+      "loss": 0.7691,
+      "step": 31567
+    },
+    {
+      "epoch": 0.8669907445552169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.136494966502263e-07,
+      "loss": 0.838,
+      "step": 31568
+    },
+    {
+      "epoch": 0.86701820878306,
+      "grad_norm": 0.0,
+      "learning_rate": 9.132780786081685e-07,
+      "loss": 0.7158,
+      "step": 31569
+    },
+    {
+      "epoch": 0.8670456730109033,
+      "grad_norm": 0.0,
+      "learning_rate": 9.129067324637742e-07,
+      "loss": 0.8073,
+      "step": 31570
+    },
+    {
+      "epoch": 0.8670731372387466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.125354582199819e-07,
+      "loss": 0.7168,
+      "step": 31571
+    },
+    {
+      "epoch": 0.8671006014665897,
+      "grad_norm": 0.0,
+      "learning_rate": 9.121642558797284e-07,
+      "loss": 0.8002,
+      "step": 31572
+    },
+    {
+      "epoch": 0.867128065694433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.117931254459511e-07,
+      "loss": 0.8959,
+      "step": 31573
+    },
+    {
+      "epoch": 0.8671555299222762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.114220669215879e-07,
+      "loss": 0.7742,
+      "step": 31574
+    },
+    {
+      "epoch": 0.8671829941501195,
+      "grad_norm": 0.0,
+      "learning_rate": 9.110510803095751e-07,
+      "loss": 0.8762,
+      "step": 31575
+    },
+    {
+      "epoch": 0.8672104583779627,
+      "grad_norm": 0.0,
+      "learning_rate": 9.10680165612845e-07,
+      "loss": 0.6947,
+      "step": 31576
+    },
+    {
+      "epoch": 0.8672379226058059,
+      "grad_norm": 0.0,
+      "learning_rate": 9.10309322834334e-07,
+      "loss": 0.7372,
+      "step": 31577
+    },
+    {
+      "epoch": 0.8672653868336492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.099385519769788e-07,
+      "loss": 0.69,
+      "step": 31578
+    },
+    {
+      "epoch": 0.8672928510614925,
+      "grad_norm": 0.0,
+      "learning_rate": 9.095678530437091e-07,
+      "loss": 0.856,
+      "step": 31579
+    },
+    {
+      "epoch": 0.8673203152893356,
+      "grad_norm": 0.0,
+      "learning_rate": 9.091972260374604e-07,
+      "loss": 0.7535,
+      "step": 31580
+    },
+    {
+      "epoch": 0.8673477795171789,
+      "grad_norm": 0.0,
+      "learning_rate": 9.088266709611659e-07,
+      "loss": 0.741,
+      "step": 31581
+    },
+    {
+      "epoch": 0.8673752437450221,
+      "grad_norm": 0.0,
+      "learning_rate": 9.084561878177545e-07,
+      "loss": 0.7997,
+      "step": 31582
+    },
+    {
+      "epoch": 0.8674027079728653,
+      "grad_norm": 0.0,
+      "learning_rate": 9.080857766101591e-07,
+      "loss": 0.8428,
+      "step": 31583
+    },
+    {
+      "epoch": 0.8674301722007086,
+      "grad_norm": 0.0,
+      "learning_rate": 9.077154373413111e-07,
+      "loss": 0.8573,
+      "step": 31584
+    },
+    {
+      "epoch": 0.8674576364285518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.073451700141411e-07,
+      "loss": 0.7501,
+      "step": 31585
+    },
+    {
+      "epoch": 0.8674851006563951,
+      "grad_norm": 0.0,
+      "learning_rate": 9.069749746315804e-07,
+      "loss": 0.743,
+      "step": 31586
+    },
+    {
+      "epoch": 0.8675125648842382,
+      "grad_norm": 0.0,
+      "learning_rate": 9.066048511965553e-07,
+      "loss": 0.8555,
+      "step": 31587
+    },
+    {
+      "epoch": 0.8675400291120815,
+      "grad_norm": 0.0,
+      "learning_rate": 9.062347997119936e-07,
+      "loss": 0.7721,
+      "step": 31588
+    },
+    {
+      "epoch": 0.8675674933399248,
+      "grad_norm": 0.0,
+      "learning_rate": 9.05864820180824e-07,
+      "loss": 0.7599,
+      "step": 31589
+    },
+    {
+      "epoch": 0.8675949575677679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.054949126059753e-07,
+      "loss": 0.8781,
+      "step": 31590
+    },
+    {
+      "epoch": 0.8676224217956112,
+      "grad_norm": 0.0,
+      "learning_rate": 9.051250769903741e-07,
+      "loss": 0.7759,
+      "step": 31591
+    },
+    {
+      "epoch": 0.8676498860234545,
+      "grad_norm": 0.0,
+      "learning_rate": 9.047553133369468e-07,
+      "loss": 0.7217,
+      "step": 31592
+    },
+    {
+      "epoch": 0.8676773502512977,
+      "grad_norm": 0.0,
+      "learning_rate": 9.04385621648618e-07,
+      "loss": 0.7961,
+      "step": 31593
+    },
+    {
+      "epoch": 0.8677048144791409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.040160019283128e-07,
+      "loss": 0.7937,
+      "step": 31594
+    },
+    {
+      "epoch": 0.8677322787069841,
+      "grad_norm": 0.0,
+      "learning_rate": 9.036464541789591e-07,
+      "loss": 0.8434,
+      "step": 31595
+    },
+    {
+      "epoch": 0.8677597429348274,
+      "grad_norm": 0.0,
+      "learning_rate": 9.032769784034756e-07,
+      "loss": 0.7843,
+      "step": 31596
+    },
+    {
+      "epoch": 0.8677872071626707,
+      "grad_norm": 0.0,
+      "learning_rate": 9.029075746047888e-07,
+      "loss": 0.8594,
+      "step": 31597
+    },
+    {
+      "epoch": 0.8678146713905138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.025382427858231e-07,
+      "loss": 0.7639,
+      "step": 31598
+    },
+    {
+      "epoch": 0.8678421356183571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.021689829494973e-07,
+      "loss": 0.801,
+      "step": 31599
+    },
+    {
+      "epoch": 0.8678695998462004,
+      "grad_norm": 0.0,
+      "learning_rate": 9.017997950987334e-07,
+      "loss": 0.7716,
+      "step": 31600
+    },
+    {
+      "epoch": 0.8678970640740435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.014306792364546e-07,
+      "loss": 0.8445,
+      "step": 31601
+    },
+    {
+      "epoch": 0.8679245283018868,
+      "grad_norm": 0.0,
+      "learning_rate": 9.010616353655832e-07,
+      "loss": 0.8562,
+      "step": 31602
+    },
+    {
+      "epoch": 0.86795199252973,
+      "grad_norm": 0.0,
+      "learning_rate": 9.006926634890334e-07,
+      "loss": 0.8289,
+      "step": 31603
+    },
+    {
+      "epoch": 0.8679794567575733,
+      "grad_norm": 0.0,
+      "learning_rate": 9.003237636097306e-07,
+      "loss": 0.8183,
+      "step": 31604
+    },
+    {
+      "epoch": 0.8680069209854165,
+      "grad_norm": 0.0,
+      "learning_rate": 8.999549357305893e-07,
+      "loss": 0.8257,
+      "step": 31605
+    },
+    {
+      "epoch": 0.8680343852132597,
+      "grad_norm": 0.0,
+      "learning_rate": 8.995861798545302e-07,
+      "loss": 0.7893,
+      "step": 31606
+    },
+    {
+      "epoch": 0.868061849441103,
+      "grad_norm": 0.0,
+      "learning_rate": 8.992174959844691e-07,
+      "loss": 0.7078,
+      "step": 31607
+    },
+    {
+      "epoch": 0.8680893136689461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.988488841233256e-07,
+      "loss": 0.7901,
+      "step": 31608
+    },
+    {
+      "epoch": 0.8681167778967894,
+      "grad_norm": 0.0,
+      "learning_rate": 8.984803442740142e-07,
+      "loss": 0.7755,
+      "step": 31609
+    },
+    {
+      "epoch": 0.8681442421246327,
+      "grad_norm": 0.0,
+      "learning_rate": 8.981118764394548e-07,
+      "loss": 0.8311,
+      "step": 31610
+    },
+    {
+      "epoch": 0.8681717063524759,
+      "grad_norm": 0.0,
+      "learning_rate": 8.977434806225593e-07,
+      "loss": 0.8009,
+      "step": 31611
+    },
+    {
+      "epoch": 0.8681991705803191,
+      "grad_norm": 0.0,
+      "learning_rate": 8.973751568262412e-07,
+      "loss": 0.7494,
+      "step": 31612
+    },
+    {
+      "epoch": 0.8682266348081624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.97006905053418e-07,
+      "loss": 0.7815,
+      "step": 31613
+    },
+    {
+      "epoch": 0.8682540990360056,
+      "grad_norm": 0.0,
+      "learning_rate": 8.966387253070008e-07,
+      "loss": 0.7918,
+      "step": 31614
+    },
+    {
+      "epoch": 0.8682815632638489,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96270617589905e-07,
+      "loss": 0.774,
+      "step": 31615
+    },
+    {
+      "epoch": 0.868309027491692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.959025819050449e-07,
+      "loss": 0.8191,
+      "step": 31616
+    },
+    {
+      "epoch": 0.8683364917195353,
+      "grad_norm": 0.0,
+      "learning_rate": 8.955346182553282e-07,
+      "loss": 0.7536,
+      "step": 31617
+    },
+    {
+      "epoch": 0.8683639559473786,
+      "grad_norm": 0.0,
+      "learning_rate": 8.951667266436693e-07,
+      "loss": 0.7735,
+      "step": 31618
+    },
+    {
+      "epoch": 0.8683914201752218,
+      "grad_norm": 0.0,
+      "learning_rate": 8.947989070729768e-07,
+      "loss": 0.7874,
+      "step": 31619
+    },
+    {
+      "epoch": 0.868418884403065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.944311595461619e-07,
+      "loss": 0.7886,
+      "step": 31620
+    },
+    {
+      "epoch": 0.8684463486309082,
+      "grad_norm": 0.0,
+      "learning_rate": 8.940634840661355e-07,
+      "loss": 0.8416,
+      "step": 31621
+    },
+    {
+      "epoch": 0.8684738128587515,
+      "grad_norm": 0.0,
+      "learning_rate": 8.936958806358076e-07,
+      "loss": 0.8064,
+      "step": 31622
+    },
+    {
+      "epoch": 0.8685012770865947,
+      "grad_norm": 0.0,
+      "learning_rate": 8.933283492580846e-07,
+      "loss": 0.7865,
+      "step": 31623
+    },
+    {
+      "epoch": 0.8685287413144379,
+      "grad_norm": 0.0,
+      "learning_rate": 8.929608899358744e-07,
+      "loss": 0.8261,
+      "step": 31624
+    },
+    {
+      "epoch": 0.8685562055422812,
+      "grad_norm": 0.0,
+      "learning_rate": 8.925935026720844e-07,
+      "loss": 0.865,
+      "step": 31625
+    },
+    {
+      "epoch": 0.8685836697701245,
+      "grad_norm": 0.0,
+      "learning_rate": 8.922261874696259e-07,
+      "loss": 0.7714,
+      "step": 31626
+    },
+    {
+      "epoch": 0.8686111339979676,
+      "grad_norm": 0.0,
+      "learning_rate": 8.918589443313985e-07,
+      "loss": 0.7244,
+      "step": 31627
+    },
+    {
+      "epoch": 0.8686385982258109,
+      "grad_norm": 0.0,
+      "learning_rate": 8.914917732603146e-07,
+      "loss": 0.8202,
+      "step": 31628
+    },
+    {
+      "epoch": 0.8686660624536541,
+      "grad_norm": 0.0,
+      "learning_rate": 8.911246742592739e-07,
+      "loss": 0.8427,
+      "step": 31629
+    },
+    {
+      "epoch": 0.8686935266814974,
+      "grad_norm": 0.0,
+      "learning_rate": 8.90757647331183e-07,
+      "loss": 0.9003,
+      "step": 31630
+    },
+    {
+      "epoch": 0.8687209909093406,
+      "grad_norm": 0.0,
+      "learning_rate": 8.903906924789451e-07,
+      "loss": 0.8175,
+      "step": 31631
+    },
+    {
+      "epoch": 0.8687484551371838,
+      "grad_norm": 0.0,
+      "learning_rate": 8.900238097054659e-07,
+      "loss": 0.7628,
+      "step": 31632
+    },
+    {
+      "epoch": 0.8687759193650271,
+      "grad_norm": 0.0,
+      "learning_rate": 8.896569990136472e-07,
+      "loss": 0.802,
+      "step": 31633
+    },
+    {
+      "epoch": 0.8688033835928702,
+      "grad_norm": 0.0,
+      "learning_rate": 8.892902604063935e-07,
+      "loss": 0.7999,
+      "step": 31634
+    },
+    {
+      "epoch": 0.8688308478207135,
+      "grad_norm": 0.0,
+      "learning_rate": 8.889235938866037e-07,
+      "loss": 0.7891,
+      "step": 31635
+    },
+    {
+      "epoch": 0.8688583120485568,
+      "grad_norm": 0.0,
+      "learning_rate": 8.885569994571775e-07,
+      "loss": 0.8387,
+      "step": 31636
+    },
+    {
+      "epoch": 0.8688857762764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.881904771210192e-07,
+      "loss": 0.6938,
+      "step": 31637
+    },
+    {
+      "epoch": 0.8689132405042432,
+      "grad_norm": 0.0,
+      "learning_rate": 8.878240268810257e-07,
+      "loss": 0.7171,
+      "step": 31638
+    },
+    {
+      "epoch": 0.8689407047320865,
+      "grad_norm": 0.0,
+      "learning_rate": 8.874576487400999e-07,
+      "loss": 0.8348,
+      "step": 31639
+    },
+    {
+      "epoch": 0.8689681689599297,
+      "grad_norm": 0.0,
+      "learning_rate": 8.870913427011397e-07,
+      "loss": 0.7882,
+      "step": 31640
+    },
+    {
+      "epoch": 0.868995633187773,
+      "grad_norm": 0.0,
+      "learning_rate": 8.867251087670414e-07,
+      "loss": 0.7753,
+      "step": 31641
+    },
+    {
+      "epoch": 0.8690230974156161,
+      "grad_norm": 0.0,
+      "learning_rate": 8.863589469407064e-07,
+      "loss": 0.8436,
+      "step": 31642
+    },
+    {
+      "epoch": 0.8690505616434594,
+      "grad_norm": 0.0,
+      "learning_rate": 8.859928572250265e-07,
+      "loss": 0.8368,
+      "step": 31643
+    },
+    {
+      "epoch": 0.8690780258713027,
+      "grad_norm": 0.0,
+      "learning_rate": 8.856268396229028e-07,
+      "loss": 0.7814,
+      "step": 31644
+    },
+    {
+      "epoch": 0.8691054900991458,
+      "grad_norm": 0.0,
+      "learning_rate": 8.852608941372299e-07,
+      "loss": 0.7753,
+      "step": 31645
+    },
+    {
+      "epoch": 0.8691329543269891,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84895020770905e-07,
+      "loss": 0.7509,
+      "step": 31646
+    },
+    {
+      "epoch": 0.8691604185548323,
+      "grad_norm": 0.0,
+      "learning_rate": 8.845292195268196e-07,
+      "loss": 0.8729,
+      "step": 31647
+    },
+    {
+      "epoch": 0.8691878827826756,
+      "grad_norm": 0.0,
+      "learning_rate": 8.841634904078711e-07,
+      "loss": 0.7791,
+      "step": 31648
+    },
+    {
+      "epoch": 0.8692153470105188,
+      "grad_norm": 0.0,
+      "learning_rate": 8.837978334169516e-07,
+      "loss": 0.8743,
+      "step": 31649
+    },
+    {
+      "epoch": 0.869242811238362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.834322485569558e-07,
+      "loss": 0.7693,
+      "step": 31650
+    },
+    {
+      "epoch": 0.8692702754662053,
+      "grad_norm": 0.0,
+      "learning_rate": 8.830667358307765e-07,
+      "loss": 0.8766,
+      "step": 31651
+    },
+    {
+      "epoch": 0.8692977396940486,
+      "grad_norm": 0.0,
+      "learning_rate": 8.827012952413028e-07,
+      "loss": 0.8415,
+      "step": 31652
+    },
+    {
+      "epoch": 0.8693252039218917,
+      "grad_norm": 0.0,
+      "learning_rate": 8.823359267914278e-07,
+      "loss": 0.8526,
+      "step": 31653
+    },
+    {
+      "epoch": 0.869352668149735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.819706304840436e-07,
+      "loss": 0.8403,
+      "step": 31654
+    },
+    {
+      "epoch": 0.8693801323775782,
+      "grad_norm": 0.0,
+      "learning_rate": 8.816054063220391e-07,
+      "loss": 0.8014,
+      "step": 31655
+    },
+    {
+      "epoch": 0.8694075966054214,
+      "grad_norm": 0.0,
+      "learning_rate": 8.812402543083043e-07,
+      "loss": 0.8396,
+      "step": 31656
+    },
+    {
+      "epoch": 0.8694350608332647,
+      "grad_norm": 0.0,
+      "learning_rate": 8.808751744457322e-07,
+      "loss": 0.8067,
+      "step": 31657
+    },
+    {
+      "epoch": 0.8694625250611079,
+      "grad_norm": 0.0,
+      "learning_rate": 8.805101667372052e-07,
+      "loss": 0.8683,
+      "step": 31658
+    },
+    {
+      "epoch": 0.8694899892889512,
+      "grad_norm": 0.0,
+      "learning_rate": 8.801452311856163e-07,
+      "loss": 0.7187,
+      "step": 31659
+    },
+    {
+      "epoch": 0.8695174535167943,
+      "grad_norm": 0.0,
+      "learning_rate": 8.797803677938488e-07,
+      "loss": 0.8392,
+      "step": 31660
+    },
+    {
+      "epoch": 0.8695449177446376,
+      "grad_norm": 0.0,
+      "learning_rate": 8.794155765647927e-07,
+      "loss": 0.7604,
+      "step": 31661
+    },
+    {
+      "epoch": 0.8695723819724809,
+      "grad_norm": 0.0,
+      "learning_rate": 8.790508575013323e-07,
+      "loss": 0.8794,
+      "step": 31662
+    },
+    {
+      "epoch": 0.869599846200324,
+      "grad_norm": 0.0,
+      "learning_rate": 8.786862106063576e-07,
+      "loss": 0.843,
+      "step": 31663
+    },
+    {
+      "epoch": 0.8696273104281673,
+      "grad_norm": 0.0,
+      "learning_rate": 8.783216358827485e-07,
+      "loss": 0.8208,
+      "step": 31664
+    },
+    {
+      "epoch": 0.8696547746560106,
+      "grad_norm": 0.0,
+      "learning_rate": 8.779571333333925e-07,
+      "loss": 0.9685,
+      "step": 31665
+    },
+    {
+      "epoch": 0.8696822388838538,
+      "grad_norm": 0.0,
+      "learning_rate": 8.775927029611753e-07,
+      "loss": 0.8892,
+      "step": 31666
+    },
+    {
+      "epoch": 0.869709703111697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.772283447689755e-07,
+      "loss": 0.9017,
+      "step": 31667
+    },
+    {
+      "epoch": 0.8697371673395402,
+      "grad_norm": 0.0,
+      "learning_rate": 8.768640587596788e-07,
+      "loss": 0.7599,
+      "step": 31668
+    },
+    {
+      "epoch": 0.8697646315673835,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764998449361705e-07,
+      "loss": 0.8504,
+      "step": 31669
+    },
+    {
+      "epoch": 0.8697920957952268,
+      "grad_norm": 0.0,
+      "learning_rate": 8.761357033013274e-07,
+      "loss": 0.7821,
+      "step": 31670
+    },
+    {
+      "epoch": 0.8698195600230699,
+      "grad_norm": 0.0,
+      "learning_rate": 8.757716338580325e-07,
+      "loss": 0.7211,
+      "step": 31671
+    },
+    {
+      "epoch": 0.8698470242509132,
+      "grad_norm": 0.0,
+      "learning_rate": 8.754076366091668e-07,
+      "loss": 0.8575,
+      "step": 31672
+    },
+    {
+      "epoch": 0.8698744884787564,
+      "grad_norm": 0.0,
+      "learning_rate": 8.750437115576105e-07,
+      "loss": 0.7618,
+      "step": 31673
+    },
+    {
+      "epoch": 0.8699019527065996,
+      "grad_norm": 0.0,
+      "learning_rate": 8.746798587062444e-07,
+      "loss": 0.8128,
+      "step": 31674
+    },
+    {
+      "epoch": 0.8699294169344429,
+      "grad_norm": 0.0,
+      "learning_rate": 8.743160780579473e-07,
+      "loss": 0.8027,
+      "step": 31675
+    },
+    {
+      "epoch": 0.8699568811622861,
+      "grad_norm": 0.0,
+      "learning_rate": 8.739523696155938e-07,
+      "loss": 0.7862,
+      "step": 31676
+    },
+    {
+      "epoch": 0.8699843453901294,
+      "grad_norm": 0.0,
+      "learning_rate": 8.735887333820636e-07,
+      "loss": 0.7229,
+      "step": 31677
+    },
+    {
+      "epoch": 0.8700118096179726,
+      "grad_norm": 0.0,
+      "learning_rate": 8.732251693602356e-07,
+      "loss": 0.8312,
+      "step": 31678
+    },
+    {
+      "epoch": 0.8700392738458158,
+      "grad_norm": 0.0,
+      "learning_rate": 8.728616775529852e-07,
+      "loss": 0.7685,
+      "step": 31679
+    },
+    {
+      "epoch": 0.8700667380736591,
+      "grad_norm": 0.0,
+      "learning_rate": 8.72498257963188e-07,
+      "loss": 0.8387,
+      "step": 31680
+    },
+    {
+      "epoch": 0.8700942023015023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.721349105937226e-07,
+      "loss": 0.869,
+      "step": 31681
+    },
+    {
+      "epoch": 0.8701216665293455,
+      "grad_norm": 0.0,
+      "learning_rate": 8.717716354474604e-07,
+      "loss": 0.9266,
+      "step": 31682
+    },
+    {
+      "epoch": 0.8701491307571888,
+      "grad_norm": 0.0,
+      "learning_rate": 8.714084325272776e-07,
+      "loss": 0.8134,
+      "step": 31683
+    },
+    {
+      "epoch": 0.870176594985032,
+      "grad_norm": 0.0,
+      "learning_rate": 8.710453018360465e-07,
+      "loss": 0.7138,
+      "step": 31684
+    },
+    {
+      "epoch": 0.8702040592128752,
+      "grad_norm": 0.0,
+      "learning_rate": 8.706822433766404e-07,
+      "loss": 0.8758,
+      "step": 31685
+    },
+    {
+      "epoch": 0.8702315234407184,
+      "grad_norm": 0.0,
+      "learning_rate": 8.703192571519337e-07,
+      "loss": 0.8123,
+      "step": 31686
+    },
+    {
+      "epoch": 0.8702589876685617,
+      "grad_norm": 0.0,
+      "learning_rate": 8.699563431647984e-07,
+      "loss": 0.7175,
+      "step": 31687
+    },
+    {
+      "epoch": 0.870286451896405,
+      "grad_norm": 0.0,
+      "learning_rate": 8.695935014181045e-07,
+      "loss": 0.8602,
+      "step": 31688
+    },
+    {
+      "epoch": 0.8703139161242481,
+      "grad_norm": 0.0,
+      "learning_rate": 8.69230731914722e-07,
+      "loss": 0.7584,
+      "step": 31689
+    },
+    {
+      "epoch": 0.8703413803520914,
+      "grad_norm": 0.0,
+      "learning_rate": 8.688680346575262e-07,
+      "loss": 0.8944,
+      "step": 31690
+    },
+    {
+      "epoch": 0.8703688445799347,
+      "grad_norm": 0.0,
+      "learning_rate": 8.685054096493806e-07,
+      "loss": 0.8346,
+      "step": 31691
+    },
+    {
+      "epoch": 0.8703963088077779,
+      "grad_norm": 0.0,
+      "learning_rate": 8.681428568931583e-07,
+      "loss": 0.7718,
+      "step": 31692
+    },
+    {
+      "epoch": 0.8704237730356211,
+      "grad_norm": 0.0,
+      "learning_rate": 8.677803763917281e-07,
+      "loss": 0.7236,
+      "step": 31693
+    },
+    {
+      "epoch": 0.8704512372634643,
+      "grad_norm": 0.0,
+      "learning_rate": 8.674179681479555e-07,
+      "loss": 0.7729,
+      "step": 31694
+    },
+    {
+      "epoch": 0.8704787014913076,
+      "grad_norm": 0.0,
+      "learning_rate": 8.670556321647095e-07,
+      "loss": 0.8147,
+      "step": 31695
+    },
+    {
+      "epoch": 0.8705061657191508,
+      "grad_norm": 0.0,
+      "learning_rate": 8.666933684448576e-07,
+      "loss": 0.8417,
+      "step": 31696
+    },
+    {
+      "epoch": 0.870533629946994,
+      "grad_norm": 0.0,
+      "learning_rate": 8.663311769912653e-07,
+      "loss": 0.8608,
+      "step": 31697
+    },
+    {
+      "epoch": 0.8705610941748373,
+      "grad_norm": 0.0,
+      "learning_rate": 8.659690578068003e-07,
+      "loss": 0.7498,
+      "step": 31698
+    },
+    {
+      "epoch": 0.8705885584026805,
+      "grad_norm": 0.0,
+      "learning_rate": 8.656070108943271e-07,
+      "loss": 0.7844,
+      "step": 31699
+    },
+    {
+      "epoch": 0.8706160226305237,
+      "grad_norm": 0.0,
+      "learning_rate": 8.652450362567067e-07,
+      "loss": 0.7482,
+      "step": 31700
+    },
+    {
+      "epoch": 0.870643486858367,
+      "grad_norm": 0.0,
+      "learning_rate": 8.648831338968067e-07,
+      "loss": 0.7731,
+      "step": 31701
+    },
+    {
+      "epoch": 0.8706709510862102,
+      "grad_norm": 0.0,
+      "learning_rate": 8.645213038174905e-07,
+      "loss": 0.8282,
+      "step": 31702
+    },
+    {
+      "epoch": 0.8706984153140535,
+      "grad_norm": 0.0,
+      "learning_rate": 8.64159546021619e-07,
+      "loss": 0.7939,
+      "step": 31703
+    },
+    {
+      "epoch": 0.8707258795418967,
+      "grad_norm": 0.0,
+      "learning_rate": 8.637978605120578e-07,
+      "loss": 0.752,
+      "step": 31704
+    },
+    {
+      "epoch": 0.8707533437697399,
+      "grad_norm": 0.0,
+      "learning_rate": 8.63436247291668e-07,
+      "loss": 0.8562,
+      "step": 31705
+    },
+    {
+      "epoch": 0.8707808079975832,
+      "grad_norm": 0.0,
+      "learning_rate": 8.630747063633071e-07,
+      "loss": 0.7575,
+      "step": 31706
+    },
+    {
+      "epoch": 0.8708082722254263,
+      "grad_norm": 0.0,
+      "learning_rate": 8.627132377298408e-07,
+      "loss": 0.7219,
+      "step": 31707
+    },
+    {
+      "epoch": 0.8708357364532696,
+      "grad_norm": 0.0,
+      "learning_rate": 8.623518413941256e-07,
+      "loss": 0.7683,
+      "step": 31708
+    },
+    {
+      "epoch": 0.8708632006811129,
+      "grad_norm": 0.0,
+      "learning_rate": 8.619905173590215e-07,
+      "loss": 0.799,
+      "step": 31709
+    },
+    {
+      "epoch": 0.8708906649089561,
+      "grad_norm": 0.0,
+      "learning_rate": 8.616292656273905e-07,
+      "loss": 0.8502,
+      "step": 31710
+    },
+    {
+      "epoch": 0.8709181291367993,
+      "grad_norm": 0.0,
+      "learning_rate": 8.612680862020862e-07,
+      "loss": 0.8105,
+      "step": 31711
+    },
+    {
+      "epoch": 0.8709455933646425,
+      "grad_norm": 0.0,
+      "learning_rate": 8.609069790859692e-07,
+      "loss": 0.8066,
+      "step": 31712
+    },
+    {
+      "epoch": 0.8709730575924858,
+      "grad_norm": 0.0,
+      "learning_rate": 8.605459442818964e-07,
+      "loss": 0.8292,
+      "step": 31713
+    },
+    {
+      "epoch": 0.871000521820329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.601849817927266e-07,
+      "loss": 0.8071,
+      "step": 31714
+    },
+    {
+      "epoch": 0.8710279860481722,
+      "grad_norm": 0.0,
+      "learning_rate": 8.598240916213108e-07,
+      "loss": 0.8093,
+      "step": 31715
+    },
+    {
+      "epoch": 0.8710554502760155,
+      "grad_norm": 0.0,
+      "learning_rate": 8.594632737705099e-07,
+      "loss": 0.8067,
+      "step": 31716
+    },
+    {
+      "epoch": 0.8710829145038588,
+      "grad_norm": 0.0,
+      "learning_rate": 8.591025282431742e-07,
+      "loss": 0.7328,
+      "step": 31717
+    },
+    {
+      "epoch": 0.8711103787317019,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5874185504216e-07,
+      "loss": 0.7737,
+      "step": 31718
+    },
+    {
+      "epoch": 0.8711378429595452,
+      "grad_norm": 0.0,
+      "learning_rate": 8.583812541703218e-07,
+      "loss": 0.8184,
+      "step": 31719
+    },
+    {
+      "epoch": 0.8711653071873884,
+      "grad_norm": 0.0,
+      "learning_rate": 8.580207256305117e-07,
+      "loss": 0.7778,
+      "step": 31720
+    },
+    {
+      "epoch": 0.8711927714152317,
+      "grad_norm": 0.0,
+      "learning_rate": 8.576602694255831e-07,
+      "loss": 0.7872,
+      "step": 31721
+    },
+    {
+      "epoch": 0.8712202356430749,
+      "grad_norm": 0.0,
+      "learning_rate": 8.572998855583891e-07,
+      "loss": 0.9567,
+      "step": 31722
+    },
+    {
+      "epoch": 0.8712476998709181,
+      "grad_norm": 0.0,
+      "learning_rate": 8.56939574031781e-07,
+      "loss": 0.8384,
+      "step": 31723
+    },
+    {
+      "epoch": 0.8712751640987614,
+      "grad_norm": 0.0,
+      "learning_rate": 8.565793348486062e-07,
+      "loss": 0.782,
+      "step": 31724
+    },
+    {
+      "epoch": 0.8713026283266045,
+      "grad_norm": 0.0,
+      "learning_rate": 8.562191680117171e-07,
+      "loss": 0.7271,
+      "step": 31725
+    },
+    {
+      "epoch": 0.8713300925544478,
+      "grad_norm": 0.0,
+      "learning_rate": 8.558590735239646e-07,
+      "loss": 0.8071,
+      "step": 31726
+    },
+    {
+      "epoch": 0.8713575567822911,
+      "grad_norm": 0.0,
+      "learning_rate": 8.554990513881978e-07,
+      "loss": 0.8512,
+      "step": 31727
+    },
+    {
+      "epoch": 0.8713850210101343,
+      "grad_norm": 0.0,
+      "learning_rate": 8.551391016072652e-07,
+      "loss": 0.7591,
+      "step": 31728
+    },
+    {
+      "epoch": 0.8714124852379775,
+      "grad_norm": 0.0,
+      "learning_rate": 8.547792241840136e-07,
+      "loss": 0.7911,
+      "step": 31729
+    },
+    {
+      "epoch": 0.8714399494658208,
+      "grad_norm": 0.0,
+      "learning_rate": 8.544194191212906e-07,
+      "loss": 0.7131,
+      "step": 31730
+    },
+    {
+      "epoch": 0.871467413693664,
+      "grad_norm": 0.0,
+      "learning_rate": 8.540596864219464e-07,
+      "loss": 0.7569,
+      "step": 31731
+    },
+    {
+      "epoch": 0.8714948779215073,
+      "grad_norm": 0.0,
+      "learning_rate": 8.537000260888217e-07,
+      "loss": 0.8086,
+      "step": 31732
+    },
+    {
+      "epoch": 0.8715223421493504,
+      "grad_norm": 0.0,
+      "learning_rate": 8.533404381247668e-07,
+      "loss": 0.7897,
+      "step": 31733
+    },
+    {
+      "epoch": 0.8715498063771937,
+      "grad_norm": 0.0,
+      "learning_rate": 8.529809225326257e-07,
+      "loss": 0.8038,
+      "step": 31734
+    },
+    {
+      "epoch": 0.871577270605037,
+      "grad_norm": 0.0,
+      "learning_rate": 8.52621479315242e-07,
+      "loss": 0.7927,
+      "step": 31735
+    },
+    {
+      "epoch": 0.8716047348328801,
+      "grad_norm": 0.0,
+      "learning_rate": 8.522621084754613e-07,
+      "loss": 0.8516,
+      "step": 31736
+    },
+    {
+      "epoch": 0.8716321990607234,
+      "grad_norm": 0.0,
+      "learning_rate": 8.519028100161242e-07,
+      "loss": 0.8124,
+      "step": 31737
+    },
+    {
+      "epoch": 0.8716596632885666,
+      "grad_norm": 0.0,
+      "learning_rate": 8.515435839400787e-07,
+      "loss": 0.9441,
+      "step": 31738
+    },
+    {
+      "epoch": 0.8716871275164099,
+      "grad_norm": 0.0,
+      "learning_rate": 8.511844302501627e-07,
+      "loss": 0.8277,
+      "step": 31739
+    },
+    {
+      "epoch": 0.8717145917442531,
+      "grad_norm": 0.0,
+      "learning_rate": 8.508253489492202e-07,
+      "loss": 0.8321,
+      "step": 31740
+    },
+    {
+      "epoch": 0.8717420559720963,
+      "grad_norm": 0.0,
+      "learning_rate": 8.504663400400903e-07,
+      "loss": 0.8094,
+      "step": 31741
+    },
+    {
+      "epoch": 0.8717695201999396,
+      "grad_norm": 0.0,
+      "learning_rate": 8.501074035256152e-07,
+      "loss": 0.8041,
+      "step": 31742
+    },
+    {
+      "epoch": 0.8717969844277829,
+      "grad_norm": 0.0,
+      "learning_rate": 8.497485394086346e-07,
+      "loss": 0.787,
+      "step": 31743
+    },
+    {
+      "epoch": 0.871824448655626,
+      "grad_norm": 0.0,
+      "learning_rate": 8.493897476919876e-07,
+      "loss": 0.8097,
+      "step": 31744
+    },
+    {
+      "epoch": 0.8718519128834693,
+      "grad_norm": 0.0,
+      "learning_rate": 8.490310283785142e-07,
+      "loss": 0.8341,
+      "step": 31745
+    },
+    {
+      "epoch": 0.8718793771113125,
+      "grad_norm": 0.0,
+      "learning_rate": 8.486723814710529e-07,
+      "loss": 0.7785,
+      "step": 31746
+    },
+    {
+      "epoch": 0.8719068413391557,
+      "grad_norm": 0.0,
+      "learning_rate": 8.483138069724417e-07,
+      "loss": 0.7248,
+      "step": 31747
+    },
+    {
+      "epoch": 0.871934305566999,
+      "grad_norm": 0.0,
+      "learning_rate": 8.479553048855138e-07,
+      "loss": 0.8311,
+      "step": 31748
+    },
+    {
+      "epoch": 0.8719617697948422,
+      "grad_norm": 0.0,
+      "learning_rate": 8.475968752131092e-07,
+      "loss": 0.8601,
+      "step": 31749
+    },
+    {
+      "epoch": 0.8719892340226855,
+      "grad_norm": 0.0,
+      "learning_rate": 8.472385179580634e-07,
+      "loss": 0.7948,
+      "step": 31750
+    },
+    {
+      "epoch": 0.8720166982505286,
+      "grad_norm": 0.0,
+      "learning_rate": 8.468802331232118e-07,
+      "loss": 0.8358,
+      "step": 31751
+    },
+    {
+      "epoch": 0.8720441624783719,
+      "grad_norm": 0.0,
+      "learning_rate": 8.465220207113911e-07,
+      "loss": 0.7346,
+      "step": 31752
+    },
+    {
+      "epoch": 0.8720716267062152,
+      "grad_norm": 0.0,
+      "learning_rate": 8.461638807254325e-07,
+      "loss": 0.7731,
+      "step": 31753
+    },
+    {
+      "epoch": 0.8720990909340584,
+      "grad_norm": 0.0,
+      "learning_rate": 8.458058131681712e-07,
+      "loss": 0.8894,
+      "step": 31754
+    },
+    {
+      "epoch": 0.8721265551619016,
+      "grad_norm": 0.0,
+      "learning_rate": 8.454478180424419e-07,
+      "loss": 0.7779,
+      "step": 31755
+    },
+    {
+      "epoch": 0.8721540193897449,
+      "grad_norm": 0.0,
+      "learning_rate": 8.450898953510733e-07,
+      "loss": 0.86,
+      "step": 31756
+    },
+    {
+      "epoch": 0.8721814836175881,
+      "grad_norm": 0.0,
+      "learning_rate": 8.447320450968999e-07,
+      "loss": 0.8539,
+      "step": 31757
+    },
+    {
+      "epoch": 0.8722089478454313,
+      "grad_norm": 0.0,
+      "learning_rate": 8.443742672827548e-07,
+      "loss": 0.7477,
+      "step": 31758
+    },
+    {
+      "epoch": 0.8722364120732745,
+      "grad_norm": 0.0,
+      "learning_rate": 8.440165619114659e-07,
+      "loss": 0.7787,
+      "step": 31759
+    },
+    {
+      "epoch": 0.8722638763011178,
+      "grad_norm": 0.0,
+      "learning_rate": 8.436589289858643e-07,
+      "loss": 0.8682,
+      "step": 31760
+    },
+    {
+      "epoch": 0.8722913405289611,
+      "grad_norm": 0.0,
+      "learning_rate": 8.433013685087799e-07,
+      "loss": 0.7165,
+      "step": 31761
+    },
+    {
+      "epoch": 0.8723188047568042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.429438804830437e-07,
+      "loss": 0.7975,
+      "step": 31762
+    },
+    {
+      "epoch": 0.8723462689846475,
+      "grad_norm": 0.0,
+      "learning_rate": 8.425864649114813e-07,
+      "loss": 0.8235,
+      "step": 31763
+    },
+    {
+      "epoch": 0.8723737332124907,
+      "grad_norm": 0.0,
+      "learning_rate": 8.422291217969236e-07,
+      "loss": 0.8271,
+      "step": 31764
+    },
+    {
+      "epoch": 0.872401197440334,
+      "grad_norm": 0.0,
+      "learning_rate": 8.418718511421942e-07,
+      "loss": 0.8777,
+      "step": 31765
+    },
+    {
+      "epoch": 0.8724286616681772,
+      "grad_norm": 0.0,
+      "learning_rate": 8.415146529501228e-07,
+      "loss": 0.7928,
+      "step": 31766
+    },
+    {
+      "epoch": 0.8724561258960204,
+      "grad_norm": 0.0,
+      "learning_rate": 8.41157527223535e-07,
+      "loss": 0.9141,
+      "step": 31767
+    },
+    {
+      "epoch": 0.8724835901238637,
+      "grad_norm": 0.0,
+      "learning_rate": 8.408004739652564e-07,
+      "loss": 0.8736,
+      "step": 31768
+    },
+    {
+      "epoch": 0.872511054351707,
+      "grad_norm": 0.0,
+      "learning_rate": 8.404434931781124e-07,
+      "loss": 0.6846,
+      "step": 31769
+    },
+    {
+      "epoch": 0.8725385185795501,
+      "grad_norm": 0.0,
+      "learning_rate": 8.400865848649297e-07,
+      "loss": 0.7664,
+      "step": 31770
+    },
+    {
+      "epoch": 0.8725659828073934,
+      "grad_norm": 0.0,
+      "learning_rate": 8.397297490285305e-07,
+      "loss": 0.8046,
+      "step": 31771
+    },
+    {
+      "epoch": 0.8725934470352366,
+      "grad_norm": 0.0,
+      "learning_rate": 8.393729856717359e-07,
+      "loss": 0.6972,
+      "step": 31772
+    },
+    {
+      "epoch": 0.8726209112630798,
+      "grad_norm": 0.0,
+      "learning_rate": 8.390162947973701e-07,
+      "loss": 0.9056,
+      "step": 31773
+    },
+    {
+      "epoch": 0.8726483754909231,
+      "grad_norm": 0.0,
+      "learning_rate": 8.386596764082566e-07,
+      "loss": 0.7837,
+      "step": 31774
+    },
+    {
+      "epoch": 0.8726758397187663,
+      "grad_norm": 0.0,
+      "learning_rate": 8.383031305072175e-07,
+      "loss": 0.7739,
+      "step": 31775
+    },
+    {
+      "epoch": 0.8727033039466096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.379466570970718e-07,
+      "loss": 0.6929,
+      "step": 31776
+    },
+    {
+      "epoch": 0.8727307681744527,
+      "grad_norm": 0.0,
+      "learning_rate": 8.375902561806415e-07,
+      "loss": 0.7329,
+      "step": 31777
+    },
+    {
+      "epoch": 0.872758232402296,
+      "grad_norm": 0.0,
+      "learning_rate": 8.372339277607466e-07,
+      "loss": 0.7752,
+      "step": 31778
+    },
+    {
+      "epoch": 0.8727856966301393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.368776718402061e-07,
+      "loss": 0.8517,
+      "step": 31779
+    },
+    {
+      "epoch": 0.8728131608579824,
+      "grad_norm": 0.0,
+      "learning_rate": 8.365214884218375e-07,
+      "loss": 0.9285,
+      "step": 31780
+    },
+    {
+      "epoch": 0.8728406250858257,
+      "grad_norm": 0.0,
+      "learning_rate": 8.361653775084633e-07,
+      "loss": 0.792,
+      "step": 31781
+    },
+    {
+      "epoch": 0.872868089313669,
+      "grad_norm": 0.0,
+      "learning_rate": 8.358093391028965e-07,
+      "loss": 0.8346,
+      "step": 31782
+    },
+    {
+      "epoch": 0.8728955535415122,
+      "grad_norm": 0.0,
+      "learning_rate": 8.354533732079562e-07,
+      "loss": 0.8451,
+      "step": 31783
+    },
+    {
+      "epoch": 0.8729230177693554,
+      "grad_norm": 0.0,
+      "learning_rate": 8.350974798264577e-07,
+      "loss": 0.7575,
+      "step": 31784
+    },
+    {
+      "epoch": 0.8729504819971986,
+      "grad_norm": 0.0,
+      "learning_rate": 8.34741658961219e-07,
+      "loss": 0.7733,
+      "step": 31785
+    },
+    {
+      "epoch": 0.8729779462250419,
+      "grad_norm": 0.0,
+      "learning_rate": 8.343859106150565e-07,
+      "loss": 0.7997,
+      "step": 31786
+    },
+    {
+      "epoch": 0.8730054104528852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.340302347907825e-07,
+      "loss": 0.8542,
+      "step": 31787
+    },
+    {
+      "epoch": 0.8730328746807283,
+      "grad_norm": 0.0,
+      "learning_rate": 8.336746314912103e-07,
+      "loss": 0.8112,
+      "step": 31788
+    },
+    {
+      "epoch": 0.8730603389085716,
+      "grad_norm": 0.0,
+      "learning_rate": 8.333191007191555e-07,
+      "loss": 0.8158,
+      "step": 31789
+    },
+    {
+      "epoch": 0.8730878031364149,
+      "grad_norm": 0.0,
+      "learning_rate": 8.329636424774301e-07,
+      "loss": 0.8105,
+      "step": 31790
+    },
+    {
+      "epoch": 0.873115267364258,
+      "grad_norm": 0.0,
+      "learning_rate": 8.326082567688476e-07,
+      "loss": 0.7742,
+      "step": 31791
+    },
+    {
+      "epoch": 0.8731427315921013,
+      "grad_norm": 0.0,
+      "learning_rate": 8.322529435962201e-07,
+      "loss": 0.8211,
+      "step": 31792
+    },
+    {
+      "epoch": 0.8731701958199445,
+      "grad_norm": 0.0,
+      "learning_rate": 8.318977029623598e-07,
+      "loss": 0.8379,
+      "step": 31793
+    },
+    {
+      "epoch": 0.8731976600477878,
+      "grad_norm": 0.0,
+      "learning_rate": 8.315425348700745e-07,
+      "loss": 0.814,
+      "step": 31794
+    },
+    {
+      "epoch": 0.873225124275631,
+      "grad_norm": 0.0,
+      "learning_rate": 8.311874393221786e-07,
+      "loss": 0.8164,
+      "step": 31795
+    },
+    {
+      "epoch": 0.8732525885034742,
+      "grad_norm": 0.0,
+      "learning_rate": 8.308324163214764e-07,
+      "loss": 0.8355,
+      "step": 31796
+    },
+    {
+      "epoch": 0.8732800527313175,
+      "grad_norm": 0.0,
+      "learning_rate": 8.304774658707814e-07,
+      "loss": 0.7963,
+      "step": 31797
+    },
+    {
+      "epoch": 0.8733075169591606,
+      "grad_norm": 0.0,
+      "learning_rate": 8.301225879728992e-07,
+      "loss": 0.8961,
+      "step": 31798
+    },
+    {
+      "epoch": 0.8733349811870039,
+      "grad_norm": 0.0,
+      "learning_rate": 8.297677826306417e-07,
+      "loss": 0.8007,
+      "step": 31799
+    },
+    {
+      "epoch": 0.8733624454148472,
+      "grad_norm": 0.0,
+      "learning_rate": 8.294130498468123e-07,
+      "loss": 0.8426,
+      "step": 31800
+    },
+    {
+      "epoch": 0.8733899096426904,
+      "grad_norm": 0.0,
+      "learning_rate": 8.290583896242188e-07,
+      "loss": 0.8231,
+      "step": 31801
+    },
+    {
+      "epoch": 0.8734173738705336,
+      "grad_norm": 0.0,
+      "learning_rate": 8.28703801965669e-07,
+      "loss": 0.8929,
+      "step": 31802
+    },
+    {
+      "epoch": 0.8734448380983769,
+      "grad_norm": 0.0,
+      "learning_rate": 8.283492868739662e-07,
+      "loss": 0.728,
+      "step": 31803
+    },
+    {
+      "epoch": 0.8734723023262201,
+      "grad_norm": 0.0,
+      "learning_rate": 8.279948443519159e-07,
+      "loss": 0.8053,
+      "step": 31804
+    },
+    {
+      "epoch": 0.8734997665540634,
+      "grad_norm": 0.0,
+      "learning_rate": 8.276404744023258e-07,
+      "loss": 0.8352,
+      "step": 31805
+    },
+    {
+      "epoch": 0.8735272307819065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.272861770279961e-07,
+      "loss": 0.7909,
+      "step": 31806
+    },
+    {
+      "epoch": 0.8735546950097498,
+      "grad_norm": 0.0,
+      "learning_rate": 8.269319522317298e-07,
+      "loss": 0.9256,
+      "step": 31807
+    },
+    {
+      "epoch": 0.8735821592375931,
+      "grad_norm": 0.0,
+      "learning_rate": 8.265778000163327e-07,
+      "loss": 0.911,
+      "step": 31808
+    },
+    {
+      "epoch": 0.8736096234654362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.262237203846046e-07,
+      "loss": 0.8409,
+      "step": 31809
+    },
+    {
+      "epoch": 0.8736370876932795,
+      "grad_norm": 0.0,
+      "learning_rate": 8.2586971333935e-07,
+      "loss": 0.8047,
+      "step": 31810
+    },
+    {
+      "epoch": 0.8736645519211227,
+      "grad_norm": 0.0,
+      "learning_rate": 8.255157788833678e-07,
+      "loss": 0.8267,
+      "step": 31811
+    },
+    {
+      "epoch": 0.873692016148966,
+      "grad_norm": 0.0,
+      "learning_rate": 8.25161917019457e-07,
+      "loss": 0.7782,
+      "step": 31812
+    },
+    {
+      "epoch": 0.8737194803768092,
+      "grad_norm": 0.0,
+      "learning_rate": 8.248081277504183e-07,
+      "loss": 0.789,
+      "step": 31813
+    },
+    {
+      "epoch": 0.8737469446046524,
+      "grad_norm": 0.0,
+      "learning_rate": 8.244544110790531e-07,
+      "loss": 0.8224,
+      "step": 31814
+    },
+    {
+      "epoch": 0.8737744088324957,
+      "grad_norm": 0.0,
+      "learning_rate": 8.24100767008158e-07,
+      "loss": 0.9058,
+      "step": 31815
+    },
+    {
+      "epoch": 0.873801873060339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.23747195540533e-07,
+      "loss": 0.767,
+      "step": 31816
+    },
+    {
+      "epoch": 0.8738293372881821,
+      "grad_norm": 0.0,
+      "learning_rate": 8.233936966789757e-07,
+      "loss": 0.7215,
+      "step": 31817
+    },
+    {
+      "epoch": 0.8738568015160254,
+      "grad_norm": 0.0,
+      "learning_rate": 8.230402704262807e-07,
+      "loss": 0.8182,
+      "step": 31818
+    },
+    {
+      "epoch": 0.8738842657438686,
+      "grad_norm": 0.0,
+      "learning_rate": 8.226869167852469e-07,
+      "loss": 0.732,
+      "step": 31819
+    },
+    {
+      "epoch": 0.8739117299717118,
+      "grad_norm": 0.0,
+      "learning_rate": 8.223336357586687e-07,
+      "loss": 0.7816,
+      "step": 31820
+    },
+    {
+      "epoch": 0.8739391941995551,
+      "grad_norm": 0.0,
+      "learning_rate": 8.219804273493404e-07,
+      "loss": 0.8453,
+      "step": 31821
+    },
+    {
+      "epoch": 0.8739666584273983,
+      "grad_norm": 0.0,
+      "learning_rate": 8.216272915600587e-07,
+      "loss": 0.8275,
+      "step": 31822
+    },
+    {
+      "epoch": 0.8739941226552416,
+      "grad_norm": 0.0,
+      "learning_rate": 8.212742283936193e-07,
+      "loss": 0.869,
+      "step": 31823
+    },
+    {
+      "epoch": 0.8740215868830847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.209212378528119e-07,
+      "loss": 0.7295,
+      "step": 31824
+    },
+    {
+      "epoch": 0.874049051110928,
+      "grad_norm": 0.0,
+      "learning_rate": 8.205683199404313e-07,
+      "loss": 0.8314,
+      "step": 31825
+    },
+    {
+      "epoch": 0.8740765153387713,
+      "grad_norm": 0.0,
+      "learning_rate": 8.202154746592717e-07,
+      "loss": 0.881,
+      "step": 31826
+    },
+    {
+      "epoch": 0.8741039795666145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.198627020121219e-07,
+      "loss": 0.8128,
+      "step": 31827
+    },
+    {
+      "epoch": 0.8741314437944577,
+      "grad_norm": 0.0,
+      "learning_rate": 8.195100020017732e-07,
+      "loss": 0.7298,
+      "step": 31828
+    },
+    {
+      "epoch": 0.874158908022301,
+      "grad_norm": 0.0,
+      "learning_rate": 8.191573746310211e-07,
+      "loss": 0.8804,
+      "step": 31829
+    },
+    {
+      "epoch": 0.8741863722501442,
+      "grad_norm": 0.0,
+      "learning_rate": 8.188048199026488e-07,
+      "loss": 0.7587,
+      "step": 31830
+    },
+    {
+      "epoch": 0.8742138364779874,
+      "grad_norm": 0.0,
+      "learning_rate": 8.184523378194497e-07,
+      "loss": 0.8618,
+      "step": 31831
+    },
+    {
+      "epoch": 0.8742413007058306,
+      "grad_norm": 0.0,
+      "learning_rate": 8.180999283842128e-07,
+      "loss": 0.9055,
+      "step": 31832
+    },
+    {
+      "epoch": 0.8742687649336739,
+      "grad_norm": 0.0,
+      "learning_rate": 8.177475915997246e-07,
+      "loss": 0.7379,
+      "step": 31833
+    },
+    {
+      "epoch": 0.8742962291615172,
+      "grad_norm": 0.0,
+      "learning_rate": 8.173953274687774e-07,
+      "loss": 0.7762,
+      "step": 31834
+    },
+    {
+      "epoch": 0.8743236933893603,
+      "grad_norm": 0.0,
+      "learning_rate": 8.170431359941555e-07,
+      "loss": 0.829,
+      "step": 31835
+    },
+    {
+      "epoch": 0.8743511576172036,
+      "grad_norm": 0.0,
+      "learning_rate": 8.166910171786424e-07,
+      "loss": 0.8466,
+      "step": 31836
+    },
+    {
+      "epoch": 0.8743786218450468,
+      "grad_norm": 0.0,
+      "learning_rate": 8.163389710250281e-07,
+      "loss": 0.8276,
+      "step": 31837
+    },
+    {
+      "epoch": 0.87440608607289,
+      "grad_norm": 0.0,
+      "learning_rate": 8.159869975360979e-07,
+      "loss": 0.7882,
+      "step": 31838
+    },
+    {
+      "epoch": 0.8744335503007333,
+      "grad_norm": 0.0,
+      "learning_rate": 8.156350967146354e-07,
+      "loss": 0.7245,
+      "step": 31839
+    },
+    {
+      "epoch": 0.8744610145285765,
+      "grad_norm": 0.0,
+      "learning_rate": 8.152832685634271e-07,
+      "loss": 0.7683,
+      "step": 31840
+    },
+    {
+      "epoch": 0.8744884787564198,
+      "grad_norm": 0.0,
+      "learning_rate": 8.149315130852542e-07,
+      "loss": 0.8732,
+      "step": 31841
+    },
+    {
+      "epoch": 0.874515942984263,
+      "grad_norm": 0.0,
+      "learning_rate": 8.145798302829011e-07,
+      "loss": 0.8301,
+      "step": 31842
+    },
+    {
+      "epoch": 0.8745434072121062,
+      "grad_norm": 0.0,
+      "learning_rate": 8.142282201591523e-07,
+      "loss": 0.7637,
+      "step": 31843
+    },
+    {
+      "epoch": 0.8745708714399495,
+      "grad_norm": 0.0,
+      "learning_rate": 8.138766827167854e-07,
+      "loss": 0.6963,
+      "step": 31844
+    },
+    {
+      "epoch": 0.8745983356677927,
+      "grad_norm": 0.0,
+      "learning_rate": 8.135252179585851e-07,
+      "loss": 0.7582,
+      "step": 31845
+    },
+    {
+      "epoch": 0.8746257998956359,
+      "grad_norm": 0.0,
+      "learning_rate": 8.131738258873345e-07,
+      "loss": 0.8142,
+      "step": 31846
+    },
+    {
+      "epoch": 0.8746532641234792,
+      "grad_norm": 0.0,
+      "learning_rate": 8.128225065058081e-07,
+      "loss": 0.8374,
+      "step": 31847
+    },
+    {
+      "epoch": 0.8746807283513224,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124712598167894e-07,
+      "loss": 0.8319,
+      "step": 31848
+    },
+    {
+      "epoch": 0.8747081925791657,
+      "grad_norm": 0.0,
+      "learning_rate": 8.121200858230571e-07,
+      "loss": 0.8269,
+      "step": 31849
+    },
+    {
+      "epoch": 0.8747356568070088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.117689845273912e-07,
+      "loss": 0.8394,
+      "step": 31850
+    },
+    {
+      "epoch": 0.8747631210348521,
+      "grad_norm": 0.0,
+      "learning_rate": 8.114179559325663e-07,
+      "loss": 0.8021,
+      "step": 31851
+    },
+    {
+      "epoch": 0.8747905852626954,
+      "grad_norm": 0.0,
+      "learning_rate": 8.110670000413645e-07,
+      "loss": 0.7513,
+      "step": 31852
+    },
+    {
+      "epoch": 0.8748180494905385,
+      "grad_norm": 0.0,
+      "learning_rate": 8.107161168565569e-07,
+      "loss": 0.7078,
+      "step": 31853
+    },
+    {
+      "epoch": 0.8748455137183818,
+      "grad_norm": 0.0,
+      "learning_rate": 8.103653063809247e-07,
+      "loss": 0.8855,
+      "step": 31854
+    },
+    {
+      "epoch": 0.8748729779462251,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1001456861724e-07,
+      "loss": 0.8553,
+      "step": 31855
+    },
+    {
+      "epoch": 0.8749004421740683,
+      "grad_norm": 0.0,
+      "learning_rate": 8.096639035682818e-07,
+      "loss": 0.9178,
+      "step": 31856
+    },
+    {
+      "epoch": 0.8749279064019115,
+      "grad_norm": 0.0,
+      "learning_rate": 8.093133112368212e-07,
+      "loss": 0.8271,
+      "step": 31857
+    },
+    {
+      "epoch": 0.8749553706297547,
+      "grad_norm": 0.0,
+      "learning_rate": 8.089627916256371e-07,
+      "loss": 0.825,
+      "step": 31858
+    },
+    {
+      "epoch": 0.874982834857598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.086123447374983e-07,
+      "loss": 0.8476,
+      "step": 31859
+    },
+    {
+      "epoch": 0.8750102990854413,
+      "grad_norm": 0.0,
+      "learning_rate": 8.082619705751782e-07,
+      "loss": 0.749,
+      "step": 31860
+    },
+    {
+      "epoch": 0.8750377633132844,
+      "grad_norm": 0.0,
+      "learning_rate": 8.07911669141449e-07,
+      "loss": 0.801,
+      "step": 31861
+    },
+    {
+      "epoch": 0.8750652275411277,
+      "grad_norm": 0.0,
+      "learning_rate": 8.075614404390841e-07,
+      "loss": 0.8195,
+      "step": 31862
+    },
+    {
+      "epoch": 0.8750926917689709,
+      "grad_norm": 0.0,
+      "learning_rate": 8.072112844708535e-07,
+      "loss": 0.7401,
+      "step": 31863
+    },
+    {
+      "epoch": 0.8751201559968141,
+      "grad_norm": 0.0,
+      "learning_rate": 8.068612012395305e-07,
+      "loss": 0.8514,
+      "step": 31864
+    },
+    {
+      "epoch": 0.8751476202246574,
+      "grad_norm": 0.0,
+      "learning_rate": 8.065111907478806e-07,
+      "loss": 0.7392,
+      "step": 31865
+    },
+    {
+      "epoch": 0.8751750844525006,
+      "grad_norm": 0.0,
+      "learning_rate": 8.061612529986751e-07,
+      "loss": 0.8066,
+      "step": 31866
+    },
+    {
+      "epoch": 0.8752025486803439,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05811387994685e-07,
+      "loss": 0.8358,
+      "step": 31867
+    },
+    {
+      "epoch": 0.8752300129081871,
+      "grad_norm": 0.0,
+      "learning_rate": 8.054615957386746e-07,
+      "loss": 0.7615,
+      "step": 31868
+    },
+    {
+      "epoch": 0.8752574771360303,
+      "grad_norm": 0.0,
+      "learning_rate": 8.051118762334142e-07,
+      "loss": 0.803,
+      "step": 31869
+    },
+    {
+      "epoch": 0.8752849413638736,
+      "grad_norm": 0.0,
+      "learning_rate": 8.047622294816726e-07,
+      "loss": 0.7998,
+      "step": 31870
+    },
+    {
+      "epoch": 0.8753124055917167,
+      "grad_norm": 0.0,
+      "learning_rate": 8.04412655486212e-07,
+      "loss": 0.8315,
+      "step": 31871
+    },
+    {
+      "epoch": 0.87533986981956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.040631542498001e-07,
+      "loss": 0.8596,
+      "step": 31872
+    },
+    {
+      "epoch": 0.8753673340474033,
+      "grad_norm": 0.0,
+      "learning_rate": 8.037137257752026e-07,
+      "loss": 0.7912,
+      "step": 31873
+    },
+    {
+      "epoch": 0.8753947982752465,
+      "grad_norm": 0.0,
+      "learning_rate": 8.033643700651872e-07,
+      "loss": 0.8121,
+      "step": 31874
+    },
+    {
+      "epoch": 0.8754222625030897,
+      "grad_norm": 0.0,
+      "learning_rate": 8.030150871225129e-07,
+      "loss": 0.7435,
+      "step": 31875
+    },
+    {
+      "epoch": 0.8754497267309329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.026658769499474e-07,
+      "loss": 0.856,
+      "step": 31876
+    },
+    {
+      "epoch": 0.8754771909587762,
+      "grad_norm": 0.0,
+      "learning_rate": 8.023167395502507e-07,
+      "loss": 0.7472,
+      "step": 31877
+    },
+    {
+      "epoch": 0.8755046551866195,
+      "grad_norm": 0.0,
+      "learning_rate": 8.019676749261862e-07,
+      "loss": 0.7548,
+      "step": 31878
+    },
+    {
+      "epoch": 0.8755321194144626,
+      "grad_norm": 0.0,
+      "learning_rate": 8.016186830805172e-07,
+      "loss": 0.8642,
+      "step": 31879
+    },
+    {
+      "epoch": 0.8755595836423059,
+      "grad_norm": 0.0,
+      "learning_rate": 8.012697640160039e-07,
+      "loss": 0.8746,
+      "step": 31880
+    },
+    {
+      "epoch": 0.8755870478701492,
+      "grad_norm": 0.0,
+      "learning_rate": 8.009209177354083e-07,
+      "loss": 0.7557,
+      "step": 31881
+    },
+    {
+      "epoch": 0.8756145120979923,
+      "grad_norm": 0.0,
+      "learning_rate": 8.005721442414904e-07,
+      "loss": 0.7865,
+      "step": 31882
+    },
+    {
+      "epoch": 0.8756419763258356,
+      "grad_norm": 0.0,
+      "learning_rate": 8.002234435370094e-07,
+      "loss": 0.7149,
+      "step": 31883
+    },
+    {
+      "epoch": 0.8756694405536788,
+      "grad_norm": 0.0,
+      "learning_rate": 7.998748156247227e-07,
+      "loss": 0.7366,
+      "step": 31884
+    },
+    {
+      "epoch": 0.8756969047815221,
+      "grad_norm": 0.0,
+      "learning_rate": 7.995262605073905e-07,
+      "loss": 0.7534,
+      "step": 31885
+    },
+    {
+      "epoch": 0.8757243690093653,
+      "grad_norm": 0.0,
+      "learning_rate": 7.991777781877696e-07,
+      "loss": 0.7965,
+      "step": 31886
+    },
+    {
+      "epoch": 0.8757518332372085,
+      "grad_norm": 0.0,
+      "learning_rate": 7.988293686686188e-07,
+      "loss": 0.6903,
+      "step": 31887
+    },
+    {
+      "epoch": 0.8757792974650518,
+      "grad_norm": 0.0,
+      "learning_rate": 7.984810319526959e-07,
+      "loss": 0.8976,
+      "step": 31888
+    },
+    {
+      "epoch": 0.875806761692895,
+      "grad_norm": 0.0,
+      "learning_rate": 7.981327680427542e-07,
+      "loss": 0.8787,
+      "step": 31889
+    },
+    {
+      "epoch": 0.8758342259207382,
+      "grad_norm": 0.0,
+      "learning_rate": 7.977845769415493e-07,
+      "loss": 0.7173,
+      "step": 31890
+    },
+    {
+      "epoch": 0.8758616901485815,
+      "grad_norm": 0.0,
+      "learning_rate": 7.974364586518401e-07,
+      "loss": 0.8211,
+      "step": 31891
+    },
+    {
+      "epoch": 0.8758891543764247,
+      "grad_norm": 0.0,
+      "learning_rate": 7.970884131763768e-07,
+      "loss": 0.7957,
+      "step": 31892
+    },
+    {
+      "epoch": 0.875916618604268,
+      "grad_norm": 0.0,
+      "learning_rate": 7.967404405179157e-07,
+      "loss": 0.8449,
+      "step": 31893
+    },
+    {
+      "epoch": 0.8759440828321112,
+      "grad_norm": 0.0,
+      "learning_rate": 7.963925406792084e-07,
+      "loss": 0.7414,
+      "step": 31894
+    },
+    {
+      "epoch": 0.8759715470599544,
+      "grad_norm": 0.0,
+      "learning_rate": 7.960447136630078e-07,
+      "loss": 0.9066,
+      "step": 31895
+    },
+    {
+      "epoch": 0.8759990112877977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.956969594720665e-07,
+      "loss": 0.7532,
+      "step": 31896
+    },
+    {
+      "epoch": 0.8760264755156408,
+      "grad_norm": 0.0,
+      "learning_rate": 7.953492781091366e-07,
+      "loss": 0.7238,
+      "step": 31897
+    },
+    {
+      "epoch": 0.8760539397434841,
+      "grad_norm": 0.0,
+      "learning_rate": 7.950016695769713e-07,
+      "loss": 0.7095,
+      "step": 31898
+    },
+    {
+      "epoch": 0.8760814039713274,
+      "grad_norm": 0.0,
+      "learning_rate": 7.946541338783176e-07,
+      "loss": 0.8415,
+      "step": 31899
+    },
+    {
+      "epoch": 0.8761088681991706,
+      "grad_norm": 0.0,
+      "learning_rate": 7.943066710159241e-07,
+      "loss": 0.8176,
+      "step": 31900
+    },
+    {
+      "epoch": 0.8761363324270138,
+      "grad_norm": 0.0,
+      "learning_rate": 7.939592809925422e-07,
+      "loss": 0.8871,
+      "step": 31901
+    },
+    {
+      "epoch": 0.876163796654857,
+      "grad_norm": 0.0,
+      "learning_rate": 7.936119638109219e-07,
+      "loss": 0.781,
+      "step": 31902
+    },
+    {
+      "epoch": 0.8761912608827003,
+      "grad_norm": 0.0,
+      "learning_rate": 7.932647194738085e-07,
+      "loss": 0.7721,
+      "step": 31903
+    },
+    {
+      "epoch": 0.8762187251105436,
+      "grad_norm": 0.0,
+      "learning_rate": 7.929175479839512e-07,
+      "loss": 0.8499,
+      "step": 31904
+    },
+    {
+      "epoch": 0.8762461893383867,
+      "grad_norm": 0.0,
+      "learning_rate": 7.925704493440989e-07,
+      "loss": 0.8633,
+      "step": 31905
+    },
+    {
+      "epoch": 0.87627365356623,
+      "grad_norm": 0.0,
+      "learning_rate": 7.922234235569947e-07,
+      "loss": 0.8049,
+      "step": 31906
+    },
+    {
+      "epoch": 0.8763011177940733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.918764706253867e-07,
+      "loss": 0.7192,
+      "step": 31907
+    },
+    {
+      "epoch": 0.8763285820219164,
+      "grad_norm": 0.0,
+      "learning_rate": 7.91529590552017e-07,
+      "loss": 0.6696,
+      "step": 31908
+    },
+    {
+      "epoch": 0.8763560462497597,
+      "grad_norm": 0.0,
+      "learning_rate": 7.911827833396324e-07,
+      "loss": 0.8752,
+      "step": 31909
+    },
+    {
+      "epoch": 0.8763835104776029,
+      "grad_norm": 0.0,
+      "learning_rate": 7.90836048990976e-07,
+      "loss": 0.801,
+      "step": 31910
+    },
+    {
+      "epoch": 0.8764109747054462,
+      "grad_norm": 0.0,
+      "learning_rate": 7.904893875087949e-07,
+      "loss": 0.8589,
+      "step": 31911
+    },
+    {
+      "epoch": 0.8764384389332894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.901427988958266e-07,
+      "loss": 0.8176,
+      "step": 31912
+    },
+    {
+      "epoch": 0.8764659031611326,
+      "grad_norm": 0.0,
+      "learning_rate": 7.897962831548167e-07,
+      "loss": 0.8123,
+      "step": 31913
+    },
+    {
+      "epoch": 0.8764933673889759,
+      "grad_norm": 0.0,
+      "learning_rate": 7.894498402885053e-07,
+      "loss": 0.7281,
+      "step": 31914
+    },
+    {
+      "epoch": 0.876520831616819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.891034702996381e-07,
+      "loss": 0.7726,
+      "step": 31915
+    },
+    {
+      "epoch": 0.8765482958446623,
+      "grad_norm": 0.0,
+      "learning_rate": 7.887571731909493e-07,
+      "loss": 0.8302,
+      "step": 31916
+    },
+    {
+      "epoch": 0.8765757600725056,
+      "grad_norm": 0.0,
+      "learning_rate": 7.884109489651836e-07,
+      "loss": 0.8215,
+      "step": 31917
+    },
+    {
+      "epoch": 0.8766032243003488,
+      "grad_norm": 0.0,
+      "learning_rate": 7.880647976250766e-07,
+      "loss": 0.7149,
+      "step": 31918
+    },
+    {
+      "epoch": 0.876630688528192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.877187191733704e-07,
+      "loss": 0.7728,
+      "step": 31919
+    },
+    {
+      "epoch": 0.8766581527560353,
+      "grad_norm": 0.0,
+      "learning_rate": 7.873727136128007e-07,
+      "loss": 0.7922,
+      "step": 31920
+    },
+    {
+      "epoch": 0.8766856169838785,
+      "grad_norm": 0.0,
+      "learning_rate": 7.870267809461085e-07,
+      "loss": 0.7646,
+      "step": 31921
+    },
+    {
+      "epoch": 0.8767130812117218,
+      "grad_norm": 0.0,
+      "learning_rate": 7.866809211760296e-07,
+      "loss": 0.718,
+      "step": 31922
+    },
+    {
+      "epoch": 0.8767405454395649,
+      "grad_norm": 0.0,
+      "learning_rate": 7.863351343053005e-07,
+      "loss": 0.8279,
+      "step": 31923
+    },
+    {
+      "epoch": 0.8767680096674082,
+      "grad_norm": 0.0,
+      "learning_rate": 7.859894203366548e-07,
+      "loss": 0.7768,
+      "step": 31924
+    },
+    {
+      "epoch": 0.8767954738952515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.85643779272831e-07,
+      "loss": 0.7428,
+      "step": 31925
+    },
+    {
+      "epoch": 0.8768229381230946,
+      "grad_norm": 0.0,
+      "learning_rate": 7.852982111165619e-07,
+      "loss": 0.7963,
+      "step": 31926
+    },
+    {
+      "epoch": 0.8768504023509379,
+      "grad_norm": 0.0,
+      "learning_rate": 7.849527158705828e-07,
+      "loss": 0.7631,
+      "step": 31927
+    },
+    {
+      "epoch": 0.8768778665787811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.846072935376281e-07,
+      "loss": 0.8648,
+      "step": 31928
+    },
+    {
+      "epoch": 0.8769053308066244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.842619441204313e-07,
+      "loss": 0.7288,
+      "step": 31929
+    },
+    {
+      "epoch": 0.8769327950344676,
+      "grad_norm": 0.0,
+      "learning_rate": 7.839166676217224e-07,
+      "loss": 0.8039,
+      "step": 31930
+    },
+    {
+      "epoch": 0.8769602592623108,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83571464044236e-07,
+      "loss": 0.758,
+      "step": 31931
+    },
+    {
+      "epoch": 0.8769877234901541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.832263333907008e-07,
+      "loss": 0.7942,
+      "step": 31932
+    },
+    {
+      "epoch": 0.8770151877179974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.828812756638493e-07,
+      "loss": 0.8949,
+      "step": 31933
+    },
+    {
+      "epoch": 0.8770426519458405,
+      "grad_norm": 0.0,
+      "learning_rate": 7.825362908664125e-07,
+      "loss": 0.8611,
+      "step": 31934
+    },
+    {
+      "epoch": 0.8770701161736838,
+      "grad_norm": 0.0,
+      "learning_rate": 7.821913790011193e-07,
+      "loss": 0.7888,
+      "step": 31935
+    },
+    {
+      "epoch": 0.877097580401527,
+      "grad_norm": 0.0,
+      "learning_rate": 7.818465400706976e-07,
+      "loss": 0.7344,
+      "step": 31936
+    },
+    {
+      "epoch": 0.8771250446293702,
+      "grad_norm": 0.0,
+      "learning_rate": 7.815017740778763e-07,
+      "loss": 0.8296,
+      "step": 31937
+    },
+    {
+      "epoch": 0.8771525088572135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.811570810253855e-07,
+      "loss": 0.7817,
+      "step": 31938
+    },
+    {
+      "epoch": 0.8771799730850567,
+      "grad_norm": 0.0,
+      "learning_rate": 7.808124609159496e-07,
+      "loss": 0.8347,
+      "step": 31939
+    },
+    {
+      "epoch": 0.8772074373129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.804679137522974e-07,
+      "loss": 0.7915,
+      "step": 31940
+    },
+    {
+      "epoch": 0.8772349015407431,
+      "grad_norm": 0.0,
+      "learning_rate": 7.801234395371549e-07,
+      "loss": 0.7291,
+      "step": 31941
+    },
+    {
+      "epoch": 0.8772623657685864,
+      "grad_norm": 0.0,
+      "learning_rate": 7.797790382732462e-07,
+      "loss": 0.7948,
+      "step": 31942
+    },
+    {
+      "epoch": 0.8772898299964297,
+      "grad_norm": 0.0,
+      "learning_rate": 7.79434709963297e-07,
+      "loss": 0.7726,
+      "step": 31943
+    },
+    {
+      "epoch": 0.8773172942242728,
+      "grad_norm": 0.0,
+      "learning_rate": 7.790904546100319e-07,
+      "loss": 0.8051,
+      "step": 31944
+    },
+    {
+      "epoch": 0.8773447584521161,
+      "grad_norm": 0.0,
+      "learning_rate": 7.787462722161754e-07,
+      "loss": 0.8029,
+      "step": 31945
+    },
+    {
+      "epoch": 0.8773722226799594,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784021627844518e-07,
+      "loss": 0.7763,
+      "step": 31946
+    },
+    {
+      "epoch": 0.8773996869078026,
+      "grad_norm": 0.0,
+      "learning_rate": 7.780581263175823e-07,
+      "loss": 0.7399,
+      "step": 31947
+    },
+    {
+      "epoch": 0.8774271511356458,
+      "grad_norm": 0.0,
+      "learning_rate": 7.777141628182872e-07,
+      "loss": 0.8362,
+      "step": 31948
+    },
+    {
+      "epoch": 0.877454615363489,
+      "grad_norm": 0.0,
+      "learning_rate": 7.773702722892907e-07,
+      "loss": 0.8619,
+      "step": 31949
+    },
+    {
+      "epoch": 0.8774820795913323,
+      "grad_norm": 0.0,
+      "learning_rate": 7.770264547333128e-07,
+      "loss": 0.874,
+      "step": 31950
+    },
+    {
+      "epoch": 0.8775095438191756,
+      "grad_norm": 0.0,
+      "learning_rate": 7.766827101530739e-07,
+      "loss": 0.8543,
+      "step": 31951
+    },
+    {
+      "epoch": 0.8775370080470187,
+      "grad_norm": 0.0,
+      "learning_rate": 7.763390385512959e-07,
+      "loss": 0.8525,
+      "step": 31952
+    },
+    {
+      "epoch": 0.877564472274862,
+      "grad_norm": 0.0,
+      "learning_rate": 7.759954399306957e-07,
+      "loss": 0.7866,
+      "step": 31953
+    },
+    {
+      "epoch": 0.8775919365027052,
+      "grad_norm": 0.0,
+      "learning_rate": 7.756519142939912e-07,
+      "loss": 0.8336,
+      "step": 31954
+    },
+    {
+      "epoch": 0.8776194007305484,
+      "grad_norm": 0.0,
+      "learning_rate": 7.753084616439044e-07,
+      "loss": 0.8084,
+      "step": 31955
+    },
+    {
+      "epoch": 0.8776468649583917,
+      "grad_norm": 0.0,
+      "learning_rate": 7.749650819831489e-07,
+      "loss": 0.9005,
+      "step": 31956
+    },
+    {
+      "epoch": 0.8776743291862349,
+      "grad_norm": 0.0,
+      "learning_rate": 7.746217753144425e-07,
+      "loss": 0.8406,
+      "step": 31957
+    },
+    {
+      "epoch": 0.8777017934140782,
+      "grad_norm": 0.0,
+      "learning_rate": 7.742785416405041e-07,
+      "loss": 0.7752,
+      "step": 31958
+    },
+    {
+      "epoch": 0.8777292576419214,
+      "grad_norm": 0.0,
+      "learning_rate": 7.739353809640459e-07,
+      "loss": 0.8317,
+      "step": 31959
+    },
+    {
+      "epoch": 0.8777567218697646,
+      "grad_norm": 0.0,
+      "learning_rate": 7.735922932877848e-07,
+      "loss": 0.7531,
+      "step": 31960
+    },
+    {
+      "epoch": 0.8777841860976079,
+      "grad_norm": 0.0,
+      "learning_rate": 7.732492786144353e-07,
+      "loss": 0.8893,
+      "step": 31961
+    },
+    {
+      "epoch": 0.877811650325451,
+      "grad_norm": 0.0,
+      "learning_rate": 7.729063369467138e-07,
+      "loss": 0.7953,
+      "step": 31962
+    },
+    {
+      "epoch": 0.8778391145532943,
+      "grad_norm": 0.0,
+      "learning_rate": 7.725634682873295e-07,
+      "loss": 0.7478,
+      "step": 31963
+    },
+    {
+      "epoch": 0.8778665787811376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.722206726389991e-07,
+      "loss": 0.7887,
+      "step": 31964
+    },
+    {
+      "epoch": 0.8778940430089808,
+      "grad_norm": 0.0,
+      "learning_rate": 7.718779500044316e-07,
+      "loss": 0.7901,
+      "step": 31965
+    },
+    {
+      "epoch": 0.877921507236824,
+      "grad_norm": 0.0,
+      "learning_rate": 7.715353003863402e-07,
+      "loss": 0.8674,
+      "step": 31966
+    },
+    {
+      "epoch": 0.8779489714646672,
+      "grad_norm": 0.0,
+      "learning_rate": 7.711927237874361e-07,
+      "loss": 0.8962,
+      "step": 31967
+    },
+    {
+      "epoch": 0.8779764356925105,
+      "grad_norm": 0.0,
+      "learning_rate": 7.708502202104307e-07,
+      "loss": 0.8045,
+      "step": 31968
+    },
+    {
+      "epoch": 0.8780038999203538,
+      "grad_norm": 0.0,
+      "learning_rate": 7.705077896580326e-07,
+      "loss": 0.8841,
+      "step": 31969
+    },
+    {
+      "epoch": 0.8780313641481969,
+      "grad_norm": 0.0,
+      "learning_rate": 7.701654321329543e-07,
+      "loss": 0.703,
+      "step": 31970
+    },
+    {
+      "epoch": 0.8780588283760402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.698231476379026e-07,
+      "loss": 0.8634,
+      "step": 31971
+    },
+    {
+      "epoch": 0.8780862926038835,
+      "grad_norm": 0.0,
+      "learning_rate": 7.69480936175584e-07,
+      "loss": 0.8022,
+      "step": 31972
+    },
+    {
+      "epoch": 0.8781137568317267,
+      "grad_norm": 0.0,
+      "learning_rate": 7.691387977487075e-07,
+      "loss": 0.7525,
+      "step": 31973
+    },
+    {
+      "epoch": 0.8781412210595699,
+      "grad_norm": 0.0,
+      "learning_rate": 7.687967323599798e-07,
+      "loss": 0.7477,
+      "step": 31974
+    },
+    {
+      "epoch": 0.8781686852874131,
+      "grad_norm": 0.0,
+      "learning_rate": 7.684547400121089e-07,
+      "loss": 0.7287,
+      "step": 31975
+    },
+    {
+      "epoch": 0.8781961495152564,
+      "grad_norm": 0.0,
+      "learning_rate": 7.681128207078015e-07,
+      "loss": 0.8002,
+      "step": 31976
+    },
+    {
+      "epoch": 0.8782236137430997,
+      "grad_norm": 0.0,
+      "learning_rate": 7.677709744497608e-07,
+      "loss": 0.7631,
+      "step": 31977
+    },
+    {
+      "epoch": 0.8782510779709428,
+      "grad_norm": 0.0,
+      "learning_rate": 7.674292012406914e-07,
+      "loss": 0.7941,
+      "step": 31978
+    },
+    {
+      "epoch": 0.8782785421987861,
+      "grad_norm": 0.0,
+      "learning_rate": 7.670875010833001e-07,
+      "loss": 0.8855,
+      "step": 31979
+    },
+    {
+      "epoch": 0.8783060064266293,
+      "grad_norm": 0.0,
+      "learning_rate": 7.66745873980288e-07,
+      "loss": 0.7328,
+      "step": 31980
+    },
+    {
+      "epoch": 0.8783334706544725,
+      "grad_norm": 0.0,
+      "learning_rate": 7.664043199343585e-07,
+      "loss": 0.8333,
+      "step": 31981
+    },
+    {
+      "epoch": 0.8783609348823158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.660628389482172e-07,
+      "loss": 0.8187,
+      "step": 31982
+    },
+    {
+      "epoch": 0.878388399110159,
+      "grad_norm": 0.0,
+      "learning_rate": 7.657214310245608e-07,
+      "loss": 0.7613,
+      "step": 31983
+    },
+    {
+      "epoch": 0.8784158633380023,
+      "grad_norm": 0.0,
+      "learning_rate": 7.65380096166094e-07,
+      "loss": 0.8509,
+      "step": 31984
+    },
+    {
+      "epoch": 0.8784433275658455,
+      "grad_norm": 0.0,
+      "learning_rate": 7.650388343755177e-07,
+      "loss": 0.6997,
+      "step": 31985
+    },
+    {
+      "epoch": 0.8784707917936887,
+      "grad_norm": 0.0,
+      "learning_rate": 7.646976456555321e-07,
+      "loss": 0.8324,
+      "step": 31986
+    },
+    {
+      "epoch": 0.878498256021532,
+      "grad_norm": 0.0,
+      "learning_rate": 7.643565300088351e-07,
+      "loss": 0.8968,
+      "step": 31987
+    },
+    {
+      "epoch": 0.8785257202493751,
+      "grad_norm": 0.0,
+      "learning_rate": 7.640154874381278e-07,
+      "loss": 0.8227,
+      "step": 31988
+    },
+    {
+      "epoch": 0.8785531844772184,
+      "grad_norm": 0.0,
+      "learning_rate": 7.636745179461069e-07,
+      "loss": 0.7618,
+      "step": 31989
+    },
+    {
+      "epoch": 0.8785806487050617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.633336215354704e-07,
+      "loss": 0.7739,
+      "step": 31990
+    },
+    {
+      "epoch": 0.8786081129329049,
+      "grad_norm": 0.0,
+      "learning_rate": 7.629927982089158e-07,
+      "loss": 0.7437,
+      "step": 31991
+    },
+    {
+      "epoch": 0.8786355771607481,
+      "grad_norm": 0.0,
+      "learning_rate": 7.626520479691412e-07,
+      "loss": 0.7623,
+      "step": 31992
+    },
+    {
+      "epoch": 0.8786630413885914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.623113708188412e-07,
+      "loss": 0.8485,
+      "step": 31993
+    },
+    {
+      "epoch": 0.8786905056164346,
+      "grad_norm": 0.0,
+      "learning_rate": 7.619707667607146e-07,
+      "loss": 0.9051,
+      "step": 31994
+    },
+    {
+      "epoch": 0.8787179698442779,
+      "grad_norm": 0.0,
+      "learning_rate": 7.616302357974525e-07,
+      "loss": 0.7306,
+      "step": 31995
+    },
+    {
+      "epoch": 0.878745434072121,
+      "grad_norm": 0.0,
+      "learning_rate": 7.612897779317496e-07,
+      "loss": 0.7324,
+      "step": 31996
+    },
+    {
+      "epoch": 0.8787728982999643,
+      "grad_norm": 0.0,
+      "learning_rate": 7.609493931663003e-07,
+      "loss": 0.7266,
+      "step": 31997
+    },
+    {
+      "epoch": 0.8788003625278076,
+      "grad_norm": 0.0,
+      "learning_rate": 7.606090815037992e-07,
+      "loss": 0.8009,
+      "step": 31998
+    },
+    {
+      "epoch": 0.8788278267556507,
+      "grad_norm": 0.0,
+      "learning_rate": 7.602688429469374e-07,
+      "loss": 0.7743,
+      "step": 31999
+    },
+    {
+      "epoch": 0.878855290983494,
+      "grad_norm": 0.0,
+      "learning_rate": 7.599286774984094e-07,
+      "loss": 0.8224,
+      "step": 32000
+    },
+    {
+      "epoch": 0.8788827552113372,
+      "grad_norm": 0.0,
+      "learning_rate": 7.595885851609031e-07,
+      "loss": 0.8646,
+      "step": 32001
+    },
+    {
+      "epoch": 0.8789102194391805,
+      "grad_norm": 0.0,
+      "learning_rate": 7.592485659371107e-07,
+      "loss": 0.7518,
+      "step": 32002
+    },
+    {
+      "epoch": 0.8789376836670237,
+      "grad_norm": 0.0,
+      "learning_rate": 7.589086198297257e-07,
+      "loss": 0.7516,
+      "step": 32003
+    },
+    {
+      "epoch": 0.8789651478948669,
+      "grad_norm": 0.0,
+      "learning_rate": 7.585687468414338e-07,
+      "loss": 0.907,
+      "step": 32004
+    },
+    {
+      "epoch": 0.8789926121227102,
+      "grad_norm": 0.0,
+      "learning_rate": 7.582289469749248e-07,
+      "loss": 0.8804,
+      "step": 32005
+    },
+    {
+      "epoch": 0.8790200763505535,
+      "grad_norm": 0.0,
+      "learning_rate": 7.578892202328913e-07,
+      "loss": 0.7807,
+      "step": 32006
+    },
+    {
+      "epoch": 0.8790475405783966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.575495666180155e-07,
+      "loss": 0.8148,
+      "step": 32007
+    },
+    {
+      "epoch": 0.8790750048062399,
+      "grad_norm": 0.0,
+      "learning_rate": 7.572099861329885e-07,
+      "loss": 0.7191,
+      "step": 32008
+    },
+    {
+      "epoch": 0.8791024690340831,
+      "grad_norm": 0.0,
+      "learning_rate": 7.568704787804959e-07,
+      "loss": 0.8843,
+      "step": 32009
+    },
+    {
+      "epoch": 0.8791299332619263,
+      "grad_norm": 0.0,
+      "learning_rate": 7.565310445632268e-07,
+      "loss": 0.7685,
+      "step": 32010
+    },
+    {
+      "epoch": 0.8791573974897696,
+      "grad_norm": 0.0,
+      "learning_rate": 7.561916834838623e-07,
+      "loss": 0.8462,
+      "step": 32011
+    },
+    {
+      "epoch": 0.8791848617176128,
+      "grad_norm": 0.0,
+      "learning_rate": 7.558523955450914e-07,
+      "loss": 0.8704,
+      "step": 32012
+    },
+    {
+      "epoch": 0.8792123259454561,
+      "grad_norm": 0.0,
+      "learning_rate": 7.555131807495964e-07,
+      "loss": 0.8053,
+      "step": 32013
+    },
+    {
+      "epoch": 0.8792397901732992,
+      "grad_norm": 0.0,
+      "learning_rate": 7.551740391000617e-07,
+      "loss": 0.8076,
+      "step": 32014
+    },
+    {
+      "epoch": 0.8792672544011425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.548349705991709e-07,
+      "loss": 0.8066,
+      "step": 32015
+    },
+    {
+      "epoch": 0.8792947186289858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.544959752496073e-07,
+      "loss": 0.8088,
+      "step": 32016
+    },
+    {
+      "epoch": 0.879322182856829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.541570530540543e-07,
+      "loss": 0.7818,
+      "step": 32017
+    },
+    {
+      "epoch": 0.8793496470846722,
+      "grad_norm": 0.0,
+      "learning_rate": 7.538182040151909e-07,
+      "loss": 0.7568,
+      "step": 32018
+    },
+    {
+      "epoch": 0.8793771113125155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.534794281357017e-07,
+      "loss": 0.7739,
+      "step": 32019
+    },
+    {
+      "epoch": 0.8794045755403587,
+      "grad_norm": 0.0,
+      "learning_rate": 7.531407254182621e-07,
+      "loss": 0.8678,
+      "step": 32020
+    },
+    {
+      "epoch": 0.879432039768202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.528020958655558e-07,
+      "loss": 0.8018,
+      "step": 32021
+    },
+    {
+      "epoch": 0.8794595039960451,
+      "grad_norm": 0.0,
+      "learning_rate": 7.524635394802605e-07,
+      "loss": 0.78,
+      "step": 32022
+    },
+    {
+      "epoch": 0.8794869682238884,
+      "grad_norm": 0.0,
+      "learning_rate": 7.521250562650584e-07,
+      "loss": 0.8407,
+      "step": 32023
+    },
+    {
+      "epoch": 0.8795144324517317,
+      "grad_norm": 0.0,
+      "learning_rate": 7.517866462226231e-07,
+      "loss": 0.8126,
+      "step": 32024
+    },
+    {
+      "epoch": 0.8795418966795748,
+      "grad_norm": 0.0,
+      "learning_rate": 7.514483093556346e-07,
+      "loss": 0.8064,
+      "step": 32025
+    },
+    {
+      "epoch": 0.8795693609074181,
+      "grad_norm": 0.0,
+      "learning_rate": 7.511100456667686e-07,
+      "loss": 0.8418,
+      "step": 32026
+    },
+    {
+      "epoch": 0.8795968251352613,
+      "grad_norm": 0.0,
+      "learning_rate": 7.507718551587051e-07,
+      "loss": 0.8148,
+      "step": 32027
+    },
+    {
+      "epoch": 0.8796242893631046,
+      "grad_norm": 0.0,
+      "learning_rate": 7.504337378341165e-07,
+      "loss": 0.8214,
+      "step": 32028
+    },
+    {
+      "epoch": 0.8796517535909478,
+      "grad_norm": 0.0,
+      "learning_rate": 7.500956936956793e-07,
+      "loss": 0.8832,
+      "step": 32029
+    },
+    {
+      "epoch": 0.879679217818791,
+      "grad_norm": 0.0,
+      "learning_rate": 7.497577227460673e-07,
+      "loss": 0.7426,
+      "step": 32030
+    },
+    {
+      "epoch": 0.8797066820466343,
+      "grad_norm": 0.0,
+      "learning_rate": 7.494198249879548e-07,
+      "loss": 0.7617,
+      "step": 32031
+    },
+    {
+      "epoch": 0.8797341462744775,
+      "grad_norm": 0.0,
+      "learning_rate": 7.490820004240162e-07,
+      "loss": 0.8322,
+      "step": 32032
+    },
+    {
+      "epoch": 0.8797616105023207,
+      "grad_norm": 0.0,
+      "learning_rate": 7.487442490569252e-07,
+      "loss": 0.8012,
+      "step": 32033
+    },
+    {
+      "epoch": 0.879789074730164,
+      "grad_norm": 0.0,
+      "learning_rate": 7.48406570889354e-07,
+      "loss": 0.8674,
+      "step": 32034
+    },
+    {
+      "epoch": 0.8798165389580072,
+      "grad_norm": 0.0,
+      "learning_rate": 7.480689659239714e-07,
+      "loss": 0.8614,
+      "step": 32035
+    },
+    {
+      "epoch": 0.8798440031858504,
+      "grad_norm": 0.0,
+      "learning_rate": 7.477314341634545e-07,
+      "loss": 0.788,
+      "step": 32036
+    },
+    {
+      "epoch": 0.8798714674136937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.473939756104675e-07,
+      "loss": 0.7948,
+      "step": 32037
+    },
+    {
+      "epoch": 0.8798989316415369,
+      "grad_norm": 0.0,
+      "learning_rate": 7.470565902676841e-07,
+      "loss": 0.7375,
+      "step": 32038
+    },
+    {
+      "epoch": 0.8799263958693802,
+      "grad_norm": 0.0,
+      "learning_rate": 7.46719278137773e-07,
+      "loss": 0.7831,
+      "step": 32039
+    },
+    {
+      "epoch": 0.8799538600972233,
+      "grad_norm": 0.0,
+      "learning_rate": 7.463820392234022e-07,
+      "loss": 0.9138,
+      "step": 32040
+    },
+    {
+      "epoch": 0.8799813243250666,
+      "grad_norm": 0.0,
+      "learning_rate": 7.460448735272441e-07,
+      "loss": 0.7284,
+      "step": 32041
+    },
+    {
+      "epoch": 0.8800087885529099,
+      "grad_norm": 0.0,
+      "learning_rate": 7.457077810519619e-07,
+      "loss": 0.8568,
+      "step": 32042
+    },
+    {
+      "epoch": 0.880036252780753,
+      "grad_norm": 0.0,
+      "learning_rate": 7.453707618002259e-07,
+      "loss": 0.8038,
+      "step": 32043
+    },
+    {
+      "epoch": 0.8800637170085963,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450338157747006e-07,
+      "loss": 0.8835,
+      "step": 32044
+    },
+    {
+      "epoch": 0.8800911812364396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.446969429780515e-07,
+      "loss": 0.8097,
+      "step": 32045
+    },
+    {
+      "epoch": 0.8801186454642828,
+      "grad_norm": 0.0,
+      "learning_rate": 7.443601434129455e-07,
+      "loss": 0.9132,
+      "step": 32046
+    },
+    {
+      "epoch": 0.880146109692126,
+      "grad_norm": 0.0,
+      "learning_rate": 7.440234170820492e-07,
+      "loss": 0.7506,
+      "step": 32047
+    },
+    {
+      "epoch": 0.8801735739199692,
+      "grad_norm": 0.0,
+      "learning_rate": 7.43686763988024e-07,
+      "loss": 0.9066,
+      "step": 32048
+    },
+    {
+      "epoch": 0.8802010381478125,
+      "grad_norm": 0.0,
+      "learning_rate": 7.433501841335344e-07,
+      "loss": 0.6926,
+      "step": 32049
+    },
+    {
+      "epoch": 0.8802285023756558,
+      "grad_norm": 0.0,
+      "learning_rate": 7.430136775212437e-07,
+      "loss": 0.8344,
+      "step": 32050
+    },
+    {
+      "epoch": 0.8802559666034989,
+      "grad_norm": 0.0,
+      "learning_rate": 7.426772441538155e-07,
+      "loss": 0.7878,
+      "step": 32051
+    },
+    {
+      "epoch": 0.8802834308313422,
+      "grad_norm": 0.0,
+      "learning_rate": 7.423408840339108e-07,
+      "loss": 0.7887,
+      "step": 32052
+    },
+    {
+      "epoch": 0.8803108950591854,
+      "grad_norm": 0.0,
+      "learning_rate": 7.42004597164191e-07,
+      "loss": 0.8884,
+      "step": 32053
+    },
+    {
+      "epoch": 0.8803383592870286,
+      "grad_norm": 0.0,
+      "learning_rate": 7.416683835473171e-07,
+      "loss": 0.861,
+      "step": 32054
+    },
+    {
+      "epoch": 0.8803658235148719,
+      "grad_norm": 0.0,
+      "learning_rate": 7.413322431859482e-07,
+      "loss": 0.8577,
+      "step": 32055
+    },
+    {
+      "epoch": 0.8803932877427151,
+      "grad_norm": 0.0,
+      "learning_rate": 7.409961760827444e-07,
+      "loss": 0.6882,
+      "step": 32056
+    },
+    {
+      "epoch": 0.8804207519705584,
+      "grad_norm": 0.0,
+      "learning_rate": 7.406601822403669e-07,
+      "loss": 0.7127,
+      "step": 32057
+    },
+    {
+      "epoch": 0.8804482161984016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.403242616614736e-07,
+      "loss": 0.84,
+      "step": 32058
+    },
+    {
+      "epoch": 0.8804756804262448,
+      "grad_norm": 0.0,
+      "learning_rate": 7.399884143487224e-07,
+      "loss": 0.8225,
+      "step": 32059
+    },
+    {
+      "epoch": 0.8805031446540881,
+      "grad_norm": 0.0,
+      "learning_rate": 7.396526403047677e-07,
+      "loss": 0.7866,
+      "step": 32060
+    },
+    {
+      "epoch": 0.8805306088819312,
+      "grad_norm": 0.0,
+      "learning_rate": 7.393169395322686e-07,
+      "loss": 0.8178,
+      "step": 32061
+    },
+    {
+      "epoch": 0.8805580731097745,
+      "grad_norm": 0.0,
+      "learning_rate": 7.389813120338807e-07,
+      "loss": 0.7268,
+      "step": 32062
+    },
+    {
+      "epoch": 0.8805855373376178,
+      "grad_norm": 0.0,
+      "learning_rate": 7.386457578122597e-07,
+      "loss": 0.7573,
+      "step": 32063
+    },
+    {
+      "epoch": 0.880613001565461,
+      "grad_norm": 0.0,
+      "learning_rate": 7.383102768700612e-07,
+      "loss": 0.8708,
+      "step": 32064
+    },
+    {
+      "epoch": 0.8806404657933042,
+      "grad_norm": 0.0,
+      "learning_rate": 7.37974869209942e-07,
+      "loss": 0.7897,
+      "step": 32065
+    },
+    {
+      "epoch": 0.8806679300211474,
+      "grad_norm": 0.0,
+      "learning_rate": 7.376395348345511e-07,
+      "loss": 0.7199,
+      "step": 32066
+    },
+    {
+      "epoch": 0.8806953942489907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.373042737465453e-07,
+      "loss": 0.8195,
+      "step": 32067
+    },
+    {
+      "epoch": 0.880722858476834,
+      "grad_norm": 0.0,
+      "learning_rate": 7.369690859485757e-07,
+      "loss": 0.8171,
+      "step": 32068
+    },
+    {
+      "epoch": 0.8807503227046771,
+      "grad_norm": 0.0,
+      "learning_rate": 7.366339714432935e-07,
+      "loss": 0.8192,
+      "step": 32069
+    },
+    {
+      "epoch": 0.8807777869325204,
+      "grad_norm": 0.0,
+      "learning_rate": 7.362989302333523e-07,
+      "loss": 0.7957,
+      "step": 32070
+    },
+    {
+      "epoch": 0.8808052511603637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.359639623214043e-07,
+      "loss": 0.851,
+      "step": 32071
+    },
+    {
+      "epoch": 0.8808327153882068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.356290677100952e-07,
+      "loss": 0.7885,
+      "step": 32072
+    },
+    {
+      "epoch": 0.8808601796160501,
+      "grad_norm": 0.0,
+      "learning_rate": 7.352942464020785e-07,
+      "loss": 0.6897,
+      "step": 32073
+    },
+    {
+      "epoch": 0.8808876438438933,
+      "grad_norm": 0.0,
+      "learning_rate": 7.349594984000019e-07,
+      "loss": 0.8931,
+      "step": 32074
+    },
+    {
+      "epoch": 0.8809151080717366,
+      "grad_norm": 0.0,
+      "learning_rate": 7.346248237065168e-07,
+      "loss": 0.7236,
+      "step": 32075
+    },
+    {
+      "epoch": 0.8809425722995798,
+      "grad_norm": 0.0,
+      "learning_rate": 7.342902223242698e-07,
+      "loss": 0.8026,
+      "step": 32076
+    },
+    {
+      "epoch": 0.880970036527423,
+      "grad_norm": 0.0,
+      "learning_rate": 7.339556942559056e-07,
+      "loss": 0.83,
+      "step": 32077
+    },
+    {
+      "epoch": 0.8809975007552663,
+      "grad_norm": 0.0,
+      "learning_rate": 7.336212395040731e-07,
+      "loss": 0.9023,
+      "step": 32078
+    },
+    {
+      "epoch": 0.8810249649831094,
+      "grad_norm": 0.0,
+      "learning_rate": 7.332868580714181e-07,
+      "loss": 0.8045,
+      "step": 32079
+    },
+    {
+      "epoch": 0.8810524292109527,
+      "grad_norm": 0.0,
+      "learning_rate": 7.329525499605882e-07,
+      "loss": 0.9673,
+      "step": 32080
+    },
+    {
+      "epoch": 0.881079893438796,
+      "grad_norm": 0.0,
+      "learning_rate": 7.326183151742272e-07,
+      "loss": 0.9001,
+      "step": 32081
+    },
+    {
+      "epoch": 0.8811073576666392,
+      "grad_norm": 0.0,
+      "learning_rate": 7.322841537149816e-07,
+      "loss": 0.706,
+      "step": 32082
+    },
+    {
+      "epoch": 0.8811348218944824,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319500655854927e-07,
+      "loss": 0.8622,
+      "step": 32083
+    },
+    {
+      "epoch": 0.8811622861223257,
+      "grad_norm": 0.0,
+      "learning_rate": 7.316160507884041e-07,
+      "loss": 0.8544,
+      "step": 32084
+    },
+    {
+      "epoch": 0.8811897503501689,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3128210932636e-07,
+      "loss": 0.8893,
+      "step": 32085
+    },
+    {
+      "epoch": 0.8812172145780122,
+      "grad_norm": 0.0,
+      "learning_rate": 7.309482412020019e-07,
+      "loss": 0.7745,
+      "step": 32086
+    },
+    {
+      "epoch": 0.8812446788058553,
+      "grad_norm": 0.0,
+      "learning_rate": 7.30614446417972e-07,
+      "loss": 0.7353,
+      "step": 32087
+    },
+    {
+      "epoch": 0.8812721430336986,
+      "grad_norm": 0.0,
+      "learning_rate": 7.302807249769117e-07,
+      "loss": 0.777,
+      "step": 32088
+    },
+    {
+      "epoch": 0.8812996072615419,
+      "grad_norm": 0.0,
+      "learning_rate": 7.299470768814609e-07,
+      "loss": 0.879,
+      "step": 32089
+    },
+    {
+      "epoch": 0.881327071489385,
+      "grad_norm": 0.0,
+      "learning_rate": 7.296135021342588e-07,
+      "loss": 0.8606,
+      "step": 32090
+    },
+    {
+      "epoch": 0.8813545357172283,
+      "grad_norm": 0.0,
+      "learning_rate": 7.292800007379475e-07,
+      "loss": 0.8425,
+      "step": 32091
+    },
+    {
+      "epoch": 0.8813819999450715,
+      "grad_norm": 0.0,
+      "learning_rate": 7.289465726951628e-07,
+      "loss": 0.7934,
+      "step": 32092
+    },
+    {
+      "epoch": 0.8814094641729148,
+      "grad_norm": 0.0,
+      "learning_rate": 7.286132180085437e-07,
+      "loss": 0.817,
+      "step": 32093
+    },
+    {
+      "epoch": 0.881436928400758,
+      "grad_norm": 0.0,
+      "learning_rate": 7.282799366807302e-07,
+      "loss": 0.7779,
+      "step": 32094
+    },
+    {
+      "epoch": 0.8814643926286012,
+      "grad_norm": 0.0,
+      "learning_rate": 7.279467287143549e-07,
+      "loss": 0.8235,
+      "step": 32095
+    },
+    {
+      "epoch": 0.8814918568564445,
+      "grad_norm": 0.0,
+      "learning_rate": 7.276135941120577e-07,
+      "loss": 0.7823,
+      "step": 32096
+    },
+    {
+      "epoch": 0.8815193210842878,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272805328764732e-07,
+      "loss": 0.8015,
+      "step": 32097
+    },
+    {
+      "epoch": 0.8815467853121309,
+      "grad_norm": 0.0,
+      "learning_rate": 7.269475450102382e-07,
+      "loss": 0.6723,
+      "step": 32098
+    },
+    {
+      "epoch": 0.8815742495399742,
+      "grad_norm": 0.0,
+      "learning_rate": 7.266146305159849e-07,
+      "loss": 0.8113,
+      "step": 32099
+    },
+    {
+      "epoch": 0.8816017137678174,
+      "grad_norm": 0.0,
+      "learning_rate": 7.262817893963503e-07,
+      "loss": 0.7591,
+      "step": 32100
+    },
+    {
+      "epoch": 0.8816291779956607,
+      "grad_norm": 0.0,
+      "learning_rate": 7.259490216539644e-07,
+      "loss": 0.7681,
+      "step": 32101
+    },
+    {
+      "epoch": 0.8816566422235039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.256163272914629e-07,
+      "loss": 0.8567,
+      "step": 32102
+    },
+    {
+      "epoch": 0.8816841064513471,
+      "grad_norm": 0.0,
+      "learning_rate": 7.252837063114782e-07,
+      "loss": 0.7723,
+      "step": 32103
+    },
+    {
+      "epoch": 0.8817115706791904,
+      "grad_norm": 0.0,
+      "learning_rate": 7.249511587166403e-07,
+      "loss": 0.8211,
+      "step": 32104
+    },
+    {
+      "epoch": 0.8817390349070335,
+      "grad_norm": 0.0,
+      "learning_rate": 7.246186845095826e-07,
+      "loss": 0.7593,
+      "step": 32105
+    },
+    {
+      "epoch": 0.8817664991348768,
+      "grad_norm": 0.0,
+      "learning_rate": 7.242862836929365e-07,
+      "loss": 0.7595,
+      "step": 32106
+    },
+    {
+      "epoch": 0.8817939633627201,
+      "grad_norm": 0.0,
+      "learning_rate": 7.239539562693298e-07,
+      "loss": 0.829,
+      "step": 32107
+    },
+    {
+      "epoch": 0.8818214275905633,
+      "grad_norm": 0.0,
+      "learning_rate": 7.236217022413905e-07,
+      "loss": 0.8234,
+      "step": 32108
+    },
+    {
+      "epoch": 0.8818488918184065,
+      "grad_norm": 0.0,
+      "learning_rate": 7.232895216117508e-07,
+      "loss": 0.7797,
+      "step": 32109
+    },
+    {
+      "epoch": 0.8818763560462498,
+      "grad_norm": 0.0,
+      "learning_rate": 7.229574143830365e-07,
+      "loss": 0.8721,
+      "step": 32110
+    },
+    {
+      "epoch": 0.881903820274093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.226253805578775e-07,
+      "loss": 0.8846,
+      "step": 32111
+    },
+    {
+      "epoch": 0.8819312845019363,
+      "grad_norm": 0.0,
+      "learning_rate": 7.22293420138902e-07,
+      "loss": 0.8657,
+      "step": 32112
+    },
+    {
+      "epoch": 0.8819587487297794,
+      "grad_norm": 0.0,
+      "learning_rate": 7.219615331287333e-07,
+      "loss": 0.86,
+      "step": 32113
+    },
+    {
+      "epoch": 0.8819862129576227,
+      "grad_norm": 0.0,
+      "learning_rate": 7.216297195299982e-07,
+      "loss": 0.8058,
+      "step": 32114
+    },
+    {
+      "epoch": 0.882013677185466,
+      "grad_norm": 0.0,
+      "learning_rate": 7.212979793453257e-07,
+      "loss": 0.7358,
+      "step": 32115
+    },
+    {
+      "epoch": 0.8820411414133091,
+      "grad_norm": 0.0,
+      "learning_rate": 7.209663125773347e-07,
+      "loss": 0.8354,
+      "step": 32116
+    },
+    {
+      "epoch": 0.8820686056411524,
+      "grad_norm": 0.0,
+      "learning_rate": 7.206347192286533e-07,
+      "loss": 0.8024,
+      "step": 32117
+    },
+    {
+      "epoch": 0.8820960698689956,
+      "grad_norm": 0.0,
+      "learning_rate": 7.20303199301906e-07,
+      "loss": 0.8943,
+      "step": 32118
+    },
+    {
+      "epoch": 0.8821235340968389,
+      "grad_norm": 0.0,
+      "learning_rate": 7.19971752799713e-07,
+      "loss": 0.8769,
+      "step": 32119
+    },
+    {
+      "epoch": 0.8821509983246821,
+      "grad_norm": 0.0,
+      "learning_rate": 7.196403797246976e-07,
+      "loss": 0.752,
+      "step": 32120
+    },
+    {
+      "epoch": 0.8821784625525253,
+      "grad_norm": 0.0,
+      "learning_rate": 7.193090800794833e-07,
+      "loss": 0.7746,
+      "step": 32121
+    },
+    {
+      "epoch": 0.8822059267803686,
+      "grad_norm": 0.0,
+      "learning_rate": 7.189778538666914e-07,
+      "loss": 0.7701,
+      "step": 32122
+    },
+    {
+      "epoch": 0.8822333910082119,
+      "grad_norm": 0.0,
+      "learning_rate": 7.186467010889409e-07,
+      "loss": 0.8023,
+      "step": 32123
+    },
+    {
+      "epoch": 0.882260855236055,
+      "grad_norm": 0.0,
+      "learning_rate": 7.183156217488541e-07,
+      "loss": 0.7619,
+      "step": 32124
+    },
+    {
+      "epoch": 0.8822883194638983,
+      "grad_norm": 0.0,
+      "learning_rate": 7.179846158490478e-07,
+      "loss": 0.7892,
+      "step": 32125
+    },
+    {
+      "epoch": 0.8823157836917415,
+      "grad_norm": 0.0,
+      "learning_rate": 7.176536833921433e-07,
+      "loss": 0.8211,
+      "step": 32126
+    },
+    {
+      "epoch": 0.8823432479195847,
+      "grad_norm": 0.0,
+      "learning_rate": 7.173228243807584e-07,
+      "loss": 0.8624,
+      "step": 32127
+    },
+    {
+      "epoch": 0.882370712147428,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1699203881751e-07,
+      "loss": 0.7926,
+      "step": 32128
+    },
+    {
+      "epoch": 0.8823981763752712,
+      "grad_norm": 0.0,
+      "learning_rate": 7.166613267050182e-07,
+      "loss": 0.8766,
+      "step": 32129
+    },
+    {
+      "epoch": 0.8824256406031145,
+      "grad_norm": 0.0,
+      "learning_rate": 7.163306880458987e-07,
+      "loss": 0.8187,
+      "step": 32130
+    },
+    {
+      "epoch": 0.8824531048309576,
+      "grad_norm": 0.0,
+      "learning_rate": 7.160001228427659e-07,
+      "loss": 0.8218,
+      "step": 32131
+    },
+    {
+      "epoch": 0.8824805690588009,
+      "grad_norm": 0.0,
+      "learning_rate": 7.156696310982359e-07,
+      "loss": 0.7197,
+      "step": 32132
+    },
+    {
+      "epoch": 0.8825080332866442,
+      "grad_norm": 0.0,
+      "learning_rate": 7.15339212814925e-07,
+      "loss": 0.7313,
+      "step": 32133
+    },
+    {
+      "epoch": 0.8825354975144873,
+      "grad_norm": 0.0,
+      "learning_rate": 7.150088679954458e-07,
+      "loss": 0.7182,
+      "step": 32134
+    },
+    {
+      "epoch": 0.8825629617423306,
+      "grad_norm": 0.0,
+      "learning_rate": 7.146785966424141e-07,
+      "loss": 0.801,
+      "step": 32135
+    },
+    {
+      "epoch": 0.8825904259701739,
+      "grad_norm": 0.0,
+      "learning_rate": 7.143483987584421e-07,
+      "loss": 0.8377,
+      "step": 32136
+    },
+    {
+      "epoch": 0.8826178901980171,
+      "grad_norm": 0.0,
+      "learning_rate": 7.14018274346141e-07,
+      "loss": 0.8388,
+      "step": 32137
+    },
+    {
+      "epoch": 0.8826453544258603,
+      "grad_norm": 0.0,
+      "learning_rate": 7.136882234081255e-07,
+      "loss": 0.6957,
+      "step": 32138
+    },
+    {
+      "epoch": 0.8826728186537035,
+      "grad_norm": 0.0,
+      "learning_rate": 7.133582459470067e-07,
+      "loss": 0.7427,
+      "step": 32139
+    },
+    {
+      "epoch": 0.8827002828815468,
+      "grad_norm": 0.0,
+      "learning_rate": 7.130283419653938e-07,
+      "loss": 0.8308,
+      "step": 32140
+    },
+    {
+      "epoch": 0.8827277471093901,
+      "grad_norm": 0.0,
+      "learning_rate": 7.126985114658991e-07,
+      "loss": 0.6871,
+      "step": 32141
+    },
+    {
+      "epoch": 0.8827552113372332,
+      "grad_norm": 0.0,
+      "learning_rate": 7.123687544511304e-07,
+      "loss": 0.9133,
+      "step": 32142
+    },
+    {
+      "epoch": 0.8827826755650765,
+      "grad_norm": 0.0,
+      "learning_rate": 7.120390709236968e-07,
+      "loss": 0.813,
+      "step": 32143
+    },
+    {
+      "epoch": 0.8828101397929197,
+      "grad_norm": 0.0,
+      "learning_rate": 7.117094608862086e-07,
+      "loss": 0.776,
+      "step": 32144
+    },
+    {
+      "epoch": 0.882837604020763,
+      "grad_norm": 0.0,
+      "learning_rate": 7.113799243412734e-07,
+      "loss": 0.7845,
+      "step": 32145
+    },
+    {
+      "epoch": 0.8828650682486062,
+      "grad_norm": 0.0,
+      "learning_rate": 7.110504612914992e-07,
+      "loss": 0.7738,
+      "step": 32146
+    },
+    {
+      "epoch": 0.8828925324764494,
+      "grad_norm": 0.0,
+      "learning_rate": 7.107210717394908e-07,
+      "loss": 0.7086,
+      "step": 32147
+    },
+    {
+      "epoch": 0.8829199967042927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.10391755687857e-07,
+      "loss": 0.8586,
+      "step": 32148
+    },
+    {
+      "epoch": 0.8829474609321359,
+      "grad_norm": 0.0,
+      "learning_rate": 7.100625131392003e-07,
+      "loss": 0.8278,
+      "step": 32149
+    },
+    {
+      "epoch": 0.8829749251599791,
+      "grad_norm": 0.0,
+      "learning_rate": 7.097333440961273e-07,
+      "loss": 0.8268,
+      "step": 32150
+    },
+    {
+      "epoch": 0.8830023893878224,
+      "grad_norm": 0.0,
+      "learning_rate": 7.09404248561244e-07,
+      "loss": 0.8242,
+      "step": 32151
+    },
+    {
+      "epoch": 0.8830298536156655,
+      "grad_norm": 0.0,
+      "learning_rate": 7.090752265371525e-07,
+      "loss": 0.8538,
+      "step": 32152
+    },
+    {
+      "epoch": 0.8830573178435088,
+      "grad_norm": 0.0,
+      "learning_rate": 7.087462780264575e-07,
+      "loss": 0.7503,
+      "step": 32153
+    },
+    {
+      "epoch": 0.8830847820713521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.084174030317603e-07,
+      "loss": 0.7621,
+      "step": 32154
+    },
+    {
+      "epoch": 0.8831122462991953,
+      "grad_norm": 0.0,
+      "learning_rate": 7.080886015556654e-07,
+      "loss": 0.7162,
+      "step": 32155
+    },
+    {
+      "epoch": 0.8831397105270385,
+      "grad_norm": 0.0,
+      "learning_rate": 7.077598736007707e-07,
+      "loss": 0.8361,
+      "step": 32156
+    },
+    {
+      "epoch": 0.8831671747548817,
+      "grad_norm": 0.0,
+      "learning_rate": 7.074312191696809e-07,
+      "loss": 0.812,
+      "step": 32157
+    },
+    {
+      "epoch": 0.883194638982725,
+      "grad_norm": 0.0,
+      "learning_rate": 7.071026382649937e-07,
+      "loss": 0.8499,
+      "step": 32158
+    },
+    {
+      "epoch": 0.8832221032105683,
+      "grad_norm": 0.0,
+      "learning_rate": 7.067741308893117e-07,
+      "loss": 0.7583,
+      "step": 32159
+    },
+    {
+      "epoch": 0.8832495674384114,
+      "grad_norm": 0.0,
+      "learning_rate": 7.064456970452305e-07,
+      "loss": 0.8466,
+      "step": 32160
+    },
+    {
+      "epoch": 0.8832770316662547,
+      "grad_norm": 0.0,
+      "learning_rate": 7.061173367353513e-07,
+      "loss": 0.7594,
+      "step": 32161
+    },
+    {
+      "epoch": 0.883304495894098,
+      "grad_norm": 0.0,
+      "learning_rate": 7.057890499622722e-07,
+      "loss": 0.8637,
+      "step": 32162
+    },
+    {
+      "epoch": 0.8833319601219412,
+      "grad_norm": 0.0,
+      "learning_rate": 7.05460836728592e-07,
+      "loss": 0.8302,
+      "step": 32163
+    },
+    {
+      "epoch": 0.8833594243497844,
+      "grad_norm": 0.0,
+      "learning_rate": 7.051326970369043e-07,
+      "loss": 0.8089,
+      "step": 32164
+    },
+    {
+      "epoch": 0.8833868885776276,
+      "grad_norm": 0.0,
+      "learning_rate": 7.048046308898094e-07,
+      "loss": 0.8474,
+      "step": 32165
+    },
+    {
+      "epoch": 0.8834143528054709,
+      "grad_norm": 0.0,
+      "learning_rate": 7.044766382898982e-07,
+      "loss": 0.8698,
+      "step": 32166
+    },
+    {
+      "epoch": 0.8834418170333141,
+      "grad_norm": 0.0,
+      "learning_rate": 7.04148719239769e-07,
+      "loss": 0.803,
+      "step": 32167
+    },
+    {
+      "epoch": 0.8834692812611573,
+      "grad_norm": 0.0,
+      "learning_rate": 7.038208737420161e-07,
+      "loss": 0.8696,
+      "step": 32168
+    },
+    {
+      "epoch": 0.8834967454890006,
+      "grad_norm": 0.0,
+      "learning_rate": 7.034931017992341e-07,
+      "loss": 0.7755,
+      "step": 32169
+    },
+    {
+      "epoch": 0.8835242097168438,
+      "grad_norm": 0.0,
+      "learning_rate": 7.031654034140168e-07,
+      "loss": 0.8117,
+      "step": 32170
+    },
+    {
+      "epoch": 0.883551673944687,
+      "grad_norm": 0.0,
+      "learning_rate": 7.02837778588954e-07,
+      "loss": 0.8244,
+      "step": 32171
+    },
+    {
+      "epoch": 0.8835791381725303,
+      "grad_norm": 0.0,
+      "learning_rate": 7.025102273266415e-07,
+      "loss": 0.7594,
+      "step": 32172
+    },
+    {
+      "epoch": 0.8836066024003735,
+      "grad_norm": 0.0,
+      "learning_rate": 7.021827496296684e-07,
+      "loss": 0.7124,
+      "step": 32173
+    },
+    {
+      "epoch": 0.8836340666282168,
+      "grad_norm": 0.0,
+      "learning_rate": 7.018553455006272e-07,
+      "loss": 0.6667,
+      "step": 32174
+    },
+    {
+      "epoch": 0.88366153085606,
+      "grad_norm": 0.0,
+      "learning_rate": 7.015280149421089e-07,
+      "loss": 0.8682,
+      "step": 32175
+    },
+    {
+      "epoch": 0.8836889950839032,
+      "grad_norm": 0.0,
+      "learning_rate": 7.012007579567015e-07,
+      "loss": 0.8644,
+      "step": 32176
+    },
+    {
+      "epoch": 0.8837164593117465,
+      "grad_norm": 0.0,
+      "learning_rate": 7.008735745469985e-07,
+      "loss": 0.8418,
+      "step": 32177
+    },
+    {
+      "epoch": 0.8837439235395896,
+      "grad_norm": 0.0,
+      "learning_rate": 7.005464647155835e-07,
+      "loss": 0.8932,
+      "step": 32178
+    },
+    {
+      "epoch": 0.8837713877674329,
+      "grad_norm": 0.0,
+      "learning_rate": 7.002194284650488e-07,
+      "loss": 0.6742,
+      "step": 32179
+    },
+    {
+      "epoch": 0.8837988519952762,
+      "grad_norm": 0.0,
+      "learning_rate": 6.998924657979789e-07,
+      "loss": 0.8229,
+      "step": 32180
+    },
+    {
+      "epoch": 0.8838263162231194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.995655767169629e-07,
+      "loss": 0.7891,
+      "step": 32181
+    },
+    {
+      "epoch": 0.8838537804509626,
+      "grad_norm": 0.0,
+      "learning_rate": 6.992387612245866e-07,
+      "loss": 0.8706,
+      "step": 32182
+    },
+    {
+      "epoch": 0.8838812446788059,
+      "grad_norm": 0.0,
+      "learning_rate": 6.989120193234366e-07,
+      "loss": 0.6863,
+      "step": 32183
+    },
+    {
+      "epoch": 0.8839087089066491,
+      "grad_norm": 0.0,
+      "learning_rate": 6.985853510160967e-07,
+      "loss": 0.7458,
+      "step": 32184
+    },
+    {
+      "epoch": 0.8839361731344924,
+      "grad_norm": 0.0,
+      "learning_rate": 6.982587563051523e-07,
+      "loss": 0.7448,
+      "step": 32185
+    },
+    {
+      "epoch": 0.8839636373623355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.979322351931883e-07,
+      "loss": 0.7625,
+      "step": 32186
+    },
+    {
+      "epoch": 0.8839911015901788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.97605787682789e-07,
+      "loss": 0.8627,
+      "step": 32187
+    },
+    {
+      "epoch": 0.8840185658180221,
+      "grad_norm": 0.0,
+      "learning_rate": 6.972794137765337e-07,
+      "loss": 0.7806,
+      "step": 32188
+    },
+    {
+      "epoch": 0.8840460300458652,
+      "grad_norm": 0.0,
+      "learning_rate": 6.969531134770102e-07,
+      "loss": 0.7622,
+      "step": 32189
+    },
+    {
+      "epoch": 0.8840734942737085,
+      "grad_norm": 0.0,
+      "learning_rate": 6.966268867867942e-07,
+      "loss": 0.7049,
+      "step": 32190
+    },
+    {
+      "epoch": 0.8841009585015517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.963007337084715e-07,
+      "loss": 0.8877,
+      "step": 32191
+    },
+    {
+      "epoch": 0.884128422729395,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95974654244621e-07,
+      "loss": 0.9514,
+      "step": 32192
+    },
+    {
+      "epoch": 0.8841558869572382,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95648648397822e-07,
+      "loss": 0.8658,
+      "step": 32193
+    },
+    {
+      "epoch": 0.8841833511850814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.953227161706577e-07,
+      "loss": 0.8214,
+      "step": 32194
+    },
+    {
+      "epoch": 0.8842108154129247,
+      "grad_norm": 0.0,
+      "learning_rate": 6.949968575657029e-07,
+      "loss": 0.8559,
+      "step": 32195
+    },
+    {
+      "epoch": 0.884238279640768,
+      "grad_norm": 0.0,
+      "learning_rate": 6.946710725855388e-07,
+      "loss": 0.711,
+      "step": 32196
+    },
+    {
+      "epoch": 0.8842657438686111,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9434536123274e-07,
+      "loss": 0.7475,
+      "step": 32197
+    },
+    {
+      "epoch": 0.8842932080964544,
+      "grad_norm": 0.0,
+      "learning_rate": 6.940197235098844e-07,
+      "loss": 0.8224,
+      "step": 32198
+    },
+    {
+      "epoch": 0.8843206723242976,
+      "grad_norm": 0.0,
+      "learning_rate": 6.936941594195512e-07,
+      "loss": 0.8214,
+      "step": 32199
+    },
+    {
+      "epoch": 0.8843481365521408,
+      "grad_norm": 0.0,
+      "learning_rate": 6.933686689643149e-07,
+      "loss": 0.8379,
+      "step": 32200
+    },
+    {
+      "epoch": 0.8843756007799841,
+      "grad_norm": 0.0,
+      "learning_rate": 6.930432521467501e-07,
+      "loss": 0.8223,
+      "step": 32201
+    },
+    {
+      "epoch": 0.8844030650078273,
+      "grad_norm": 0.0,
+      "learning_rate": 6.927179089694313e-07,
+      "loss": 0.9007,
+      "step": 32202
+    },
+    {
+      "epoch": 0.8844305292356706,
+      "grad_norm": 0.0,
+      "learning_rate": 6.923926394349367e-07,
+      "loss": 0.8647,
+      "step": 32203
+    },
+    {
+      "epoch": 0.8844579934635137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.920674435458341e-07,
+      "loss": 0.8349,
+      "step": 32204
+    },
+    {
+      "epoch": 0.884485457691357,
+      "grad_norm": 0.0,
+      "learning_rate": 6.917423213047003e-07,
+      "loss": 0.9144,
+      "step": 32205
+    },
+    {
+      "epoch": 0.8845129219192003,
+      "grad_norm": 0.0,
+      "learning_rate": 6.914172727141078e-07,
+      "loss": 0.7157,
+      "step": 32206
+    },
+    {
+      "epoch": 0.8845403861470434,
+      "grad_norm": 0.0,
+      "learning_rate": 6.910922977766266e-07,
+      "loss": 0.8595,
+      "step": 32207
+    },
+    {
+      "epoch": 0.8845678503748867,
+      "grad_norm": 0.0,
+      "learning_rate": 6.907673964948291e-07,
+      "loss": 0.8668,
+      "step": 32208
+    },
+    {
+      "epoch": 0.88459531460273,
+      "grad_norm": 0.0,
+      "learning_rate": 6.904425688712856e-07,
+      "loss": 0.7934,
+      "step": 32209
+    },
+    {
+      "epoch": 0.8846227788305732,
+      "grad_norm": 0.0,
+      "learning_rate": 6.901178149085663e-07,
+      "loss": 0.7267,
+      "step": 32210
+    },
+    {
+      "epoch": 0.8846502430584164,
+      "grad_norm": 0.0,
+      "learning_rate": 6.897931346092435e-07,
+      "loss": 0.8113,
+      "step": 32211
+    },
+    {
+      "epoch": 0.8846777072862596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.894685279758839e-07,
+      "loss": 0.761,
+      "step": 32212
+    },
+    {
+      "epoch": 0.8847051715141029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.891439950110535e-07,
+      "loss": 0.7646,
+      "step": 32213
+    },
+    {
+      "epoch": 0.8847326357419462,
+      "grad_norm": 0.0,
+      "learning_rate": 6.888195357173233e-07,
+      "loss": 0.8107,
+      "step": 32214
+    },
+    {
+      "epoch": 0.8847600999697893,
+      "grad_norm": 0.0,
+      "learning_rate": 6.884951500972592e-07,
+      "loss": 0.7534,
+      "step": 32215
+    },
+    {
+      "epoch": 0.8847875641976326,
+      "grad_norm": 0.0,
+      "learning_rate": 6.881708381534291e-07,
+      "loss": 0.7692,
+      "step": 32216
+    },
+    {
+      "epoch": 0.8848150284254758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.878465998883976e-07,
+      "loss": 0.7881,
+      "step": 32217
+    },
+    {
+      "epoch": 0.884842492653319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.875224353047328e-07,
+      "loss": 0.9211,
+      "step": 32218
+    },
+    {
+      "epoch": 0.8848699568811623,
+      "grad_norm": 0.0,
+      "learning_rate": 6.871983444049979e-07,
+      "loss": 0.8751,
+      "step": 32219
+    },
+    {
+      "epoch": 0.8848974211090055,
+      "grad_norm": 0.0,
+      "learning_rate": 6.868743271917555e-07,
+      "loss": 0.739,
+      "step": 32220
+    },
+    {
+      "epoch": 0.8849248853368488,
+      "grad_norm": 0.0,
+      "learning_rate": 6.865503836675714e-07,
+      "loss": 0.8064,
+      "step": 32221
+    },
+    {
+      "epoch": 0.884952349564692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.862265138350078e-07,
+      "loss": 0.8217,
+      "step": 32222
+    },
+    {
+      "epoch": 0.8849798137925352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.859027176966281e-07,
+      "loss": 0.8186,
+      "step": 32223
+    },
+    {
+      "epoch": 0.8850072780203785,
+      "grad_norm": 0.0,
+      "learning_rate": 6.855789952549951e-07,
+      "loss": 0.8401,
+      "step": 32224
+    },
+    {
+      "epoch": 0.8850347422482217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.852553465126687e-07,
+      "loss": 0.8944,
+      "step": 32225
+    },
+    {
+      "epoch": 0.8850622064760649,
+      "grad_norm": 0.0,
+      "learning_rate": 6.849317714722093e-07,
+      "loss": 0.8126,
+      "step": 32226
+    },
+    {
+      "epoch": 0.8850896707039082,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8460827013618e-07,
+      "loss": 0.8393,
+      "step": 32227
+    },
+    {
+      "epoch": 0.8851171349317514,
+      "grad_norm": 0.0,
+      "learning_rate": 6.842848425071358e-07,
+      "loss": 0.745,
+      "step": 32228
+    },
+    {
+      "epoch": 0.8851445991595946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8396148858764e-07,
+      "loss": 0.7423,
+      "step": 32229
+    },
+    {
+      "epoch": 0.8851720633874378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.836382083802507e-07,
+      "loss": 0.8702,
+      "step": 32230
+    },
+    {
+      "epoch": 0.8851995276152811,
+      "grad_norm": 0.0,
+      "learning_rate": 6.833150018875223e-07,
+      "loss": 0.8595,
+      "step": 32231
+    },
+    {
+      "epoch": 0.8852269918431244,
+      "grad_norm": 0.0,
+      "learning_rate": 6.829918691120152e-07,
+      "loss": 0.8579,
+      "step": 32232
+    },
+    {
+      "epoch": 0.8852544560709675,
+      "grad_norm": 0.0,
+      "learning_rate": 6.826688100562862e-07,
+      "loss": 0.7938,
+      "step": 32233
+    },
+    {
+      "epoch": 0.8852819202988108,
+      "grad_norm": 0.0,
+      "learning_rate": 6.823458247228898e-07,
+      "loss": 0.7633,
+      "step": 32234
+    },
+    {
+      "epoch": 0.8853093845266541,
+      "grad_norm": 0.0,
+      "learning_rate": 6.820229131143852e-07,
+      "loss": 0.6898,
+      "step": 32235
+    },
+    {
+      "epoch": 0.8853368487544973,
+      "grad_norm": 0.0,
+      "learning_rate": 6.817000752333236e-07,
+      "loss": 0.8609,
+      "step": 32236
+    },
+    {
+      "epoch": 0.8853643129823405,
+      "grad_norm": 0.0,
+      "learning_rate": 6.813773110822597e-07,
+      "loss": 0.7851,
+      "step": 32237
+    },
+    {
+      "epoch": 0.8853917772101837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.810546206637469e-07,
+      "loss": 0.8265,
+      "step": 32238
+    },
+    {
+      "epoch": 0.885419241438027,
+      "grad_norm": 0.0,
+      "learning_rate": 6.807320039803412e-07,
+      "loss": 0.7795,
+      "step": 32239
+    },
+    {
+      "epoch": 0.8854467056658702,
+      "grad_norm": 0.0,
+      "learning_rate": 6.804094610345935e-07,
+      "loss": 0.7917,
+      "step": 32240
+    },
+    {
+      "epoch": 0.8854741698937134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.800869918290565e-07,
+      "loss": 0.8286,
+      "step": 32241
+    },
+    {
+      "epoch": 0.8855016341215567,
+      "grad_norm": 0.0,
+      "learning_rate": 6.797645963662825e-07,
+      "loss": 0.7636,
+      "step": 32242
+    },
+    {
+      "epoch": 0.8855290983493999,
+      "grad_norm": 0.0,
+      "learning_rate": 6.794422746488205e-07,
+      "loss": 0.7233,
+      "step": 32243
+    },
+    {
+      "epoch": 0.8855565625772431,
+      "grad_norm": 0.0,
+      "learning_rate": 6.79120026679222e-07,
+      "loss": 0.8162,
+      "step": 32244
+    },
+    {
+      "epoch": 0.8855840268050864,
+      "grad_norm": 0.0,
+      "learning_rate": 6.787978524600347e-07,
+      "loss": 0.7275,
+      "step": 32245
+    },
+    {
+      "epoch": 0.8856114910329296,
+      "grad_norm": 0.0,
+      "learning_rate": 6.784757519938101e-07,
+      "loss": 0.8314,
+      "step": 32246
+    },
+    {
+      "epoch": 0.8856389552607729,
+      "grad_norm": 0.0,
+      "learning_rate": 6.781537252830961e-07,
+      "loss": 0.8106,
+      "step": 32247
+    },
+    {
+      "epoch": 0.8856664194886161,
+      "grad_norm": 0.0,
+      "learning_rate": 6.778317723304429e-07,
+      "loss": 0.7876,
+      "step": 32248
+    },
+    {
+      "epoch": 0.8856938837164593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.775098931383939e-07,
+      "loss": 0.7913,
+      "step": 32249
+    },
+    {
+      "epoch": 0.8857213479443026,
+      "grad_norm": 0.0,
+      "learning_rate": 6.771880877094972e-07,
+      "loss": 0.8414,
+      "step": 32250
+    },
+    {
+      "epoch": 0.8857488121721457,
+      "grad_norm": 0.0,
+      "learning_rate": 6.768663560463017e-07,
+      "loss": 0.737,
+      "step": 32251
+    },
+    {
+      "epoch": 0.885776276399989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.765446981513501e-07,
+      "loss": 0.8586,
+      "step": 32252
+    },
+    {
+      "epoch": 0.8858037406278323,
+      "grad_norm": 0.0,
+      "learning_rate": 6.762231140271891e-07,
+      "loss": 0.7186,
+      "step": 32253
+    },
+    {
+      "epoch": 0.8858312048556755,
+      "grad_norm": 0.0,
+      "learning_rate": 6.75901603676361e-07,
+      "loss": 0.741,
+      "step": 32254
+    },
+    {
+      "epoch": 0.8858586690835187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.755801671014117e-07,
+      "loss": 0.8362,
+      "step": 32255
+    },
+    {
+      "epoch": 0.8858861333113619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.752588043048836e-07,
+      "loss": 0.6991,
+      "step": 32256
+    },
+    {
+      "epoch": 0.8859135975392052,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7493751528932e-07,
+      "loss": 0.7631,
+      "step": 32257
+    },
+    {
+      "epoch": 0.8859410617670485,
+      "grad_norm": 0.0,
+      "learning_rate": 6.746163000572647e-07,
+      "loss": 0.77,
+      "step": 32258
+    },
+    {
+      "epoch": 0.8859685259948916,
+      "grad_norm": 0.0,
+      "learning_rate": 6.742951586112556e-07,
+      "loss": 0.6869,
+      "step": 32259
+    },
+    {
+      "epoch": 0.8859959902227349,
+      "grad_norm": 0.0,
+      "learning_rate": 6.739740909538373e-07,
+      "loss": 0.6565,
+      "step": 32260
+    },
+    {
+      "epoch": 0.8860234544505782,
+      "grad_norm": 0.0,
+      "learning_rate": 6.736530970875476e-07,
+      "loss": 0.8602,
+      "step": 32261
+    },
+    {
+      "epoch": 0.8860509186784213,
+      "grad_norm": 0.0,
+      "learning_rate": 6.733321770149271e-07,
+      "loss": 0.861,
+      "step": 32262
+    },
+    {
+      "epoch": 0.8860783829062646,
+      "grad_norm": 0.0,
+      "learning_rate": 6.730113307385144e-07,
+      "loss": 0.7571,
+      "step": 32263
+    },
+    {
+      "epoch": 0.8861058471341078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7269055826085e-07,
+      "loss": 0.8123,
+      "step": 32264
+    },
+    {
+      "epoch": 0.8861333113619511,
+      "grad_norm": 0.0,
+      "learning_rate": 6.723698595844718e-07,
+      "loss": 0.819,
+      "step": 32265
+    },
+    {
+      "epoch": 0.8861607755897943,
+      "grad_norm": 0.0,
+      "learning_rate": 6.720492347119156e-07,
+      "loss": 0.8204,
+      "step": 32266
+    },
+    {
+      "epoch": 0.8861882398176375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.717286836457193e-07,
+      "loss": 0.7849,
+      "step": 32267
+    },
+    {
+      "epoch": 0.8862157040454808,
+      "grad_norm": 0.0,
+      "learning_rate": 6.714082063884187e-07,
+      "loss": 0.8254,
+      "step": 32268
+    },
+    {
+      "epoch": 0.886243168273324,
+      "grad_norm": 0.0,
+      "learning_rate": 6.710878029425504e-07,
+      "loss": 0.8203,
+      "step": 32269
+    },
+    {
+      "epoch": 0.8862706325011672,
+      "grad_norm": 0.0,
+      "learning_rate": 6.707674733106484e-07,
+      "loss": 0.8446,
+      "step": 32270
+    },
+    {
+      "epoch": 0.8862980967290105,
+      "grad_norm": 0.0,
+      "learning_rate": 6.704472174952492e-07,
+      "loss": 0.8115,
+      "step": 32271
+    },
+    {
+      "epoch": 0.8863255609568537,
+      "grad_norm": 0.0,
+      "learning_rate": 6.701270354988843e-07,
+      "loss": 0.7603,
+      "step": 32272
+    },
+    {
+      "epoch": 0.8863530251846969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.698069273240881e-07,
+      "loss": 0.8138,
+      "step": 32273
+    },
+    {
+      "epoch": 0.8863804894125402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.694868929733944e-07,
+      "loss": 0.8239,
+      "step": 32274
+    },
+    {
+      "epoch": 0.8864079536403834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.691669324493355e-07,
+      "loss": 0.8608,
+      "step": 32275
+    },
+    {
+      "epoch": 0.8864354178682267,
+      "grad_norm": 0.0,
+      "learning_rate": 6.688470457544405e-07,
+      "loss": 0.764,
+      "step": 32276
+    },
+    {
+      "epoch": 0.8864628820960698,
+      "grad_norm": 0.0,
+      "learning_rate": 6.68527232891244e-07,
+      "loss": 0.8227,
+      "step": 32277
+    },
+    {
+      "epoch": 0.8864903463239131,
+      "grad_norm": 0.0,
+      "learning_rate": 6.68207493862274e-07,
+      "loss": 0.7685,
+      "step": 32278
+    },
+    {
+      "epoch": 0.8865178105517564,
+      "grad_norm": 0.0,
+      "learning_rate": 6.678878286700608e-07,
+      "loss": 0.8081,
+      "step": 32279
+    },
+    {
+      "epoch": 0.8865452747795995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.675682373171333e-07,
+      "loss": 0.8726,
+      "step": 32280
+    },
+    {
+      "epoch": 0.8865727390074428,
+      "grad_norm": 0.0,
+      "learning_rate": 6.672487198060207e-07,
+      "loss": 0.8484,
+      "step": 32281
+    },
+    {
+      "epoch": 0.886600203235286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.669292761392543e-07,
+      "loss": 0.7347,
+      "step": 32282
+    },
+    {
+      "epoch": 0.8866276674631293,
+      "grad_norm": 0.0,
+      "learning_rate": 6.666099063193566e-07,
+      "loss": 0.8348,
+      "step": 32283
+    },
+    {
+      "epoch": 0.8866551316909725,
+      "grad_norm": 0.0,
+      "learning_rate": 6.662906103488575e-07,
+      "loss": 0.9074,
+      "step": 32284
+    },
+    {
+      "epoch": 0.8866825959188157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.65971388230282e-07,
+      "loss": 0.8368,
+      "step": 32285
+    },
+    {
+      "epoch": 0.886710060146659,
+      "grad_norm": 0.0,
+      "learning_rate": 6.656522399661557e-07,
+      "loss": 0.8479,
+      "step": 32286
+    },
+    {
+      "epoch": 0.8867375243745023,
+      "grad_norm": 0.0,
+      "learning_rate": 6.653331655590056e-07,
+      "loss": 0.8075,
+      "step": 32287
+    },
+    {
+      "epoch": 0.8867649886023454,
+      "grad_norm": 0.0,
+      "learning_rate": 6.65014165011355e-07,
+      "loss": 0.9029,
+      "step": 32288
+    },
+    {
+      "epoch": 0.8867924528301887,
+      "grad_norm": 0.0,
+      "learning_rate": 6.646952383257299e-07,
+      "loss": 0.7873,
+      "step": 32289
+    },
+    {
+      "epoch": 0.8868199170580319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.643763855046503e-07,
+      "loss": 0.8357,
+      "step": 32290
+    },
+    {
+      "epoch": 0.8868473812858751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.640576065506432e-07,
+      "loss": 0.7991,
+      "step": 32291
+    },
+    {
+      "epoch": 0.8868748455137184,
+      "grad_norm": 0.0,
+      "learning_rate": 6.637389014662266e-07,
+      "loss": 0.7301,
+      "step": 32292
+    },
+    {
+      "epoch": 0.8869023097415616,
+      "grad_norm": 0.0,
+      "learning_rate": 6.634202702539249e-07,
+      "loss": 0.8473,
+      "step": 32293
+    },
+    {
+      "epoch": 0.8869297739694049,
+      "grad_norm": 0.0,
+      "learning_rate": 6.631017129162586e-07,
+      "loss": 0.8759,
+      "step": 32294
+    },
+    {
+      "epoch": 0.886957238197248,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62783229455749e-07,
+      "loss": 0.7565,
+      "step": 32295
+    },
+    {
+      "epoch": 0.8869847024250913,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62464819874915e-07,
+      "loss": 0.7489,
+      "step": 32296
+    },
+    {
+      "epoch": 0.8870121666529346,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62146484176276e-07,
+      "loss": 0.7763,
+      "step": 32297
+    },
+    {
+      "epoch": 0.8870396308807778,
+      "grad_norm": 0.0,
+      "learning_rate": 6.618282223623507e-07,
+      "loss": 0.7619,
+      "step": 32298
+    },
+    {
+      "epoch": 0.887067095108621,
+      "grad_norm": 0.0,
+      "learning_rate": 6.615100344356596e-07,
+      "loss": 0.8439,
+      "step": 32299
+    },
+    {
+      "epoch": 0.8870945593364643,
+      "grad_norm": 0.0,
+      "learning_rate": 6.611919203987172e-07,
+      "loss": 0.8268,
+      "step": 32300
+    },
+    {
+      "epoch": 0.8871220235643075,
+      "grad_norm": 0.0,
+      "learning_rate": 6.608738802540427e-07,
+      "loss": 0.7709,
+      "step": 32301
+    },
+    {
+      "epoch": 0.8871494877921507,
+      "grad_norm": 0.0,
+      "learning_rate": 6.605559140041506e-07,
+      "loss": 0.8051,
+      "step": 32302
+    },
+    {
+      "epoch": 0.8871769520199939,
+      "grad_norm": 0.0,
+      "learning_rate": 6.602380216515581e-07,
+      "loss": 0.729,
+      "step": 32303
+    },
+    {
+      "epoch": 0.8872044162478372,
+      "grad_norm": 0.0,
+      "learning_rate": 6.599202031987806e-07,
+      "loss": 0.7963,
+      "step": 32304
+    },
+    {
+      "epoch": 0.8872318804756805,
+      "grad_norm": 0.0,
+      "learning_rate": 6.596024586483319e-07,
+      "loss": 0.7462,
+      "step": 32305
+    },
+    {
+      "epoch": 0.8872593447035236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.592847880027275e-07,
+      "loss": 0.9088,
+      "step": 32306
+    },
+    {
+      "epoch": 0.8872868089313669,
+      "grad_norm": 0.0,
+      "learning_rate": 6.589671912644791e-07,
+      "loss": 0.8076,
+      "step": 32307
+    },
+    {
+      "epoch": 0.8873142731592101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.586496684361033e-07,
+      "loss": 0.7702,
+      "step": 32308
+    },
+    {
+      "epoch": 0.8873417373870534,
+      "grad_norm": 0.0,
+      "learning_rate": 6.583322195201069e-07,
+      "loss": 0.7504,
+      "step": 32309
+    },
+    {
+      "epoch": 0.8873692016148966,
+      "grad_norm": 0.0,
+      "learning_rate": 6.580148445190049e-07,
+      "loss": 0.8685,
+      "step": 32310
+    },
+    {
+      "epoch": 0.8873966658427398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.576975434353084e-07,
+      "loss": 0.7195,
+      "step": 32311
+    },
+    {
+      "epoch": 0.8874241300705831,
+      "grad_norm": 0.0,
+      "learning_rate": 6.573803162715275e-07,
+      "loss": 0.8576,
+      "step": 32312
+    },
+    {
+      "epoch": 0.8874515942984263,
+      "grad_norm": 0.0,
+      "learning_rate": 6.570631630301738e-07,
+      "loss": 0.7731,
+      "step": 32313
+    },
+    {
+      "epoch": 0.8874790585262695,
+      "grad_norm": 0.0,
+      "learning_rate": 6.567460837137552e-07,
+      "loss": 0.8458,
+      "step": 32314
+    },
+    {
+      "epoch": 0.8875065227541128,
+      "grad_norm": 0.0,
+      "learning_rate": 6.564290783247818e-07,
+      "loss": 0.8439,
+      "step": 32315
+    },
+    {
+      "epoch": 0.887533986981956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.561121468657583e-07,
+      "loss": 0.8308,
+      "step": 32316
+    },
+    {
+      "epoch": 0.8875614512097992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.557952893391961e-07,
+      "loss": 0.7818,
+      "step": 32317
+    },
+    {
+      "epoch": 0.8875889154376425,
+      "grad_norm": 0.0,
+      "learning_rate": 6.55478505747602e-07,
+      "loss": 0.8438,
+      "step": 32318
+    },
+    {
+      "epoch": 0.8876163796654857,
+      "grad_norm": 0.0,
+      "learning_rate": 6.551617960934809e-07,
+      "loss": 0.7512,
+      "step": 32319
+    },
+    {
+      "epoch": 0.887643843893329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.548451603793393e-07,
+      "loss": 0.816,
+      "step": 32320
+    },
+    {
+      "epoch": 0.8876713081211721,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545285986076821e-07,
+      "loss": 0.9161,
+      "step": 32321
+    },
+    {
+      "epoch": 0.8876987723490154,
+      "grad_norm": 0.0,
+      "learning_rate": 6.542121107810151e-07,
+      "loss": 0.824,
+      "step": 32322
+    },
+    {
+      "epoch": 0.8877262365768587,
+      "grad_norm": 0.0,
+      "learning_rate": 6.53895696901844e-07,
+      "loss": 0.8191,
+      "step": 32323
+    },
+    {
+      "epoch": 0.8877537008047018,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5357935697267e-07,
+      "loss": 0.8442,
+      "step": 32324
+    },
+    {
+      "epoch": 0.8877811650325451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.532630909959959e-07,
+      "loss": 0.8243,
+      "step": 32325
+    },
+    {
+      "epoch": 0.8878086292603884,
+      "grad_norm": 0.0,
+      "learning_rate": 6.529468989743237e-07,
+      "loss": 0.7798,
+      "step": 32326
+    },
+    {
+      "epoch": 0.8878360934882316,
+      "grad_norm": 0.0,
+      "learning_rate": 6.526307809101573e-07,
+      "loss": 0.8521,
+      "step": 32327
+    },
+    {
+      "epoch": 0.8878635577160748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.523147368059957e-07,
+      "loss": 0.8763,
+      "step": 32328
+    },
+    {
+      "epoch": 0.887891021943918,
+      "grad_norm": 0.0,
+      "learning_rate": 6.519987666643424e-07,
+      "loss": 0.8058,
+      "step": 32329
+    },
+    {
+      "epoch": 0.8879184861717613,
+      "grad_norm": 0.0,
+      "learning_rate": 6.516828704876965e-07,
+      "loss": 0.7731,
+      "step": 32330
+    },
+    {
+      "epoch": 0.8879459503996046,
+      "grad_norm": 0.0,
+      "learning_rate": 6.51367048278555e-07,
+      "loss": 0.7712,
+      "step": 32331
+    },
+    {
+      "epoch": 0.8879734146274477,
+      "grad_norm": 0.0,
+      "learning_rate": 6.510513000394214e-07,
+      "loss": 0.789,
+      "step": 32332
+    },
+    {
+      "epoch": 0.888000878855291,
+      "grad_norm": 0.0,
+      "learning_rate": 6.507356257727893e-07,
+      "loss": 0.8075,
+      "step": 32333
+    },
+    {
+      "epoch": 0.8880283430831342,
+      "grad_norm": 0.0,
+      "learning_rate": 6.504200254811577e-07,
+      "loss": 0.7666,
+      "step": 32334
+    },
+    {
+      "epoch": 0.8880558073109774,
+      "grad_norm": 0.0,
+      "learning_rate": 6.501044991670257e-07,
+      "loss": 0.8468,
+      "step": 32335
+    },
+    {
+      "epoch": 0.8880832715388207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.497890468328894e-07,
+      "loss": 0.7736,
+      "step": 32336
+    },
+    {
+      "epoch": 0.8881107357666639,
+      "grad_norm": 0.0,
+      "learning_rate": 6.49473668481243e-07,
+      "loss": 0.7767,
+      "step": 32337
+    },
+    {
+      "epoch": 0.8881381999945072,
+      "grad_norm": 0.0,
+      "learning_rate": 6.491583641145838e-07,
+      "loss": 0.8196,
+      "step": 32338
+    },
+    {
+      "epoch": 0.8881656642223504,
+      "grad_norm": 0.0,
+      "learning_rate": 6.488431337354062e-07,
+      "loss": 0.6812,
+      "step": 32339
+    },
+    {
+      "epoch": 0.8881931284501936,
+      "grad_norm": 0.0,
+      "learning_rate": 6.485279773462028e-07,
+      "loss": 0.7632,
+      "step": 32340
+    },
+    {
+      "epoch": 0.8882205926780369,
+      "grad_norm": 0.0,
+      "learning_rate": 6.482128949494681e-07,
+      "loss": 0.8357,
+      "step": 32341
+    },
+    {
+      "epoch": 0.88824805690588,
+      "grad_norm": 0.0,
+      "learning_rate": 6.47897886547697e-07,
+      "loss": 0.8733,
+      "step": 32342
+    },
+    {
+      "epoch": 0.8882755211337233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.475829521433796e-07,
+      "loss": 0.8039,
+      "step": 32343
+    },
+    {
+      "epoch": 0.8883029853615666,
+      "grad_norm": 0.0,
+      "learning_rate": 6.472680917390084e-07,
+      "loss": 0.8421,
+      "step": 32344
+    },
+    {
+      "epoch": 0.8883304495894098,
+      "grad_norm": 0.0,
+      "learning_rate": 6.469533053370747e-07,
+      "loss": 0.8085,
+      "step": 32345
+    },
+    {
+      "epoch": 0.888357913817253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.466385929400698e-07,
+      "loss": 0.7942,
+      "step": 32346
+    },
+    {
+      "epoch": 0.8883853780450962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.463239545504852e-07,
+      "loss": 0.8433,
+      "step": 32347
+    },
+    {
+      "epoch": 0.8884128422729395,
+      "grad_norm": 0.0,
+      "learning_rate": 6.460093901708087e-07,
+      "loss": 0.7908,
+      "step": 32348
+    },
+    {
+      "epoch": 0.8884403065007828,
+      "grad_norm": 0.0,
+      "learning_rate": 6.456948998035273e-07,
+      "loss": 0.8236,
+      "step": 32349
+    },
+    {
+      "epoch": 0.8884677707286259,
+      "grad_norm": 0.0,
+      "learning_rate": 6.453804834511313e-07,
+      "loss": 0.8496,
+      "step": 32350
+    },
+    {
+      "epoch": 0.8884952349564692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.450661411161097e-07,
+      "loss": 0.8093,
+      "step": 32351
+    },
+    {
+      "epoch": 0.8885226991843125,
+      "grad_norm": 0.0,
+      "learning_rate": 6.447518728009472e-07,
+      "loss": 0.7852,
+      "step": 32352
+    },
+    {
+      "epoch": 0.8885501634121556,
+      "grad_norm": 0.0,
+      "learning_rate": 6.444376785081319e-07,
+      "loss": 0.7735,
+      "step": 32353
+    },
+    {
+      "epoch": 0.8885776276399989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.441235582401517e-07,
+      "loss": 0.7708,
+      "step": 32354
+    },
+    {
+      "epoch": 0.8886050918678421,
+      "grad_norm": 0.0,
+      "learning_rate": 6.438095119994892e-07,
+      "loss": 0.7746,
+      "step": 32355
+    },
+    {
+      "epoch": 0.8886325560956854,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4349553978863e-07,
+      "loss": 0.7915,
+      "step": 32356
+    },
+    {
+      "epoch": 0.8886600203235286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.431816416100578e-07,
+      "loss": 0.7891,
+      "step": 32357
+    },
+    {
+      "epoch": 0.8886874845513718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.428678174662573e-07,
+      "loss": 0.7906,
+      "step": 32358
+    },
+    {
+      "epoch": 0.8887149487792151,
+      "grad_norm": 0.0,
+      "learning_rate": 6.425540673597108e-07,
+      "loss": 0.8849,
+      "step": 32359
+    },
+    {
+      "epoch": 0.8887424130070583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.42240391292902e-07,
+      "loss": 0.7599,
+      "step": 32360
+    },
+    {
+      "epoch": 0.8887698772349015,
+      "grad_norm": 0.0,
+      "learning_rate": 6.419267892683112e-07,
+      "loss": 0.7458,
+      "step": 32361
+    },
+    {
+      "epoch": 0.8887973414627448,
+      "grad_norm": 0.0,
+      "learning_rate": 6.416132612884207e-07,
+      "loss": 0.8993,
+      "step": 32362
+    },
+    {
+      "epoch": 0.888824805690588,
+      "grad_norm": 0.0,
+      "learning_rate": 6.41299807355713e-07,
+      "loss": 0.8562,
+      "step": 32363
+    },
+    {
+      "epoch": 0.8888522699184312,
+      "grad_norm": 0.0,
+      "learning_rate": 6.409864274726651e-07,
+      "loss": 0.8268,
+      "step": 32364
+    },
+    {
+      "epoch": 0.8888797341462745,
+      "grad_norm": 0.0,
+      "learning_rate": 6.406731216417572e-07,
+      "loss": 0.813,
+      "step": 32365
+    },
+    {
+      "epoch": 0.8889071983741177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.403598898654705e-07,
+      "loss": 0.757,
+      "step": 32366
+    },
+    {
+      "epoch": 0.888934662601961,
+      "grad_norm": 0.0,
+      "learning_rate": 6.400467321462811e-07,
+      "loss": 0.6872,
+      "step": 32367
+    },
+    {
+      "epoch": 0.8889621268298041,
+      "grad_norm": 0.0,
+      "learning_rate": 6.397336484866668e-07,
+      "loss": 0.7993,
+      "step": 32368
+    },
+    {
+      "epoch": 0.8889895910576474,
+      "grad_norm": 0.0,
+      "learning_rate": 6.394206388891066e-07,
+      "loss": 0.8049,
+      "step": 32369
+    },
+    {
+      "epoch": 0.8890170552854907,
+      "grad_norm": 0.0,
+      "learning_rate": 6.391077033560755e-07,
+      "loss": 0.7472,
+      "step": 32370
+    },
+    {
+      "epoch": 0.8890445195133339,
+      "grad_norm": 0.0,
+      "learning_rate": 6.387948418900524e-07,
+      "loss": 0.8126,
+      "step": 32371
+    },
+    {
+      "epoch": 0.8890719837411771,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3848205449351e-07,
+      "loss": 0.8195,
+      "step": 32372
+    },
+    {
+      "epoch": 0.8890994479690203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.381693411689227e-07,
+      "loss": 0.8227,
+      "step": 32373
+    },
+    {
+      "epoch": 0.8891269121968636,
+      "grad_norm": 0.0,
+      "learning_rate": 6.378567019187665e-07,
+      "loss": 0.8259,
+      "step": 32374
+    },
+    {
+      "epoch": 0.8891543764247068,
+      "grad_norm": 0.0,
+      "learning_rate": 6.375441367455138e-07,
+      "loss": 0.7613,
+      "step": 32375
+    },
+    {
+      "epoch": 0.88918184065255,
+      "grad_norm": 0.0,
+      "learning_rate": 6.372316456516392e-07,
+      "loss": 0.7537,
+      "step": 32376
+    },
+    {
+      "epoch": 0.8892093048803933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.369192286396165e-07,
+      "loss": 0.6844,
+      "step": 32377
+    },
+    {
+      "epoch": 0.8892367691082366,
+      "grad_norm": 0.0,
+      "learning_rate": 6.366068857119135e-07,
+      "loss": 0.8427,
+      "step": 32378
+    },
+    {
+      "epoch": 0.8892642333360797,
+      "grad_norm": 0.0,
+      "learning_rate": 6.362946168710049e-07,
+      "loss": 0.8479,
+      "step": 32379
+    },
+    {
+      "epoch": 0.889291697563923,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3598242211936e-07,
+      "loss": 0.8059,
+      "step": 32380
+    },
+    {
+      "epoch": 0.8893191617917662,
+      "grad_norm": 0.0,
+      "learning_rate": 6.356703014594489e-07,
+      "loss": 0.7901,
+      "step": 32381
+    },
+    {
+      "epoch": 0.8893466260196095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.35358254893742e-07,
+      "loss": 0.7769,
+      "step": 32382
+    },
+    {
+      "epoch": 0.8893740902474527,
+      "grad_norm": 0.0,
+      "learning_rate": 6.350462824247094e-07,
+      "loss": 0.837,
+      "step": 32383
+    },
+    {
+      "epoch": 0.8894015544752959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.34734384054817e-07,
+      "loss": 0.7185,
+      "step": 32384
+    },
+    {
+      "epoch": 0.8894290187031392,
+      "grad_norm": 0.0,
+      "learning_rate": 6.344225597865338e-07,
+      "loss": 0.8583,
+      "step": 32385
+    },
+    {
+      "epoch": 0.8894564829309825,
+      "grad_norm": 0.0,
+      "learning_rate": 6.34110809622327e-07,
+      "loss": 0.8041,
+      "step": 32386
+    },
+    {
+      "epoch": 0.8894839471588256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.337991335646654e-07,
+      "loss": 0.876,
+      "step": 32387
+    },
+    {
+      "epoch": 0.8895114113866689,
+      "grad_norm": 0.0,
+      "learning_rate": 6.334875316160105e-07,
+      "loss": 0.8572,
+      "step": 32388
+    },
+    {
+      "epoch": 0.8895388756145121,
+      "grad_norm": 0.0,
+      "learning_rate": 6.331760037788337e-07,
+      "loss": 0.85,
+      "step": 32389
+    },
+    {
+      "epoch": 0.8895663398423553,
+      "grad_norm": 0.0,
+      "learning_rate": 6.328645500555941e-07,
+      "loss": 0.8771,
+      "step": 32390
+    },
+    {
+      "epoch": 0.8895938040701986,
+      "grad_norm": 0.0,
+      "learning_rate": 6.325531704487586e-07,
+      "loss": 0.7702,
+      "step": 32391
+    },
+    {
+      "epoch": 0.8896212682980418,
+      "grad_norm": 0.0,
+      "learning_rate": 6.32241864960792e-07,
+      "loss": 0.775,
+      "step": 32392
+    },
+    {
+      "epoch": 0.889648732525885,
+      "grad_norm": 0.0,
+      "learning_rate": 6.319306335941555e-07,
+      "loss": 0.7231,
+      "step": 32393
+    },
+    {
+      "epoch": 0.8896761967537282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.316194763513128e-07,
+      "loss": 0.7933,
+      "step": 32394
+    },
+    {
+      "epoch": 0.8897036609815715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.313083932347275e-07,
+      "loss": 0.7935,
+      "step": 32395
+    },
+    {
+      "epoch": 0.8897311252094148,
+      "grad_norm": 0.0,
+      "learning_rate": 6.309973842468597e-07,
+      "loss": 0.8156,
+      "step": 32396
+    },
+    {
+      "epoch": 0.8897585894372579,
+      "grad_norm": 0.0,
+      "learning_rate": 6.306864493901676e-07,
+      "loss": 0.8198,
+      "step": 32397
+    },
+    {
+      "epoch": 0.8897860536651012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.303755886671137e-07,
+      "loss": 0.77,
+      "step": 32398
+    },
+    {
+      "epoch": 0.8898135178929445,
+      "grad_norm": 0.0,
+      "learning_rate": 6.30064802080158e-07,
+      "loss": 0.7871,
+      "step": 32399
+    },
+    {
+      "epoch": 0.8898409821207877,
+      "grad_norm": 0.0,
+      "learning_rate": 6.297540896317589e-07,
+      "loss": 0.6899,
+      "step": 32400
+    },
+    {
+      "epoch": 0.8898684463486309,
+      "grad_norm": 0.0,
+      "learning_rate": 6.294434513243764e-07,
+      "loss": 0.7732,
+      "step": 32401
+    },
+    {
+      "epoch": 0.8898959105764741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.291328871604663e-07,
+      "loss": 0.7092,
+      "step": 32402
+    },
+    {
+      "epoch": 0.8899233748043174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.288223971424878e-07,
+      "loss": 0.8177,
+      "step": 32403
+    },
+    {
+      "epoch": 0.8899508390321607,
+      "grad_norm": 0.0,
+      "learning_rate": 6.285119812728946e-07,
+      "loss": 0.706,
+      "step": 32404
+    },
+    {
+      "epoch": 0.8899783032600038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.282016395541457e-07,
+      "loss": 0.7376,
+      "step": 32405
+    },
+    {
+      "epoch": 0.8900057674878471,
+      "grad_norm": 0.0,
+      "learning_rate": 6.278913719886948e-07,
+      "loss": 0.8059,
+      "step": 32406
+    },
+    {
+      "epoch": 0.8900332317156903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.275811785789998e-07,
+      "loss": 0.853,
+      "step": 32407
+    },
+    {
+      "epoch": 0.8900606959435335,
+      "grad_norm": 0.0,
+      "learning_rate": 6.272710593275111e-07,
+      "loss": 0.7382,
+      "step": 32408
+    },
+    {
+      "epoch": 0.8900881601713768,
+      "grad_norm": 0.0,
+      "learning_rate": 6.269610142366855e-07,
+      "loss": 0.7928,
+      "step": 32409
+    },
+    {
+      "epoch": 0.89011562439922,
+      "grad_norm": 0.0,
+      "learning_rate": 6.266510433089734e-07,
+      "loss": 0.7868,
+      "step": 32410
+    },
+    {
+      "epoch": 0.8901430886270633,
+      "grad_norm": 0.0,
+      "learning_rate": 6.263411465468317e-07,
+      "loss": 0.845,
+      "step": 32411
+    },
+    {
+      "epoch": 0.8901705528549065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.260313239527083e-07,
+      "loss": 0.8322,
+      "step": 32412
+    },
+    {
+      "epoch": 0.8901980170827497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.25721575529058e-07,
+      "loss": 0.7036,
+      "step": 32413
+    },
+    {
+      "epoch": 0.890225481310593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.254119012783278e-07,
+      "loss": 0.773,
+      "step": 32414
+    },
+    {
+      "epoch": 0.8902529455384361,
+      "grad_norm": 0.0,
+      "learning_rate": 6.251023012029712e-07,
+      "loss": 0.8557,
+      "step": 32415
+    },
+    {
+      "epoch": 0.8902804097662794,
+      "grad_norm": 0.0,
+      "learning_rate": 6.24792775305435e-07,
+      "loss": 0.7445,
+      "step": 32416
+    },
+    {
+      "epoch": 0.8903078739941227,
+      "grad_norm": 0.0,
+      "learning_rate": 6.244833235881709e-07,
+      "loss": 0.8008,
+      "step": 32417
+    },
+    {
+      "epoch": 0.8903353382219659,
+      "grad_norm": 0.0,
+      "learning_rate": 6.241739460536278e-07,
+      "loss": 0.7738,
+      "step": 32418
+    },
+    {
+      "epoch": 0.8903628024498091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.238646427042505e-07,
+      "loss": 0.8042,
+      "step": 32419
+    },
+    {
+      "epoch": 0.8903902666776523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.235554135424904e-07,
+      "loss": 0.8195,
+      "step": 32420
+    },
+    {
+      "epoch": 0.8904177309054956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.232462585707899e-07,
+      "loss": 0.8403,
+      "step": 32421
+    },
+    {
+      "epoch": 0.8904451951333389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.229371777915972e-07,
+      "loss": 0.6744,
+      "step": 32422
+    },
+    {
+      "epoch": 0.890472659361182,
+      "grad_norm": 0.0,
+      "learning_rate": 6.226281712073579e-07,
+      "loss": 0.8645,
+      "step": 32423
+    },
+    {
+      "epoch": 0.8905001235890253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.223192388205169e-07,
+      "loss": 0.8533,
+      "step": 32424
+    },
+    {
+      "epoch": 0.8905275878168686,
+      "grad_norm": 0.0,
+      "learning_rate": 6.220103806335199e-07,
+      "loss": 0.8348,
+      "step": 32425
+    },
+    {
+      "epoch": 0.8905550520447117,
+      "grad_norm": 0.0,
+      "learning_rate": 6.217015966488083e-07,
+      "loss": 0.887,
+      "step": 32426
+    },
+    {
+      "epoch": 0.890582516272555,
+      "grad_norm": 0.0,
+      "learning_rate": 6.21392886868828e-07,
+      "loss": 0.8615,
+      "step": 32427
+    },
+    {
+      "epoch": 0.8906099805003982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.21084251296018e-07,
+      "loss": 0.7941,
+      "step": 32428
+    },
+    {
+      "epoch": 0.8906374447282415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.207756899328233e-07,
+      "loss": 0.8674,
+      "step": 32429
+    },
+    {
+      "epoch": 0.8906649089560847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.204672027816838e-07,
+      "loss": 0.8181,
+      "step": 32430
+    },
+    {
+      "epoch": 0.8906923731839279,
+      "grad_norm": 0.0,
+      "learning_rate": 6.201587898450434e-07,
+      "loss": 0.8029,
+      "step": 32431
+    },
+    {
+      "epoch": 0.8907198374117712,
+      "grad_norm": 0.0,
+      "learning_rate": 6.198504511253378e-07,
+      "loss": 0.825,
+      "step": 32432
+    },
+    {
+      "epoch": 0.8907473016396144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.195421866250095e-07,
+      "loss": 0.7957,
+      "step": 32433
+    },
+    {
+      "epoch": 0.8907747658674576,
+      "grad_norm": 0.0,
+      "learning_rate": 6.192339963464966e-07,
+      "loss": 0.8824,
+      "step": 32434
+    },
+    {
+      "epoch": 0.8908022300953009,
+      "grad_norm": 0.0,
+      "learning_rate": 6.189258802922404e-07,
+      "loss": 0.6779,
+      "step": 32435
+    },
+    {
+      "epoch": 0.8908296943231441,
+      "grad_norm": 0.0,
+      "learning_rate": 6.186178384646757e-07,
+      "loss": 0.9058,
+      "step": 32436
+    },
+    {
+      "epoch": 0.8908571585509873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.183098708662394e-07,
+      "loss": 0.7154,
+      "step": 32437
+    },
+    {
+      "epoch": 0.8908846227788306,
+      "grad_norm": 0.0,
+      "learning_rate": 6.180019774993696e-07,
+      "loss": 0.8449,
+      "step": 32438
+    },
+    {
+      "epoch": 0.8909120870066738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.17694158366502e-07,
+      "loss": 0.7739,
+      "step": 32439
+    },
+    {
+      "epoch": 0.8909395512345171,
+      "grad_norm": 0.0,
+      "learning_rate": 6.173864134700724e-07,
+      "loss": 0.8721,
+      "step": 32440
+    },
+    {
+      "epoch": 0.8909670154623602,
+      "grad_norm": 0.0,
+      "learning_rate": 6.170787428125169e-07,
+      "loss": 0.889,
+      "step": 32441
+    },
+    {
+      "epoch": 0.8909944796902035,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1677114639627e-07,
+      "loss": 0.7583,
+      "step": 32442
+    },
+    {
+      "epoch": 0.8910219439180468,
+      "grad_norm": 0.0,
+      "learning_rate": 6.164636242237631e-07,
+      "loss": 0.7901,
+      "step": 32443
+    },
+    {
+      "epoch": 0.89104940814589,
+      "grad_norm": 0.0,
+      "learning_rate": 6.161561762974322e-07,
+      "loss": 0.6753,
+      "step": 32444
+    },
+    {
+      "epoch": 0.8910768723737332,
+      "grad_norm": 0.0,
+      "learning_rate": 6.158488026197074e-07,
+      "loss": 0.8236,
+      "step": 32445
+    },
+    {
+      "epoch": 0.8911043366015764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.155415031930212e-07,
+      "loss": 0.8089,
+      "step": 32446
+    },
+    {
+      "epoch": 0.8911318008294197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.152342780198062e-07,
+      "loss": 0.7689,
+      "step": 32447
+    },
+    {
+      "epoch": 0.891159265057263,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149271271024948e-07,
+      "loss": 0.7932,
+      "step": 32448
+    },
+    {
+      "epoch": 0.8911867292851061,
+      "grad_norm": 0.0,
+      "learning_rate": 6.146200504435141e-07,
+      "loss": 0.823,
+      "step": 32449
+    },
+    {
+      "epoch": 0.8912141935129494,
+      "grad_norm": 0.0,
+      "learning_rate": 6.143130480452941e-07,
+      "loss": 0.8077,
+      "step": 32450
+    },
+    {
+      "epoch": 0.8912416577407927,
+      "grad_norm": 0.0,
+      "learning_rate": 6.140061199102676e-07,
+      "loss": 0.7979,
+      "step": 32451
+    },
+    {
+      "epoch": 0.8912691219686358,
+      "grad_norm": 0.0,
+      "learning_rate": 6.13699266040858e-07,
+      "loss": 0.9023,
+      "step": 32452
+    },
+    {
+      "epoch": 0.8912965861964791,
+      "grad_norm": 0.0,
+      "learning_rate": 6.133924864394957e-07,
+      "loss": 0.7783,
+      "step": 32453
+    },
+    {
+      "epoch": 0.8913240504243223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.130857811086099e-07,
+      "loss": 0.7615,
+      "step": 32454
+    },
+    {
+      "epoch": 0.8913515146521656,
+      "grad_norm": 0.0,
+      "learning_rate": 6.127791500506241e-07,
+      "loss": 0.7758,
+      "step": 32455
+    },
+    {
+      "epoch": 0.8913789788800088,
+      "grad_norm": 0.0,
+      "learning_rate": 6.124725932679665e-07,
+      "loss": 0.7128,
+      "step": 32456
+    },
+    {
+      "epoch": 0.891406443107852,
+      "grad_norm": 0.0,
+      "learning_rate": 6.121661107630616e-07,
+      "loss": 0.7932,
+      "step": 32457
+    },
+    {
+      "epoch": 0.8914339073356953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.118597025383344e-07,
+      "loss": 0.8064,
+      "step": 32458
+    },
+    {
+      "epoch": 0.8914613715635384,
+      "grad_norm": 0.0,
+      "learning_rate": 6.115533685962116e-07,
+      "loss": 0.7444,
+      "step": 32459
+    },
+    {
+      "epoch": 0.8914888357913817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.112471089391159e-07,
+      "loss": 0.7909,
+      "step": 32460
+    },
+    {
+      "epoch": 0.891516300019225,
+      "grad_norm": 0.0,
+      "learning_rate": 6.109409235694674e-07,
+      "loss": 0.8522,
+      "step": 32461
+    },
+    {
+      "epoch": 0.8915437642470682,
+      "grad_norm": 0.0,
+      "learning_rate": 6.106348124896922e-07,
+      "loss": 0.9097,
+      "step": 32462
+    },
+    {
+      "epoch": 0.8915712284749114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.103287757022114e-07,
+      "loss": 0.7471,
+      "step": 32463
+    },
+    {
+      "epoch": 0.8915986927027547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.100228132094465e-07,
+      "loss": 0.7492,
+      "step": 32464
+    },
+    {
+      "epoch": 0.8916261569305979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.097169250138179e-07,
+      "loss": 0.7711,
+      "step": 32465
+    },
+    {
+      "epoch": 0.8916536211584412,
+      "grad_norm": 0.0,
+      "learning_rate": 6.094111111177481e-07,
+      "loss": 0.7573,
+      "step": 32466
+    },
+    {
+      "epoch": 0.8916810853862843,
+      "grad_norm": 0.0,
+      "learning_rate": 6.091053715236539e-07,
+      "loss": 0.7846,
+      "step": 32467
+    },
+    {
+      "epoch": 0.8917085496141276,
+      "grad_norm": 0.0,
+      "learning_rate": 6.087997062339567e-07,
+      "loss": 0.7681,
+      "step": 32468
+    },
+    {
+      "epoch": 0.8917360138419709,
+      "grad_norm": 0.0,
+      "learning_rate": 6.084941152510737e-07,
+      "loss": 0.7663,
+      "step": 32469
+    },
+    {
+      "epoch": 0.891763478069814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.081885985774216e-07,
+      "loss": 0.7869,
+      "step": 32470
+    },
+    {
+      "epoch": 0.8917909422976573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.078831562154208e-07,
+      "loss": 0.8336,
+      "step": 32471
+    },
+    {
+      "epoch": 0.8918184065255005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.075777881674871e-07,
+      "loss": 0.828,
+      "step": 32472
+    },
+    {
+      "epoch": 0.8918458707533438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.072724944360354e-07,
+      "loss": 0.7149,
+      "step": 32473
+    },
+    {
+      "epoch": 0.891873334981187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.069672750234822e-07,
+      "loss": 0.7545,
+      "step": 32474
+    },
+    {
+      "epoch": 0.8919007992090302,
+      "grad_norm": 0.0,
+      "learning_rate": 6.066621299322429e-07,
+      "loss": 0.8161,
+      "step": 32475
+    },
+    {
+      "epoch": 0.8919282634368735,
+      "grad_norm": 0.0,
+      "learning_rate": 6.063570591647306e-07,
+      "loss": 0.8943,
+      "step": 32476
+    },
+    {
+      "epoch": 0.8919557276647168,
+      "grad_norm": 0.0,
+      "learning_rate": 6.060520627233602e-07,
+      "loss": 0.7573,
+      "step": 32477
+    },
+    {
+      "epoch": 0.8919831918925599,
+      "grad_norm": 0.0,
+      "learning_rate": 6.057471406105464e-07,
+      "loss": 0.8252,
+      "step": 32478
+    },
+    {
+      "epoch": 0.8920106561204032,
+      "grad_norm": 0.0,
+      "learning_rate": 6.054422928286984e-07,
+      "loss": 0.7751,
+      "step": 32479
+    },
+    {
+      "epoch": 0.8920381203482464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0513751938023e-07,
+      "loss": 0.8926,
+      "step": 32480
+    },
+    {
+      "epoch": 0.8920655845760896,
+      "grad_norm": 0.0,
+      "learning_rate": 6.048328202675524e-07,
+      "loss": 0.7699,
+      "step": 32481
+    },
+    {
+      "epoch": 0.8920930488039329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04528195493077e-07,
+      "loss": 0.8453,
+      "step": 32482
+    },
+    {
+      "epoch": 0.8921205130317761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.042236450592165e-07,
+      "loss": 0.7508,
+      "step": 32483
+    },
+    {
+      "epoch": 0.8921479772596194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.039191689683766e-07,
+      "loss": 0.8787,
+      "step": 32484
+    },
+    {
+      "epoch": 0.8921754414874625,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036147672229675e-07,
+      "loss": 0.8106,
+      "step": 32485
+    },
+    {
+      "epoch": 0.8922029057153058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.033104398253975e-07,
+      "loss": 0.8006,
+      "step": 32486
+    },
+    {
+      "epoch": 0.8922303699431491,
+      "grad_norm": 0.0,
+      "learning_rate": 6.030061867780757e-07,
+      "loss": 0.7663,
+      "step": 32487
+    },
+    {
+      "epoch": 0.8922578341709922,
+      "grad_norm": 0.0,
+      "learning_rate": 6.02702008083409e-07,
+      "loss": 0.6894,
+      "step": 32488
+    },
+    {
+      "epoch": 0.8922852983988355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.023979037438044e-07,
+      "loss": 0.8575,
+      "step": 32489
+    },
+    {
+      "epoch": 0.8923127626266788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.020938737616688e-07,
+      "loss": 0.8624,
+      "step": 32490
+    },
+    {
+      "epoch": 0.892340226854522,
+      "grad_norm": 0.0,
+      "learning_rate": 6.01789918139406e-07,
+      "loss": 0.8214,
+      "step": 32491
+    },
+    {
+      "epoch": 0.8923676910823652,
+      "grad_norm": 0.0,
+      "learning_rate": 6.014860368794229e-07,
+      "loss": 0.8991,
+      "step": 32492
+    },
+    {
+      "epoch": 0.8923951553102084,
+      "grad_norm": 0.0,
+      "learning_rate": 6.011822299841208e-07,
+      "loss": 0.7876,
+      "step": 32493
+    },
+    {
+      "epoch": 0.8924226195380517,
+      "grad_norm": 0.0,
+      "learning_rate": 6.008784974559068e-07,
+      "loss": 0.8132,
+      "step": 32494
+    },
+    {
+      "epoch": 0.892450083765895,
+      "grad_norm": 0.0,
+      "learning_rate": 6.005748392971844e-07,
+      "loss": 0.761,
+      "step": 32495
+    },
+    {
+      "epoch": 0.8924775479937381,
+      "grad_norm": 0.0,
+      "learning_rate": 6.002712555103529e-07,
+      "loss": 0.7108,
+      "step": 32496
+    },
+    {
+      "epoch": 0.8925050122215814,
+      "grad_norm": 0.0,
+      "learning_rate": 5.99967746097816e-07,
+      "loss": 0.7705,
+      "step": 32497
+    },
+    {
+      "epoch": 0.8925324764494246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.99664311061976e-07,
+      "loss": 0.6769,
+      "step": 32498
+    },
+    {
+      "epoch": 0.8925599406772678,
+      "grad_norm": 0.0,
+      "learning_rate": 5.993609504052345e-07,
+      "loss": 0.7881,
+      "step": 32499
+    },
+    {
+      "epoch": 0.8925874049051111,
+      "grad_norm": 0.0,
+      "learning_rate": 5.990576641299894e-07,
+      "loss": 0.7809,
+      "step": 32500
+    },
+    {
+      "epoch": 0.8926148691329543,
+      "grad_norm": 0.0,
+      "learning_rate": 5.987544522386435e-07,
+      "loss": 0.7856,
+      "step": 32501
+    },
+    {
+      "epoch": 0.8926423333607976,
+      "grad_norm": 0.0,
+      "learning_rate": 5.984513147335924e-07,
+      "loss": 0.7799,
+      "step": 32502
+    },
+    {
+      "epoch": 0.8926697975886408,
+      "grad_norm": 0.0,
+      "learning_rate": 5.981482516172354e-07,
+      "loss": 0.8109,
+      "step": 32503
+    },
+    {
+      "epoch": 0.892697261816484,
+      "grad_norm": 0.0,
+      "learning_rate": 5.978452628919729e-07,
+      "loss": 0.8259,
+      "step": 32504
+    },
+    {
+      "epoch": 0.8927247260443273,
+      "grad_norm": 0.0,
+      "learning_rate": 5.975423485601994e-07,
+      "loss": 0.8656,
+      "step": 32505
+    },
+    {
+      "epoch": 0.8927521902721705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.972395086243132e-07,
+      "loss": 0.8771,
+      "step": 32506
+    },
+    {
+      "epoch": 0.8927796545000137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.969367430867124e-07,
+      "loss": 0.8113,
+      "step": 32507
+    },
+    {
+      "epoch": 0.892807118727857,
+      "grad_norm": 0.0,
+      "learning_rate": 5.966340519497904e-07,
+      "loss": 0.811,
+      "step": 32508
+    },
+    {
+      "epoch": 0.8928345829557002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.963314352159399e-07,
+      "loss": 0.7902,
+      "step": 32509
+    },
+    {
+      "epoch": 0.8928620471835435,
+      "grad_norm": 0.0,
+      "learning_rate": 5.96028892887559e-07,
+      "loss": 0.8144,
+      "step": 32510
+    },
+    {
+      "epoch": 0.8928895114113866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.957264249670391e-07,
+      "loss": 0.7734,
+      "step": 32511
+    },
+    {
+      "epoch": 0.8929169756392299,
+      "grad_norm": 0.0,
+      "learning_rate": 5.95424031456776e-07,
+      "loss": 0.8512,
+      "step": 32512
+    },
+    {
+      "epoch": 0.8929444398670732,
+      "grad_norm": 0.0,
+      "learning_rate": 5.951217123591613e-07,
+      "loss": 0.8068,
+      "step": 32513
+    },
+    {
+      "epoch": 0.8929719040949163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.94819467676585e-07,
+      "loss": 0.7298,
+      "step": 32514
+    },
+    {
+      "epoch": 0.8929993683227596,
+      "grad_norm": 0.0,
+      "learning_rate": 5.945172974114422e-07,
+      "loss": 0.7715,
+      "step": 32515
+    },
+    {
+      "epoch": 0.8930268325506029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.942152015661207e-07,
+      "loss": 0.779,
+      "step": 32516
+    },
+    {
+      "epoch": 0.893054296778446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.93913180143012e-07,
+      "loss": 0.8223,
+      "step": 32517
+    },
+    {
+      "epoch": 0.8930817610062893,
+      "grad_norm": 0.0,
+      "learning_rate": 5.936112331445054e-07,
+      "loss": 0.8181,
+      "step": 32518
+    },
+    {
+      "epoch": 0.8931092252341325,
+      "grad_norm": 0.0,
+      "learning_rate": 5.933093605729922e-07,
+      "loss": 0.7126,
+      "step": 32519
+    },
+    {
+      "epoch": 0.8931366894619758,
+      "grad_norm": 0.0,
+      "learning_rate": 5.930075624308574e-07,
+      "loss": 0.7811,
+      "step": 32520
+    },
+    {
+      "epoch": 0.893164153689819,
+      "grad_norm": 0.0,
+      "learning_rate": 5.927058387204909e-07,
+      "loss": 0.813,
+      "step": 32521
+    },
+    {
+      "epoch": 0.8931916179176622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9240418944428e-07,
+      "loss": 0.7852,
+      "step": 32522
+    },
+    {
+      "epoch": 0.8932190821455055,
+      "grad_norm": 0.0,
+      "learning_rate": 5.921026146046116e-07,
+      "loss": 0.8328,
+      "step": 32523
+    },
+    {
+      "epoch": 0.8932465463733487,
+      "grad_norm": 0.0,
+      "learning_rate": 5.918011142038715e-07,
+      "loss": 0.8585,
+      "step": 32524
+    },
+    {
+      "epoch": 0.8932740106011919,
+      "grad_norm": 0.0,
+      "learning_rate": 5.914996882444468e-07,
+      "loss": 0.906,
+      "step": 32525
+    },
+    {
+      "epoch": 0.8933014748290352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.911983367287188e-07,
+      "loss": 0.8104,
+      "step": 32526
+    },
+    {
+      "epoch": 0.8933289390568784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.908970596590746e-07,
+      "loss": 0.877,
+      "step": 32527
+    },
+    {
+      "epoch": 0.8933564032847217,
+      "grad_norm": 0.0,
+      "learning_rate": 5.905958570378978e-07,
+      "loss": 0.7846,
+      "step": 32528
+    },
+    {
+      "epoch": 0.8933838675125649,
+      "grad_norm": 0.0,
+      "learning_rate": 5.902947288675709e-07,
+      "loss": 0.7663,
+      "step": 32529
+    },
+    {
+      "epoch": 0.8934113317404081,
+      "grad_norm": 0.0,
+      "learning_rate": 5.899936751504775e-07,
+      "loss": 0.7097,
+      "step": 32530
+    },
+    {
+      "epoch": 0.8934387959682514,
+      "grad_norm": 0.0,
+      "learning_rate": 5.896926958890004e-07,
+      "loss": 0.7722,
+      "step": 32531
+    },
+    {
+      "epoch": 0.8934662601960945,
+      "grad_norm": 0.0,
+      "learning_rate": 5.893917910855196e-07,
+      "loss": 0.9569,
+      "step": 32532
+    },
+    {
+      "epoch": 0.8934937244239378,
+      "grad_norm": 0.0,
+      "learning_rate": 5.890909607424155e-07,
+      "loss": 0.7795,
+      "step": 32533
+    },
+    {
+      "epoch": 0.8935211886517811,
+      "grad_norm": 0.0,
+      "learning_rate": 5.887902048620686e-07,
+      "loss": 0.8717,
+      "step": 32534
+    },
+    {
+      "epoch": 0.8935486528796243,
+      "grad_norm": 0.0,
+      "learning_rate": 5.884895234468601e-07,
+      "loss": 0.8004,
+      "step": 32535
+    },
+    {
+      "epoch": 0.8935761171074675,
+      "grad_norm": 0.0,
+      "learning_rate": 5.881889164991672e-07,
+      "loss": 0.7891,
+      "step": 32536
+    },
+    {
+      "epoch": 0.8936035813353107,
+      "grad_norm": 0.0,
+      "learning_rate": 5.878883840213712e-07,
+      "loss": 0.7871,
+      "step": 32537
+    },
+    {
+      "epoch": 0.893631045563154,
+      "grad_norm": 0.0,
+      "learning_rate": 5.875879260158469e-07,
+      "loss": 0.7531,
+      "step": 32538
+    },
+    {
+      "epoch": 0.8936585097909973,
+      "grad_norm": 0.0,
+      "learning_rate": 5.872875424849734e-07,
+      "loss": 0.9435,
+      "step": 32539
+    },
+    {
+      "epoch": 0.8936859740188404,
+      "grad_norm": 0.0,
+      "learning_rate": 5.869872334311267e-07,
+      "loss": 0.7642,
+      "step": 32540
+    },
+    {
+      "epoch": 0.8937134382466837,
+      "grad_norm": 0.0,
+      "learning_rate": 5.866869988566825e-07,
+      "loss": 0.7085,
+      "step": 32541
+    },
+    {
+      "epoch": 0.893740902474527,
+      "grad_norm": 0.0,
+      "learning_rate": 5.863868387640181e-07,
+      "loss": 0.7504,
+      "step": 32542
+    },
+    {
+      "epoch": 0.8937683667023701,
+      "grad_norm": 0.0,
+      "learning_rate": 5.860867531555082e-07,
+      "loss": 0.8607,
+      "step": 32543
+    },
+    {
+      "epoch": 0.8937958309302134,
+      "grad_norm": 0.0,
+      "learning_rate": 5.857867420335251e-07,
+      "loss": 0.8198,
+      "step": 32544
+    },
+    {
+      "epoch": 0.8938232951580566,
+      "grad_norm": 0.0,
+      "learning_rate": 5.854868054004448e-07,
+      "loss": 0.7654,
+      "step": 32545
+    },
+    {
+      "epoch": 0.8938507593858999,
+      "grad_norm": 0.0,
+      "learning_rate": 5.851869432586387e-07,
+      "loss": 0.6652,
+      "step": 32546
+    },
+    {
+      "epoch": 0.8938782236137431,
+      "grad_norm": 0.0,
+      "learning_rate": 5.848871556104829e-07,
+      "loss": 0.848,
+      "step": 32547
+    },
+    {
+      "epoch": 0.8939056878415863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.845874424583442e-07,
+      "loss": 0.7559,
+      "step": 32548
+    },
+    {
+      "epoch": 0.8939331520694296,
+      "grad_norm": 0.0,
+      "learning_rate": 5.842878038045996e-07,
+      "loss": 0.786,
+      "step": 32549
+    },
+    {
+      "epoch": 0.8939606162972727,
+      "grad_norm": 0.0,
+      "learning_rate": 5.83988239651615e-07,
+      "loss": 0.7104,
+      "step": 32550
+    },
+    {
+      "epoch": 0.893988080525116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.83688750001763e-07,
+      "loss": 0.8054,
+      "step": 32551
+    },
+    {
+      "epoch": 0.8940155447529593,
+      "grad_norm": 0.0,
+      "learning_rate": 5.833893348574127e-07,
+      "loss": 0.7726,
+      "step": 32552
+    },
+    {
+      "epoch": 0.8940430089808025,
+      "grad_norm": 0.0,
+      "learning_rate": 5.830899942209345e-07,
+      "loss": 0.8093,
+      "step": 32553
+    },
+    {
+      "epoch": 0.8940704732086457,
+      "grad_norm": 0.0,
+      "learning_rate": 5.827907280946976e-07,
+      "loss": 0.713,
+      "step": 32554
+    },
+    {
+      "epoch": 0.894097937436489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824915364810657e-07,
+      "loss": 0.8015,
+      "step": 32555
+    },
+    {
+      "epoch": 0.8941254016643322,
+      "grad_norm": 0.0,
+      "learning_rate": 5.821924193824114e-07,
+      "loss": 0.778,
+      "step": 32556
+    },
+    {
+      "epoch": 0.8941528658921755,
+      "grad_norm": 0.0,
+      "learning_rate": 5.818933768010971e-07,
+      "loss": 0.828,
+      "step": 32557
+    },
+    {
+      "epoch": 0.8941803301200186,
+      "grad_norm": 0.0,
+      "learning_rate": 5.815944087394898e-07,
+      "loss": 0.8459,
+      "step": 32558
+    },
+    {
+      "epoch": 0.8942077943478619,
+      "grad_norm": 0.0,
+      "learning_rate": 5.812955151999566e-07,
+      "loss": 0.8059,
+      "step": 32559
+    },
+    {
+      "epoch": 0.8942352585757052,
+      "grad_norm": 0.0,
+      "learning_rate": 5.809966961848635e-07,
+      "loss": 0.7972,
+      "step": 32560
+    },
+    {
+      "epoch": 0.8942627228035483,
+      "grad_norm": 0.0,
+      "learning_rate": 5.806979516965706e-07,
+      "loss": 0.7944,
+      "step": 32561
+    },
+    {
+      "epoch": 0.8942901870313916,
+      "grad_norm": 0.0,
+      "learning_rate": 5.803992817374449e-07,
+      "loss": 0.7488,
+      "step": 32562
+    },
+    {
+      "epoch": 0.8943176512592348,
+      "grad_norm": 0.0,
+      "learning_rate": 5.801006863098502e-07,
+      "loss": 0.746,
+      "step": 32563
+    },
+    {
+      "epoch": 0.8943451154870781,
+      "grad_norm": 0.0,
+      "learning_rate": 5.798021654161467e-07,
+      "loss": 0.783,
+      "step": 32564
+    },
+    {
+      "epoch": 0.8943725797149213,
+      "grad_norm": 0.0,
+      "learning_rate": 5.795037190586972e-07,
+      "loss": 0.7505,
+      "step": 32565
+    },
+    {
+      "epoch": 0.8944000439427645,
+      "grad_norm": 0.0,
+      "learning_rate": 5.79205347239864e-07,
+      "loss": 0.6786,
+      "step": 32566
+    },
+    {
+      "epoch": 0.8944275081706078,
+      "grad_norm": 0.0,
+      "learning_rate": 5.789070499620075e-07,
+      "loss": 0.7904,
+      "step": 32567
+    },
+    {
+      "epoch": 0.8944549723984511,
+      "grad_norm": 0.0,
+      "learning_rate": 5.78608827227487e-07,
+      "loss": 0.7384,
+      "step": 32568
+    },
+    {
+      "epoch": 0.8944824366262942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.783106790386627e-07,
+      "loss": 0.82,
+      "step": 32569
+    },
+    {
+      "epoch": 0.8945099008541375,
+      "grad_norm": 0.0,
+      "learning_rate": 5.78012605397894e-07,
+      "loss": 0.8524,
+      "step": 32570
+    },
+    {
+      "epoch": 0.8945373650819807,
+      "grad_norm": 0.0,
+      "learning_rate": 5.777146063075412e-07,
+      "loss": 0.7767,
+      "step": 32571
+    },
+    {
+      "epoch": 0.894564829309824,
+      "grad_norm": 0.0,
+      "learning_rate": 5.774166817699601e-07,
+      "loss": 0.8615,
+      "step": 32572
+    },
+    {
+      "epoch": 0.8945922935376672,
+      "grad_norm": 0.0,
+      "learning_rate": 5.771188317875065e-07,
+      "loss": 0.8122,
+      "step": 32573
+    },
+    {
+      "epoch": 0.8946197577655104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.768210563625387e-07,
+      "loss": 0.7632,
+      "step": 32574
+    },
+    {
+      "epoch": 0.8946472219933537,
+      "grad_norm": 0.0,
+      "learning_rate": 5.765233554974137e-07,
+      "loss": 0.7128,
+      "step": 32575
+    },
+    {
+      "epoch": 0.894674686221197,
+      "grad_norm": 0.0,
+      "learning_rate": 5.76225729194485e-07,
+      "loss": 0.7472,
+      "step": 32576
+    },
+    {
+      "epoch": 0.8947021504490401,
+      "grad_norm": 0.0,
+      "learning_rate": 5.759281774561099e-07,
+      "loss": 0.7038,
+      "step": 32577
+    },
+    {
+      "epoch": 0.8947296146768834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.756307002846428e-07,
+      "loss": 0.8101,
+      "step": 32578
+    },
+    {
+      "epoch": 0.8947570789047266,
+      "grad_norm": 0.0,
+      "learning_rate": 5.753332976824344e-07,
+      "loss": 0.8496,
+      "step": 32579
+    },
+    {
+      "epoch": 0.8947845431325698,
+      "grad_norm": 0.0,
+      "learning_rate": 5.750359696518415e-07,
+      "loss": 0.751,
+      "step": 32580
+    },
+    {
+      "epoch": 0.8948120073604131,
+      "grad_norm": 0.0,
+      "learning_rate": 5.747387161952134e-07,
+      "loss": 0.7666,
+      "step": 32581
+    },
+    {
+      "epoch": 0.8948394715882563,
+      "grad_norm": 0.0,
+      "learning_rate": 5.744415373149037e-07,
+      "loss": 0.7964,
+      "step": 32582
+    },
+    {
+      "epoch": 0.8948669358160996,
+      "grad_norm": 0.0,
+      "learning_rate": 5.741444330132628e-07,
+      "loss": 0.7937,
+      "step": 32583
+    },
+    {
+      "epoch": 0.8948944000439427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.738474032926444e-07,
+      "loss": 0.713,
+      "step": 32584
+    },
+    {
+      "epoch": 0.894921864271786,
+      "grad_norm": 0.0,
+      "learning_rate": 5.735504481553955e-07,
+      "loss": 0.7354,
+      "step": 32585
+    },
+    {
+      "epoch": 0.8949493284996293,
+      "grad_norm": 0.0,
+      "learning_rate": 5.732535676038664e-07,
+      "loss": 0.8428,
+      "step": 32586
+    },
+    {
+      "epoch": 0.8949767927274724,
+      "grad_norm": 0.0,
+      "learning_rate": 5.729567616404086e-07,
+      "loss": 0.8151,
+      "step": 32587
+    },
+    {
+      "epoch": 0.8950042569553157,
+      "grad_norm": 0.0,
+      "learning_rate": 5.726600302673668e-07,
+      "loss": 0.7985,
+      "step": 32588
+    },
+    {
+      "epoch": 0.895031721183159,
+      "grad_norm": 0.0,
+      "learning_rate": 5.723633734870904e-07,
+      "loss": 0.7471,
+      "step": 32589
+    },
+    {
+      "epoch": 0.8950591854110022,
+      "grad_norm": 0.0,
+      "learning_rate": 5.720667913019296e-07,
+      "loss": 0.8337,
+      "step": 32590
+    },
+    {
+      "epoch": 0.8950866496388454,
+      "grad_norm": 0.0,
+      "learning_rate": 5.717702837142259e-07,
+      "loss": 0.802,
+      "step": 32591
+    },
+    {
+      "epoch": 0.8951141138666886,
+      "grad_norm": 0.0,
+      "learning_rate": 5.714738507263273e-07,
+      "loss": 0.8225,
+      "step": 32592
+    },
+    {
+      "epoch": 0.8951415780945319,
+      "grad_norm": 0.0,
+      "learning_rate": 5.711774923405811e-07,
+      "loss": 0.7003,
+      "step": 32593
+    },
+    {
+      "epoch": 0.8951690423223752,
+      "grad_norm": 0.0,
+      "learning_rate": 5.708812085593307e-07,
+      "loss": 0.7937,
+      "step": 32594
+    },
+    {
+      "epoch": 0.8951965065502183,
+      "grad_norm": 0.0,
+      "learning_rate": 5.705849993849211e-07,
+      "loss": 0.8589,
+      "step": 32595
+    },
+    {
+      "epoch": 0.8952239707780616,
+      "grad_norm": 0.0,
+      "learning_rate": 5.702888648196969e-07,
+      "loss": 0.7469,
+      "step": 32596
+    },
+    {
+      "epoch": 0.8952514350059048,
+      "grad_norm": 0.0,
+      "learning_rate": 5.699928048659975e-07,
+      "loss": 0.7366,
+      "step": 32597
+    },
+    {
+      "epoch": 0.895278899233748,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696968195261687e-07,
+      "loss": 0.8203,
+      "step": 32598
+    },
+    {
+      "epoch": 0.8953063634615913,
+      "grad_norm": 0.0,
+      "learning_rate": 5.694009088025509e-07,
+      "loss": 0.7541,
+      "step": 32599
+    },
+    {
+      "epoch": 0.8953338276894345,
+      "grad_norm": 0.0,
+      "learning_rate": 5.691050726974867e-07,
+      "loss": 0.8274,
+      "step": 32600
+    },
+    {
+      "epoch": 0.8953612919172778,
+      "grad_norm": 0.0,
+      "learning_rate": 5.688093112133164e-07,
+      "loss": 0.8536,
+      "step": 32601
+    },
+    {
+      "epoch": 0.895388756145121,
+      "grad_norm": 0.0,
+      "learning_rate": 5.685136243523814e-07,
+      "loss": 0.7681,
+      "step": 32602
+    },
+    {
+      "epoch": 0.8954162203729642,
+      "grad_norm": 0.0,
+      "learning_rate": 5.682180121170177e-07,
+      "loss": 0.887,
+      "step": 32603
+    },
+    {
+      "epoch": 0.8954436846008075,
+      "grad_norm": 0.0,
+      "learning_rate": 5.679224745095691e-07,
+      "loss": 0.8281,
+      "step": 32604
+    },
+    {
+      "epoch": 0.8954711488286506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.676270115323701e-07,
+      "loss": 0.7007,
+      "step": 32605
+    },
+    {
+      "epoch": 0.8954986130564939,
+      "grad_norm": 0.0,
+      "learning_rate": 5.67331623187759e-07,
+      "loss": 0.7521,
+      "step": 32606
+    },
+    {
+      "epoch": 0.8955260772843372,
+      "grad_norm": 0.0,
+      "learning_rate": 5.670363094780751e-07,
+      "loss": 0.8551,
+      "step": 32607
+    },
+    {
+      "epoch": 0.8955535415121804,
+      "grad_norm": 0.0,
+      "learning_rate": 5.667410704056542e-07,
+      "loss": 0.762,
+      "step": 32608
+    },
+    {
+      "epoch": 0.8955810057400236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.664459059728311e-07,
+      "loss": 0.7668,
+      "step": 32609
+    },
+    {
+      "epoch": 0.8956084699678668,
+      "grad_norm": 0.0,
+      "learning_rate": 5.661508161819429e-07,
+      "loss": 0.8386,
+      "step": 32610
+    },
+    {
+      "epoch": 0.8956359341957101,
+      "grad_norm": 0.0,
+      "learning_rate": 5.658558010353244e-07,
+      "loss": 0.9005,
+      "step": 32611
+    },
+    {
+      "epoch": 0.8956633984235534,
+      "grad_norm": 0.0,
+      "learning_rate": 5.65560860535308e-07,
+      "loss": 0.8304,
+      "step": 32612
+    },
+    {
+      "epoch": 0.8956908626513965,
+      "grad_norm": 0.0,
+      "learning_rate": 5.652659946842287e-07,
+      "loss": 0.8096,
+      "step": 32613
+    },
+    {
+      "epoch": 0.8957183268792398,
+      "grad_norm": 0.0,
+      "learning_rate": 5.649712034844212e-07,
+      "loss": 0.7736,
+      "step": 32614
+    },
+    {
+      "epoch": 0.8957457911070831,
+      "grad_norm": 0.0,
+      "learning_rate": 5.646764869382149e-07,
+      "loss": 0.833,
+      "step": 32615
+    },
+    {
+      "epoch": 0.8957732553349262,
+      "grad_norm": 0.0,
+      "learning_rate": 5.643818450479421e-07,
+      "loss": 0.8221,
+      "step": 32616
+    },
+    {
+      "epoch": 0.8958007195627695,
+      "grad_norm": 0.0,
+      "learning_rate": 5.640872778159357e-07,
+      "loss": 0.7613,
+      "step": 32617
+    },
+    {
+      "epoch": 0.8958281837906127,
+      "grad_norm": 0.0,
+      "learning_rate": 5.637927852445269e-07,
+      "loss": 0.8222,
+      "step": 32618
+    },
+    {
+      "epoch": 0.895855648018456,
+      "grad_norm": 0.0,
+      "learning_rate": 5.634983673360451e-07,
+      "loss": 0.8096,
+      "step": 32619
+    },
+    {
+      "epoch": 0.8958831122462992,
+      "grad_norm": 0.0,
+      "learning_rate": 5.632040240928194e-07,
+      "loss": 0.7009,
+      "step": 32620
+    },
+    {
+      "epoch": 0.8959105764741424,
+      "grad_norm": 0.0,
+      "learning_rate": 5.629097555171781e-07,
+      "loss": 0.8073,
+      "step": 32621
+    },
+    {
+      "epoch": 0.8959380407019857,
+      "grad_norm": 0.0,
+      "learning_rate": 5.626155616114504e-07,
+      "loss": 0.8384,
+      "step": 32622
+    },
+    {
+      "epoch": 0.8959655049298288,
+      "grad_norm": 0.0,
+      "learning_rate": 5.623214423779644e-07,
+      "loss": 0.8264,
+      "step": 32623
+    },
+    {
+      "epoch": 0.8959929691576721,
+      "grad_norm": 0.0,
+      "learning_rate": 5.620273978190461e-07,
+      "loss": 0.7918,
+      "step": 32624
+    },
+    {
+      "epoch": 0.8960204333855154,
+      "grad_norm": 0.0,
+      "learning_rate": 5.617334279370246e-07,
+      "loss": 0.8134,
+      "step": 32625
+    },
+    {
+      "epoch": 0.8960478976133586,
+      "grad_norm": 0.0,
+      "learning_rate": 5.614395327342237e-07,
+      "loss": 0.7425,
+      "step": 32626
+    },
+    {
+      "epoch": 0.8960753618412018,
+      "grad_norm": 0.0,
+      "learning_rate": 5.611457122129683e-07,
+      "loss": 0.8829,
+      "step": 32627
+    },
+    {
+      "epoch": 0.8961028260690451,
+      "grad_norm": 0.0,
+      "learning_rate": 5.608519663755863e-07,
+      "loss": 0.7645,
+      "step": 32628
+    },
+    {
+      "epoch": 0.8961302902968883,
+      "grad_norm": 0.0,
+      "learning_rate": 5.605582952243982e-07,
+      "loss": 0.7515,
+      "step": 32629
+    },
+    {
+      "epoch": 0.8961577545247316,
+      "grad_norm": 0.0,
+      "learning_rate": 5.602646987617288e-07,
+      "loss": 0.7961,
+      "step": 32630
+    },
+    {
+      "epoch": 0.8961852187525747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.599711769899041e-07,
+      "loss": 0.7898,
+      "step": 32631
+    },
+    {
+      "epoch": 0.896212682980418,
+      "grad_norm": 0.0,
+      "learning_rate": 5.596777299112422e-07,
+      "loss": 0.8446,
+      "step": 32632
+    },
+    {
+      "epoch": 0.8962401472082613,
+      "grad_norm": 0.0,
+      "learning_rate": 5.593843575280678e-07,
+      "loss": 0.7798,
+      "step": 32633
+    },
+    {
+      "epoch": 0.8962676114361044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.590910598427002e-07,
+      "loss": 0.6841,
+      "step": 32634
+    },
+    {
+      "epoch": 0.8962950756639477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.587978368574631e-07,
+      "loss": 0.7908,
+      "step": 32635
+    },
+    {
+      "epoch": 0.8963225398917909,
+      "grad_norm": 0.0,
+      "learning_rate": 5.585046885746726e-07,
+      "loss": 0.7984,
+      "step": 32636
+    },
+    {
+      "epoch": 0.8963500041196342,
+      "grad_norm": 0.0,
+      "learning_rate": 5.582116149966533e-07,
+      "loss": 0.8238,
+      "step": 32637
+    },
+    {
+      "epoch": 0.8963774683474774,
+      "grad_norm": 0.0,
+      "learning_rate": 5.579186161257178e-07,
+      "loss": 0.8015,
+      "step": 32638
+    },
+    {
+      "epoch": 0.8964049325753206,
+      "grad_norm": 0.0,
+      "learning_rate": 5.576256919641887e-07,
+      "loss": 0.8353,
+      "step": 32639
+    },
+    {
+      "epoch": 0.8964323968031639,
+      "grad_norm": 0.0,
+      "learning_rate": 5.573328425143832e-07,
+      "loss": 0.8296,
+      "step": 32640
+    },
+    {
+      "epoch": 0.8964598610310072,
+      "grad_norm": 0.0,
+      "learning_rate": 5.570400677786169e-07,
+      "loss": 0.7761,
+      "step": 32641
+    },
+    {
+      "epoch": 0.8964873252588503,
+      "grad_norm": 0.0,
+      "learning_rate": 5.567473677592083e-07,
+      "loss": 0.7943,
+      "step": 32642
+    },
+    {
+      "epoch": 0.8965147894866936,
+      "grad_norm": 0.0,
+      "learning_rate": 5.564547424584732e-07,
+      "loss": 0.8309,
+      "step": 32643
+    },
+    {
+      "epoch": 0.8965422537145368,
+      "grad_norm": 0.0,
+      "learning_rate": 5.561621918787275e-07,
+      "loss": 0.7904,
+      "step": 32644
+    },
+    {
+      "epoch": 0.89656971794238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.558697160222826e-07,
+      "loss": 0.7793,
+      "step": 32645
+    },
+    {
+      "epoch": 0.8965971821702233,
+      "grad_norm": 0.0,
+      "learning_rate": 5.555773148914545e-07,
+      "loss": 0.6668,
+      "step": 32646
+    },
+    {
+      "epoch": 0.8966246463980665,
+      "grad_norm": 0.0,
+      "learning_rate": 5.552849884885569e-07,
+      "loss": 0.8752,
+      "step": 32647
+    },
+    {
+      "epoch": 0.8966521106259098,
+      "grad_norm": 0.0,
+      "learning_rate": 5.549927368159036e-07,
+      "loss": 0.7932,
+      "step": 32648
+    },
+    {
+      "epoch": 0.8966795748537529,
+      "grad_norm": 0.0,
+      "learning_rate": 5.547005598758071e-07,
+      "loss": 0.8963,
+      "step": 32649
+    },
+    {
+      "epoch": 0.8967070390815962,
+      "grad_norm": 0.0,
+      "learning_rate": 5.544084576705777e-07,
+      "loss": 0.7818,
+      "step": 32650
+    },
+    {
+      "epoch": 0.8967345033094395,
+      "grad_norm": 0.0,
+      "learning_rate": 5.541164302025281e-07,
+      "loss": 0.9111,
+      "step": 32651
+    },
+    {
+      "epoch": 0.8967619675372827,
+      "grad_norm": 0.0,
+      "learning_rate": 5.538244774739687e-07,
+      "loss": 0.8016,
+      "step": 32652
+    },
+    {
+      "epoch": 0.8967894317651259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.535325994872076e-07,
+      "loss": 0.7991,
+      "step": 32653
+    },
+    {
+      "epoch": 0.8968168959929692,
+      "grad_norm": 0.0,
+      "learning_rate": 5.532407962445563e-07,
+      "loss": 0.8333,
+      "step": 32654
+    },
+    {
+      "epoch": 0.8968443602208124,
+      "grad_norm": 0.0,
+      "learning_rate": 5.529490677483252e-07,
+      "loss": 0.8145,
+      "step": 32655
+    },
+    {
+      "epoch": 0.8968718244486557,
+      "grad_norm": 0.0,
+      "learning_rate": 5.526574140008178e-07,
+      "loss": 0.8145,
+      "step": 32656
+    },
+    {
+      "epoch": 0.8968992886764988,
+      "grad_norm": 0.0,
+      "learning_rate": 5.523658350043448e-07,
+      "loss": 0.8493,
+      "step": 32657
+    },
+    {
+      "epoch": 0.8969267529043421,
+      "grad_norm": 0.0,
+      "learning_rate": 5.520743307612142e-07,
+      "loss": 0.7946,
+      "step": 32658
+    },
+    {
+      "epoch": 0.8969542171321854,
+      "grad_norm": 0.0,
+      "learning_rate": 5.517829012737308e-07,
+      "loss": 0.8548,
+      "step": 32659
+    },
+    {
+      "epoch": 0.8969816813600285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.514915465442005e-07,
+      "loss": 0.8581,
+      "step": 32660
+    },
+    {
+      "epoch": 0.8970091455878718,
+      "grad_norm": 0.0,
+      "learning_rate": 5.512002665749305e-07,
+      "loss": 0.8468,
+      "step": 32661
+    },
+    {
+      "epoch": 0.897036609815715,
+      "grad_norm": 0.0,
+      "learning_rate": 5.509090613682233e-07,
+      "loss": 0.8189,
+      "step": 32662
+    },
+    {
+      "epoch": 0.8970640740435583,
+      "grad_norm": 0.0,
+      "learning_rate": 5.506179309263826e-07,
+      "loss": 0.7721,
+      "step": 32663
+    },
+    {
+      "epoch": 0.8970915382714015,
+      "grad_norm": 0.0,
+      "learning_rate": 5.503268752517144e-07,
+      "loss": 0.7419,
+      "step": 32664
+    },
+    {
+      "epoch": 0.8971190024992447,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5003589434652e-07,
+      "loss": 0.8998,
+      "step": 32665
+    },
+    {
+      "epoch": 0.897146466727088,
+      "grad_norm": 0.0,
+      "learning_rate": 5.497449882131011e-07,
+      "loss": 0.8309,
+      "step": 32666
+    },
+    {
+      "epoch": 0.8971739309549313,
+      "grad_norm": 0.0,
+      "learning_rate": 5.494541568537637e-07,
+      "loss": 0.8277,
+      "step": 32667
+    },
+    {
+      "epoch": 0.8972013951827744,
+      "grad_norm": 0.0,
+      "learning_rate": 5.491634002708047e-07,
+      "loss": 0.7873,
+      "step": 32668
+    },
+    {
+      "epoch": 0.8972288594106177,
+      "grad_norm": 0.0,
+      "learning_rate": 5.488727184665243e-07,
+      "loss": 0.9634,
+      "step": 32669
+    },
+    {
+      "epoch": 0.8972563236384609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.485821114432243e-07,
+      "loss": 0.7233,
+      "step": 32670
+    },
+    {
+      "epoch": 0.8972837878663041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.48291579203204e-07,
+      "loss": 0.8737,
+      "step": 32671
+    },
+    {
+      "epoch": 0.8973112520941474,
+      "grad_norm": 0.0,
+      "learning_rate": 5.480011217487613e-07,
+      "loss": 0.8602,
+      "step": 32672
+    },
+    {
+      "epoch": 0.8973387163219906,
+      "grad_norm": 0.0,
+      "learning_rate": 5.477107390821968e-07,
+      "loss": 0.8322,
+      "step": 32673
+    },
+    {
+      "epoch": 0.8973661805498339,
+      "grad_norm": 0.0,
+      "learning_rate": 5.474204312058051e-07,
+      "loss": 0.8197,
+      "step": 32674
+    },
+    {
+      "epoch": 0.897393644777677,
+      "grad_norm": 0.0,
+      "learning_rate": 5.471301981218857e-07,
+      "loss": 0.8831,
+      "step": 32675
+    },
+    {
+      "epoch": 0.8974211090055203,
+      "grad_norm": 0.0,
+      "learning_rate": 5.468400398327334e-07,
+      "loss": 0.8337,
+      "step": 32676
+    },
+    {
+      "epoch": 0.8974485732333636,
+      "grad_norm": 0.0,
+      "learning_rate": 5.465499563406429e-07,
+      "loss": 0.8157,
+      "step": 32677
+    },
+    {
+      "epoch": 0.8974760374612067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.462599476479135e-07,
+      "loss": 0.7862,
+      "step": 32678
+    },
+    {
+      "epoch": 0.89750350168905,
+      "grad_norm": 0.0,
+      "learning_rate": 5.459700137568358e-07,
+      "loss": 0.7659,
+      "step": 32679
+    },
+    {
+      "epoch": 0.8975309659168933,
+      "grad_norm": 0.0,
+      "learning_rate": 5.456801546697055e-07,
+      "loss": 0.8219,
+      "step": 32680
+    },
+    {
+      "epoch": 0.8975584301447365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.453903703888164e-07,
+      "loss": 0.7931,
+      "step": 32681
+    },
+    {
+      "epoch": 0.8975858943725797,
+      "grad_norm": 0.0,
+      "learning_rate": 5.451006609164611e-07,
+      "loss": 0.7108,
+      "step": 32682
+    },
+    {
+      "epoch": 0.8976133586004229,
+      "grad_norm": 0.0,
+      "learning_rate": 5.448110262549322e-07,
+      "loss": 0.8352,
+      "step": 32683
+    },
+    {
+      "epoch": 0.8976408228282662,
+      "grad_norm": 0.0,
+      "learning_rate": 5.445214664065223e-07,
+      "loss": 0.7673,
+      "step": 32684
+    },
+    {
+      "epoch": 0.8976682870561095,
+      "grad_norm": 0.0,
+      "learning_rate": 5.442319813735197e-07,
+      "loss": 0.7,
+      "step": 32685
+    },
+    {
+      "epoch": 0.8976957512839526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.439425711582158e-07,
+      "loss": 0.7724,
+      "step": 32686
+    },
+    {
+      "epoch": 0.8977232155117959,
+      "grad_norm": 0.0,
+      "learning_rate": 5.436532357629009e-07,
+      "loss": 0.7767,
+      "step": 32687
+    },
+    {
+      "epoch": 0.8977506797396391,
+      "grad_norm": 0.0,
+      "learning_rate": 5.433639751898656e-07,
+      "loss": 0.7393,
+      "step": 32688
+    },
+    {
+      "epoch": 0.8977781439674823,
+      "grad_norm": 0.0,
+      "learning_rate": 5.430747894413979e-07,
+      "loss": 0.8149,
+      "step": 32689
+    },
+    {
+      "epoch": 0.8978056081953256,
+      "grad_norm": 0.0,
+      "learning_rate": 5.427856785197861e-07,
+      "loss": 0.8503,
+      "step": 32690
+    },
+    {
+      "epoch": 0.8978330724231688,
+      "grad_norm": 0.0,
+      "learning_rate": 5.424966424273159e-07,
+      "loss": 0.7764,
+      "step": 32691
+    },
+    {
+      "epoch": 0.8978605366510121,
+      "grad_norm": 0.0,
+      "learning_rate": 5.42207681166278e-07,
+      "loss": 0.7778,
+      "step": 32692
+    },
+    {
+      "epoch": 0.8978880008788553,
+      "grad_norm": 0.0,
+      "learning_rate": 5.419187947389548e-07,
+      "loss": 0.8562,
+      "step": 32693
+    },
+    {
+      "epoch": 0.8979154651066985,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416299831476346e-07,
+      "loss": 0.8983,
+      "step": 32694
+    },
+    {
+      "epoch": 0.8979429293345418,
+      "grad_norm": 0.0,
+      "learning_rate": 5.413412463946e-07,
+      "loss": 0.7907,
+      "step": 32695
+    },
+    {
+      "epoch": 0.897970393562385,
+      "grad_norm": 0.0,
+      "learning_rate": 5.410525844821402e-07,
+      "loss": 0.8978,
+      "step": 32696
+    },
+    {
+      "epoch": 0.8979978577902282,
+      "grad_norm": 0.0,
+      "learning_rate": 5.407639974125345e-07,
+      "loss": 0.7644,
+      "step": 32697
+    },
+    {
+      "epoch": 0.8980253220180715,
+      "grad_norm": 0.0,
+      "learning_rate": 5.404754851880689e-07,
+      "loss": 0.814,
+      "step": 32698
+    },
+    {
+      "epoch": 0.8980527862459147,
+      "grad_norm": 0.0,
+      "learning_rate": 5.401870478110261e-07,
+      "loss": 0.7165,
+      "step": 32699
+    },
+    {
+      "epoch": 0.898080250473758,
+      "grad_norm": 0.0,
+      "learning_rate": 5.398986852836863e-07,
+      "loss": 0.7573,
+      "step": 32700
+    },
+    {
+      "epoch": 0.8981077147016011,
+      "grad_norm": 0.0,
+      "learning_rate": 5.396103976083334e-07,
+      "loss": 0.7755,
+      "step": 32701
+    },
+    {
+      "epoch": 0.8981351789294444,
+      "grad_norm": 0.0,
+      "learning_rate": 5.393221847872487e-07,
+      "loss": 0.8122,
+      "step": 32702
+    },
+    {
+      "epoch": 0.8981626431572877,
+      "grad_norm": 0.0,
+      "learning_rate": 5.390340468227095e-07,
+      "loss": 0.8157,
+      "step": 32703
+    },
+    {
+      "epoch": 0.8981901073851308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.387459837169984e-07,
+      "loss": 0.7913,
+      "step": 32704
+    },
+    {
+      "epoch": 0.8982175716129741,
+      "grad_norm": 0.0,
+      "learning_rate": 5.384579954723934e-07,
+      "loss": 0.86,
+      "step": 32705
+    },
+    {
+      "epoch": 0.8982450358408174,
+      "grad_norm": 0.0,
+      "learning_rate": 5.38170082091175e-07,
+      "loss": 0.8156,
+      "step": 32706
+    },
+    {
+      "epoch": 0.8982725000686606,
+      "grad_norm": 0.0,
+      "learning_rate": 5.378822435756203e-07,
+      "loss": 0.869,
+      "step": 32707
+    },
+    {
+      "epoch": 0.8982999642965038,
+      "grad_norm": 0.0,
+      "learning_rate": 5.375944799280075e-07,
+      "loss": 0.8005,
+      "step": 32708
+    },
+    {
+      "epoch": 0.898327428524347,
+      "grad_norm": 0.0,
+      "learning_rate": 5.373067911506103e-07,
+      "loss": 0.8797,
+      "step": 32709
+    },
+    {
+      "epoch": 0.8983548927521903,
+      "grad_norm": 0.0,
+      "learning_rate": 5.370191772457079e-07,
+      "loss": 0.8253,
+      "step": 32710
+    },
+    {
+      "epoch": 0.8983823569800335,
+      "grad_norm": 0.0,
+      "learning_rate": 5.367316382155752e-07,
+      "loss": 0.8403,
+      "step": 32711
+    },
+    {
+      "epoch": 0.8984098212078767,
+      "grad_norm": 0.0,
+      "learning_rate": 5.364441740624871e-07,
+      "loss": 0.8438,
+      "step": 32712
+    },
+    {
+      "epoch": 0.89843728543572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.361567847887183e-07,
+      "loss": 0.8479,
+      "step": 32713
+    },
+    {
+      "epoch": 0.8984647496635632,
+      "grad_norm": 0.0,
+      "learning_rate": 5.358694703965451e-07,
+      "loss": 0.9311,
+      "step": 32714
+    },
+    {
+      "epoch": 0.8984922138914064,
+      "grad_norm": 0.0,
+      "learning_rate": 5.355822308882374e-07,
+      "loss": 0.8568,
+      "step": 32715
+    },
+    {
+      "epoch": 0.8985196781192497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.352950662660705e-07,
+      "loss": 0.7486,
+      "step": 32716
+    },
+    {
+      "epoch": 0.8985471423470929,
+      "grad_norm": 0.0,
+      "learning_rate": 5.350079765323135e-07,
+      "loss": 0.8316,
+      "step": 32717
+    },
+    {
+      "epoch": 0.8985746065749362,
+      "grad_norm": 0.0,
+      "learning_rate": 5.347209616892402e-07,
+      "loss": 0.7633,
+      "step": 32718
+    },
+    {
+      "epoch": 0.8986020708027794,
+      "grad_norm": 0.0,
+      "learning_rate": 5.344340217391208e-07,
+      "loss": 0.7321,
+      "step": 32719
+    },
+    {
+      "epoch": 0.8986295350306226,
+      "grad_norm": 0.0,
+      "learning_rate": 5.341471566842282e-07,
+      "loss": 0.7108,
+      "step": 32720
+    },
+    {
+      "epoch": 0.8986569992584659,
+      "grad_norm": 0.0,
+      "learning_rate": 5.338603665268282e-07,
+      "loss": 0.7902,
+      "step": 32721
+    },
+    {
+      "epoch": 0.898684463486309,
+      "grad_norm": 0.0,
+      "learning_rate": 5.335736512691914e-07,
+      "loss": 0.7986,
+      "step": 32722
+    },
+    {
+      "epoch": 0.8987119277141523,
+      "grad_norm": 0.0,
+      "learning_rate": 5.33287010913588e-07,
+      "loss": 0.8459,
+      "step": 32723
+    },
+    {
+      "epoch": 0.8987393919419956,
+      "grad_norm": 0.0,
+      "learning_rate": 5.33000445462284e-07,
+      "loss": 0.9163,
+      "step": 32724
+    },
+    {
+      "epoch": 0.8987668561698388,
+      "grad_norm": 0.0,
+      "learning_rate": 5.327139549175475e-07,
+      "loss": 0.8528,
+      "step": 32725
+    },
+    {
+      "epoch": 0.898794320397682,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32427539281647e-07,
+      "loss": 0.7119,
+      "step": 32726
+    },
+    {
+      "epoch": 0.8988217846255252,
+      "grad_norm": 0.0,
+      "learning_rate": 5.321411985568459e-07,
+      "loss": 0.8032,
+      "step": 32727
+    },
+    {
+      "epoch": 0.8988492488533685,
+      "grad_norm": 0.0,
+      "learning_rate": 5.318549327454115e-07,
+      "loss": 0.7811,
+      "step": 32728
+    },
+    {
+      "epoch": 0.8988767130812118,
+      "grad_norm": 0.0,
+      "learning_rate": 5.315687418496074e-07,
+      "loss": 0.6992,
+      "step": 32729
+    },
+    {
+      "epoch": 0.8989041773090549,
+      "grad_norm": 0.0,
+      "learning_rate": 5.312826258716997e-07,
+      "loss": 0.8869,
+      "step": 32730
+    },
+    {
+      "epoch": 0.8989316415368982,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309965848139531e-07,
+      "loss": 0.795,
+      "step": 32731
+    },
+    {
+      "epoch": 0.8989591057647415,
+      "grad_norm": 0.0,
+      "learning_rate": 5.307106186786293e-07,
+      "loss": 0.7995,
+      "step": 32732
+    },
+    {
+      "epoch": 0.8989865699925846,
+      "grad_norm": 0.0,
+      "learning_rate": 5.304247274679897e-07,
+      "loss": 0.7913,
+      "step": 32733
+    },
+    {
+      "epoch": 0.8990140342204279,
+      "grad_norm": 0.0,
+      "learning_rate": 5.301389111842981e-07,
+      "loss": 0.7657,
+      "step": 32734
+    },
+    {
+      "epoch": 0.8990414984482711,
+      "grad_norm": 0.0,
+      "learning_rate": 5.298531698298159e-07,
+      "loss": 0.8075,
+      "step": 32735
+    },
+    {
+      "epoch": 0.8990689626761144,
+      "grad_norm": 0.0,
+      "learning_rate": 5.295675034068027e-07,
+      "loss": 0.855,
+      "step": 32736
+    },
+    {
+      "epoch": 0.8990964269039576,
+      "grad_norm": 0.0,
+      "learning_rate": 5.29281911917523e-07,
+      "loss": 0.7108,
+      "step": 32737
+    },
+    {
+      "epoch": 0.8991238911318008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.289963953642307e-07,
+      "loss": 0.8214,
+      "step": 32738
+    },
+    {
+      "epoch": 0.8991513553596441,
+      "grad_norm": 0.0,
+      "learning_rate": 5.287109537491875e-07,
+      "loss": 0.7662,
+      "step": 32739
+    },
+    {
+      "epoch": 0.8991788195874872,
+      "grad_norm": 0.0,
+      "learning_rate": 5.284255870746547e-07,
+      "loss": 0.812,
+      "step": 32740
+    },
+    {
+      "epoch": 0.8992062838153305,
+      "grad_norm": 0.0,
+      "learning_rate": 5.281402953428849e-07,
+      "loss": 0.8038,
+      "step": 32741
+    },
+    {
+      "epoch": 0.8992337480431738,
+      "grad_norm": 0.0,
+      "learning_rate": 5.278550785561387e-07,
+      "loss": 0.8528,
+      "step": 32742
+    },
+    {
+      "epoch": 0.899261212271017,
+      "grad_norm": 0.0,
+      "learning_rate": 5.275699367166742e-07,
+      "loss": 0.8991,
+      "step": 32743
+    },
+    {
+      "epoch": 0.8992886764988602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.27284869826744e-07,
+      "loss": 0.8093,
+      "step": 32744
+    },
+    {
+      "epoch": 0.8993161407267035,
+      "grad_norm": 0.0,
+      "learning_rate": 5.269998778886043e-07,
+      "loss": 0.8811,
+      "step": 32745
+    },
+    {
+      "epoch": 0.8993436049545467,
+      "grad_norm": 0.0,
+      "learning_rate": 5.267149609045119e-07,
+      "loss": 0.8477,
+      "step": 32746
+    },
+    {
+      "epoch": 0.89937106918239,
+      "grad_norm": 0.0,
+      "learning_rate": 5.264301188767218e-07,
+      "loss": 0.8132,
+      "step": 32747
+    },
+    {
+      "epoch": 0.8993985334102331,
+      "grad_norm": 0.0,
+      "learning_rate": 5.261453518074844e-07,
+      "loss": 0.8019,
+      "step": 32748
+    },
+    {
+      "epoch": 0.8994259976380764,
+      "grad_norm": 0.0,
+      "learning_rate": 5.258606596990567e-07,
+      "loss": 0.7946,
+      "step": 32749
+    },
+    {
+      "epoch": 0.8994534618659197,
+      "grad_norm": 0.0,
+      "learning_rate": 5.255760425536882e-07,
+      "loss": 0.8257,
+      "step": 32750
+    },
+    {
+      "epoch": 0.8994809260937628,
+      "grad_norm": 0.0,
+      "learning_rate": 5.252915003736314e-07,
+      "loss": 0.8216,
+      "step": 32751
+    },
+    {
+      "epoch": 0.8995083903216061,
+      "grad_norm": 0.0,
+      "learning_rate": 5.25007033161139e-07,
+      "loss": 0.7653,
+      "step": 32752
+    },
+    {
+      "epoch": 0.8995358545494493,
+      "grad_norm": 0.0,
+      "learning_rate": 5.247226409184603e-07,
+      "loss": 0.9227,
+      "step": 32753
+    },
+    {
+      "epoch": 0.8995633187772926,
+      "grad_norm": 0.0,
+      "learning_rate": 5.244383236478467e-07,
+      "loss": 0.8794,
+      "step": 32754
+    },
+    {
+      "epoch": 0.8995907830051358,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2415408135155e-07,
+      "loss": 0.8598,
+      "step": 32755
+    },
+    {
+      "epoch": 0.899618247232979,
+      "grad_norm": 0.0,
+      "learning_rate": 5.238699140318159e-07,
+      "loss": 0.7921,
+      "step": 32756
+    },
+    {
+      "epoch": 0.8996457114608223,
+      "grad_norm": 0.0,
+      "learning_rate": 5.235858216908929e-07,
+      "loss": 0.8541,
+      "step": 32757
+    },
+    {
+      "epoch": 0.8996731756886656,
+      "grad_norm": 0.0,
+      "learning_rate": 5.23301804331029e-07,
+      "loss": 0.8035,
+      "step": 32758
+    },
+    {
+      "epoch": 0.8997006399165087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.230178619544723e-07,
+      "loss": 0.6834,
+      "step": 32759
+    },
+    {
+      "epoch": 0.899728104144352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.227339945634702e-07,
+      "loss": 0.8549,
+      "step": 32760
+    },
+    {
+      "epoch": 0.8997555683721952,
+      "grad_norm": 0.0,
+      "learning_rate": 5.224502021602684e-07,
+      "loss": 0.7515,
+      "step": 32761
+    },
+    {
+      "epoch": 0.8997830326000384,
+      "grad_norm": 0.0,
+      "learning_rate": 5.221664847471108e-07,
+      "loss": 0.7149,
+      "step": 32762
+    },
+    {
+      "epoch": 0.8998104968278817,
+      "grad_norm": 0.0,
+      "learning_rate": 5.218828423262423e-07,
+      "loss": 0.7836,
+      "step": 32763
+    },
+    {
+      "epoch": 0.8998379610557249,
+      "grad_norm": 0.0,
+      "learning_rate": 5.21599274899911e-07,
+      "loss": 0.8596,
+      "step": 32764
+    },
+    {
+      "epoch": 0.8998654252835682,
+      "grad_norm": 0.0,
+      "learning_rate": 5.213157824703563e-07,
+      "loss": 0.8594,
+      "step": 32765
+    },
+    {
+      "epoch": 0.8998928895114113,
+      "grad_norm": 0.0,
+      "learning_rate": 5.210323650398219e-07,
+      "loss": 0.8124,
+      "step": 32766
+    },
+    {
+      "epoch": 0.8999203537392546,
+      "grad_norm": 0.0,
+      "learning_rate": 5.207490226105527e-07,
+      "loss": 0.7413,
+      "step": 32767
+    },
+    {
+      "epoch": 0.8999478179670979,
+      "grad_norm": 0.0,
+      "learning_rate": 5.204657551847891e-07,
+      "loss": 0.7275,
+      "step": 32768
+    },
+    {
+      "epoch": 0.899975282194941,
+      "grad_norm": 0.0,
+      "learning_rate": 5.201825627647716e-07,
+      "loss": 0.7585,
+      "step": 32769
+    },
+    {
+      "epoch": 0.9000027464227843,
+      "grad_norm": 0.0,
+      "learning_rate": 5.198994453527418e-07,
+      "loss": 0.8171,
+      "step": 32770
+    },
+    {
+      "epoch": 0.9000302106506276,
+      "grad_norm": 0.0,
+      "learning_rate": 5.19616402950941e-07,
+      "loss": 0.7571,
+      "step": 32771
+    },
+    {
+      "epoch": 0.9000576748784708,
+      "grad_norm": 0.0,
+      "learning_rate": 5.193334355616065e-07,
+      "loss": 0.7361,
+      "step": 32772
+    },
+    {
+      "epoch": 0.900085139106314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.190505431869786e-07,
+      "loss": 0.9064,
+      "step": 32773
+    },
+    {
+      "epoch": 0.9001126033341572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.187677258292956e-07,
+      "loss": 0.7429,
+      "step": 32774
+    },
+    {
+      "epoch": 0.9001400675620005,
+      "grad_norm": 0.0,
+      "learning_rate": 5.184849834907945e-07,
+      "loss": 0.688,
+      "step": 32775
+    },
+    {
+      "epoch": 0.9001675317898438,
+      "grad_norm": 0.0,
+      "learning_rate": 5.182023161737126e-07,
+      "loss": 0.7946,
+      "step": 32776
+    },
+    {
+      "epoch": 0.9001949960176869,
+      "grad_norm": 0.0,
+      "learning_rate": 5.179197238802868e-07,
+      "loss": 0.858,
+      "step": 32777
+    },
+    {
+      "epoch": 0.9002224602455302,
+      "grad_norm": 0.0,
+      "learning_rate": 5.176372066127533e-07,
+      "loss": 0.7417,
+      "step": 32778
+    },
+    {
+      "epoch": 0.9002499244733735,
+      "grad_norm": 0.0,
+      "learning_rate": 5.17354764373349e-07,
+      "loss": 0.8567,
+      "step": 32779
+    },
+    {
+      "epoch": 0.9002773887012167,
+      "grad_norm": 0.0,
+      "learning_rate": 5.170723971643077e-07,
+      "loss": 0.7211,
+      "step": 32780
+    },
+    {
+      "epoch": 0.9003048529290599,
+      "grad_norm": 0.0,
+      "learning_rate": 5.167901049878599e-07,
+      "loss": 0.8436,
+      "step": 32781
+    },
+    {
+      "epoch": 0.9003323171569031,
+      "grad_norm": 0.0,
+      "learning_rate": 5.165078878462437e-07,
+      "loss": 0.746,
+      "step": 32782
+    },
+    {
+      "epoch": 0.9003597813847464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.162257457416897e-07,
+      "loss": 0.8972,
+      "step": 32783
+    },
+    {
+      "epoch": 0.9003872456125896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.159436786764316e-07,
+      "loss": 0.813,
+      "step": 32784
+    },
+    {
+      "epoch": 0.9004147098404328,
+      "grad_norm": 0.0,
+      "learning_rate": 5.156616866527031e-07,
+      "loss": 0.8316,
+      "step": 32785
+    },
+    {
+      "epoch": 0.9004421740682761,
+      "grad_norm": 0.0,
+      "learning_rate": 5.153797696727303e-07,
+      "loss": 0.7982,
+      "step": 32786
+    },
+    {
+      "epoch": 0.9004696382961193,
+      "grad_norm": 0.0,
+      "learning_rate": 5.15097927738748e-07,
+      "loss": 0.8,
+      "step": 32787
+    },
+    {
+      "epoch": 0.9004971025239625,
+      "grad_norm": 0.0,
+      "learning_rate": 5.148161608529856e-07,
+      "loss": 0.8582,
+      "step": 32788
+    },
+    {
+      "epoch": 0.9005245667518058,
+      "grad_norm": 0.0,
+      "learning_rate": 5.145344690176712e-07,
+      "loss": 0.7461,
+      "step": 32789
+    },
+    {
+      "epoch": 0.900552030979649,
+      "grad_norm": 0.0,
+      "learning_rate": 5.142528522350332e-07,
+      "loss": 0.8571,
+      "step": 32790
+    },
+    {
+      "epoch": 0.9005794952074923,
+      "grad_norm": 0.0,
+      "learning_rate": 5.139713105073041e-07,
+      "loss": 0.7708,
+      "step": 32791
+    },
+    {
+      "epoch": 0.9006069594353355,
+      "grad_norm": 0.0,
+      "learning_rate": 5.136898438367056e-07,
+      "loss": 0.7538,
+      "step": 32792
+    },
+    {
+      "epoch": 0.9006344236631787,
+      "grad_norm": 0.0,
+      "learning_rate": 5.134084522254689e-07,
+      "loss": 0.7939,
+      "step": 32793
+    },
+    {
+      "epoch": 0.900661887891022,
+      "grad_norm": 0.0,
+      "learning_rate": 5.131271356758183e-07,
+      "loss": 0.7918,
+      "step": 32794
+    },
+    {
+      "epoch": 0.9006893521188651,
+      "grad_norm": 0.0,
+      "learning_rate": 5.128458941899827e-07,
+      "loss": 0.7624,
+      "step": 32795
+    },
+    {
+      "epoch": 0.9007168163467084,
+      "grad_norm": 0.0,
+      "learning_rate": 5.125647277701851e-07,
+      "loss": 0.8232,
+      "step": 32796
+    },
+    {
+      "epoch": 0.9007442805745517,
+      "grad_norm": 0.0,
+      "learning_rate": 5.12283636418649e-07,
+      "loss": 0.7971,
+      "step": 32797
+    },
+    {
+      "epoch": 0.9007717448023949,
+      "grad_norm": 0.0,
+      "learning_rate": 5.120026201376005e-07,
+      "loss": 0.8187,
+      "step": 32798
+    },
+    {
+      "epoch": 0.9007992090302381,
+      "grad_norm": 0.0,
+      "learning_rate": 5.117216789292622e-07,
+      "loss": 0.8279,
+      "step": 32799
+    },
+    {
+      "epoch": 0.9008266732580813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.11440812795857e-07,
+      "loss": 0.867,
+      "step": 32800
+    },
+    {
+      "epoch": 0.9008541374859246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.111600217396084e-07,
+      "loss": 0.7616,
+      "step": 32801
+    },
+    {
+      "epoch": 0.9008816017137679,
+      "grad_norm": 0.0,
+      "learning_rate": 5.108793057627382e-07,
+      "loss": 0.8519,
+      "step": 32802
+    },
+    {
+      "epoch": 0.900909065941611,
+      "grad_norm": 0.0,
+      "learning_rate": 5.105986648674655e-07,
+      "loss": 0.7282,
+      "step": 32803
+    },
+    {
+      "epoch": 0.9009365301694543,
+      "grad_norm": 0.0,
+      "learning_rate": 5.10318099056013e-07,
+      "loss": 0.8535,
+      "step": 32804
+    },
+    {
+      "epoch": 0.9009639943972976,
+      "grad_norm": 0.0,
+      "learning_rate": 5.100376083305991e-07,
+      "loss": 0.8903,
+      "step": 32805
+    },
+    {
+      "epoch": 0.9009914586251407,
+      "grad_norm": 0.0,
+      "learning_rate": 5.09757192693443e-07,
+      "loss": 0.9291,
+      "step": 32806
+    },
+    {
+      "epoch": 0.901018922852984,
+      "grad_norm": 0.0,
+      "learning_rate": 5.094768521467641e-07,
+      "loss": 0.8207,
+      "step": 32807
+    },
+    {
+      "epoch": 0.9010463870808272,
+      "grad_norm": 0.0,
+      "learning_rate": 5.091965866927828e-07,
+      "loss": 0.8131,
+      "step": 32808
+    },
+    {
+      "epoch": 0.9010738513086705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.089163963337118e-07,
+      "loss": 0.7661,
+      "step": 32809
+    },
+    {
+      "epoch": 0.9011013155365137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.086362810717715e-07,
+      "loss": 0.8558,
+      "step": 32810
+    },
+    {
+      "epoch": 0.9011287797643569,
+      "grad_norm": 0.0,
+      "learning_rate": 5.08356240909178e-07,
+      "loss": 0.7747,
+      "step": 32811
+    },
+    {
+      "epoch": 0.9011562439922002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.080762758481472e-07,
+      "loss": 0.7734,
+      "step": 32812
+    },
+    {
+      "epoch": 0.9011837082200433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.07796385890893e-07,
+      "loss": 0.818,
+      "step": 32813
+    },
+    {
+      "epoch": 0.9012111724478866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.075165710396313e-07,
+      "loss": 0.8402,
+      "step": 32814
+    },
+    {
+      "epoch": 0.9012386366757299,
+      "grad_norm": 0.0,
+      "learning_rate": 5.072368312965748e-07,
+      "loss": 0.8186,
+      "step": 32815
+    },
+    {
+      "epoch": 0.9012661009035731,
+      "grad_norm": 0.0,
+      "learning_rate": 5.069571666639373e-07,
+      "loss": 0.8046,
+      "step": 32816
+    },
+    {
+      "epoch": 0.9012935651314163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.066775771439325e-07,
+      "loss": 0.7532,
+      "step": 32817
+    },
+    {
+      "epoch": 0.9013210293592596,
+      "grad_norm": 0.0,
+      "learning_rate": 5.063980627387721e-07,
+      "loss": 0.8123,
+      "step": 32818
+    },
+    {
+      "epoch": 0.9013484935871028,
+      "grad_norm": 0.0,
+      "learning_rate": 5.061186234506676e-07,
+      "loss": 0.7968,
+      "step": 32819
+    },
+    {
+      "epoch": 0.9013759578149461,
+      "grad_norm": 0.0,
+      "learning_rate": 5.058392592818318e-07,
+      "loss": 0.8335,
+      "step": 32820
+    },
+    {
+      "epoch": 0.9014034220427892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.055599702344716e-07,
+      "loss": 0.8122,
+      "step": 32821
+    },
+    {
+      "epoch": 0.9014308862706325,
+      "grad_norm": 0.0,
+      "learning_rate": 5.052807563107986e-07,
+      "loss": 0.7613,
+      "step": 32822
+    },
+    {
+      "epoch": 0.9014583504984758,
+      "grad_norm": 0.0,
+      "learning_rate": 5.050016175130223e-07,
+      "loss": 0.857,
+      "step": 32823
+    },
+    {
+      "epoch": 0.901485814726319,
+      "grad_norm": 0.0,
+      "learning_rate": 5.047225538433509e-07,
+      "loss": 0.8031,
+      "step": 32824
+    },
+    {
+      "epoch": 0.9015132789541622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.044435653039926e-07,
+      "loss": 0.7296,
+      "step": 32825
+    },
+    {
+      "epoch": 0.9015407431820054,
+      "grad_norm": 0.0,
+      "learning_rate": 5.041646518971566e-07,
+      "loss": 0.9137,
+      "step": 32826
+    },
+    {
+      "epoch": 0.9015682074098487,
+      "grad_norm": 0.0,
+      "learning_rate": 5.03885813625048e-07,
+      "loss": 0.8,
+      "step": 32827
+    },
+    {
+      "epoch": 0.9015956716376919,
+      "grad_norm": 0.0,
+      "learning_rate": 5.036070504898727e-07,
+      "loss": 0.7767,
+      "step": 32828
+    },
+    {
+      "epoch": 0.9016231358655351,
+      "grad_norm": 0.0,
+      "learning_rate": 5.033283624938368e-07,
+      "loss": 0.8015,
+      "step": 32829
+    },
+    {
+      "epoch": 0.9016506000933784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.03049749639145e-07,
+      "loss": 0.7688,
+      "step": 32830
+    },
+    {
+      "epoch": 0.9016780643212217,
+      "grad_norm": 0.0,
+      "learning_rate": 5.027712119280026e-07,
+      "loss": 0.8489,
+      "step": 32831
+    },
+    {
+      "epoch": 0.9017055285490648,
+      "grad_norm": 0.0,
+      "learning_rate": 5.024927493626152e-07,
+      "loss": 0.7769,
+      "step": 32832
+    },
+    {
+      "epoch": 0.9017329927769081,
+      "grad_norm": 0.0,
+      "learning_rate": 5.022143619451825e-07,
+      "loss": 0.7643,
+      "step": 32833
+    },
+    {
+      "epoch": 0.9017604570047513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.01936049677908e-07,
+      "loss": 0.873,
+      "step": 32834
+    },
+    {
+      "epoch": 0.9017879212325945,
+      "grad_norm": 0.0,
+      "learning_rate": 5.016578125629967e-07,
+      "loss": 0.7141,
+      "step": 32835
+    },
+    {
+      "epoch": 0.9018153854604378,
+      "grad_norm": 0.0,
+      "learning_rate": 5.01379650602647e-07,
+      "loss": 0.8646,
+      "step": 32836
+    },
+    {
+      "epoch": 0.901842849688281,
+      "grad_norm": 0.0,
+      "learning_rate": 5.011015637990602e-07,
+      "loss": 0.7349,
+      "step": 32837
+    },
+    {
+      "epoch": 0.9018703139161243,
+      "grad_norm": 0.0,
+      "learning_rate": 5.008235521544391e-07,
+      "loss": 0.8605,
+      "step": 32838
+    },
+    {
+      "epoch": 0.9018977781439674,
+      "grad_norm": 0.0,
+      "learning_rate": 5.00545615670981e-07,
+      "loss": 0.816,
+      "step": 32839
+    },
+    {
+      "epoch": 0.9019252423718107,
+      "grad_norm": 0.0,
+      "learning_rate": 5.002677543508849e-07,
+      "loss": 0.8367,
+      "step": 32840
+    },
+    {
+      "epoch": 0.901952706599654,
+      "grad_norm": 0.0,
+      "learning_rate": 4.999899681963505e-07,
+      "loss": 0.8307,
+      "step": 32841
+    },
+    {
+      "epoch": 0.9019801708274972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.997122572095747e-07,
+      "loss": 0.8283,
+      "step": 32842
+    },
+    {
+      "epoch": 0.9020076350553404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.994346213927581e-07,
+      "loss": 0.7141,
+      "step": 32843
+    },
+    {
+      "epoch": 0.9020350992831837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.991570607480944e-07,
+      "loss": 0.8188,
+      "step": 32844
+    },
+    {
+      "epoch": 0.9020625635110269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.988795752777787e-07,
+      "loss": 0.7178,
+      "step": 32845
+    },
+    {
+      "epoch": 0.9020900277388701,
+      "grad_norm": 0.0,
+      "learning_rate": 4.986021649840078e-07,
+      "loss": 0.7738,
+      "step": 32846
+    },
+    {
+      "epoch": 0.9021174919667133,
+      "grad_norm": 0.0,
+      "learning_rate": 4.98324829868978e-07,
+      "loss": 0.8109,
+      "step": 32847
+    },
+    {
+      "epoch": 0.9021449561945566,
+      "grad_norm": 0.0,
+      "learning_rate": 4.980475699348819e-07,
+      "loss": 0.7601,
+      "step": 32848
+    },
+    {
+      "epoch": 0.9021724204223999,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977703851839144e-07,
+      "loss": 0.8429,
+      "step": 32849
+    },
+    {
+      "epoch": 0.902199884650243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.974932756182705e-07,
+      "loss": 0.799,
+      "step": 32850
+    },
+    {
+      "epoch": 0.9022273488780863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.972162412401394e-07,
+      "loss": 0.7942,
+      "step": 32851
+    },
+    {
+      "epoch": 0.9022548131059295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.969392820517149e-07,
+      "loss": 0.7614,
+      "step": 32852
+    },
+    {
+      "epoch": 0.9022822773337728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.966623980551865e-07,
+      "loss": 0.8992,
+      "step": 32853
+    },
+    {
+      "epoch": 0.902309741561616,
+      "grad_norm": 0.0,
+      "learning_rate": 4.96385589252747e-07,
+      "loss": 0.7732,
+      "step": 32854
+    },
+    {
+      "epoch": 0.9023372057894592,
+      "grad_norm": 0.0,
+      "learning_rate": 4.961088556465865e-07,
+      "loss": 0.8512,
+      "step": 32855
+    },
+    {
+      "epoch": 0.9023646700173025,
+      "grad_norm": 0.0,
+      "learning_rate": 4.958321972388946e-07,
+      "loss": 0.8744,
+      "step": 32856
+    },
+    {
+      "epoch": 0.9023921342451457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.955556140318585e-07,
+      "loss": 0.8695,
+      "step": 32857
+    },
+    {
+      "epoch": 0.9024195984729889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.952791060276685e-07,
+      "loss": 0.7772,
+      "step": 32858
+    },
+    {
+      "epoch": 0.9024470627008322,
+      "grad_norm": 0.0,
+      "learning_rate": 4.95002673228514e-07,
+      "loss": 0.8397,
+      "step": 32859
+    },
+    {
+      "epoch": 0.9024745269286754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.947263156365778e-07,
+      "loss": 0.7919,
+      "step": 32860
+    },
+    {
+      "epoch": 0.9025019911565186,
+      "grad_norm": 0.0,
+      "learning_rate": 4.944500332540503e-07,
+      "loss": 0.8118,
+      "step": 32861
+    },
+    {
+      "epoch": 0.9025294553843619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.941738260831153e-07,
+      "loss": 0.8749,
+      "step": 32862
+    },
+    {
+      "epoch": 0.9025569196122051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.938976941259599e-07,
+      "loss": 0.887,
+      "step": 32863
+    },
+    {
+      "epoch": 0.9025843838400484,
+      "grad_norm": 0.0,
+      "learning_rate": 4.936216373847669e-07,
+      "loss": 0.8341,
+      "step": 32864
+    },
+    {
+      "epoch": 0.9026118480678915,
+      "grad_norm": 0.0,
+      "learning_rate": 4.933456558617234e-07,
+      "loss": 0.8263,
+      "step": 32865
+    },
+    {
+      "epoch": 0.9026393122957348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.930697495590109e-07,
+      "loss": 0.8583,
+      "step": 32866
+    },
+    {
+      "epoch": 0.9026667765235781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.927939184788144e-07,
+      "loss": 0.8271,
+      "step": 32867
+    },
+    {
+      "epoch": 0.9026942407514212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.925181626233145e-07,
+      "loss": 0.8051,
+      "step": 32868
+    },
+    {
+      "epoch": 0.9027217049792645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.922424819946936e-07,
+      "loss": 0.8364,
+      "step": 32869
+    },
+    {
+      "epoch": 0.9027491692071078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.919668765951325e-07,
+      "loss": 0.7906,
+      "step": 32870
+    },
+    {
+      "epoch": 0.902776633434951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.916913464268125e-07,
+      "loss": 0.7159,
+      "step": 32871
+    },
+    {
+      "epoch": 0.9028040976627942,
+      "grad_norm": 0.0,
+      "learning_rate": 4.914158914919131e-07,
+      "loss": 0.7413,
+      "step": 32872
+    },
+    {
+      "epoch": 0.9028315618906374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.911405117926172e-07,
+      "loss": 0.8192,
+      "step": 32873
+    },
+    {
+      "epoch": 0.9028590261184807,
+      "grad_norm": 0.0,
+      "learning_rate": 4.908652073310993e-07,
+      "loss": 0.7069,
+      "step": 32874
+    },
+    {
+      "epoch": 0.902886490346324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.905899781095391e-07,
+      "loss": 0.844,
+      "step": 32875
+    },
+    {
+      "epoch": 0.9029139545741671,
+      "grad_norm": 0.0,
+      "learning_rate": 4.903148241301159e-07,
+      "loss": 0.8746,
+      "step": 32876
+    },
+    {
+      "epoch": 0.9029414188020104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.900397453950035e-07,
+      "loss": 0.7977,
+      "step": 32877
+    },
+    {
+      "epoch": 0.9029688830298536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.897647419063811e-07,
+      "loss": 0.8092,
+      "step": 32878
+    },
+    {
+      "epoch": 0.9029963472576968,
+      "grad_norm": 0.0,
+      "learning_rate": 4.89489813666425e-07,
+      "loss": 0.7961,
+      "step": 32879
+    },
+    {
+      "epoch": 0.9030238114855401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.892149606773089e-07,
+      "loss": 0.7053,
+      "step": 32880
+    },
+    {
+      "epoch": 0.9030512757133833,
+      "grad_norm": 0.0,
+      "learning_rate": 4.889401829412077e-07,
+      "loss": 0.7365,
+      "step": 32881
+    },
+    {
+      "epoch": 0.9030787399412266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.886654804602963e-07,
+      "loss": 0.8542,
+      "step": 32882
+    },
+    {
+      "epoch": 0.9031062041690698,
+      "grad_norm": 0.0,
+      "learning_rate": 4.883908532367487e-07,
+      "loss": 0.8369,
+      "step": 32883
+    },
+    {
+      "epoch": 0.903133668396913,
+      "grad_norm": 0.0,
+      "learning_rate": 4.881163012727363e-07,
+      "loss": 0.8015,
+      "step": 32884
+    },
+    {
+      "epoch": 0.9031611326247563,
+      "grad_norm": 0.0,
+      "learning_rate": 4.878418245704331e-07,
+      "loss": 0.7517,
+      "step": 32885
+    },
+    {
+      "epoch": 0.9031885968525994,
+      "grad_norm": 0.0,
+      "learning_rate": 4.875674231320093e-07,
+      "loss": 0.8646,
+      "step": 32886
+    },
+    {
+      "epoch": 0.9032160610804427,
+      "grad_norm": 0.0,
+      "learning_rate": 4.872930969596368e-07,
+      "loss": 0.86,
+      "step": 32887
+    },
+    {
+      "epoch": 0.903243525308286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.870188460554848e-07,
+      "loss": 0.8312,
+      "step": 32888
+    },
+    {
+      "epoch": 0.9032709895361292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.86744670421726e-07,
+      "loss": 0.7816,
+      "step": 32889
+    },
+    {
+      "epoch": 0.9032984537639724,
+      "grad_norm": 0.0,
+      "learning_rate": 4.864705700605288e-07,
+      "loss": 0.8218,
+      "step": 32890
+    },
+    {
+      "epoch": 0.9033259179918156,
+      "grad_norm": 0.0,
+      "learning_rate": 4.861965449740624e-07,
+      "loss": 0.7914,
+      "step": 32891
+    },
+    {
+      "epoch": 0.9033533822196589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.859225951644941e-07,
+      "loss": 0.8071,
+      "step": 32892
+    },
+    {
+      "epoch": 0.9033808464475022,
+      "grad_norm": 0.0,
+      "learning_rate": 4.856487206339899e-07,
+      "loss": 0.8068,
+      "step": 32893
+    },
+    {
+      "epoch": 0.9034083106753453,
+      "grad_norm": 0.0,
+      "learning_rate": 4.853749213847192e-07,
+      "loss": 0.8855,
+      "step": 32894
+    },
+    {
+      "epoch": 0.9034357749031886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.85101197418848e-07,
+      "loss": 0.8658,
+      "step": 32895
+    },
+    {
+      "epoch": 0.9034632391310319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.848275487385412e-07,
+      "loss": 0.8036,
+      "step": 32896
+    },
+    {
+      "epoch": 0.903490703358875,
+      "grad_norm": 0.0,
+      "learning_rate": 4.845539753459661e-07,
+      "loss": 0.7996,
+      "step": 32897
+    },
+    {
+      "epoch": 0.9035181675867183,
+      "grad_norm": 0.0,
+      "learning_rate": 4.842804772432852e-07,
+      "loss": 0.75,
+      "step": 32898
+    },
+    {
+      "epoch": 0.9035456318145615,
+      "grad_norm": 0.0,
+      "learning_rate": 4.840070544326625e-07,
+      "loss": 0.7149,
+      "step": 32899
+    },
+    {
+      "epoch": 0.9035730960424048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.837337069162629e-07,
+      "loss": 0.8677,
+      "step": 32900
+    },
+    {
+      "epoch": 0.903600560270248,
+      "grad_norm": 0.0,
+      "learning_rate": 4.834604346962479e-07,
+      "loss": 0.8267,
+      "step": 32901
+    },
+    {
+      "epoch": 0.9036280244980912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.831872377747804e-07,
+      "loss": 0.8903,
+      "step": 32902
+    },
+    {
+      "epoch": 0.9036554887259345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.829141161540229e-07,
+      "loss": 0.8194,
+      "step": 32903
+    },
+    {
+      "epoch": 0.9036829529537777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.82641069836135e-07,
+      "loss": 0.8119,
+      "step": 32904
+    },
+    {
+      "epoch": 0.9037104171816209,
+      "grad_norm": 0.0,
+      "learning_rate": 4.823680988232771e-07,
+      "loss": 0.7383,
+      "step": 32905
+    },
+    {
+      "epoch": 0.9037378814094642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.820952031176096e-07,
+      "loss": 0.7008,
+      "step": 32906
+    },
+    {
+      "epoch": 0.9037653456373074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.818223827212931e-07,
+      "loss": 0.8563,
+      "step": 32907
+    },
+    {
+      "epoch": 0.9037928098651506,
+      "grad_norm": 0.0,
+      "learning_rate": 4.815496376364825e-07,
+      "loss": 0.8612,
+      "step": 32908
+    },
+    {
+      "epoch": 0.9038202740929939,
+      "grad_norm": 0.0,
+      "learning_rate": 4.812769678653407e-07,
+      "loss": 0.6911,
+      "step": 32909
+    },
+    {
+      "epoch": 0.9038477383208371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.810043734100211e-07,
+      "loss": 0.7413,
+      "step": 32910
+    },
+    {
+      "epoch": 0.9038752025486804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.807318542726824e-07,
+      "loss": 0.7438,
+      "step": 32911
+    },
+    {
+      "epoch": 0.9039026667765235,
+      "grad_norm": 0.0,
+      "learning_rate": 4.804594104554805e-07,
+      "loss": 0.6583,
+      "step": 32912
+    },
+    {
+      "epoch": 0.9039301310043668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.801870419605714e-07,
+      "loss": 0.7997,
+      "step": 32913
+    },
+    {
+      "epoch": 0.9039575952322101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.799147487901101e-07,
+      "loss": 0.8151,
+      "step": 32914
+    },
+    {
+      "epoch": 0.9039850594600533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.796425309462526e-07,
+      "loss": 0.867,
+      "step": 32915
+    },
+    {
+      "epoch": 0.9040125236878965,
+      "grad_norm": 0.0,
+      "learning_rate": 4.793703884311506e-07,
+      "loss": 0.8178,
+      "step": 32916
+    },
+    {
+      "epoch": 0.9040399879157397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.790983212469568e-07,
+      "loss": 0.8031,
+      "step": 32917
+    },
+    {
+      "epoch": 0.904067452143583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.78826329395825e-07,
+      "loss": 0.8624,
+      "step": 32918
+    },
+    {
+      "epoch": 0.9040949163714262,
+      "grad_norm": 0.0,
+      "learning_rate": 4.78554412879908e-07,
+      "loss": 0.7957,
+      "step": 32919
+    },
+    {
+      "epoch": 0.9041223805992694,
+      "grad_norm": 0.0,
+      "learning_rate": 4.782825717013583e-07,
+      "loss": 0.9164,
+      "step": 32920
+    },
+    {
+      "epoch": 0.9041498448271127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.780108058623234e-07,
+      "loss": 0.7865,
+      "step": 32921
+    },
+    {
+      "epoch": 0.904177309054956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.777391153649558e-07,
+      "loss": 0.7494,
+      "step": 32922
+    },
+    {
+      "epoch": 0.9042047732827991,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774675002114038e-07,
+      "loss": 0.7018,
+      "step": 32923
+    },
+    {
+      "epoch": 0.9042322375106424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.771959604038201e-07,
+      "loss": 0.7351,
+      "step": 32924
+    },
+    {
+      "epoch": 0.9042597017384856,
+      "grad_norm": 0.0,
+      "learning_rate": 4.769244959443486e-07,
+      "loss": 0.8201,
+      "step": 32925
+    },
+    {
+      "epoch": 0.9042871659663289,
+      "grad_norm": 0.0,
+      "learning_rate": 4.766531068351421e-07,
+      "loss": 0.8793,
+      "step": 32926
+    },
+    {
+      "epoch": 0.9043146301941721,
+      "grad_norm": 0.0,
+      "learning_rate": 4.763817930783421e-07,
+      "loss": 0.8503,
+      "step": 32927
+    },
+    {
+      "epoch": 0.9043420944220153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7611055467610026e-07,
+      "loss": 0.8221,
+      "step": 32928
+    },
+    {
+      "epoch": 0.9043695586498586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7583939163055924e-07,
+      "loss": 0.775,
+      "step": 32929
+    },
+    {
+      "epoch": 0.9043970228777017,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7556830394386745e-07,
+      "loss": 0.8087,
+      "step": 32930
+    },
+    {
+      "epoch": 0.904424487105545,
+      "grad_norm": 0.0,
+      "learning_rate": 4.752972916181686e-07,
+      "loss": 0.8158,
+      "step": 32931
+    },
+    {
+      "epoch": 0.9044519513333883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7502635465560773e-07,
+      "loss": 0.7161,
+      "step": 32932
+    },
+    {
+      "epoch": 0.9044794155612315,
+      "grad_norm": 0.0,
+      "learning_rate": 4.747554930583276e-07,
+      "loss": 0.7953,
+      "step": 32933
+    },
+    {
+      "epoch": 0.9045068797890747,
+      "grad_norm": 0.0,
+      "learning_rate": 4.744847068284708e-07,
+      "loss": 0.8462,
+      "step": 32934
+    },
+    {
+      "epoch": 0.904534344016918,
+      "grad_norm": 0.0,
+      "learning_rate": 4.742139959681813e-07,
+      "loss": 0.751,
+      "step": 32935
+    },
+    {
+      "epoch": 0.9045618082447612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.739433604796007e-07,
+      "loss": 0.8312,
+      "step": 32936
+    },
+    {
+      "epoch": 0.9045892724726045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.736728003648694e-07,
+      "loss": 0.7809,
+      "step": 32937
+    },
+    {
+      "epoch": 0.9046167367004476,
+      "grad_norm": 0.0,
+      "learning_rate": 4.734023156261314e-07,
+      "loss": 0.7893,
+      "step": 32938
+    },
+    {
+      "epoch": 0.9046442009282909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7313190626552263e-07,
+      "loss": 0.7748,
+      "step": 32939
+    },
+    {
+      "epoch": 0.9046716651561342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7286157228518593e-07,
+      "loss": 0.8136,
+      "step": 32940
+    },
+    {
+      "epoch": 0.9046991293839773,
+      "grad_norm": 0.0,
+      "learning_rate": 4.725913136872573e-07,
+      "loss": 0.8493,
+      "step": 32941
+    },
+    {
+      "epoch": 0.9047265936118206,
+      "grad_norm": 0.0,
+      "learning_rate": 4.723211304738773e-07,
+      "loss": 0.7036,
+      "step": 32942
+    },
+    {
+      "epoch": 0.9047540578396638,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7205102264718304e-07,
+      "loss": 0.7578,
+      "step": 32943
+    },
+    {
+      "epoch": 0.9047815220675071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.717809902093129e-07,
+      "loss": 0.7438,
+      "step": 32944
+    },
+    {
+      "epoch": 0.9048089862953503,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7151103316240174e-07,
+      "loss": 0.898,
+      "step": 32945
+    },
+    {
+      "epoch": 0.9048364505231935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7124115150858576e-07,
+      "loss": 0.7921,
+      "step": 32946
+    },
+    {
+      "epoch": 0.9048639147510368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7097134525000087e-07,
+      "loss": 0.7771,
+      "step": 32947
+    },
+    {
+      "epoch": 0.9048913789788801,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7070161438878323e-07,
+      "loss": 0.8171,
+      "step": 32948
+    },
+    {
+      "epoch": 0.9049188432067232,
+      "grad_norm": 0.0,
+      "learning_rate": 4.704319589270645e-07,
+      "loss": 0.8309,
+      "step": 32949
+    },
+    {
+      "epoch": 0.9049463074345665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.701623788669818e-07,
+      "loss": 0.6854,
+      "step": 32950
+    },
+    {
+      "epoch": 0.9049737716624097,
+      "grad_norm": 0.0,
+      "learning_rate": 4.698928742106634e-07,
+      "loss": 0.7439,
+      "step": 32951
+    },
+    {
+      "epoch": 0.9050012358902529,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6962344496024435e-07,
+      "loss": 0.8059,
+      "step": 32952
+    },
+    {
+      "epoch": 0.9050287001180962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6935409111785624e-07,
+      "loss": 0.8084,
+      "step": 32953
+    },
+    {
+      "epoch": 0.9050561643459394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6908481268563066e-07,
+      "loss": 0.8857,
+      "step": 32954
+    },
+    {
+      "epoch": 0.9050836285737827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.688156096656993e-07,
+      "loss": 0.8027,
+      "step": 32955
+    },
+    {
+      "epoch": 0.9051110928016258,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6854648206019037e-07,
+      "loss": 0.7682,
+      "step": 32956
+    },
+    {
+      "epoch": 0.9051385570294691,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6827742987123225e-07,
+      "loss": 0.7566,
+      "step": 32957
+    },
+    {
+      "epoch": 0.9051660212573124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6800845310095655e-07,
+      "loss": 0.7198,
+      "step": 32958
+    },
+    {
+      "epoch": 0.9051934854851555,
+      "grad_norm": 0.0,
+      "learning_rate": 4.677395517514893e-07,
+      "loss": 0.905,
+      "step": 32959
+    },
+    {
+      "epoch": 0.9052209497129988,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6747072582496e-07,
+      "loss": 0.7677,
+      "step": 32960
+    },
+    {
+      "epoch": 0.9052484139408421,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6720197532349466e-07,
+      "loss": 0.8048,
+      "step": 32961
+    },
+    {
+      "epoch": 0.9052758781686853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6693330024922046e-07,
+      "loss": 0.775,
+      "step": 32962
+    },
+    {
+      "epoch": 0.9053033423965285,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6666470060426237e-07,
+      "loss": 0.8219,
+      "step": 32963
+    },
+    {
+      "epoch": 0.9053308066243717,
+      "grad_norm": 0.0,
+      "learning_rate": 4.663961763907465e-07,
+      "loss": 0.8181,
+      "step": 32964
+    },
+    {
+      "epoch": 0.905358270852215,
+      "grad_norm": 0.0,
+      "learning_rate": 4.661277276107967e-07,
+      "loss": 0.8153,
+      "step": 32965
+    },
+    {
+      "epoch": 0.9053857350800583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.658593542665368e-07,
+      "loss": 0.8087,
+      "step": 32966
+    },
+    {
+      "epoch": 0.9054131993079014,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6559105636009075e-07,
+      "loss": 0.8287,
+      "step": 32967
+    },
+    {
+      "epoch": 0.9054406635357447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.653228338935822e-07,
+      "loss": 0.7458,
+      "step": 32968
+    },
+    {
+      "epoch": 0.905468127763588,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6505468686913305e-07,
+      "loss": 0.8172,
+      "step": 32969
+    },
+    {
+      "epoch": 0.9054955919914311,
+      "grad_norm": 0.0,
+      "learning_rate": 4.647866152888636e-07,
+      "loss": 0.8448,
+      "step": 32970
+    },
+    {
+      "epoch": 0.9055230562192744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.645186191548967e-07,
+      "loss": 0.7151,
+      "step": 32971
+    },
+    {
+      "epoch": 0.9055505204471176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.642506984693529e-07,
+      "loss": 0.7826,
+      "step": 32972
+    },
+    {
+      "epoch": 0.9055779846749609,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6398285323435043e-07,
+      "loss": 0.801,
+      "step": 32973
+    },
+    {
+      "epoch": 0.9056054489028041,
+      "grad_norm": 0.0,
+      "learning_rate": 4.63715083452011e-07,
+      "loss": 0.9194,
+      "step": 32974
+    },
+    {
+      "epoch": 0.9056329131306473,
+      "grad_norm": 0.0,
+      "learning_rate": 4.634473891244495e-07,
+      "loss": 0.8009,
+      "step": 32975
+    },
+    {
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.631797702537877e-07,
+      "loss": 0.8022,
+      "step": 32976
+    },
+    {
+      "epoch": 0.9056878415863338,
+      "grad_norm": 0.0,
+      "learning_rate": 4.629122268421404e-07,
+      "loss": 0.7702,
+      "step": 32977
+    },
+    {
+      "epoch": 0.905715305814177,
+      "grad_norm": 0.0,
+      "learning_rate": 4.626447588916272e-07,
+      "loss": 0.8389,
+      "step": 32978
+    },
+    {
+      "epoch": 0.9057427700420203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.62377366404364e-07,
+      "loss": 0.7639,
+      "step": 32979
+    },
+    {
+      "epoch": 0.9057702342698635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6211004938246483e-07,
+      "loss": 0.7961,
+      "step": 32980
+    },
+    {
+      "epoch": 0.9057976984977067,
+      "grad_norm": 0.0,
+      "learning_rate": 4.618428078280457e-07,
+      "loss": 0.8123,
+      "step": 32981
+    },
+    {
+      "epoch": 0.90582516272555,
+      "grad_norm": 0.0,
+      "learning_rate": 4.615756417432193e-07,
+      "loss": 0.6558,
+      "step": 32982
+    },
+    {
+      "epoch": 0.9058526269533932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6130855113010297e-07,
+      "loss": 0.7231,
+      "step": 32983
+    },
+    {
+      "epoch": 0.9058800911812365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6104153599080717e-07,
+      "loss": 0.8373,
+      "step": 32984
+    },
+    {
+      "epoch": 0.9059075554090796,
+      "grad_norm": 0.0,
+      "learning_rate": 4.607745963274479e-07,
+      "loss": 0.8886,
+      "step": 32985
+    },
+    {
+      "epoch": 0.9059350196369229,
+      "grad_norm": 0.0,
+      "learning_rate": 4.605077321421325e-07,
+      "loss": 0.7744,
+      "step": 32986
+    },
+    {
+      "epoch": 0.9059624838647662,
+      "grad_norm": 0.0,
+      "learning_rate": 4.602409434369759e-07,
+      "loss": 0.8392,
+      "step": 32987
+    },
+    {
+      "epoch": 0.9059899480926094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.599742302140897e-07,
+      "loss": 0.8754,
+      "step": 32988
+    },
+    {
+      "epoch": 0.9060174123204526,
+      "grad_norm": 0.0,
+      "learning_rate": 4.597075924755823e-07,
+      "loss": 0.7822,
+      "step": 32989
+    },
+    {
+      "epoch": 0.9060448765482958,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5944103022356304e-07,
+      "loss": 0.8831,
+      "step": 32990
+    },
+    {
+      "epoch": 0.9060723407761391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5917454346014357e-07,
+      "loss": 0.766,
+      "step": 32991
+    },
+    {
+      "epoch": 0.9060998050039824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5890813218743005e-07,
+      "loss": 0.7281,
+      "step": 32992
+    },
+    {
+      "epoch": 0.9061272692318255,
+      "grad_norm": 0.0,
+      "learning_rate": 4.586417964075307e-07,
+      "loss": 0.772,
+      "step": 32993
+    },
+    {
+      "epoch": 0.9061547334596688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.583755361225528e-07,
+      "loss": 0.851,
+      "step": 32994
+    },
+    {
+      "epoch": 0.9061821976875121,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5810935133460575e-07,
+      "loss": 0.8142,
+      "step": 32995
+    },
+    {
+      "epoch": 0.9062096619153552,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5784324204579233e-07,
+      "loss": 0.8147,
+      "step": 32996
+    },
+    {
+      "epoch": 0.9062371261431985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5757720825822083e-07,
+      "loss": 0.8444,
+      "step": 32997
+    },
+    {
+      "epoch": 0.9062645903710417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.57311249973994e-07,
+      "loss": 0.8999,
+      "step": 32998
+    },
+    {
+      "epoch": 0.906292054598885,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5704536719521574e-07,
+      "loss": 0.8686,
+      "step": 32999
+    },
+    {
+      "epoch": 0.9063195188267282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5677955992399217e-07,
+      "loss": 0.7925,
+      "step": 33000
+    },
+    {
+      "epoch": 0.9063469830545714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5651382816242483e-07,
+      "loss": 0.8608,
+      "step": 33001
+    },
+    {
+      "epoch": 0.9063744472824147,
+      "grad_norm": 0.0,
+      "learning_rate": 4.562481719126177e-07,
+      "loss": 0.7812,
+      "step": 33002
+    },
+    {
+      "epoch": 0.9064019115102578,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5598259117667245e-07,
+      "loss": 0.8519,
+      "step": 33003
+    },
+    {
+      "epoch": 0.9064293757381011,
+      "grad_norm": 0.0,
+      "learning_rate": 4.557170859566906e-07,
+      "loss": 0.7089,
+      "step": 33004
+    },
+    {
+      "epoch": 0.9064568399659444,
+      "grad_norm": 0.0,
+      "learning_rate": 4.554516562547706e-07,
+      "loss": 0.8019,
+      "step": 33005
+    },
+    {
+      "epoch": 0.9064843041937876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5518630207301405e-07,
+      "loss": 0.8519,
+      "step": 33006
+    },
+    {
+      "epoch": 0.9065117684216308,
+      "grad_norm": 0.0,
+      "learning_rate": 4.549210234135204e-07,
+      "loss": 0.858,
+      "step": 33007
+    },
+    {
+      "epoch": 0.9065392326494741,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546558202783902e-07,
+      "loss": 0.8175,
+      "step": 33008
+    },
+    {
+      "epoch": 0.9065666968773173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.543906926697217e-07,
+      "loss": 0.7942,
+      "step": 33009
+    },
+    {
+      "epoch": 0.9065941611051606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5412564058960996e-07,
+      "loss": 0.7789,
+      "step": 33010
+    },
+    {
+      "epoch": 0.9066216253330037,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5386066404015327e-07,
+      "loss": 0.7651,
+      "step": 33011
+    },
+    {
+      "epoch": 0.906649089560847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.535957630234511e-07,
+      "loss": 0.8607,
+      "step": 33012
+    },
+    {
+      "epoch": 0.9066765537886903,
+      "grad_norm": 0.0,
+      "learning_rate": 4.533309375415951e-07,
+      "loss": 0.6832,
+      "step": 33013
+    },
+    {
+      "epoch": 0.9067040180165334,
+      "grad_norm": 0.0,
+      "learning_rate": 4.530661875966835e-07,
+      "loss": 0.8197,
+      "step": 33014
+    },
+    {
+      "epoch": 0.9067314822443767,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5280151319081144e-07,
+      "loss": 0.8714,
+      "step": 33015
+    },
+    {
+      "epoch": 0.9067589464722199,
+      "grad_norm": 0.0,
+      "learning_rate": 4.525369143260705e-07,
+      "loss": 0.7689,
+      "step": 33016
+    },
+    {
+      "epoch": 0.9067864107000632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.522723910045568e-07,
+      "loss": 0.795,
+      "step": 33017
+    },
+    {
+      "epoch": 0.9068138749279064,
+      "grad_norm": 0.0,
+      "learning_rate": 4.520079432283608e-07,
+      "loss": 0.815,
+      "step": 33018
+    },
+    {
+      "epoch": 0.9068413391557496,
+      "grad_norm": 0.0,
+      "learning_rate": 4.517435709995788e-07,
+      "loss": 0.7401,
+      "step": 33019
+    },
+    {
+      "epoch": 0.9068688033835929,
+      "grad_norm": 0.0,
+      "learning_rate": 4.514792743202989e-07,
+      "loss": 0.8685,
+      "step": 33020
+    },
+    {
+      "epoch": 0.9068962676114362,
+      "grad_norm": 0.0,
+      "learning_rate": 4.512150531926152e-07,
+      "loss": 0.8574,
+      "step": 33021
+    },
+    {
+      "epoch": 0.9069237318392793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.509509076186147e-07,
+      "loss": 0.7443,
+      "step": 33022
+    },
+    {
+      "epoch": 0.9069511960671226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5068683760039033e-07,
+      "loss": 0.8183,
+      "step": 33023
+    },
+    {
+      "epoch": 0.9069786602949658,
+      "grad_norm": 0.0,
+      "learning_rate": 4.504228431400304e-07,
+      "loss": 0.8367,
+      "step": 33024
+    },
+    {
+      "epoch": 0.907006124522809,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5015892423962426e-07,
+      "loss": 0.7739,
+      "step": 33025
+    },
+    {
+      "epoch": 0.9070335887506523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4989508090126033e-07,
+      "loss": 0.843,
+      "step": 33026
+    },
+    {
+      "epoch": 0.9070610529784955,
+      "grad_norm": 0.0,
+      "learning_rate": 4.496313131270269e-07,
+      "loss": 0.8474,
+      "step": 33027
+    },
+    {
+      "epoch": 0.9070885172063388,
+      "grad_norm": 0.0,
+      "learning_rate": 4.493676209190101e-07,
+      "loss": 0.8397,
+      "step": 33028
+    },
+    {
+      "epoch": 0.9071159814341819,
+      "grad_norm": 0.0,
+      "learning_rate": 4.491040042792938e-07,
+      "loss": 0.6884,
+      "step": 33029
+    },
+    {
+      "epoch": 0.9071434456620252,
+      "grad_norm": 0.0,
+      "learning_rate": 4.488404632099674e-07,
+      "loss": 0.8536,
+      "step": 33030
+    },
+    {
+      "epoch": 0.9071709098898685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.485769977131149e-07,
+      "loss": 0.6913,
+      "step": 33031
+    },
+    {
+      "epoch": 0.9071983741177116,
+      "grad_norm": 0.0,
+      "learning_rate": 4.483136077908212e-07,
+      "loss": 0.7788,
+      "step": 33032
+    },
+    {
+      "epoch": 0.9072258383455549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4805029344517134e-07,
+      "loss": 0.801,
+      "step": 33033
+    },
+    {
+      "epoch": 0.9072533025733982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.477870546782459e-07,
+      "loss": 0.8303,
+      "step": 33034
+    },
+    {
+      "epoch": 0.9072807668012414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.475238914921298e-07,
+      "loss": 0.7202,
+      "step": 33035
+    },
+    {
+      "epoch": 0.9073082310290846,
+      "grad_norm": 0.0,
+      "learning_rate": 4.47260803888907e-07,
+      "loss": 0.8011,
+      "step": 33036
+    },
+    {
+      "epoch": 0.9073356952569278,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4699779187065474e-07,
+      "loss": 0.8306,
+      "step": 33037
+    },
+    {
+      "epoch": 0.9073631594847711,
+      "grad_norm": 0.0,
+      "learning_rate": 4.467348554394579e-07,
+      "loss": 0.7939,
+      "step": 33038
+    },
+    {
+      "epoch": 0.9073906237126144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.464719945973939e-07,
+      "loss": 0.7977,
+      "step": 33039
+    },
+    {
+      "epoch": 0.9074180879404575,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4620920934654533e-07,
+      "loss": 0.7915,
+      "step": 33040
+    },
+    {
+      "epoch": 0.9074455521683008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.459464996889895e-07,
+      "loss": 0.8362,
+      "step": 33041
+    },
+    {
+      "epoch": 0.907473016396144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4568386562680586e-07,
+      "loss": 0.8125,
+      "step": 33042
+    },
+    {
+      "epoch": 0.9075004806239872,
+      "grad_norm": 0.0,
+      "learning_rate": 4.454213071620739e-07,
+      "loss": 0.8459,
+      "step": 33043
+    },
+    {
+      "epoch": 0.9075279448518305,
+      "grad_norm": 0.0,
+      "learning_rate": 4.451588242968674e-07,
+      "loss": 0.9038,
+      "step": 33044
+    },
+    {
+      "epoch": 0.9075554090796737,
+      "grad_norm": 0.0,
+      "learning_rate": 4.448964170332681e-07,
+      "loss": 0.8749,
+      "step": 33045
+    },
+    {
+      "epoch": 0.907582873307517,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4463408537334774e-07,
+      "loss": 0.8902,
+      "step": 33046
+    },
+    {
+      "epoch": 0.9076103375353602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4437182931918453e-07,
+      "loss": 0.8189,
+      "step": 33047
+    },
+    {
+      "epoch": 0.9076378017632034,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4410964887285244e-07,
+      "loss": 0.8193,
+      "step": 33048
+    },
+    {
+      "epoch": 0.9076652659910467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4384754403642536e-07,
+      "loss": 0.7859,
+      "step": 33049
+    },
+    {
+      "epoch": 0.9076927302188899,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4358551481198055e-07,
+      "loss": 0.7405,
+      "step": 33050
+    },
+    {
+      "epoch": 0.9077201944467331,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4332356120158847e-07,
+      "loss": 0.8213,
+      "step": 33051
+    },
+    {
+      "epoch": 0.9077476586745764,
+      "grad_norm": 0.0,
+      "learning_rate": 4.43061683207322e-07,
+      "loss": 0.8832,
+      "step": 33052
+    },
+    {
+      "epoch": 0.9077751229024196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.427998808312528e-07,
+      "loss": 0.7777,
+      "step": 33053
+    },
+    {
+      "epoch": 0.9078025871302629,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4253815407545366e-07,
+      "loss": 0.7867,
+      "step": 33054
+    },
+    {
+      "epoch": 0.907830051358106,
+      "grad_norm": 0.0,
+      "learning_rate": 4.42276502941994e-07,
+      "loss": 0.8648,
+      "step": 33055
+    },
+    {
+      "epoch": 0.9078575155859493,
+      "grad_norm": 0.0,
+      "learning_rate": 4.420149274329466e-07,
+      "loss": 0.8051,
+      "step": 33056
+    },
+    {
+      "epoch": 0.9078849798137926,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4175342755037875e-07,
+      "loss": 0.8859,
+      "step": 33057
+    },
+    {
+      "epoch": 0.9079124440416357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4149200329635875e-07,
+      "loss": 0.7865,
+      "step": 33058
+    },
+    {
+      "epoch": 0.907939908269479,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4123065467295834e-07,
+      "loss": 0.8382,
+      "step": 33059
+    },
+    {
+      "epoch": 0.9079673724973223,
+      "grad_norm": 0.0,
+      "learning_rate": 4.409693816822436e-07,
+      "loss": 0.7537,
+      "step": 33060
+    },
+    {
+      "epoch": 0.9079948367251655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.407081843262817e-07,
+      "loss": 0.8171,
+      "step": 33061
+    },
+    {
+      "epoch": 0.9080223009530087,
+      "grad_norm": 0.0,
+      "learning_rate": 4.404470626071411e-07,
+      "loss": 0.957,
+      "step": 33062
+    },
+    {
+      "epoch": 0.9080497651808519,
+      "grad_norm": 0.0,
+      "learning_rate": 4.401860165268834e-07,
+      "loss": 0.9104,
+      "step": 33063
+    },
+    {
+      "epoch": 0.9080772294086952,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3992504608757815e-07,
+      "loss": 0.8358,
+      "step": 33064
+    },
+    {
+      "epoch": 0.9081046936365385,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3966415129128916e-07,
+      "loss": 0.8329,
+      "step": 33065
+    },
+    {
+      "epoch": 0.9081321578643816,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3940333214008146e-07,
+      "loss": 0.8754,
+      "step": 33066
+    },
+    {
+      "epoch": 0.9081596220922249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391425886360179e-07,
+      "loss": 0.7495,
+      "step": 33067
+    },
+    {
+      "epoch": 0.9081870863200681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.388819207811612e-07,
+      "loss": 0.7951,
+      "step": 33068
+    },
+    {
+      "epoch": 0.9082145505479113,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3862132857757535e-07,
+      "loss": 0.8195,
+      "step": 33069
+    },
+    {
+      "epoch": 0.9082420147757546,
+      "grad_norm": 0.0,
+      "learning_rate": 4.383608120273197e-07,
+      "loss": 0.7784,
+      "step": 33070
+    },
+    {
+      "epoch": 0.9082694790035978,
+      "grad_norm": 0.0,
+      "learning_rate": 4.381003711324583e-07,
+      "loss": 0.8321,
+      "step": 33071
+    },
+    {
+      "epoch": 0.9082969432314411,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3784000589504937e-07,
+      "loss": 0.8494,
+      "step": 33072
+    },
+    {
+      "epoch": 0.9083244074592843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3757971631715466e-07,
+      "loss": 0.7861,
+      "step": 33073
+    },
+    {
+      "epoch": 0.9083518716871275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.373195024008359e-07,
+      "loss": 0.7941,
+      "step": 33074
+    },
+    {
+      "epoch": 0.9083793359149708,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3705936414814685e-07,
+      "loss": 0.788,
+      "step": 33075
+    },
+    {
+      "epoch": 0.9084068001428139,
+      "grad_norm": 0.0,
+      "learning_rate": 4.367993015611494e-07,
+      "loss": 0.807,
+      "step": 33076
+    },
+    {
+      "epoch": 0.9084342643706572,
+      "grad_norm": 0.0,
+      "learning_rate": 4.365393146419006e-07,
+      "loss": 0.7967,
+      "step": 33077
+    },
+    {
+      "epoch": 0.9084617285985005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3627940339245556e-07,
+      "loss": 0.8257,
+      "step": 33078
+    },
+    {
+      "epoch": 0.9084891928263437,
+      "grad_norm": 0.0,
+      "learning_rate": 4.360195678148738e-07,
+      "loss": 0.7802,
+      "step": 33079
+    },
+    {
+      "epoch": 0.9085166570541869,
+      "grad_norm": 0.0,
+      "learning_rate": 4.357598079112102e-07,
+      "loss": 0.7893,
+      "step": 33080
+    },
+    {
+      "epoch": 0.9085441212820301,
+      "grad_norm": 0.0,
+      "learning_rate": 4.355001236835188e-07,
+      "loss": 0.8468,
+      "step": 33081
+    },
+    {
+      "epoch": 0.9085715855098734,
+      "grad_norm": 0.0,
+      "learning_rate": 4.352405151338557e-07,
+      "loss": 0.7366,
+      "step": 33082
+    },
+    {
+      "epoch": 0.9085990497377167,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3498098226427477e-07,
+      "loss": 0.597,
+      "step": 33083
+    },
+    {
+      "epoch": 0.9086265139655598,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3472152507683107e-07,
+      "loss": 0.7289,
+      "step": 33084
+    },
+    {
+      "epoch": 0.9086539781934031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3446214357357406e-07,
+      "loss": 0.8333,
+      "step": 33085
+    },
+    {
+      "epoch": 0.9086814424212464,
+      "grad_norm": 0.0,
+      "learning_rate": 4.342028377565588e-07,
+      "loss": 0.789,
+      "step": 33086
+    },
+    {
+      "epoch": 0.9087089066490895,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3394360762783584e-07,
+      "loss": 0.8157,
+      "step": 33087
+    },
+    {
+      "epoch": 0.9087363708769328,
+      "grad_norm": 0.0,
+      "learning_rate": 4.336844531894557e-07,
+      "loss": 0.7639,
+      "step": 33088
+    },
+    {
+      "epoch": 0.908763835104776,
+      "grad_norm": 0.0,
+      "learning_rate": 4.334253744434702e-07,
+      "loss": 0.8528,
+      "step": 33089
+    },
+    {
+      "epoch": 0.9087912993326193,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3316637139192873e-07,
+      "loss": 0.863,
+      "step": 33090
+    },
+    {
+      "epoch": 0.9088187635604625,
+      "grad_norm": 0.0,
+      "learning_rate": 4.329074440368819e-07,
+      "loss": 0.84,
+      "step": 33091
+    },
+    {
+      "epoch": 0.9088462277883057,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3264859238037584e-07,
+      "loss": 0.8047,
+      "step": 33092
+    },
+    {
+      "epoch": 0.908873692016149,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3238981642446223e-07,
+      "loss": 0.77,
+      "step": 33093
+    },
+    {
+      "epoch": 0.9089011562439921,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3213111617118495e-07,
+      "loss": 0.8202,
+      "step": 33094
+    },
+    {
+      "epoch": 0.9089286204718354,
+      "grad_norm": 0.0,
+      "learning_rate": 4.318724916225925e-07,
+      "loss": 0.7661,
+      "step": 33095
+    },
+    {
+      "epoch": 0.9089560846996787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3161394278073085e-07,
+      "loss": 0.8349,
+      "step": 33096
+    },
+    {
+      "epoch": 0.9089835489275219,
+      "grad_norm": 0.0,
+      "learning_rate": 4.313554696476474e-07,
+      "loss": 0.9126,
+      "step": 33097
+    },
+    {
+      "epoch": 0.9090110131553651,
+      "grad_norm": 0.0,
+      "learning_rate": 4.31097072225386e-07,
+      "loss": 0.8974,
+      "step": 33098
+    },
+    {
+      "epoch": 0.9090384773832084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.308387505159894e-07,
+      "loss": 0.7564,
+      "step": 33099
+    },
+    {
+      "epoch": 0.9090659416110516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.305805045215061e-07,
+      "loss": 0.8463,
+      "step": 33100
+    },
+    {
+      "epoch": 0.9090934058388949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.303223342439744e-07,
+      "loss": 0.7864,
+      "step": 33101
+    },
+    {
+      "epoch": 0.909120870066738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3006423968543934e-07,
+      "loss": 0.7711,
+      "step": 33102
+    },
+    {
+      "epoch": 0.9091483342945813,
+      "grad_norm": 0.0,
+      "learning_rate": 4.298062208479448e-07,
+      "loss": 0.7979,
+      "step": 33103
+    },
+    {
+      "epoch": 0.9091757985224246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2954827773352913e-07,
+      "loss": 0.8372,
+      "step": 33104
+    },
+    {
+      "epoch": 0.9092032627502677,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2929041034423525e-07,
+      "loss": 0.8386,
+      "step": 33105
+    },
+    {
+      "epoch": 0.909230726978111,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2903261868210255e-07,
+      "loss": 0.7968,
+      "step": 33106
+    },
+    {
+      "epoch": 0.9092581912059542,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2877490274917167e-07,
+      "loss": 0.7109,
+      "step": 33107
+    },
+    {
+      "epoch": 0.9092856554337975,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2851726254748207e-07,
+      "loss": 0.8087,
+      "step": 33108
+    },
+    {
+      "epoch": 0.9093131196616407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2825969807907096e-07,
+      "loss": 0.7354,
+      "step": 33109
+    },
+    {
+      "epoch": 0.9093405838894839,
+      "grad_norm": 0.0,
+      "learning_rate": 4.280022093459768e-07,
+      "loss": 0.8149,
+      "step": 33110
+    },
+    {
+      "epoch": 0.9093680481173272,
+      "grad_norm": 0.0,
+      "learning_rate": 4.277447963502357e-07,
+      "loss": 0.7858,
+      "step": 33111
+    },
+    {
+      "epoch": 0.9093955123451705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.27487459093886e-07,
+      "loss": 0.7937,
+      "step": 33112
+    },
+    {
+      "epoch": 0.9094229765730136,
+      "grad_norm": 0.0,
+      "learning_rate": 4.27230197578965e-07,
+      "loss": 0.7564,
+      "step": 33113
+    },
+    {
+      "epoch": 0.9094504408008569,
+      "grad_norm": 0.0,
+      "learning_rate": 4.269730118075055e-07,
+      "loss": 0.7952,
+      "step": 33114
+    },
+    {
+      "epoch": 0.9094779050287001,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2671590178154476e-07,
+      "loss": 0.7706,
+      "step": 33115
+    },
+    {
+      "epoch": 0.9095053692565434,
+      "grad_norm": 0.0,
+      "learning_rate": 4.264588675031167e-07,
+      "loss": 0.756,
+      "step": 33116
+    },
+    {
+      "epoch": 0.9095328334843866,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2620190897425304e-07,
+      "loss": 0.8552,
+      "step": 33117
+    },
+    {
+      "epoch": 0.9095602977122298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.259450261969889e-07,
+      "loss": 0.8325,
+      "step": 33118
+    },
+    {
+      "epoch": 0.9095877619400731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2568821917335577e-07,
+      "loss": 0.8573,
+      "step": 33119
+    },
+    {
+      "epoch": 0.9096152261679162,
+      "grad_norm": 0.0,
+      "learning_rate": 4.254314879053867e-07,
+      "loss": 0.7423,
+      "step": 33120
+    },
+    {
+      "epoch": 0.9096426903957595,
+      "grad_norm": 0.0,
+      "learning_rate": 4.251748323951133e-07,
+      "loss": 0.8113,
+      "step": 33121
+    },
+    {
+      "epoch": 0.9096701546236028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.24918252644565e-07,
+      "loss": 0.8557,
+      "step": 33122
+    },
+    {
+      "epoch": 0.909697618851446,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2466174865577135e-07,
+      "loss": 0.7882,
+      "step": 33123
+    },
+    {
+      "epoch": 0.9097250830792892,
+      "grad_norm": 0.0,
+      "learning_rate": 4.244053204307652e-07,
+      "loss": 0.725,
+      "step": 33124
+    },
+    {
+      "epoch": 0.9097525473071325,
+      "grad_norm": 0.0,
+      "learning_rate": 4.241489679715727e-07,
+      "loss": 0.8575,
+      "step": 33125
+    },
+    {
+      "epoch": 0.9097800115349757,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2389269128022215e-07,
+      "loss": 0.6984,
+      "step": 33126
+    },
+    {
+      "epoch": 0.909807475762819,
+      "grad_norm": 0.0,
+      "learning_rate": 4.236364903587431e-07,
+      "loss": 0.6888,
+      "step": 33127
+    },
+    {
+      "epoch": 0.9098349399906621,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2338036520916057e-07,
+      "loss": 0.7865,
+      "step": 33128
+    },
+    {
+      "epoch": 0.9098624042185054,
+      "grad_norm": 0.0,
+      "learning_rate": 4.231243158335019e-07,
+      "loss": 0.7457,
+      "step": 33129
+    },
+    {
+      "epoch": 0.9098898684463487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2286834223379427e-07,
+      "loss": 0.8445,
+      "step": 33130
+    },
+    {
+      "epoch": 0.9099173326741918,
+      "grad_norm": 0.0,
+      "learning_rate": 4.226124444120616e-07,
+      "loss": 0.8354,
+      "step": 33131
+    },
+    {
+      "epoch": 0.9099447969020351,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2235662237033017e-07,
+      "loss": 0.8602,
+      "step": 33132
+    },
+    {
+      "epoch": 0.9099722611298783,
+      "grad_norm": 0.0,
+      "learning_rate": 4.221008761106238e-07,
+      "loss": 0.7709,
+      "step": 33133
+    },
+    {
+      "epoch": 0.9099997253577216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.218452056349631e-07,
+      "loss": 0.814,
+      "step": 33134
+    },
+    {
+      "epoch": 0.9100271895855648,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2158961094537434e-07,
+      "loss": 0.8271,
+      "step": 33135
+    },
+    {
+      "epoch": 0.910054653813408,
+      "grad_norm": 0.0,
+      "learning_rate": 4.213340920438791e-07,
+      "loss": 0.7974,
+      "step": 33136
+    },
+    {
+      "epoch": 0.9100821180412513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2107864893249807e-07,
+      "loss": 0.8539,
+      "step": 33137
+    },
+    {
+      "epoch": 0.9101095822690946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2082328161325405e-07,
+      "loss": 0.7047,
+      "step": 33138
+    },
+    {
+      "epoch": 0.9101370464969377,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2056799008816763e-07,
+      "loss": 0.8015,
+      "step": 33139
+    },
+    {
+      "epoch": 0.910164510724781,
+      "grad_norm": 0.0,
+      "learning_rate": 4.203127743592583e-07,
+      "loss": 0.75,
+      "step": 33140
+    },
+    {
+      "epoch": 0.9101919749526242,
+      "grad_norm": 0.0,
+      "learning_rate": 4.200576344285434e-07,
+      "loss": 0.7513,
+      "step": 33141
+    },
+    {
+      "epoch": 0.9102194391804674,
+      "grad_norm": 0.0,
+      "learning_rate": 4.198025702980435e-07,
+      "loss": 0.8284,
+      "step": 33142
+    },
+    {
+      "epoch": 0.9102469034083107,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1954758196977695e-07,
+      "loss": 0.8032,
+      "step": 33143
+    },
+    {
+      "epoch": 0.9102743676361539,
+      "grad_norm": 0.0,
+      "learning_rate": 4.192926694457611e-07,
+      "loss": 0.8141,
+      "step": 33144
+    },
+    {
+      "epoch": 0.9103018318639972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.190378327280142e-07,
+      "loss": 0.8928,
+      "step": 33145
+    },
+    {
+      "epoch": 0.9103292960918403,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1878307181854926e-07,
+      "loss": 0.8084,
+      "step": 33146
+    },
+    {
+      "epoch": 0.9103567603196836,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1852838671938455e-07,
+      "loss": 0.7974,
+      "step": 33147
+    },
+    {
+      "epoch": 0.9103842245475269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1827377743253625e-07,
+      "loss": 0.8484,
+      "step": 33148
+    },
+    {
+      "epoch": 0.91041168877537,
+      "grad_norm": 0.0,
+      "learning_rate": 4.180192439600161e-07,
+      "loss": 0.9082,
+      "step": 33149
+    },
+    {
+      "epoch": 0.9104391530032133,
+      "grad_norm": 0.0,
+      "learning_rate": 4.177647863038392e-07,
+      "loss": 0.7579,
+      "step": 33150
+    },
+    {
+      "epoch": 0.9104666172310566,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1751040446602055e-07,
+      "loss": 0.8275,
+      "step": 33151
+    },
+    {
+      "epoch": 0.9104940814588998,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1725609844857075e-07,
+      "loss": 0.9137,
+      "step": 33152
+    },
+    {
+      "epoch": 0.910521545686743,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1700186825350154e-07,
+      "loss": 0.7287,
+      "step": 33153
+    },
+    {
+      "epoch": 0.9105490099145862,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1674771388282574e-07,
+      "loss": 0.8428,
+      "step": 33154
+    },
+    {
+      "epoch": 0.9105764741424295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1649363533855626e-07,
+      "loss": 0.7899,
+      "step": 33155
+    },
+    {
+      "epoch": 0.9106039383702728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.162396326227003e-07,
+      "loss": 0.8068,
+      "step": 33156
+    },
+    {
+      "epoch": 0.9106314025981159,
+      "grad_norm": 0.0,
+      "learning_rate": 4.159857057372696e-07,
+      "loss": 0.7817,
+      "step": 33157
+    },
+    {
+      "epoch": 0.9106588668259592,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1573185468427036e-07,
+      "loss": 0.7842,
+      "step": 33158
+    },
+    {
+      "epoch": 0.9106863310538024,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1547807946571426e-07,
+      "loss": 0.8057,
+      "step": 33159
+    },
+    {
+      "epoch": 0.9107137952816456,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1522438008360865e-07,
+      "loss": 0.7536,
+      "step": 33160
+    },
+    {
+      "epoch": 0.9107412595094889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.149707565399597e-07,
+      "loss": 0.8075,
+      "step": 33161
+    },
+    {
+      "epoch": 0.9107687237373321,
+      "grad_norm": 0.0,
+      "learning_rate": 4.147172088367768e-07,
+      "loss": 0.8169,
+      "step": 33162
+    },
+    {
+      "epoch": 0.9107961879651754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.144637369760629e-07,
+      "loss": 0.8229,
+      "step": 33163
+    },
+    {
+      "epoch": 0.9108236521930186,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1421034095982636e-07,
+      "loss": 0.8281,
+      "step": 33164
+    },
+    {
+      "epoch": 0.9108511164208618,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1395702079006895e-07,
+      "loss": 0.7753,
+      "step": 33165
+    },
+    {
+      "epoch": 0.9108785806487051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.137037764687979e-07,
+      "loss": 0.728,
+      "step": 33166
+    },
+    {
+      "epoch": 0.9109060448765482,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1345060799801495e-07,
+      "loss": 0.7648,
+      "step": 33167
+    },
+    {
+      "epoch": 0.9109335091043915,
+      "grad_norm": 0.0,
+      "learning_rate": 4.131975153797263e-07,
+      "loss": 0.8167,
+      "step": 33168
+    },
+    {
+      "epoch": 0.9109609733322348,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1294449861593033e-07,
+      "loss": 0.8283,
+      "step": 33169
+    },
+    {
+      "epoch": 0.910988437560078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.126915577086321e-07,
+      "loss": 0.8859,
+      "step": 33170
+    },
+    {
+      "epoch": 0.9110159017879212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.124386926598312e-07,
+      "loss": 0.8087,
+      "step": 33171
+    },
+    {
+      "epoch": 0.9110433660157645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.121859034715303e-07,
+      "loss": 0.81,
+      "step": 33172
+    },
+    {
+      "epoch": 0.9110708302436077,
+      "grad_norm": 0.0,
+      "learning_rate": 4.11933190145728e-07,
+      "loss": 0.7409,
+      "step": 33173
+    },
+    {
+      "epoch": 0.911098294471451,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1168055268442584e-07,
+      "loss": 0.9353,
+      "step": 33174
+    },
+    {
+      "epoch": 0.9111257586992941,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1142799108961904e-07,
+      "loss": 0.7937,
+      "step": 33175
+    },
+    {
+      "epoch": 0.9111532229271374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.111755053633093e-07,
+      "loss": 0.7572,
+      "step": 33176
+    },
+    {
+      "epoch": 0.9111806871549807,
+      "grad_norm": 0.0,
+      "learning_rate": 4.109230955074928e-07,
+      "loss": 0.678,
+      "step": 33177
+    },
+    {
+      "epoch": 0.9112081513828238,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1067076152416783e-07,
+      "loss": 0.8951,
+      "step": 33178
+    },
+    {
+      "epoch": 0.9112356156106671,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1041850341533076e-07,
+      "loss": 0.8561,
+      "step": 33179
+    },
+    {
+      "epoch": 0.9112630798385103,
+      "grad_norm": 0.0,
+      "learning_rate": 4.101663211829765e-07,
+      "loss": 0.7702,
+      "step": 33180
+    },
+    {
+      "epoch": 0.9112905440663536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.099142148291013e-07,
+      "loss": 0.7596,
+      "step": 33181
+    },
+    {
+      "epoch": 0.9113180082941968,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0966218435569916e-07,
+      "loss": 0.7932,
+      "step": 33182
+    },
+    {
+      "epoch": 0.91134547252204,
+      "grad_norm": 0.0,
+      "learning_rate": 4.094102297647651e-07,
+      "loss": 0.9008,
+      "step": 33183
+    },
+    {
+      "epoch": 0.9113729367498833,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0915835105829195e-07,
+      "loss": 0.8211,
+      "step": 33184
+    },
+    {
+      "epoch": 0.9114004009777266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.089065482382726e-07,
+      "loss": 0.7991,
+      "step": 33185
+    },
+    {
+      "epoch": 0.9114278652055697,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0865482130670207e-07,
+      "loss": 0.8108,
+      "step": 33186
+    },
+    {
+      "epoch": 0.911455329433413,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0840317026556774e-07,
+      "loss": 0.9044,
+      "step": 33187
+    },
+    {
+      "epoch": 0.9114827936612562,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0815159511686463e-07,
+      "loss": 0.7704,
+      "step": 33188
+    },
+    {
+      "epoch": 0.9115102578890995,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0790009586258005e-07,
+      "loss": 0.8243,
+      "step": 33189
+    },
+    {
+      "epoch": 0.9115377221169427,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0764867250470683e-07,
+      "loss": 0.7909,
+      "step": 33190
+    },
+    {
+      "epoch": 0.9115651863447859,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0739732504523123e-07,
+      "loss": 0.7822,
+      "step": 33191
+    },
+    {
+      "epoch": 0.9115926505726292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.07146053486146e-07,
+      "loss": 0.8205,
+      "step": 33192
+    },
+    {
+      "epoch": 0.9116201148004723,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0689485782943626e-07,
+      "loss": 0.748,
+      "step": 33193
+    },
+    {
+      "epoch": 0.9116475790283156,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0664373807708937e-07,
+      "loss": 0.8193,
+      "step": 33194
+    },
+    {
+      "epoch": 0.9116750432561589,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0639269423109473e-07,
+      "loss": 0.8356,
+      "step": 33195
+    },
+    {
+      "epoch": 0.9117025074840021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.061417262934375e-07,
+      "loss": 0.8749,
+      "step": 33196
+    },
+    {
+      "epoch": 0.9117299717118453,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0589083426610165e-07,
+      "loss": 0.8491,
+      "step": 33197
+    },
+    {
+      "epoch": 0.9117574359396886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0564001815107556e-07,
+      "loss": 0.727,
+      "step": 33198
+    },
+    {
+      "epoch": 0.9117849001675318,
+      "grad_norm": 0.0,
+      "learning_rate": 4.053892779503421e-07,
+      "loss": 0.7023,
+      "step": 33199
+    },
+    {
+      "epoch": 0.911812364395375,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0513861366588414e-07,
+      "loss": 0.7349,
+      "step": 33200
+    },
+    {
+      "epoch": 0.9118398286232182,
+      "grad_norm": 0.0,
+      "learning_rate": 4.048880252996856e-07,
+      "loss": 0.8369,
+      "step": 33201
+    },
+    {
+      "epoch": 0.9118672928510615,
+      "grad_norm": 0.0,
+      "learning_rate": 4.046375128537305e-07,
+      "loss": 0.735,
+      "step": 33202
+    },
+    {
+      "epoch": 0.9118947570789048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.043870763300017e-07,
+      "loss": 0.8583,
+      "step": 33203
+    },
+    {
+      "epoch": 0.9119222213067479,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0413671573047763e-07,
+      "loss": 0.821,
+      "step": 33204
+    },
+    {
+      "epoch": 0.9119496855345912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.038864310571422e-07,
+      "loss": 0.8123,
+      "step": 33205
+    },
+    {
+      "epoch": 0.9119771497624344,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0363622231197387e-07,
+      "loss": 0.8202,
+      "step": 33206
+    },
+    {
+      "epoch": 0.9120046139902777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0338608949695324e-07,
+      "loss": 0.7078,
+      "step": 33207
+    },
+    {
+      "epoch": 0.9120320782181209,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0313603261405876e-07,
+      "loss": 0.6799,
+      "step": 33208
+    },
+    {
+      "epoch": 0.9120595424459641,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0288605166526883e-07,
+      "loss": 0.8257,
+      "step": 33209
+    },
+    {
+      "epoch": 0.9120870066738074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0263614665256414e-07,
+      "loss": 0.8041,
+      "step": 33210
+    },
+    {
+      "epoch": 0.9121144709016507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.023863175779175e-07,
+      "loss": 0.8108,
+      "step": 33211
+    },
+    {
+      "epoch": 0.9121419351294938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0213656444331063e-07,
+      "loss": 0.7487,
+      "step": 33212
+    },
+    {
+      "epoch": 0.9121693993573371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0188688725071534e-07,
+      "loss": 0.8105,
+      "step": 33213
+    },
+    {
+      "epoch": 0.9121968635851803,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0163728600210894e-07,
+      "loss": 0.785,
+      "step": 33214
+    },
+    {
+      "epoch": 0.9122243278130235,
+      "grad_norm": 0.0,
+      "learning_rate": 4.013877606994665e-07,
+      "loss": 0.9111,
+      "step": 33215
+    },
+    {
+      "epoch": 0.9122517920408668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0113831134476313e-07,
+      "loss": 0.7676,
+      "step": 33216
+    },
+    {
+      "epoch": 0.91227925626871,
+      "grad_norm": 0.0,
+      "learning_rate": 4.008889379399716e-07,
+      "loss": 0.8654,
+      "step": 33217
+    },
+    {
+      "epoch": 0.9123067204965533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0063964048706383e-07,
+      "loss": 0.788,
+      "step": 33218
+    },
+    {
+      "epoch": 0.9123341847243964,
+      "grad_norm": 0.0,
+      "learning_rate": 4.003904189880137e-07,
+      "loss": 0.7655,
+      "step": 33219
+    },
+    {
+      "epoch": 0.9123616489522397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0014127344479515e-07,
+      "loss": 0.89,
+      "step": 33220
+    },
+    {
+      "epoch": 0.912389113180083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9989220385937667e-07,
+      "loss": 0.7715,
+      "step": 33221
+    },
+    {
+      "epoch": 0.9124165774079261,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9964321023373e-07,
+      "loss": 0.7499,
+      "step": 33222
+    },
+    {
+      "epoch": 0.9124440416357694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9939429256982356e-07,
+      "loss": 0.7787,
+      "step": 33223
+    },
+    {
+      "epoch": 0.9124715058636127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9914545086962907e-07,
+      "loss": 0.8781,
+      "step": 33224
+    },
+    {
+      "epoch": 0.9124989700914559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9889668513511393e-07,
+      "loss": 0.7551,
+      "step": 33225
+    },
+    {
+      "epoch": 0.9125264343192991,
+      "grad_norm": 0.0,
+      "learning_rate": 3.986479953682487e-07,
+      "loss": 0.8172,
+      "step": 33226
+    },
+    {
+      "epoch": 0.9125538985471423,
+      "grad_norm": 0.0,
+      "learning_rate": 3.983993815709997e-07,
+      "loss": 0.8765,
+      "step": 33227
+    },
+    {
+      "epoch": 0.9125813627749856,
+      "grad_norm": 0.0,
+      "learning_rate": 3.981508437453341e-07,
+      "loss": 0.8036,
+      "step": 33228
+    },
+    {
+      "epoch": 0.9126088270028289,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9790238189321926e-07,
+      "loss": 0.8157,
+      "step": 33229
+    },
+    {
+      "epoch": 0.912636291230672,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9765399601661816e-07,
+      "loss": 0.8078,
+      "step": 33230
+    },
+    {
+      "epoch": 0.9126637554585153,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9740568611749906e-07,
+      "loss": 0.824,
+      "step": 33231
+    },
+    {
+      "epoch": 0.9126912196863585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.971574521978261e-07,
+      "loss": 0.8036,
+      "step": 33232
+    },
+    {
+      "epoch": 0.9127186839142017,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9690929425956314e-07,
+      "loss": 0.8748,
+      "step": 33233
+    },
+    {
+      "epoch": 0.912746148142045,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9666121230467315e-07,
+      "loss": 0.8818,
+      "step": 33234
+    },
+    {
+      "epoch": 0.9127736123698882,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9641320633512003e-07,
+      "loss": 0.8961,
+      "step": 33235
+    },
+    {
+      "epoch": 0.9128010765977315,
+      "grad_norm": 0.0,
+      "learning_rate": 3.961652763528667e-07,
+      "loss": 0.7776,
+      "step": 33236
+    },
+    {
+      "epoch": 0.9128285408255747,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9591742235987273e-07,
+      "loss": 0.7991,
+      "step": 33237
+    },
+    {
+      "epoch": 0.9128560050534179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9566964435809983e-07,
+      "loss": 0.7599,
+      "step": 33238
+    },
+    {
+      "epoch": 0.9128834692812612,
+      "grad_norm": 0.0,
+      "learning_rate": 3.954219423495109e-07,
+      "loss": 0.8033,
+      "step": 33239
+    },
+    {
+      "epoch": 0.9129109335091043,
+      "grad_norm": 0.0,
+      "learning_rate": 3.951743163360633e-07,
+      "loss": 0.8724,
+      "step": 33240
+    },
+    {
+      "epoch": 0.9129383977369476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.949267663197165e-07,
+      "loss": 0.8242,
+      "step": 33241
+    },
+    {
+      "epoch": 0.9129658619647909,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9467929230242894e-07,
+      "loss": 0.7888,
+      "step": 33242
+    },
+    {
+      "epoch": 0.9129933261926341,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9443189428616024e-07,
+      "loss": 0.7943,
+      "step": 33243
+    },
+    {
+      "epoch": 0.9130207904204773,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9418457227286877e-07,
+      "loss": 0.7614,
+      "step": 33244
+    },
+    {
+      "epoch": 0.9130482546483205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9393732626450963e-07,
+      "loss": 0.7315,
+      "step": 33245
+    },
+    {
+      "epoch": 0.9130757188761638,
+      "grad_norm": 0.0,
+      "learning_rate": 3.93690156263038e-07,
+      "loss": 0.7989,
+      "step": 33246
+    },
+    {
+      "epoch": 0.9131031831040071,
+      "grad_norm": 0.0,
+      "learning_rate": 3.934430622704111e-07,
+      "loss": 0.7877,
+      "step": 33247
+    },
+    {
+      "epoch": 0.9131306473318502,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9319604428858407e-07,
+      "loss": 0.8056,
+      "step": 33248
+    },
+    {
+      "epoch": 0.9131581115596935,
+      "grad_norm": 0.0,
+      "learning_rate": 3.92949102319512e-07,
+      "loss": 0.8485,
+      "step": 33249
+    },
+    {
+      "epoch": 0.9131855757875368,
+      "grad_norm": 0.0,
+      "learning_rate": 3.927022363651467e-07,
+      "loss": 0.8017,
+      "step": 33250
+    },
+    {
+      "epoch": 0.91321304001538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.924554464274444e-07,
+      "loss": 0.8645,
+      "step": 33251
+    },
+    {
+      "epoch": 0.9132405042432232,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9220873250835566e-07,
+      "loss": 0.8278,
+      "step": 33252
+    },
+    {
+      "epoch": 0.9132679684710664,
+      "grad_norm": 0.0,
+      "learning_rate": 3.919620946098335e-07,
+      "loss": 0.7866,
+      "step": 33253
+    },
+    {
+      "epoch": 0.9132954326989097,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9171553273382843e-07,
+      "loss": 0.8715,
+      "step": 33254
+    },
+    {
+      "epoch": 0.913322896926753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9146904688229125e-07,
+      "loss": 0.797,
+      "step": 33255
+    },
+    {
+      "epoch": 0.9133503611545961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.912226370571737e-07,
+      "loss": 0.7928,
+      "step": 33256
+    },
+    {
+      "epoch": 0.9133778253824394,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9097630326042526e-07,
+      "loss": 0.8959,
+      "step": 33257
+    },
+    {
+      "epoch": 0.9134052896102826,
+      "grad_norm": 0.0,
+      "learning_rate": 3.907300454939944e-07,
+      "loss": 0.9028,
+      "step": 33258
+    },
+    {
+      "epoch": 0.9134327538381258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.904838637598285e-07,
+      "loss": 0.7449,
+      "step": 33259
+    },
+    {
+      "epoch": 0.9134602180659691,
+      "grad_norm": 0.0,
+      "learning_rate": 3.902377580598782e-07,
+      "loss": 0.7673,
+      "step": 33260
+    },
+    {
+      "epoch": 0.9134876822938123,
+      "grad_norm": 0.0,
+      "learning_rate": 3.899917283960886e-07,
+      "loss": 0.8285,
+      "step": 33261
+    },
+    {
+      "epoch": 0.9135151465216556,
+      "grad_norm": 0.0,
+      "learning_rate": 3.897457747704059e-07,
+      "loss": 0.8494,
+      "step": 33262
+    },
+    {
+      "epoch": 0.9135426107494988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8949989718477964e-07,
+      "loss": 0.7481,
+      "step": 33263
+    },
+    {
+      "epoch": 0.913570074977342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.892540956411506e-07,
+      "loss": 0.8314,
+      "step": 33264
+    },
+    {
+      "epoch": 0.9135975392051853,
+      "grad_norm": 0.0,
+      "learning_rate": 3.890083701414671e-07,
+      "loss": 0.831,
+      "step": 33265
+    },
+    {
+      "epoch": 0.9136250034330284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.887627206876721e-07,
+      "loss": 0.7601,
+      "step": 33266
+    },
+    {
+      "epoch": 0.9136524676608717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.885171472817084e-07,
+      "loss": 0.8042,
+      "step": 33267
+    },
+    {
+      "epoch": 0.913679931888715,
+      "grad_norm": 0.0,
+      "learning_rate": 3.882716499255224e-07,
+      "loss": 0.8041,
+      "step": 33268
+    },
+    {
+      "epoch": 0.9137073961165582,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8802622862105455e-07,
+      "loss": 0.8034,
+      "step": 33269
+    },
+    {
+      "epoch": 0.9137348603444014,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8778088337024453e-07,
+      "loss": 0.764,
+      "step": 33270
+    },
+    {
+      "epoch": 0.9137623245722446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8753561417503635e-07,
+      "loss": 0.8293,
+      "step": 33271
+    },
+    {
+      "epoch": 0.9137897888000879,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8729042103736957e-07,
+      "loss": 0.813,
+      "step": 33272
+    },
+    {
+      "epoch": 0.9138172530279312,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8704530395918484e-07,
+      "loss": 0.8231,
+      "step": 33273
+    },
+    {
+      "epoch": 0.9138447172557743,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8680026294242055e-07,
+      "loss": 0.8787,
+      "step": 33274
+    },
+    {
+      "epoch": 0.9138721814836176,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8655529798901857e-07,
+      "loss": 0.8479,
+      "step": 33275
+    },
+    {
+      "epoch": 0.9138996457114609,
+      "grad_norm": 0.0,
+      "learning_rate": 3.86310409100914e-07,
+      "loss": 0.8209,
+      "step": 33276
+    },
+    {
+      "epoch": 0.913927109939304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8606559628004415e-07,
+      "loss": 0.7906,
+      "step": 33277
+    },
+    {
+      "epoch": 0.9139545741671473,
+      "grad_norm": 0.0,
+      "learning_rate": 3.858208595283486e-07,
+      "loss": 0.8119,
+      "step": 33278
+    },
+    {
+      "epoch": 0.9139820383949905,
+      "grad_norm": 0.0,
+      "learning_rate": 3.855761988477613e-07,
+      "loss": 0.7268,
+      "step": 33279
+    },
+    {
+      "epoch": 0.9140095026228338,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8533161424022193e-07,
+      "loss": 0.8092,
+      "step": 33280
+    },
+    {
+      "epoch": 0.914036966850677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.850871057076611e-07,
+      "loss": 0.7611,
+      "step": 33281
+    },
+    {
+      "epoch": 0.9140644310785202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.848426732520161e-07,
+      "loss": 0.7647,
+      "step": 33282
+    },
+    {
+      "epoch": 0.9140918953063635,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8459831687521984e-07,
+      "loss": 0.7508,
+      "step": 33283
+    },
+    {
+      "epoch": 0.9141193595342066,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8435403657920754e-07,
+      "loss": 0.7884,
+      "step": 33284
+    },
+    {
+      "epoch": 0.9141468237620499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.841098323659098e-07,
+      "loss": 0.7895,
+      "step": 33285
+    },
+    {
+      "epoch": 0.9141742879898932,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8386570423726175e-07,
+      "loss": 0.8254,
+      "step": 33286
+    },
+    {
+      "epoch": 0.9142017522177364,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8362165219519075e-07,
+      "loss": 0.7865,
+      "step": 33287
+    },
+    {
+      "epoch": 0.9142292164455796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.833776762416308e-07,
+      "loss": 0.7663,
+      "step": 33288
+    },
+    {
+      "epoch": 0.9142566806734229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8313377637851256e-07,
+      "loss": 0.781,
+      "step": 33289
+    },
+    {
+      "epoch": 0.9142841449012661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.828899526077645e-07,
+      "loss": 0.8894,
+      "step": 33290
+    },
+    {
+      "epoch": 0.9143116091291094,
+      "grad_norm": 0.0,
+      "learning_rate": 3.826462049313162e-07,
+      "loss": 0.9141,
+      "step": 33291
+    },
+    {
+      "epoch": 0.9143390733569525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8240253335109835e-07,
+      "loss": 0.7449,
+      "step": 33292
+    },
+    {
+      "epoch": 0.9143665375847958,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8215893786903604e-07,
+      "loss": 0.8376,
+      "step": 33293
+    },
+    {
+      "epoch": 0.9143940018126391,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8191541848705773e-07,
+      "loss": 0.8378,
+      "step": 33294
+    },
+    {
+      "epoch": 0.9144214660404822,
+      "grad_norm": 0.0,
+      "learning_rate": 3.816719752070897e-07,
+      "loss": 0.7631,
+      "step": 33295
+    },
+    {
+      "epoch": 0.9144489302683255,
+      "grad_norm": 0.0,
+      "learning_rate": 3.814286080310592e-07,
+      "loss": 0.8844,
+      "step": 33296
+    },
+    {
+      "epoch": 0.9144763944961687,
+      "grad_norm": 0.0,
+      "learning_rate": 3.811853169608914e-07,
+      "loss": 0.8132,
+      "step": 33297
+    },
+    {
+      "epoch": 0.914503858724012,
+      "grad_norm": 0.0,
+      "learning_rate": 3.809421019985116e-07,
+      "loss": 0.8785,
+      "step": 33298
+    },
+    {
+      "epoch": 0.9145313229518552,
+      "grad_norm": 0.0,
+      "learning_rate": 3.806989631458435e-07,
+      "loss": 0.799,
+      "step": 33299
+    },
+    {
+      "epoch": 0.9145587871796984,
+      "grad_norm": 0.0,
+      "learning_rate": 3.804559004048136e-07,
+      "loss": 0.7925,
+      "step": 33300
+    },
+    {
+      "epoch": 0.9145862514075417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8021291377734026e-07,
+      "loss": 0.7523,
+      "step": 33301
+    },
+    {
+      "epoch": 0.914613715635385,
+      "grad_norm": 0.0,
+      "learning_rate": 3.799700032653486e-07,
+      "loss": 0.7963,
+      "step": 33302
+    },
+    {
+      "epoch": 0.9146411798632281,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7972716887076154e-07,
+      "loss": 0.9207,
+      "step": 33303
+    },
+    {
+      "epoch": 0.9146686440910714,
+      "grad_norm": 0.0,
+      "learning_rate": 3.794844105955009e-07,
+      "loss": 0.8159,
+      "step": 33304
+    },
+    {
+      "epoch": 0.9146961083189146,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7924172844148397e-07,
+      "loss": 0.8589,
+      "step": 33305
+    },
+    {
+      "epoch": 0.9147235725467578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7899912241063483e-07,
+      "loss": 0.763,
+      "step": 33306
+    },
+    {
+      "epoch": 0.9147510367746011,
+      "grad_norm": 0.0,
+      "learning_rate": 3.787565925048697e-07,
+      "loss": 0.931,
+      "step": 33307
+    },
+    {
+      "epoch": 0.9147785010024443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.785141387261115e-07,
+      "loss": 0.8019,
+      "step": 33308
+    },
+    {
+      "epoch": 0.9148059652302876,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7827176107627536e-07,
+      "loss": 0.7503,
+      "step": 33309
+    },
+    {
+      "epoch": 0.9148334294581307,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7802945955728087e-07,
+      "loss": 0.7386,
+      "step": 33310
+    },
+    {
+      "epoch": 0.914860893685974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7778723417104424e-07,
+      "loss": 0.7941,
+      "step": 33311
+    },
+    {
+      "epoch": 0.9148883579138173,
+      "grad_norm": 0.0,
+      "learning_rate": 3.775450849194806e-07,
+      "loss": 0.825,
+      "step": 33312
+    },
+    {
+      "epoch": 0.9149158221416605,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7730301180450955e-07,
+      "loss": 0.788,
+      "step": 33313
+    },
+    {
+      "epoch": 0.9149432863695037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7706101482804293e-07,
+      "loss": 0.7443,
+      "step": 33314
+    },
+    {
+      "epoch": 0.914970750597347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7681909399199914e-07,
+      "loss": 0.7924,
+      "step": 33315
+    },
+    {
+      "epoch": 0.9149982148251902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.765772492982889e-07,
+      "loss": 0.7135,
+      "step": 33316
+    },
+    {
+      "epoch": 0.9150256790530334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7633548074882843e-07,
+      "loss": 0.8031,
+      "step": 33317
+    },
+    {
+      "epoch": 0.9150531432808766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7609378834552846e-07,
+      "loss": 0.7527,
+      "step": 33318
+    },
+    {
+      "epoch": 0.9150806075087199,
+      "grad_norm": 0.0,
+      "learning_rate": 3.75852172090303e-07,
+      "loss": 0.8861,
+      "step": 33319
+    },
+    {
+      "epoch": 0.9151080717365632,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7561063198506165e-07,
+      "loss": 0.8708,
+      "step": 33320
+    },
+    {
+      "epoch": 0.9151355359644063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.753691680317184e-07,
+      "loss": 0.7367,
+      "step": 33321
+    },
+    {
+      "epoch": 0.9151630001922496,
+      "grad_norm": 0.0,
+      "learning_rate": 3.751277802321829e-07,
+      "loss": 0.7537,
+      "step": 33322
+    },
+    {
+      "epoch": 0.9151904644200928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7488646858836464e-07,
+      "loss": 0.7804,
+      "step": 33323
+    },
+    {
+      "epoch": 0.915217928647936,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7464523310217325e-07,
+      "loss": 0.8263,
+      "step": 33324
+    },
+    {
+      "epoch": 0.9152453928757793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.744040737755161e-07,
+      "loss": 0.8216,
+      "step": 33325
+    },
+    {
+      "epoch": 0.9152728571036225,
+      "grad_norm": 0.0,
+      "learning_rate": 3.741629906103028e-07,
+      "loss": 0.7429,
+      "step": 33326
+    },
+    {
+      "epoch": 0.9153003213314658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7392198360844065e-07,
+      "loss": 0.7799,
+      "step": 33327
+    },
+    {
+      "epoch": 0.915327785559309,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7368105277183707e-07,
+      "loss": 0.8136,
+      "step": 33328
+    },
+    {
+      "epoch": 0.9153552497871522,
+      "grad_norm": 0.0,
+      "learning_rate": 3.734401981023983e-07,
+      "loss": 0.6995,
+      "step": 33329
+    },
+    {
+      "epoch": 0.9153827140149955,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7319941960202834e-07,
+      "loss": 0.7332,
+      "step": 33330
+    },
+    {
+      "epoch": 0.9154101782428387,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7295871727263347e-07,
+      "loss": 0.7974,
+      "step": 33331
+    },
+    {
+      "epoch": 0.9154376424706819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.727180911161199e-07,
+      "loss": 0.8022,
+      "step": 33332
+    },
+    {
+      "epoch": 0.9154651066985252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.724775411343895e-07,
+      "loss": 0.8459,
+      "step": 33333
+    },
+    {
+      "epoch": 0.9154925709263684,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7223706732934627e-07,
+      "loss": 0.8182,
+      "step": 33334
+    },
+    {
+      "epoch": 0.9155200351542117,
+      "grad_norm": 0.0,
+      "learning_rate": 3.719966697028932e-07,
+      "loss": 0.8033,
+      "step": 33335
+    },
+    {
+      "epoch": 0.9155474993820548,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7175634825693085e-07,
+      "loss": 0.7982,
+      "step": 33336
+    },
+    {
+      "epoch": 0.9155749636098981,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7151610299336115e-07,
+      "loss": 0.797,
+      "step": 33337
+    },
+    {
+      "epoch": 0.9156024278377414,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7127593391408703e-07,
+      "loss": 0.719,
+      "step": 33338
+    },
+    {
+      "epoch": 0.9156298920655845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7103584102100797e-07,
+      "loss": 0.7807,
+      "step": 33339
+    },
+    {
+      "epoch": 0.9156573562934278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7079582431602254e-07,
+      "loss": 0.6688,
+      "step": 33340
+    },
+    {
+      "epoch": 0.9156848205212711,
+      "grad_norm": 0.0,
+      "learning_rate": 3.705558838010315e-07,
+      "loss": 0.7924,
+      "step": 33341
+    },
+    {
+      "epoch": 0.9157122847491143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7031601947793205e-07,
+      "loss": 0.8115,
+      "step": 33342
+    },
+    {
+      "epoch": 0.9157397489769575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.700762313486217e-07,
+      "loss": 0.8802,
+      "step": 33343
+    },
+    {
+      "epoch": 0.9157672132048007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6983651941499776e-07,
+      "loss": 0.8332,
+      "step": 33344
+    },
+    {
+      "epoch": 0.915794677432644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6959688367895985e-07,
+      "loss": 0.7708,
+      "step": 33345
+    },
+    {
+      "epoch": 0.9158221416604873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.693573241424009e-07,
+      "loss": 0.767,
+      "step": 33346
+    },
+    {
+      "epoch": 0.9158496058883304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.69117840807216e-07,
+      "loss": 0.8744,
+      "step": 33347
+    },
+    {
+      "epoch": 0.9158770701161737,
+      "grad_norm": 0.0,
+      "learning_rate": 3.688784336753037e-07,
+      "loss": 0.8364,
+      "step": 33348
+    },
+    {
+      "epoch": 0.9159045343440169,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6863910274855475e-07,
+      "loss": 0.7523,
+      "step": 33349
+    },
+    {
+      "epoch": 0.9159319985718601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.683998480288642e-07,
+      "loss": 0.8064,
+      "step": 33350
+    },
+    {
+      "epoch": 0.9159594627997034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68160669518125e-07,
+      "loss": 0.8253,
+      "step": 33351
+    },
+    {
+      "epoch": 0.9159869270275466,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6792156721823015e-07,
+      "loss": 0.8442,
+      "step": 33352
+    },
+    {
+      "epoch": 0.9160143912553899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6768254113106916e-07,
+      "loss": 0.8876,
+      "step": 33353
+    },
+    {
+      "epoch": 0.9160418554832331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.674435912585361e-07,
+      "loss": 0.8322,
+      "step": 33354
+    },
+    {
+      "epoch": 0.9160693197110763,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6720471760252065e-07,
+      "loss": 0.7122,
+      "step": 33355
+    },
+    {
+      "epoch": 0.9160967839389196,
+      "grad_norm": 0.0,
+      "learning_rate": 3.669659201649134e-07,
+      "loss": 0.8015,
+      "step": 33356
+    },
+    {
+      "epoch": 0.9161242481667627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.66727198947604e-07,
+      "loss": 0.7189,
+      "step": 33357
+    },
+    {
+      "epoch": 0.916151712394606,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6648855395247985e-07,
+      "loss": 0.8705,
+      "step": 33358
+    },
+    {
+      "epoch": 0.9161791766224493,
+      "grad_norm": 0.0,
+      "learning_rate": 3.662499851814283e-07,
+      "loss": 0.8797,
+      "step": 33359
+    },
+    {
+      "epoch": 0.9162066408502925,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6601149263634006e-07,
+      "loss": 0.8172,
+      "step": 33360
+    },
+    {
+      "epoch": 0.9162341050781357,
+      "grad_norm": 0.0,
+      "learning_rate": 3.657730763191003e-07,
+      "loss": 0.8587,
+      "step": 33361
+    },
+    {
+      "epoch": 0.916261569305979,
+      "grad_norm": 0.0,
+      "learning_rate": 3.655347362315953e-07,
+      "loss": 0.8496,
+      "step": 33362
+    },
+    {
+      "epoch": 0.9162890335338222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.652964723757135e-07,
+      "loss": 0.7949,
+      "step": 33363
+    },
+    {
+      "epoch": 0.9163164977616655,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6505828475333685e-07,
+      "loss": 0.8523,
+      "step": 33364
+    },
+    {
+      "epoch": 0.9163439619895086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6482017336635255e-07,
+      "loss": 0.764,
+      "step": 33365
+    },
+    {
+      "epoch": 0.9163714262173519,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6458213821664146e-07,
+      "loss": 0.7729,
+      "step": 33366
+    },
+    {
+      "epoch": 0.9163988904451952,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6434417930608977e-07,
+      "loss": 0.8864,
+      "step": 33367
+    },
+    {
+      "epoch": 0.9164263546730383,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6410629663657825e-07,
+      "loss": 0.8318,
+      "step": 33368
+    },
+    {
+      "epoch": 0.9164538189008816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6386849020999314e-07,
+      "loss": 0.7331,
+      "step": 33369
+    },
+    {
+      "epoch": 0.9164812831287248,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6363076002821074e-07,
+      "loss": 0.8005,
+      "step": 33370
+    },
+    {
+      "epoch": 0.9165087473565681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.633931060931151e-07,
+      "loss": 0.7832,
+      "step": 33371
+    },
+    {
+      "epoch": 0.9165362115844113,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6315552840658683e-07,
+      "loss": 0.7999,
+      "step": 33372
+    },
+    {
+      "epoch": 0.9165636758122545,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6291802697050347e-07,
+      "loss": 0.824,
+      "step": 33373
+    },
+    {
+      "epoch": 0.9165911400400978,
+      "grad_norm": 0.0,
+      "learning_rate": 3.626806017867468e-07,
+      "loss": 0.736,
+      "step": 33374
+    },
+    {
+      "epoch": 0.9166186042679411,
+      "grad_norm": 0.0,
+      "learning_rate": 3.624432528571953e-07,
+      "loss": 0.8107,
+      "step": 33375
+    },
+    {
+      "epoch": 0.9166460684957842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.622059801837241e-07,
+      "loss": 0.7131,
+      "step": 33376
+    },
+    {
+      "epoch": 0.9166735327236275,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6196878376821286e-07,
+      "loss": 0.8726,
+      "step": 33377
+    },
+    {
+      "epoch": 0.9167009969514707,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6173166361253896e-07,
+      "loss": 0.7348,
+      "step": 33378
+    },
+    {
+      "epoch": 0.916728461179314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6149461971857645e-07,
+      "loss": 0.7401,
+      "step": 33379
+    },
+    {
+      "epoch": 0.9167559254071572,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6125765208820387e-07,
+      "loss": 0.7507,
+      "step": 33380
+    },
+    {
+      "epoch": 0.9167833896350004,
+      "grad_norm": 0.0,
+      "learning_rate": 3.610207607232941e-07,
+      "loss": 0.7359,
+      "step": 33381
+    },
+    {
+      "epoch": 0.9168108538628437,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6078394562572007e-07,
+      "loss": 0.8769,
+      "step": 33382
+    },
+    {
+      "epoch": 0.9168383180906868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6054720679735813e-07,
+      "loss": 0.7505,
+      "step": 33383
+    },
+    {
+      "epoch": 0.9168657823185301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6031054424008116e-07,
+      "loss": 0.8404,
+      "step": 33384
+    },
+    {
+      "epoch": 0.9168932465463734,
+      "grad_norm": 0.0,
+      "learning_rate": 3.600739579557599e-07,
+      "loss": 0.8546,
+      "step": 33385
+    },
+    {
+      "epoch": 0.9169207107742166,
+      "grad_norm": 0.0,
+      "learning_rate": 3.598374479462685e-07,
+      "loss": 0.7756,
+      "step": 33386
+    },
+    {
+      "epoch": 0.9169481750020598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.596010142134787e-07,
+      "loss": 0.9407,
+      "step": 33387
+    },
+    {
+      "epoch": 0.9169756392299031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.59364656759259e-07,
+      "loss": 0.709,
+      "step": 33388
+    },
+    {
+      "epoch": 0.9170031034577463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.591283755854813e-07,
+      "loss": 0.8799,
+      "step": 33389
+    },
+    {
+      "epoch": 0.9170305676855895,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5889217069401294e-07,
+      "loss": 0.8028,
+      "step": 33390
+    },
+    {
+      "epoch": 0.9170580319134327,
+      "grad_norm": 0.0,
+      "learning_rate": 3.586560420867247e-07,
+      "loss": 0.8761,
+      "step": 33391
+    },
+    {
+      "epoch": 0.917085496141276,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5841998976548497e-07,
+      "loss": 0.8209,
+      "step": 33392
+    },
+    {
+      "epoch": 0.9171129603691193,
+      "grad_norm": 0.0,
+      "learning_rate": 3.581840137321624e-07,
+      "loss": 0.7209,
+      "step": 33393
+    },
+    {
+      "epoch": 0.9171404245969624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5794811398862095e-07,
+      "loss": 0.8768,
+      "step": 33394
+    },
+    {
+      "epoch": 0.9171678888248057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.577122905367292e-07,
+      "loss": 0.7435,
+      "step": 33395
+    },
+    {
+      "epoch": 0.9171953530526489,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5747654337835337e-07,
+      "loss": 0.8188,
+      "step": 33396
+    },
+    {
+      "epoch": 0.9172228172804922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5724087251535753e-07,
+      "loss": 0.9007,
+      "step": 33397
+    },
+    {
+      "epoch": 0.9172502815083354,
+      "grad_norm": 0.0,
+      "learning_rate": 3.570052779496092e-07,
+      "loss": 0.7935,
+      "step": 33398
+    },
+    {
+      "epoch": 0.9172777457361786,
+      "grad_norm": 0.0,
+      "learning_rate": 3.567697596829678e-07,
+      "loss": 0.8168,
+      "step": 33399
+    },
+    {
+      "epoch": 0.9173052099640219,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5653431771730087e-07,
+      "loss": 0.8833,
+      "step": 33400
+    },
+    {
+      "epoch": 0.9173326741918651,
+      "grad_norm": 0.0,
+      "learning_rate": 3.562989520544691e-07,
+      "loss": 0.8348,
+      "step": 33401
+    },
+    {
+      "epoch": 0.9173601384197083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5606366269633543e-07,
+      "loss": 0.8119,
+      "step": 33402
+    },
+    {
+      "epoch": 0.9173876026475516,
+      "grad_norm": 0.0,
+      "learning_rate": 3.558284496447628e-07,
+      "loss": 0.888,
+      "step": 33403
+    },
+    {
+      "epoch": 0.9174150668753948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.55593312901612e-07,
+      "loss": 0.7828,
+      "step": 33404
+    },
+    {
+      "epoch": 0.917442531103238,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5535825246874267e-07,
+      "loss": 0.7722,
+      "step": 33405
+    },
+    {
+      "epoch": 0.9174699953310813,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5512326834801324e-07,
+      "loss": 0.8066,
+      "step": 33406
+    },
+    {
+      "epoch": 0.9174974595589245,
+      "grad_norm": 0.0,
+      "learning_rate": 3.548883605412845e-07,
+      "loss": 0.8912,
+      "step": 33407
+    },
+    {
+      "epoch": 0.9175249237867678,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5465352905041604e-07,
+      "loss": 0.8758,
+      "step": 33408
+    },
+    {
+      "epoch": 0.9175523880146109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.544187738772653e-07,
+      "loss": 0.7324,
+      "step": 33409
+    },
+    {
+      "epoch": 0.9175798522424542,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5418409502368965e-07,
+      "loss": 0.7989,
+      "step": 33410
+    },
+    {
+      "epoch": 0.9176073164702975,
+      "grad_norm": 0.0,
+      "learning_rate": 3.539494924915454e-07,
+      "loss": 0.7857,
+      "step": 33411
+    },
+    {
+      "epoch": 0.9176347806981406,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5371496628268996e-07,
+      "loss": 0.7464,
+      "step": 33412
+    },
+    {
+      "epoch": 0.9176622449259839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.534805163989774e-07,
+      "loss": 0.7829,
+      "step": 33413
+    },
+    {
+      "epoch": 0.9176897091538272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.53246142842264e-07,
+      "loss": 0.8296,
+      "step": 33414
+    },
+    {
+      "epoch": 0.9177171733816704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5301184561440385e-07,
+      "loss": 0.7462,
+      "step": 33415
+    },
+    {
+      "epoch": 0.9177446376095136,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5277762471725206e-07,
+      "loss": 0.8013,
+      "step": 33416
+    },
+    {
+      "epoch": 0.9177721018373568,
+      "grad_norm": 0.0,
+      "learning_rate": 3.525434801526595e-07,
+      "loss": 0.7974,
+      "step": 33417
+    },
+    {
+      "epoch": 0.9177995660652001,
+      "grad_norm": 0.0,
+      "learning_rate": 3.52309411922479e-07,
+      "loss": 0.7484,
+      "step": 33418
+    },
+    {
+      "epoch": 0.9178270302930434,
+      "grad_norm": 0.0,
+      "learning_rate": 3.520754200285648e-07,
+      "loss": 0.8279,
+      "step": 33419
+    },
+    {
+      "epoch": 0.9178544945208865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.518415044727674e-07,
+      "loss": 0.7362,
+      "step": 33420
+    },
+    {
+      "epoch": 0.9178819587487298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5160766525693665e-07,
+      "loss": 0.8059,
+      "step": 33421
+    },
+    {
+      "epoch": 0.917909422976573,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5137390238292546e-07,
+      "loss": 0.8278,
+      "step": 33422
+    },
+    {
+      "epoch": 0.9179368872044162,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5114021585257894e-07,
+      "loss": 0.7903,
+      "step": 33423
+    },
+    {
+      "epoch": 0.9179643514322595,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5090660566774903e-07,
+      "loss": 0.8022,
+      "step": 33424
+    },
+    {
+      "epoch": 0.9179918156601027,
+      "grad_norm": 0.0,
+      "learning_rate": 3.506730718302842e-07,
+      "loss": 0.7234,
+      "step": 33425
+    },
+    {
+      "epoch": 0.918019279887946,
+      "grad_norm": 0.0,
+      "learning_rate": 3.504396143420319e-07,
+      "loss": 0.7867,
+      "step": 33426
+    },
+    {
+      "epoch": 0.9180467441157892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5020623320483835e-07,
+      "loss": 0.8804,
+      "step": 33427
+    },
+    {
+      "epoch": 0.9180742083436324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.499729284205533e-07,
+      "loss": 0.8386,
+      "step": 33428
+    },
+    {
+      "epoch": 0.9181016725714757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4973969999101963e-07,
+      "loss": 0.7784,
+      "step": 33429
+    },
+    {
+      "epoch": 0.9181291367993188,
+      "grad_norm": 0.0,
+      "learning_rate": 3.495065479180826e-07,
+      "loss": 0.8973,
+      "step": 33430
+    },
+    {
+      "epoch": 0.9181566010271621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4927347220358844e-07,
+      "loss": 0.9097,
+      "step": 33431
+    },
+    {
+      "epoch": 0.9181840652550054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4904047284938127e-07,
+      "loss": 0.8148,
+      "step": 33432
+    },
+    {
+      "epoch": 0.9182115294828486,
+      "grad_norm": 0.0,
+      "learning_rate": 3.488075498573029e-07,
+      "loss": 0.7821,
+      "step": 33433
+    },
+    {
+      "epoch": 0.9182389937106918,
+      "grad_norm": 0.0,
+      "learning_rate": 3.485747032291997e-07,
+      "loss": 0.8687,
+      "step": 33434
+    },
+    {
+      "epoch": 0.918266457938535,
+      "grad_norm": 0.0,
+      "learning_rate": 3.483419329669113e-07,
+      "loss": 0.8656,
+      "step": 33435
+    },
+    {
+      "epoch": 0.9182939221663783,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4810923907228066e-07,
+      "loss": 0.6949,
+      "step": 33436
+    },
+    {
+      "epoch": 0.9183213863942216,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4787662154714854e-07,
+      "loss": 0.7921,
+      "step": 33437
+    },
+    {
+      "epoch": 0.9183488506220647,
+      "grad_norm": 0.0,
+      "learning_rate": 3.476440803933545e-07,
+      "loss": 0.8681,
+      "step": 33438
+    },
+    {
+      "epoch": 0.918376314849908,
+      "grad_norm": 0.0,
+      "learning_rate": 3.474116156127394e-07,
+      "loss": 0.7711,
+      "step": 33439
+    },
+    {
+      "epoch": 0.9184037790777513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4717922720714394e-07,
+      "loss": 0.8084,
+      "step": 33440
+    },
+    {
+      "epoch": 0.9184312433055944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.469469151784044e-07,
+      "loss": 0.7402,
+      "step": 33441
+    },
+    {
+      "epoch": 0.9184587075334377,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4671467952835934e-07,
+      "loss": 0.8268,
+      "step": 33442
+    },
+    {
+      "epoch": 0.9184861717612809,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4648252025884845e-07,
+      "loss": 0.7247,
+      "step": 33443
+    },
+    {
+      "epoch": 0.9185136359891242,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4625043737170793e-07,
+      "loss": 0.859,
+      "step": 33444
+    },
+    {
+      "epoch": 0.9185411002169674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4601843086877194e-07,
+      "loss": 0.7638,
+      "step": 33445
+    },
+    {
+      "epoch": 0.9185685644448106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4578650075187905e-07,
+      "loss": 0.7451,
+      "step": 33446
+    },
+    {
+      "epoch": 0.9185960286726539,
+      "grad_norm": 0.0,
+      "learning_rate": 3.455546470228621e-07,
+      "loss": 0.7373,
+      "step": 33447
+    },
+    {
+      "epoch": 0.918623492900497,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4532286968355644e-07,
+      "loss": 0.7649,
+      "step": 33448
+    },
+    {
+      "epoch": 0.9186509571283403,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4509116873579606e-07,
+      "loss": 0.8315,
+      "step": 33449
+    },
+    {
+      "epoch": 0.9186784213561836,
+      "grad_norm": 0.0,
+      "learning_rate": 3.44859544181414e-07,
+      "loss": 0.7738,
+      "step": 33450
+    },
+    {
+      "epoch": 0.9187058855840268,
+      "grad_norm": 0.0,
+      "learning_rate": 3.446279960222454e-07,
+      "loss": 0.8211,
+      "step": 33451
+    },
+    {
+      "epoch": 0.91873334981187,
+      "grad_norm": 0.0,
+      "learning_rate": 3.443965242601177e-07,
+      "loss": 0.7347,
+      "step": 33452
+    },
+    {
+      "epoch": 0.9187608140397133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.441651288968673e-07,
+      "loss": 0.6789,
+      "step": 33453
+    },
+    {
+      "epoch": 0.9187882782675565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.439338099343215e-07,
+      "loss": 0.8302,
+      "step": 33454
+    },
+    {
+      "epoch": 0.9188157424953998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.437025673743111e-07,
+      "loss": 0.8131,
+      "step": 33455
+    },
+    {
+      "epoch": 0.9188432067232429,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4347140121866684e-07,
+      "loss": 0.8135,
+      "step": 33456
+    },
+    {
+      "epoch": 0.9188706709510862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.432403114692173e-07,
+      "loss": 0.8035,
+      "step": 33457
+    },
+    {
+      "epoch": 0.9188981351789295,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4300929812779217e-07,
+      "loss": 0.7457,
+      "step": 33458
+    },
+    {
+      "epoch": 0.9189255994067727,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4277836119621766e-07,
+      "loss": 0.7974,
+      "step": 33459
+    },
+    {
+      "epoch": 0.9189530636346159,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4254750067632234e-07,
+      "loss": 0.7966,
+      "step": 33460
+    },
+    {
+      "epoch": 0.9189805278624591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4231671656993036e-07,
+      "loss": 0.7991,
+      "step": 33461
+    },
+    {
+      "epoch": 0.9190079920903024,
+      "grad_norm": 0.0,
+      "learning_rate": 3.420860088788702e-07,
+      "loss": 0.6904,
+      "step": 33462
+    },
+    {
+      "epoch": 0.9190354563181456,
+      "grad_norm": 0.0,
+      "learning_rate": 3.418553776049682e-07,
+      "loss": 0.8076,
+      "step": 33463
+    },
+    {
+      "epoch": 0.9190629205459888,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4162482275004517e-07,
+      "loss": 0.7926,
+      "step": 33464
+    },
+    {
+      "epoch": 0.9190903847738321,
+      "grad_norm": 0.0,
+      "learning_rate": 3.413943443159296e-07,
+      "loss": 0.8761,
+      "step": 33465
+    },
+    {
+      "epoch": 0.9191178490016754,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4116394230444214e-07,
+      "loss": 0.7265,
+      "step": 33466
+    },
+    {
+      "epoch": 0.9191453132295185,
+      "grad_norm": 0.0,
+      "learning_rate": 3.409336167174071e-07,
+      "loss": 0.7297,
+      "step": 33467
+    },
+    {
+      "epoch": 0.9191727774573618,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4070336755664846e-07,
+      "loss": 0.834,
+      "step": 33468
+    },
+    {
+      "epoch": 0.919200241685205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.404731948239859e-07,
+      "loss": 0.9019,
+      "step": 33469
+    },
+    {
+      "epoch": 0.9192277059130483,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4024309852124126e-07,
+      "loss": 0.7719,
+      "step": 33470
+    },
+    {
+      "epoch": 0.9192551701408915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4001307865023313e-07,
+      "loss": 0.7949,
+      "step": 33471
+    },
+    {
+      "epoch": 0.9192826343687347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.397831352127845e-07,
+      "loss": 0.8124,
+      "step": 33472
+    },
+    {
+      "epoch": 0.919310098596578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.395532682107139e-07,
+      "loss": 0.8614,
+      "step": 33473
+    },
+    {
+      "epoch": 0.9193375628244211,
+      "grad_norm": 0.0,
+      "learning_rate": 3.393234776458398e-07,
+      "loss": 0.8478,
+      "step": 33474
+    },
+    {
+      "epoch": 0.9193650270522644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3909376351998203e-07,
+      "loss": 0.7851,
+      "step": 33475
+    },
+    {
+      "epoch": 0.9193924912801077,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3886412583495456e-07,
+      "loss": 0.845,
+      "step": 33476
+    },
+    {
+      "epoch": 0.9194199555079509,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3863456459257816e-07,
+      "loss": 0.7774,
+      "step": 33477
+    },
+    {
+      "epoch": 0.9194474197357941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3840507979466695e-07,
+      "loss": 0.9656,
+      "step": 33478
+    },
+    {
+      "epoch": 0.9194748839636374,
+      "grad_norm": 0.0,
+      "learning_rate": 3.381756714430362e-07,
+      "loss": 0.9191,
+      "step": 33479
+    },
+    {
+      "epoch": 0.9195023481914806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.379463395395033e-07,
+      "loss": 0.8128,
+      "step": 33480
+    },
+    {
+      "epoch": 0.9195298124193239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.377170840858823e-07,
+      "loss": 0.795,
+      "step": 33481
+    },
+    {
+      "epoch": 0.919557276647167,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3748790508398523e-07,
+      "loss": 0.7822,
+      "step": 33482
+    },
+    {
+      "epoch": 0.9195847408750103,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3725880253562715e-07,
+      "loss": 0.8695,
+      "step": 33483
+    },
+    {
+      "epoch": 0.9196122051028536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.370297764426211e-07,
+      "loss": 0.9009,
+      "step": 33484
+    },
+    {
+      "epoch": 0.9196396693306967,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3680082680677795e-07,
+      "loss": 0.8585,
+      "step": 33485
+    },
+    {
+      "epoch": 0.91966713355854,
+      "grad_norm": 0.0,
+      "learning_rate": 3.365719536299095e-07,
+      "loss": 0.8851,
+      "step": 33486
+    },
+    {
+      "epoch": 0.9196945977863832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3634315691382755e-07,
+      "loss": 0.819,
+      "step": 33487
+    },
+    {
+      "epoch": 0.9197220620142265,
+      "grad_norm": 0.0,
+      "learning_rate": 3.361144366603419e-07,
+      "loss": 0.8331,
+      "step": 33488
+    },
+    {
+      "epoch": 0.9197495262420697,
+      "grad_norm": 0.0,
+      "learning_rate": 3.358857928712611e-07,
+      "loss": 0.7314,
+      "step": 33489
+    },
+    {
+      "epoch": 0.9197769904699129,
+      "grad_norm": 0.0,
+      "learning_rate": 3.356572255483959e-07,
+      "loss": 0.781,
+      "step": 33490
+    },
+    {
+      "epoch": 0.9198044546977562,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3542873469355475e-07,
+      "loss": 0.8529,
+      "step": 33491
+    },
+    {
+      "epoch": 0.9198319189255995,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3520032030854634e-07,
+      "loss": 0.683,
+      "step": 33492
+    },
+    {
+      "epoch": 0.9198593831534426,
+      "grad_norm": 0.0,
+      "learning_rate": 3.349719823951758e-07,
+      "loss": 0.8552,
+      "step": 33493
+    },
+    {
+      "epoch": 0.9198868473812859,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3474372095525066e-07,
+      "loss": 0.9613,
+      "step": 33494
+    },
+    {
+      "epoch": 0.9199143116091291,
+      "grad_norm": 0.0,
+      "learning_rate": 3.345155359905761e-07,
+      "loss": 0.7754,
+      "step": 33495
+    },
+    {
+      "epoch": 0.9199417758369723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3428742750295953e-07,
+      "loss": 0.8495,
+      "step": 33496
+    },
+    {
+      "epoch": 0.9199692400648156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3405939549420507e-07,
+      "loss": 0.8207,
+      "step": 33497
+    },
+    {
+      "epoch": 0.9199967042926588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.338314399661169e-07,
+      "loss": 0.8514,
+      "step": 33498
+    },
+    {
+      "epoch": 0.9200241685205021,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3360356092049905e-07,
+      "loss": 0.8765,
+      "step": 33499
+    },
+    {
+      "epoch": 0.9200516327483452,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3337575835915346e-07,
+      "loss": 0.7614,
+      "step": 33500
+    },
+    {
+      "epoch": 0.9200790969761885,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3314803228388537e-07,
+      "loss": 0.8045,
+      "step": 33501
+    },
+    {
+      "epoch": 0.9201065612040318,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3292038269649217e-07,
+      "loss": 0.7182,
+      "step": 33502
+    },
+    {
+      "epoch": 0.920134025431875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3269280959877804e-07,
+      "loss": 0.7683,
+      "step": 33503
+    },
+    {
+      "epoch": 0.9201614896597182,
+      "grad_norm": 0.0,
+      "learning_rate": 3.324653129925437e-07,
+      "loss": 0.9004,
+      "step": 33504
+    },
+    {
+      "epoch": 0.9201889538875615,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3223789287958886e-07,
+      "loss": 0.8396,
+      "step": 33505
+    },
+    {
+      "epoch": 0.9202164181154047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3201054926171206e-07,
+      "loss": 0.8088,
+      "step": 33506
+    },
+    {
+      "epoch": 0.9202438823432479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.31783282140713e-07,
+      "loss": 0.7229,
+      "step": 33507
+    },
+    {
+      "epoch": 0.9202713465710911,
+      "grad_norm": 0.0,
+      "learning_rate": 3.315560915183913e-07,
+      "loss": 0.8102,
+      "step": 33508
+    },
+    {
+      "epoch": 0.9202988107989344,
+      "grad_norm": 0.0,
+      "learning_rate": 3.313289773965411e-07,
+      "loss": 0.7832,
+      "step": 33509
+    },
+    {
+      "epoch": 0.9203262750267777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.31101939776961e-07,
+      "loss": 0.8061,
+      "step": 33510
+    },
+    {
+      "epoch": 0.9203537392546208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3087497866144957e-07,
+      "loss": 0.7939,
+      "step": 33511
+    },
+    {
+      "epoch": 0.9203812034824641,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3064809405179975e-07,
+      "loss": 0.8914,
+      "step": 33512
+    },
+    {
+      "epoch": 0.9204086677103073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.304212859498068e-07,
+      "loss": 0.7623,
+      "step": 33513
+    },
+    {
+      "epoch": 0.9204361319381505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.301945543572671e-07,
+      "loss": 0.8836,
+      "step": 33514
+    },
+    {
+      "epoch": 0.9204635961659938,
+      "grad_norm": 0.0,
+      "learning_rate": 3.299678992759736e-07,
+      "loss": 0.8131,
+      "step": 33515
+    },
+    {
+      "epoch": 0.920491060393837,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2974132070772155e-07,
+      "loss": 0.7557,
+      "step": 33516
+    },
+    {
+      "epoch": 0.9205185246216803,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2951481865430177e-07,
+      "loss": 0.7802,
+      "step": 33517
+    },
+    {
+      "epoch": 0.9205459888495235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2928839311750507e-07,
+      "loss": 0.7962,
+      "step": 33518
+    },
+    {
+      "epoch": 0.9205734530773667,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2906204409912547e-07,
+      "loss": 0.7699,
+      "step": 33519
+    },
+    {
+      "epoch": 0.92060091730521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.288357716009527e-07,
+      "loss": 0.9176,
+      "step": 33520
+    },
+    {
+      "epoch": 0.9206283815330532,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2860957562477645e-07,
+      "loss": 0.7764,
+      "step": 33521
+    },
+    {
+      "epoch": 0.9206558457608964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2838345617239087e-07,
+      "loss": 0.8161,
+      "step": 33522
+    },
+    {
+      "epoch": 0.9206833099887397,
+      "grad_norm": 0.0,
+      "learning_rate": 3.281574132455789e-07,
+      "loss": 0.7875,
+      "step": 33523
+    },
+    {
+      "epoch": 0.9207107742165829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2793144684613255e-07,
+      "loss": 0.782,
+      "step": 33524
+    },
+    {
+      "epoch": 0.9207382384444261,
+      "grad_norm": 0.0,
+      "learning_rate": 3.277055569758403e-07,
+      "loss": 0.823,
+      "step": 33525
+    },
+    {
+      "epoch": 0.9207657026722693,
+      "grad_norm": 0.0,
+      "learning_rate": 3.274797436364874e-07,
+      "loss": 0.7701,
+      "step": 33526
+    },
+    {
+      "epoch": 0.9207931669001126,
+      "grad_norm": 0.0,
+      "learning_rate": 3.272540068298624e-07,
+      "loss": 0.8836,
+      "step": 33527
+    },
+    {
+      "epoch": 0.9208206311279559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.270283465577506e-07,
+      "loss": 0.8115,
+      "step": 33528
+    },
+    {
+      "epoch": 0.920848095355799,
+      "grad_norm": 0.0,
+      "learning_rate": 3.268027628219361e-07,
+      "loss": 0.7784,
+      "step": 33529
+    },
+    {
+      "epoch": 0.9208755595836423,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2657725562420637e-07,
+      "loss": 0.8171,
+      "step": 33530
+    },
+    {
+      "epoch": 0.9209030238114856,
+      "grad_norm": 0.0,
+      "learning_rate": 3.263518249663433e-07,
+      "loss": 0.8853,
+      "step": 33531
+    },
+    {
+      "epoch": 0.9209304880393288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.261264708501333e-07,
+      "loss": 0.8503,
+      "step": 33532
+    },
+    {
+      "epoch": 0.920957952267172,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2590119327735703e-07,
+      "loss": 0.8538,
+      "step": 33533
+    },
+    {
+      "epoch": 0.9209854164950152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2567599224979876e-07,
+      "loss": 0.7251,
+      "step": 33534
+    },
+    {
+      "epoch": 0.9210128807228585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.254508677692392e-07,
+      "loss": 0.7611,
+      "step": 33535
+    },
+    {
+      "epoch": 0.9210403449507018,
+      "grad_norm": 0.0,
+      "learning_rate": 3.252258198374592e-07,
+      "loss": 0.7253,
+      "step": 33536
+    },
+    {
+      "epoch": 0.9210678091785449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.250008484562406e-07,
+      "loss": 0.8195,
+      "step": 33537
+    },
+    {
+      "epoch": 0.9210952734063882,
+      "grad_norm": 0.0,
+      "learning_rate": 3.24775953627362e-07,
+      "loss": 0.833,
+      "step": 33538
+    },
+    {
+      "epoch": 0.9211227376342314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.245511353526054e-07,
+      "loss": 0.812,
+      "step": 33539
+    },
+    {
+      "epoch": 0.9211502018620746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.243263936337482e-07,
+      "loss": 0.7758,
+      "step": 33540
+    },
+    {
+      "epoch": 0.9211776660899179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.241017284725689e-07,
+      "loss": 0.7423,
+      "step": 33541
+    },
+    {
+      "epoch": 0.9212051303177611,
+      "grad_norm": 0.0,
+      "learning_rate": 3.23877139870844e-07,
+      "loss": 0.7807,
+      "step": 33542
+    },
+    {
+      "epoch": 0.9212325945456044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.236526278303509e-07,
+      "loss": 0.8776,
+      "step": 33543
+    },
+    {
+      "epoch": 0.9212600587734476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.234281923528659e-07,
+      "loss": 0.8285,
+      "step": 33544
+    },
+    {
+      "epoch": 0.9212875230012908,
+      "grad_norm": 0.0,
+      "learning_rate": 3.232038334401666e-07,
+      "loss": 0.8974,
+      "step": 33545
+    },
+    {
+      "epoch": 0.9213149872291341,
+      "grad_norm": 0.0,
+      "learning_rate": 3.22979551094027e-07,
+      "loss": 0.7871,
+      "step": 33546
+    },
+    {
+      "epoch": 0.9213424514569772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2275534531622134e-07,
+      "loss": 0.7449,
+      "step": 33547
+    },
+    {
+      "epoch": 0.9213699156848205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.225312161085237e-07,
+      "loss": 0.7994,
+      "step": 33548
+    },
+    {
+      "epoch": 0.9213973799126638,
+      "grad_norm": 0.0,
+      "learning_rate": 3.223071634727093e-07,
+      "loss": 0.8395,
+      "step": 33549
+    },
+    {
+      "epoch": 0.921424844140507,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2208318741054787e-07,
+      "loss": 0.8434,
+      "step": 33550
+    },
+    {
+      "epoch": 0.9214523083683502,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2185928792381247e-07,
+      "loss": 0.786,
+      "step": 33551
+    },
+    {
+      "epoch": 0.9214797725961934,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2163546501427724e-07,
+      "loss": 0.7635,
+      "step": 33552
+    },
+    {
+      "epoch": 0.9215072368240367,
+      "grad_norm": 0.0,
+      "learning_rate": 3.214117186837107e-07,
+      "loss": 0.7494,
+      "step": 33553
+    },
+    {
+      "epoch": 0.92153470105188,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2118804893388256e-07,
+      "loss": 0.7979,
+      "step": 33554
+    },
+    {
+      "epoch": 0.9215621652797231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2096445576656476e-07,
+      "loss": 0.8793,
+      "step": 33555
+    },
+    {
+      "epoch": 0.9215896295075664,
+      "grad_norm": 0.0,
+      "learning_rate": 3.207409391835259e-07,
+      "loss": 0.8356,
+      "step": 33556
+    },
+    {
+      "epoch": 0.9216170937354097,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2051749918653343e-07,
+      "loss": 0.8156,
+      "step": 33557
+    },
+    {
+      "epoch": 0.9216445579632528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.20294135777357e-07,
+      "loss": 0.7947,
+      "step": 33558
+    },
+    {
+      "epoch": 0.9216720221910961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2007084895776195e-07,
+      "loss": 0.7337,
+      "step": 33559
+    },
+    {
+      "epoch": 0.9216994864189393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1984763872951576e-07,
+      "loss": 0.777,
+      "step": 33560
+    },
+    {
+      "epoch": 0.9217269506467826,
+      "grad_norm": 0.0,
+      "learning_rate": 3.196245050943847e-07,
+      "loss": 0.9257,
+      "step": 33561
+    },
+    {
+      "epoch": 0.9217544148746258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.194014480541352e-07,
+      "loss": 0.9109,
+      "step": 33562
+    },
+    {
+      "epoch": 0.921781879102469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1917846761053026e-07,
+      "loss": 0.7649,
+      "step": 33563
+    },
+    {
+      "epoch": 0.9218093433303123,
+      "grad_norm": 0.0,
+      "learning_rate": 3.189555637653374e-07,
+      "loss": 0.7059,
+      "step": 33564
+    },
+    {
+      "epoch": 0.9218368075581556,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1873273652031743e-07,
+      "loss": 0.8523,
+      "step": 33565
+    },
+    {
+      "epoch": 0.9218642717859987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.185099858772334e-07,
+      "loss": 0.8581,
+      "step": 33566
+    },
+    {
+      "epoch": 0.921891736013842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.182873118378482e-07,
+      "loss": 0.8547,
+      "step": 33567
+    },
+    {
+      "epoch": 0.9219192002416852,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1806471440392394e-07,
+      "loss": 0.7615,
+      "step": 33568
+    },
+    {
+      "epoch": 0.9219466644695284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.178421935772236e-07,
+      "loss": 0.7529,
+      "step": 33569
+    },
+    {
+      "epoch": 0.9219741286973717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1761974935950566e-07,
+      "loss": 0.817,
+      "step": 33570
+    },
+    {
+      "epoch": 0.9220015929252149,
+      "grad_norm": 0.0,
+      "learning_rate": 3.173973817525311e-07,
+      "loss": 0.8484,
+      "step": 33571
+    },
+    {
+      "epoch": 0.9220290571530582,
+      "grad_norm": 0.0,
+      "learning_rate": 3.171750907580595e-07,
+      "loss": 0.7521,
+      "step": 33572
+    },
+    {
+      "epoch": 0.9220565213809013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.169528763778484e-07,
+      "loss": 0.8981,
+      "step": 33573
+    },
+    {
+      "epoch": 0.9220839856087446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1673073861365645e-07,
+      "loss": 0.8089,
+      "step": 33574
+    },
+    {
+      "epoch": 0.9221114498365879,
+      "grad_norm": 0.0,
+      "learning_rate": 3.165086774672432e-07,
+      "loss": 0.7194,
+      "step": 33575
+    },
+    {
+      "epoch": 0.922138914064431,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1628669294036407e-07,
+      "loss": 0.7247,
+      "step": 33576
+    },
+    {
+      "epoch": 0.9221663782922743,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1606478503477534e-07,
+      "loss": 0.7991,
+      "step": 33577
+    },
+    {
+      "epoch": 0.9221938425201176,
+      "grad_norm": 0.0,
+      "learning_rate": 3.158429537522334e-07,
+      "loss": 0.788,
+      "step": 33578
+    },
+    {
+      "epoch": 0.9222213067479608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.156211990944924e-07,
+      "loss": 0.8527,
+      "step": 33579
+    },
+    {
+      "epoch": 0.922248770975804,
+      "grad_norm": 0.0,
+      "learning_rate": 3.153995210633098e-07,
+      "loss": 0.8845,
+      "step": 33580
+    },
+    {
+      "epoch": 0.9222762352036472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1517791966043764e-07,
+      "loss": 0.8341,
+      "step": 33581
+    },
+    {
+      "epoch": 0.9223036994314905,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1495639488762775e-07,
+      "loss": 0.8038,
+      "step": 33582
+    },
+    {
+      "epoch": 0.9223311636593338,
+      "grad_norm": 0.0,
+      "learning_rate": 3.147349467466343e-07,
+      "loss": 0.7993,
+      "step": 33583
+    },
+    {
+      "epoch": 0.9223586278871769,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1451357523921034e-07,
+      "loss": 0.6799,
+      "step": 33584
+    },
+    {
+      "epoch": 0.9223860921150202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.142922803671067e-07,
+      "loss": 0.7965,
+      "step": 33585
+    },
+    {
+      "epoch": 0.9224135563428634,
+      "grad_norm": 0.0,
+      "learning_rate": 3.140710621320731e-07,
+      "loss": 0.7665,
+      "step": 33586
+    },
+    {
+      "epoch": 0.9224410205707066,
+      "grad_norm": 0.0,
+      "learning_rate": 3.138499205358636e-07,
+      "loss": 0.7934,
+      "step": 33587
+    },
+    {
+      "epoch": 0.9224684847985499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1362885558022473e-07,
+      "loss": 0.8307,
+      "step": 33588
+    },
+    {
+      "epoch": 0.9224959490263931,
+      "grad_norm": 0.0,
+      "learning_rate": 3.134078672669072e-07,
+      "loss": 0.7615,
+      "step": 33589
+    },
+    {
+      "epoch": 0.9225234132542364,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1318695559765743e-07,
+      "loss": 0.8466,
+      "step": 33590
+    },
+    {
+      "epoch": 0.9225508774820796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1296612057422516e-07,
+      "loss": 0.727,
+      "step": 33591
+    },
+    {
+      "epoch": 0.9225783417099228,
+      "grad_norm": 0.0,
+      "learning_rate": 3.127453621983567e-07,
+      "loss": 0.8003,
+      "step": 33592
+    },
+    {
+      "epoch": 0.9226058059377661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1252468047180183e-07,
+      "loss": 0.8772,
+      "step": 33593
+    },
+    {
+      "epoch": 0.9226332701656093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.123040753963025e-07,
+      "loss": 0.7659,
+      "step": 33594
+    },
+    {
+      "epoch": 0.9226607343934525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.12083546973605e-07,
+      "loss": 0.7581,
+      "step": 33595
+    },
+    {
+      "epoch": 0.9226881986212958,
+      "grad_norm": 0.0,
+      "learning_rate": 3.11863095205458e-07,
+      "loss": 0.7378,
+      "step": 33596
+    },
+    {
+      "epoch": 0.922715662849139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.116427200936012e-07,
+      "loss": 0.7197,
+      "step": 33597
+    },
+    {
+      "epoch": 0.9227431270769823,
+      "grad_norm": 0.0,
+      "learning_rate": 3.114224216397799e-07,
+      "loss": 0.7801,
+      "step": 33598
+    },
+    {
+      "epoch": 0.9227705913048254,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1120219984573817e-07,
+      "loss": 0.8415,
+      "step": 33599
+    },
+    {
+      "epoch": 0.9227980555326687,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1098205471321697e-07,
+      "loss": 0.782,
+      "step": 33600
+    },
+    {
+      "epoch": 0.922825519760512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1076198624395927e-07,
+      "loss": 0.6884,
+      "step": 33601
+    },
+    {
+      "epoch": 0.9228529839883551,
+      "grad_norm": 0.0,
+      "learning_rate": 3.105419944397059e-07,
+      "loss": 0.7356,
+      "step": 33602
+    },
+    {
+      "epoch": 0.9228804482161984,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103220793021966e-07,
+      "loss": 0.6867,
+      "step": 33603
+    },
+    {
+      "epoch": 0.9229079124440417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.101022408331744e-07,
+      "loss": 0.8015,
+      "step": 33604
+    },
+    {
+      "epoch": 0.9229353766718849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0988247903437684e-07,
+      "loss": 0.7291,
+      "step": 33605
+    },
+    {
+      "epoch": 0.9229628408997281,
+      "grad_norm": 0.0,
+      "learning_rate": 3.096627939075414e-07,
+      "loss": 0.8025,
+      "step": 33606
+    },
+    {
+      "epoch": 0.9229903051275713,
+      "grad_norm": 0.0,
+      "learning_rate": 3.094431854544078e-07,
+      "loss": 0.8258,
+      "step": 33607
+    },
+    {
+      "epoch": 0.9230177693554146,
+      "grad_norm": 0.0,
+      "learning_rate": 3.092236536767135e-07,
+      "loss": 0.7317,
+      "step": 33608
+    },
+    {
+      "epoch": 0.9230452335832579,
+      "grad_norm": 0.0,
+      "learning_rate": 3.09004198576196e-07,
+      "loss": 0.7833,
+      "step": 33609
+    },
+    {
+      "epoch": 0.923072697811101,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0878482015459066e-07,
+      "loss": 0.748,
+      "step": 33610
+    },
+    {
+      "epoch": 0.9231001620389443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.085655184136338e-07,
+      "loss": 0.759,
+      "step": 33611
+    },
+    {
+      "epoch": 0.9231276262667875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0834629335506073e-07,
+      "loss": 0.8949,
+      "step": 33612
+    },
+    {
+      "epoch": 0.9231550904946307,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0812714498060783e-07,
+      "loss": 0.8125,
+      "step": 33613
+    },
+    {
+      "epoch": 0.923182554722474,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0790807329200477e-07,
+      "loss": 0.8259,
+      "step": 33614
+    },
+    {
+      "epoch": 0.9232100189503172,
+      "grad_norm": 0.0,
+      "learning_rate": 3.07689078290988e-07,
+      "loss": 0.8221,
+      "step": 33615
+    },
+    {
+      "epoch": 0.9232374831781605,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0747015997928955e-07,
+      "loss": 0.9359,
+      "step": 33616
+    },
+    {
+      "epoch": 0.9232649474060037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.072513183586434e-07,
+      "loss": 0.8379,
+      "step": 33617
+    },
+    {
+      "epoch": 0.9232924116338469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.070325534307772e-07,
+      "loss": 0.8028,
+      "step": 33618
+    },
+    {
+      "epoch": 0.9233198758616902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.068138651974251e-07,
+      "loss": 0.7402,
+      "step": 33619
+    },
+    {
+      "epoch": 0.9233473400895333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.065952536603167e-07,
+      "loss": 0.6838,
+      "step": 33620
+    },
+    {
+      "epoch": 0.9233748043173766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.063767188211808e-07,
+      "loss": 0.8499,
+      "step": 33621
+    },
+    {
+      "epoch": 0.9234022685452199,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0615826068174815e-07,
+      "loss": 0.7774,
+      "step": 33622
+    },
+    {
+      "epoch": 0.9234297327730631,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0593987924374626e-07,
+      "loss": 0.7933,
+      "step": 33623
+    },
+    {
+      "epoch": 0.9234571970009063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.057215745089026e-07,
+      "loss": 0.8227,
+      "step": 33624
+    },
+    {
+      "epoch": 0.9234846612287495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0550334647894477e-07,
+      "loss": 0.8133,
+      "step": 33625
+    },
+    {
+      "epoch": 0.9235121254565928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0528519515560015e-07,
+      "loss": 0.7162,
+      "step": 33626
+    },
+    {
+      "epoch": 0.9235395896844361,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0506712054059416e-07,
+      "loss": 0.7726,
+      "step": 33627
+    },
+    {
+      "epoch": 0.9235670539122792,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0484912263565315e-07,
+      "loss": 0.9416,
+      "step": 33628
+    },
+    {
+      "epoch": 0.9235945181401225,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0463120144250236e-07,
+      "loss": 0.7405,
+      "step": 33629
+    },
+    {
+      "epoch": 0.9236219823679658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0441335696286376e-07,
+      "loss": 0.847,
+      "step": 33630
+    },
+    {
+      "epoch": 0.9236494465958089,
+      "grad_norm": 0.0,
+      "learning_rate": 3.041955891984627e-07,
+      "loss": 0.8449,
+      "step": 33631
+    },
+    {
+      "epoch": 0.9236769108236522,
+      "grad_norm": 0.0,
+      "learning_rate": 3.039778981510211e-07,
+      "loss": 0.7296,
+      "step": 33632
+    },
+    {
+      "epoch": 0.9237043750514954,
+      "grad_norm": 0.0,
+      "learning_rate": 3.037602838222631e-07,
+      "loss": 0.7021,
+      "step": 33633
+    },
+    {
+      "epoch": 0.9237318392793387,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0354274621390847e-07,
+      "loss": 0.838,
+      "step": 33634
+    },
+    {
+      "epoch": 0.9237593035071819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.033252853276825e-07,
+      "loss": 0.8044,
+      "step": 33635
+    },
+    {
+      "epoch": 0.9237867677350251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.031079011653004e-07,
+      "loss": 0.7752,
+      "step": 33636
+    },
+    {
+      "epoch": 0.9238142319628684,
+      "grad_norm": 0.0,
+      "learning_rate": 3.028905937284865e-07,
+      "loss": 0.746,
+      "step": 33637
+    },
+    {
+      "epoch": 0.9238416961907115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.026733630189571e-07,
+      "loss": 0.7622,
+      "step": 33638
+    },
+    {
+      "epoch": 0.9238691604185548,
+      "grad_norm": 0.0,
+      "learning_rate": 3.024562090384331e-07,
+      "loss": 0.8484,
+      "step": 33639
+    },
+    {
+      "epoch": 0.9238966246463981,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0223913178863305e-07,
+      "loss": 0.8015,
+      "step": 33640
+    },
+    {
+      "epoch": 0.9239240888742413,
+      "grad_norm": 0.0,
+      "learning_rate": 3.020221312712723e-07,
+      "loss": 0.8414,
+      "step": 33641
+    },
+    {
+      "epoch": 0.9239515531020845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.018052074880684e-07,
+      "loss": 0.7693,
+      "step": 33642
+    },
+    {
+      "epoch": 0.9239790173299278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.015883604407399e-07,
+      "loss": 0.8547,
+      "step": 33643
+    },
+    {
+      "epoch": 0.924006481557771,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0137159013100106e-07,
+      "loss": 0.8648,
+      "step": 33644
+    },
+    {
+      "epoch": 0.9240339457856143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0115489656056597e-07,
+      "loss": 0.7727,
+      "step": 33645
+    },
+    {
+      "epoch": 0.9240614100134574,
+      "grad_norm": 0.0,
+      "learning_rate": 3.009382797311522e-07,
+      "loss": 0.8391,
+      "step": 33646
+    },
+    {
+      "epoch": 0.9240888742413007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.007217396444695e-07,
+      "loss": 0.8329,
+      "step": 33647
+    },
+    {
+      "epoch": 0.924116338469144,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0050527630223427e-07,
+      "loss": 0.841,
+      "step": 33648
+    },
+    {
+      "epoch": 0.9241438026969871,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0028888970615844e-07,
+      "loss": 0.7225,
+      "step": 33649
+    },
+    {
+      "epoch": 0.9241712669248304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.00072579857954e-07,
+      "loss": 0.7132,
+      "step": 33650
+    },
+    {
+      "epoch": 0.9241987311526736,
+      "grad_norm": 0.0,
+      "learning_rate": 2.99856346759333e-07,
+      "loss": 0.9156,
+      "step": 33651
+    },
+    {
+      "epoch": 0.9242261953805169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.996401904120072e-07,
+      "loss": 0.8469,
+      "step": 33652
+    },
+    {
+      "epoch": 0.9242536596083601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9942411081768653e-07,
+      "loss": 0.8209,
+      "step": 33653
+    },
+    {
+      "epoch": 0.9242811238362033,
+      "grad_norm": 0.0,
+      "learning_rate": 2.992081079780773e-07,
+      "loss": 0.9739,
+      "step": 33654
+    },
+    {
+      "epoch": 0.9243085880640466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9899218189489267e-07,
+      "loss": 0.7692,
+      "step": 33655
+    },
+    {
+      "epoch": 0.9243360522918899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9877633256983894e-07,
+      "loss": 0.7703,
+      "step": 33656
+    },
+    {
+      "epoch": 0.924363516519733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9856056000462487e-07,
+      "loss": 0.8428,
+      "step": 33657
+    },
+    {
+      "epoch": 0.9243909807475763,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9834486420095785e-07,
+      "loss": 0.7642,
+      "step": 33658
+    },
+    {
+      "epoch": 0.9244184449754195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.981292451605444e-07,
+      "loss": 0.8262,
+      "step": 33659
+    },
+    {
+      "epoch": 0.9244459092032628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.979137028850898e-07,
+      "loss": 0.7324,
+      "step": 33660
+    },
+    {
+      "epoch": 0.924473373431106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9769823737630157e-07,
+      "loss": 0.7393,
+      "step": 33661
+    },
+    {
+      "epoch": 0.9245008376589492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.974828486358827e-07,
+      "loss": 0.7615,
+      "step": 33662
+    },
+    {
+      "epoch": 0.9245283018867925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9726753666553753e-07,
+      "loss": 0.8363,
+      "step": 33663
+    },
+    {
+      "epoch": 0.9245557661146356,
+      "grad_norm": 0.0,
+      "learning_rate": 2.970523014669713e-07,
+      "loss": 0.8299,
+      "step": 33664
+    },
+    {
+      "epoch": 0.9245832303424789,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9683714304188373e-07,
+      "loss": 0.8109,
+      "step": 33665
+    },
+    {
+      "epoch": 0.9246106945703222,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9662206139198014e-07,
+      "loss": 0.8731,
+      "step": 33666
+    },
+    {
+      "epoch": 0.9246381587981654,
+      "grad_norm": 0.0,
+      "learning_rate": 2.964070565189614e-07,
+      "loss": 0.8238,
+      "step": 33667
+    },
+    {
+      "epoch": 0.9246656230260086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.961921284245306e-07,
+      "loss": 0.7589,
+      "step": 33668
+    },
+    {
+      "epoch": 0.9246930872538519,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9597727711038417e-07,
+      "loss": 0.8159,
+      "step": 33669
+    },
+    {
+      "epoch": 0.9247205514816951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.957625025782274e-07,
+      "loss": 0.823,
+      "step": 33670
+    },
+    {
+      "epoch": 0.9247480157095384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.955478048297544e-07,
+      "loss": 0.8342,
+      "step": 33671
+    },
+    {
+      "epoch": 0.9247754799373815,
+      "grad_norm": 0.0,
+      "learning_rate": 2.953331838666662e-07,
+      "loss": 0.8197,
+      "step": 33672
+    },
+    {
+      "epoch": 0.9248029441652248,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9511863969066135e-07,
+      "loss": 0.8782,
+      "step": 33673
+    },
+    {
+      "epoch": 0.9248304083930681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.949041723034374e-07,
+      "loss": 0.9161,
+      "step": 33674
+    },
+    {
+      "epoch": 0.9248578726209112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.946897817066907e-07,
+      "loss": 0.7704,
+      "step": 33675
+    },
+    {
+      "epoch": 0.9248853368487545,
+      "grad_norm": 0.0,
+      "learning_rate": 2.944754679021189e-07,
+      "loss": 0.6708,
+      "step": 33676
+    },
+    {
+      "epoch": 0.9249128010765977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.942612308914172e-07,
+      "loss": 0.752,
+      "step": 33677
+    },
+    {
+      "epoch": 0.924940265304441,
+      "grad_norm": 0.0,
+      "learning_rate": 2.940470706762788e-07,
+      "loss": 0.8386,
+      "step": 33678
+    },
+    {
+      "epoch": 0.9249677295322842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.938329872584e-07,
+      "loss": 0.7773,
+      "step": 33679
+    },
+    {
+      "epoch": 0.9249951937601274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9361898063947405e-07,
+      "loss": 0.7771,
+      "step": 33680
+    },
+    {
+      "epoch": 0.9250226579879707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9340505082119497e-07,
+      "loss": 0.7915,
+      "step": 33681
+    },
+    {
+      "epoch": 0.925050122215814,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9319119780525595e-07,
+      "loss": 0.7512,
+      "step": 33682
+    },
+    {
+      "epoch": 0.9250775864436571,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9297742159334783e-07,
+      "loss": 0.929,
+      "step": 33683
+    },
+    {
+      "epoch": 0.9251050506715004,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9276372218716155e-07,
+      "loss": 0.7368,
+      "step": 33684
+    },
+    {
+      "epoch": 0.9251325148993436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.925500995883912e-07,
+      "loss": 0.8688,
+      "step": 33685
+    },
+    {
+      "epoch": 0.9251599791271868,
+      "grad_norm": 0.0,
+      "learning_rate": 2.923365537987222e-07,
+      "loss": 0.7913,
+      "step": 33686
+    },
+    {
+      "epoch": 0.9251874433550301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9212308481984755e-07,
+      "loss": 0.737,
+      "step": 33687
+    },
+    {
+      "epoch": 0.9252149075828733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9190969265345703e-07,
+      "loss": 0.8586,
+      "step": 33688
+    },
+    {
+      "epoch": 0.9252423718107166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9169637730123603e-07,
+      "loss": 0.8118,
+      "step": 33689
+    },
+    {
+      "epoch": 0.9252698360385597,
+      "grad_norm": 0.0,
+      "learning_rate": 2.914831387648731e-07,
+      "loss": 0.8701,
+      "step": 33690
+    },
+    {
+      "epoch": 0.925297300266403,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9126997704605695e-07,
+      "loss": 0.7644,
+      "step": 33691
+    },
+    {
+      "epoch": 0.9253247644942463,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9105689214647515e-07,
+      "loss": 0.8285,
+      "step": 33692
+    },
+    {
+      "epoch": 0.9253522287220894,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9084388406781074e-07,
+      "loss": 0.8104,
+      "step": 33693
+    },
+    {
+      "epoch": 0.9253796929499327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9063095281175123e-07,
+      "loss": 0.8377,
+      "step": 33694
+    },
+    {
+      "epoch": 0.925407157177776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.904180983799798e-07,
+      "loss": 0.8428,
+      "step": 33695
+    },
+    {
+      "epoch": 0.9254346214056192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9020532077418173e-07,
+      "loss": 0.7055,
+      "step": 33696
+    },
+    {
+      "epoch": 0.9254620856334624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.899926199960401e-07,
+      "loss": 0.8154,
+      "step": 33697
+    },
+    {
+      "epoch": 0.9254895498613056,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8977999604723805e-07,
+      "loss": 0.7029,
+      "step": 33698
+    },
+    {
+      "epoch": 0.9255170140891489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8956744892945976e-07,
+      "loss": 0.8946,
+      "step": 33699
+    },
+    {
+      "epoch": 0.9255444783169922,
+      "grad_norm": 0.0,
+      "learning_rate": 2.893549786443839e-07,
+      "loss": 0.8271,
+      "step": 33700
+    },
+    {
+      "epoch": 0.9255719425448353,
+      "grad_norm": 0.0,
+      "learning_rate": 2.891425851936935e-07,
+      "loss": 0.7698,
+      "step": 33701
+    },
+    {
+      "epoch": 0.9255994067726786,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8893026857906846e-07,
+      "loss": 0.8104,
+      "step": 33702
+    },
+    {
+      "epoch": 0.9256268710005218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.887180288021885e-07,
+      "loss": 0.8953,
+      "step": 33703
+    },
+    {
+      "epoch": 0.925654335228365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.885058658647322e-07,
+      "loss": 0.7969,
+      "step": 33704
+    },
+    {
+      "epoch": 0.9256817994562083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.882937797683816e-07,
+      "loss": 0.7961,
+      "step": 33705
+    },
+    {
+      "epoch": 0.9257092636840515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.88081770514812e-07,
+      "loss": 0.7,
+      "step": 33706
+    },
+    {
+      "epoch": 0.9257367279118948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.878698381056999e-07,
+      "loss": 0.7856,
+      "step": 33707
+    },
+    {
+      "epoch": 0.925764192139738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8765798254272504e-07,
+      "loss": 0.8786,
+      "step": 33708
+    },
+    {
+      "epoch": 0.9257916563675812,
+      "grad_norm": 0.0,
+      "learning_rate": 2.874462038275627e-07,
+      "loss": 0.8723,
+      "step": 33709
+    },
+    {
+      "epoch": 0.9258191205954245,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8723450196188715e-07,
+      "loss": 0.8363,
+      "step": 33710
+    },
+    {
+      "epoch": 0.9258465848232676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8702287694737706e-07,
+      "loss": 0.8586,
+      "step": 33711
+    },
+    {
+      "epoch": 0.9258740490511109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8681132878570216e-07,
+      "loss": 0.837,
+      "step": 33712
+    },
+    {
+      "epoch": 0.9259015132789542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8659985747853894e-07,
+      "loss": 0.8093,
+      "step": 33713
+    },
+    {
+      "epoch": 0.9259289775067974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8638846302755927e-07,
+      "loss": 0.8255,
+      "step": 33714
+    },
+    {
+      "epoch": 0.9259564417346406,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8617714543443754e-07,
+      "loss": 0.828,
+      "step": 33715
+    },
+    {
+      "epoch": 0.9259839059624838,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8596590470084674e-07,
+      "loss": 0.8016,
+      "step": 33716
+    },
+    {
+      "epoch": 0.9260113701903271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8575474082845557e-07,
+      "loss": 0.8177,
+      "step": 33717
+    },
+    {
+      "epoch": 0.9260388344181704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8554365381893487e-07,
+      "loss": 0.8231,
+      "step": 33718
+    },
+    {
+      "epoch": 0.9260662986460135,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8533264367395673e-07,
+      "loss": 0.757,
+      "step": 33719
+    },
+    {
+      "epoch": 0.9260937628738568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8512171039518863e-07,
+      "loss": 0.7481,
+      "step": 33720
+    },
+    {
+      "epoch": 0.9261212271017001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.849108539843015e-07,
+      "loss": 0.8434,
+      "step": 33721
+    },
+    {
+      "epoch": 0.9261486913295432,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8470007444296287e-07,
+      "loss": 0.8279,
+      "step": 33722
+    },
+    {
+      "epoch": 0.9261761555573865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.844893717728414e-07,
+      "loss": 0.8889,
+      "step": 33723
+    },
+    {
+      "epoch": 0.9262036197852297,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8427874597560244e-07,
+      "loss": 0.8741,
+      "step": 33724
+    },
+    {
+      "epoch": 0.926231084013073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.840681970529147e-07,
+      "loss": 0.9496,
+      "step": 33725
+    },
+    {
+      "epoch": 0.9262585482409162,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8385772500644116e-07,
+      "loss": 0.7889,
+      "step": 33726
+    },
+    {
+      "epoch": 0.9262860124687594,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836473298378495e-07,
+      "loss": 0.7153,
+      "step": 33727
+    },
+    {
+      "epoch": 0.9263134766966027,
+      "grad_norm": 0.0,
+      "learning_rate": 2.83437011548805e-07,
+      "loss": 0.7871,
+      "step": 33728
+    },
+    {
+      "epoch": 0.9263409409244459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8322677014097075e-07,
+      "loss": 0.8523,
+      "step": 33729
+    },
+    {
+      "epoch": 0.9263684051522891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8301660561600886e-07,
+      "loss": 0.8017,
+      "step": 33730
+    },
+    {
+      "epoch": 0.9263958693801324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8280651797558454e-07,
+      "loss": 0.8243,
+      "step": 33731
+    },
+    {
+      "epoch": 0.9264233336079756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.825965072213588e-07,
+      "loss": 0.8183,
+      "step": 33732
+    },
+    {
+      "epoch": 0.9264507978358189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.823865733549935e-07,
+      "loss": 0.8134,
+      "step": 33733
+    },
+    {
+      "epoch": 0.9264782620636621,
+      "grad_norm": 0.0,
+      "learning_rate": 2.821767163781497e-07,
+      "loss": 0.7924,
+      "step": 33734
+    },
+    {
+      "epoch": 0.9265057262915053,
+      "grad_norm": 0.0,
+      "learning_rate": 2.819669362924893e-07,
+      "loss": 0.9261,
+      "step": 33735
+    },
+    {
+      "epoch": 0.9265331905193486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8175723309966986e-07,
+      "loss": 0.9212,
+      "step": 33736
+    },
+    {
+      "epoch": 0.9265606547471917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8154760680135116e-07,
+      "loss": 0.8353,
+      "step": 33737
+    },
+    {
+      "epoch": 0.926588118975035,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8133805739919307e-07,
+      "loss": 0.7829,
+      "step": 33738
+    },
+    {
+      "epoch": 0.9266155832028783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8112858489485307e-07,
+      "loss": 0.7865,
+      "step": 33739
+    },
+    {
+      "epoch": 0.9266430474307215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8091918928998874e-07,
+      "loss": 0.7927,
+      "step": 33740
+    },
+    {
+      "epoch": 0.9266705116585647,
+      "grad_norm": 0.0,
+      "learning_rate": 2.807098705862565e-07,
+      "loss": 0.7744,
+      "step": 33741
+    },
+    {
+      "epoch": 0.9266979758864079,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8050062878531294e-07,
+      "loss": 0.7361,
+      "step": 33742
+    },
+    {
+      "epoch": 0.9267254401142512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.802914638888121e-07,
+      "loss": 0.7913,
+      "step": 33743
+    },
+    {
+      "epoch": 0.9267529043420945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8008237589841057e-07,
+      "loss": 0.8671,
+      "step": 33744
+    },
+    {
+      "epoch": 0.9267803685699376,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7987336481576256e-07,
+      "loss": 0.7011,
+      "step": 33745
+    },
+    {
+      "epoch": 0.9268078327977809,
+      "grad_norm": 0.0,
+      "learning_rate": 2.796644306425211e-07,
+      "loss": 0.6893,
+      "step": 33746
+    },
+    {
+      "epoch": 0.9268352970256242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.794555733803417e-07,
+      "loss": 0.7805,
+      "step": 33747
+    },
+    {
+      "epoch": 0.9268627612534673,
+      "grad_norm": 0.0,
+      "learning_rate": 2.79246793030874e-07,
+      "loss": 0.7463,
+      "step": 33748
+    },
+    {
+      "epoch": 0.9268902254813106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7903808959577226e-07,
+      "loss": 0.7633,
+      "step": 33749
+    },
+    {
+      "epoch": 0.9269176897091538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7882946307668525e-07,
+      "loss": 0.8554,
+      "step": 33750
+    },
+    {
+      "epoch": 0.9269451539369971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7862091347526597e-07,
+      "loss": 0.9123,
+      "step": 33751
+    },
+    {
+      "epoch": 0.9269726181648403,
+      "grad_norm": 0.0,
+      "learning_rate": 2.784124407931632e-07,
+      "loss": 0.7552,
+      "step": 33752
+    },
+    {
+      "epoch": 0.9270000823926835,
+      "grad_norm": 0.0,
+      "learning_rate": 2.782040450320278e-07,
+      "loss": 0.7617,
+      "step": 33753
+    },
+    {
+      "epoch": 0.9270275466205268,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7799572619350736e-07,
+      "loss": 0.6811,
+      "step": 33754
+    },
+    {
+      "epoch": 0.92705501084837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.777874842792494e-07,
+      "loss": 0.8493,
+      "step": 33755
+    },
+    {
+      "epoch": 0.9270824750762132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.775793192909049e-07,
+      "loss": 0.7377,
+      "step": 33756
+    },
+    {
+      "epoch": 0.9271099393040565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.77371231230118e-07,
+      "loss": 0.8302,
+      "step": 33757
+    },
+    {
+      "epoch": 0.9271374035318997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7716322009853636e-07,
+      "loss": 0.7822,
+      "step": 33758
+    },
+    {
+      "epoch": 0.9271648677597429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.769552858978064e-07,
+      "loss": 0.7592,
+      "step": 33759
+    },
+    {
+      "epoch": 0.9271923319875862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.767474286295713e-07,
+      "loss": 0.741,
+      "step": 33760
+    },
+    {
+      "epoch": 0.9272197962154294,
+      "grad_norm": 0.0,
+      "learning_rate": 2.765396482954763e-07,
+      "loss": 0.7839,
+      "step": 33761
+    },
+    {
+      "epoch": 0.9272472604432727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7633194489716687e-07,
+      "loss": 0.821,
+      "step": 33762
+    },
+    {
+      "epoch": 0.9272747246711158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.761243184362861e-07,
+      "loss": 0.8578,
+      "step": 33763
+    },
+    {
+      "epoch": 0.9273021888989591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7591676891447707e-07,
+      "loss": 0.7818,
+      "step": 33764
+    },
+    {
+      "epoch": 0.9273296531268024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7570929633338075e-07,
+      "loss": 0.819,
+      "step": 33765
+    },
+    {
+      "epoch": 0.9273571173546455,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7550190069463913e-07,
+      "loss": 0.746,
+      "step": 33766
+    },
+    {
+      "epoch": 0.9273845815824888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7529458199989314e-07,
+      "loss": 0.8123,
+      "step": 33767
+    },
+    {
+      "epoch": 0.9274120458103321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.750873402507825e-07,
+      "loss": 0.9123,
+      "step": 33768
+    },
+    {
+      "epoch": 0.9274395100381753,
+      "grad_norm": 0.0,
+      "learning_rate": 2.748801754489494e-07,
+      "loss": 0.8106,
+      "step": 33769
+    },
+    {
+      "epoch": 0.9274669742660185,
+      "grad_norm": 0.0,
+      "learning_rate": 2.746730875960313e-07,
+      "loss": 0.8854,
+      "step": 33770
+    },
+    {
+      "epoch": 0.9274944384938617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.744660766936669e-07,
+      "loss": 0.8022,
+      "step": 33771
+    },
+    {
+      "epoch": 0.927521902721705,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7425914274349374e-07,
+      "loss": 0.7878,
+      "step": 33772
+    },
+    {
+      "epoch": 0.9275493669495483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.740522857471506e-07,
+      "loss": 0.7238,
+      "step": 33773
+    },
+    {
+      "epoch": 0.9275768311773914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7384550570627167e-07,
+      "loss": 0.7602,
+      "step": 33774
+    },
+    {
+      "epoch": 0.9276042954052347,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7363880262249563e-07,
+      "loss": 0.6573,
+      "step": 33775
+    },
+    {
+      "epoch": 0.9276317596330779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.734321764974579e-07,
+      "loss": 0.78,
+      "step": 33776
+    },
+    {
+      "epoch": 0.9276592238609211,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7322562733279044e-07,
+      "loss": 0.8438,
+      "step": 33777
+    },
+    {
+      "epoch": 0.9276866880887644,
+      "grad_norm": 0.0,
+      "learning_rate": 2.730191551301309e-07,
+      "loss": 0.8062,
+      "step": 33778
+    },
+    {
+      "epoch": 0.9277141523166076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7281275989111124e-07,
+      "loss": 0.7894,
+      "step": 33779
+    },
+    {
+      "epoch": 0.9277416165444509,
+      "grad_norm": 0.0,
+      "learning_rate": 2.726064416173657e-07,
+      "loss": 0.7977,
+      "step": 33780
+    },
+    {
+      "epoch": 0.9277690807722941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7240020031052526e-07,
+      "loss": 0.8314,
+      "step": 33781
+    },
+    {
+      "epoch": 0.9277965450001373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.721940359722242e-07,
+      "loss": 0.7523,
+      "step": 33782
+    },
+    {
+      "epoch": 0.9278240092279806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7198794860408993e-07,
+      "loss": 0.8127,
+      "step": 33783
+    },
+    {
+      "epoch": 0.9278514734558237,
+      "grad_norm": 0.0,
+      "learning_rate": 2.717819382077558e-07,
+      "loss": 0.8092,
+      "step": 33784
+    },
+    {
+      "epoch": 0.927878937683667,
+      "grad_norm": 0.0,
+      "learning_rate": 2.715760047848515e-07,
+      "loss": 0.8127,
+      "step": 33785
+    },
+    {
+      "epoch": 0.9279064019115103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.713701483370057e-07,
+      "loss": 0.7878,
+      "step": 33786
+    },
+    {
+      "epoch": 0.9279338661393535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.711643688658483e-07,
+      "loss": 0.8057,
+      "step": 33787
+    },
+    {
+      "epoch": 0.9279613303671967,
+      "grad_norm": 0.0,
+      "learning_rate": 2.709586663730079e-07,
+      "loss": 0.8396,
+      "step": 33788
+    },
+    {
+      "epoch": 0.9279887945950399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.707530408601111e-07,
+      "loss": 0.7869,
+      "step": 33789
+    },
+    {
+      "epoch": 0.9280162588228832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.705474923287843e-07,
+      "loss": 0.9095,
+      "step": 33790
+    },
+    {
+      "epoch": 0.9280437230507265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.703420207806551e-07,
+      "loss": 0.8097,
+      "step": 33791
+    },
+    {
+      "epoch": 0.9280711872785696,
+      "grad_norm": 0.0,
+      "learning_rate": 2.701366262173477e-07,
+      "loss": 0.8567,
+      "step": 33792
+    },
+    {
+      "epoch": 0.9280986515064129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.699313086404887e-07,
+      "loss": 0.7947,
+      "step": 33793
+    },
+    {
+      "epoch": 0.9281261157342562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.697260680517033e-07,
+      "loss": 0.7774,
+      "step": 33794
+    },
+    {
+      "epoch": 0.9281535799620994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.695209044526126e-07,
+      "loss": 0.7916,
+      "step": 33795
+    },
+    {
+      "epoch": 0.9281810441899426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6931581784484294e-07,
+      "loss": 0.766,
+      "step": 33796
+    },
+    {
+      "epoch": 0.9282085084177858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6911080823001643e-07,
+      "loss": 0.8711,
+      "step": 33797
+    },
+    {
+      "epoch": 0.9282359726456291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.689058756097529e-07,
+      "loss": 0.905,
+      "step": 33798
+    },
+    {
+      "epoch": 0.9282634368734723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.687010199856765e-07,
+      "loss": 0.771,
+      "step": 33799
+    },
+    {
+      "epoch": 0.9282909011013155,
+      "grad_norm": 0.0,
+      "learning_rate": 2.684962413594072e-07,
+      "loss": 0.7411,
+      "step": 33800
+    },
+    {
+      "epoch": 0.9283183653291588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.682915397325647e-07,
+      "loss": 0.8193,
+      "step": 33801
+    },
+    {
+      "epoch": 0.928345829557002,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6808691510677e-07,
+      "loss": 0.7642,
+      "step": 33802
+    },
+    {
+      "epoch": 0.9283732937848452,
+      "grad_norm": 0.0,
+      "learning_rate": 2.678823674836406e-07,
+      "loss": 0.8608,
+      "step": 33803
+    },
+    {
+      "epoch": 0.9284007580126885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.676778968647975e-07,
+      "loss": 0.863,
+      "step": 33804
+    },
+    {
+      "epoch": 0.9284282222405317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.674735032518549e-07,
+      "loss": 0.8068,
+      "step": 33805
+    },
+    {
+      "epoch": 0.928455686468375,
+      "grad_norm": 0.0,
+      "learning_rate": 2.672691866464339e-07,
+      "loss": 0.8565,
+      "step": 33806
+    },
+    {
+      "epoch": 0.9284831506962182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6706494705014743e-07,
+      "loss": 0.8208,
+      "step": 33807
+    },
+    {
+      "epoch": 0.9285106149240614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6686078446461317e-07,
+      "loss": 0.8448,
+      "step": 33808
+    },
+    {
+      "epoch": 0.9285380791519047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.666566988914465e-07,
+      "loss": 0.8332,
+      "step": 33809
+    },
+    {
+      "epoch": 0.9285655433797478,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6645269033226283e-07,
+      "loss": 0.6982,
+      "step": 33810
+    },
+    {
+      "epoch": 0.9285930076075911,
+      "grad_norm": 0.0,
+      "learning_rate": 2.662487587886764e-07,
+      "loss": 0.8268,
+      "step": 33811
+    },
+    {
+      "epoch": 0.9286204718354344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6604490426230034e-07,
+      "loss": 0.7763,
+      "step": 33812
+    },
+    {
+      "epoch": 0.9286479360632776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6584112675474893e-07,
+      "loss": 0.8107,
+      "step": 33813
+    },
+    {
+      "epoch": 0.9286754002911208,
+      "grad_norm": 0.0,
+      "learning_rate": 2.656374262676309e-07,
+      "loss": 0.8402,
+      "step": 33814
+    },
+    {
+      "epoch": 0.928702864518964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.654338028025605e-07,
+      "loss": 0.8783,
+      "step": 33815
+    },
+    {
+      "epoch": 0.9287303287468073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.652302563611486e-07,
+      "loss": 0.8015,
+      "step": 33816
+    },
+    {
+      "epoch": 0.9287577929746506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6502678694500626e-07,
+      "loss": 0.8356,
+      "step": 33817
+    },
+    {
+      "epoch": 0.9287852572024937,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6482339455574323e-07,
+      "loss": 0.7358,
+      "step": 33818
+    },
+    {
+      "epoch": 0.928812721430337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6462007919496826e-07,
+      "loss": 0.8481,
+      "step": 33819
+    },
+    {
+      "epoch": 0.9288401856581803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6441684086429e-07,
+      "loss": 0.7679,
+      "step": 33820
+    },
+    {
+      "epoch": 0.9288676498860234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6421367956531827e-07,
+      "loss": 0.8543,
+      "step": 33821
+    },
+    {
+      "epoch": 0.9288951141138667,
+      "grad_norm": 0.0,
+      "learning_rate": 2.640105952996574e-07,
+      "loss": 0.7935,
+      "step": 33822
+    },
+    {
+      "epoch": 0.9289225783417099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.638075880689184e-07,
+      "loss": 0.7724,
+      "step": 33823
+    },
+    {
+      "epoch": 0.9289500425695532,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6360465787470314e-07,
+      "loss": 0.7719,
+      "step": 33824
+    },
+    {
+      "epoch": 0.9289775067973964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6340180471861933e-07,
+      "loss": 0.779,
+      "step": 33825
+    },
+    {
+      "epoch": 0.9290049710252396,
+      "grad_norm": 0.0,
+      "learning_rate": 2.631990286022723e-07,
+      "loss": 0.7079,
+      "step": 33826
+    },
+    {
+      "epoch": 0.9290324352530829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.629963295272664e-07,
+      "loss": 0.7004,
+      "step": 33827
+    },
+    {
+      "epoch": 0.929059899480926,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6279370749520583e-07,
+      "loss": 0.753,
+      "step": 33828
+    },
+    {
+      "epoch": 0.9290873637087693,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6259116250769265e-07,
+      "loss": 0.7348,
+      "step": 33829
+    },
+    {
+      "epoch": 0.9291148279366126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.623886945663312e-07,
+      "loss": 0.8659,
+      "step": 33830
+    },
+    {
+      "epoch": 0.9291422921644558,
+      "grad_norm": 0.0,
+      "learning_rate": 2.621863036727212e-07,
+      "loss": 0.8553,
+      "step": 33831
+    },
+    {
+      "epoch": 0.929169756392299,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6198398982846486e-07,
+      "loss": 0.8463,
+      "step": 33832
+    },
+    {
+      "epoch": 0.9291972206201423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6178175303516406e-07,
+      "loss": 0.7719,
+      "step": 33833
+    },
+    {
+      "epoch": 0.9292246848479855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6157959329441764e-07,
+      "loss": 0.8093,
+      "step": 33834
+    },
+    {
+      "epoch": 0.9292521490758288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6137751060782756e-07,
+      "loss": 0.7576,
+      "step": 33835
+    },
+    {
+      "epoch": 0.9292796133036719,
+      "grad_norm": 0.0,
+      "learning_rate": 2.611755049769893e-07,
+      "loss": 0.842,
+      "step": 33836
+    },
+    {
+      "epoch": 0.9293070775315152,
+      "grad_norm": 0.0,
+      "learning_rate": 2.609735764035037e-07,
+      "loss": 0.8049,
+      "step": 33837
+    },
+    {
+      "epoch": 0.9293345417593585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.607717248889674e-07,
+      "loss": 0.8325,
+      "step": 33838
+    },
+    {
+      "epoch": 0.9293620059872016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6056995043497793e-07,
+      "loss": 0.8356,
+      "step": 33839
+    },
+    {
+      "epoch": 0.9293894702150449,
+      "grad_norm": 0.0,
+      "learning_rate": 2.603682530431317e-07,
+      "loss": 0.7363,
+      "step": 33840
+    },
+    {
+      "epoch": 0.9294169344428881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.601666327150243e-07,
+      "loss": 0.8474,
+      "step": 33841
+    },
+    {
+      "epoch": 0.9294443986707314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5996508945225206e-07,
+      "loss": 0.7967,
+      "step": 33842
+    },
+    {
+      "epoch": 0.9294718628985746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5976362325640826e-07,
+      "loss": 0.7666,
+      "step": 33843
+    },
+    {
+      "epoch": 0.9294993271264178,
+      "grad_norm": 0.0,
+      "learning_rate": 2.595622341290871e-07,
+      "loss": 0.7581,
+      "step": 33844
+    },
+    {
+      "epoch": 0.9295267913542611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.59360922071884e-07,
+      "loss": 0.8039,
+      "step": 33845
+    },
+    {
+      "epoch": 0.9295542555821044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.591596870863899e-07,
+      "loss": 0.8423,
+      "step": 33846
+    },
+    {
+      "epoch": 0.9295817198099475,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5895852917419804e-07,
+      "loss": 0.6892,
+      "step": 33847
+    },
+    {
+      "epoch": 0.9296091840377908,
+      "grad_norm": 0.0,
+      "learning_rate": 2.587574483368993e-07,
+      "loss": 0.8328,
+      "step": 33848
+    },
+    {
+      "epoch": 0.929636648265634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5855644457608466e-07,
+      "loss": 0.7474,
+      "step": 33849
+    },
+    {
+      "epoch": 0.9296641124934772,
+      "grad_norm": 0.0,
+      "learning_rate": 2.58355517893345e-07,
+      "loss": 0.8028,
+      "step": 33850
+    },
+    {
+      "epoch": 0.9296915767213205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5815466829026915e-07,
+      "loss": 0.8679,
+      "step": 33851
+    },
+    {
+      "epoch": 0.9297190409491637,
+      "grad_norm": 0.0,
+      "learning_rate": 2.579538957684491e-07,
+      "loss": 0.741,
+      "step": 33852
+    },
+    {
+      "epoch": 0.929746505177007,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5775320032947136e-07,
+      "loss": 0.8527,
+      "step": 33853
+    },
+    {
+      "epoch": 0.9297739694048501,
+      "grad_norm": 0.0,
+      "learning_rate": 2.575525819749236e-07,
+      "loss": 0.8859,
+      "step": 33854
+    },
+    {
+      "epoch": 0.9298014336326934,
+      "grad_norm": 0.0,
+      "learning_rate": 2.573520407063923e-07,
+      "loss": 0.8055,
+      "step": 33855
+    },
+    {
+      "epoch": 0.9298288978605367,
+      "grad_norm": 0.0,
+      "learning_rate": 2.571515765254673e-07,
+      "loss": 0.7858,
+      "step": 33856
+    },
+    {
+      "epoch": 0.9298563620883799,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5695118943373175e-07,
+      "loss": 0.8749,
+      "step": 33857
+    },
+    {
+      "epoch": 0.9298838263162231,
+      "grad_norm": 0.0,
+      "learning_rate": 2.567508794327733e-07,
+      "loss": 0.7927,
+      "step": 33858
+    },
+    {
+      "epoch": 0.9299112905440664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5655064652417625e-07,
+      "loss": 0.8668,
+      "step": 33859
+    },
+    {
+      "epoch": 0.9299387547719096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.563504907095238e-07,
+      "loss": 0.7999,
+      "step": 33860
+    },
+    {
+      "epoch": 0.9299662189997528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5615041199040236e-07,
+      "loss": 0.7629,
+      "step": 33861
+    },
+    {
+      "epoch": 0.929993683227596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5595041036839185e-07,
+      "loss": 0.7839,
+      "step": 33862
+    },
+    {
+      "epoch": 0.9300211474554393,
+      "grad_norm": 0.0,
+      "learning_rate": 2.557504858450766e-07,
+      "loss": 0.7787,
+      "step": 33863
+    },
+    {
+      "epoch": 0.9300486116832826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.555506384220374e-07,
+      "loss": 0.722,
+      "step": 33864
+    },
+    {
+      "epoch": 0.9300760759111257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5535086810085763e-07,
+      "loss": 0.7555,
+      "step": 33865
+    },
+    {
+      "epoch": 0.930103540138969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.551511748831148e-07,
+      "loss": 0.8016,
+      "step": 33866
+    },
+    {
+      "epoch": 0.9301310043668122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.54951558770391e-07,
+      "loss": 0.8271,
+      "step": 33867
+    },
+    {
+      "epoch": 0.9301584685946555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.547520197642661e-07,
+      "loss": 0.9069,
+      "step": 33868
+    },
+    {
+      "epoch": 0.9301859328224987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5455255786631885e-07,
+      "loss": 0.7872,
+      "step": 33869
+    },
+    {
+      "epoch": 0.9302133970503419,
+      "grad_norm": 0.0,
+      "learning_rate": 2.543531730781257e-07,
+      "loss": 0.7945,
+      "step": 33870
+    },
+    {
+      "epoch": 0.9302408612781852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.541538654012654e-07,
+      "loss": 0.7599,
+      "step": 33871
+    },
+    {
+      "epoch": 0.9302683255060284,
+      "grad_norm": 0.0,
+      "learning_rate": 2.539546348373145e-07,
+      "loss": 0.7602,
+      "step": 33872
+    },
+    {
+      "epoch": 0.9302957897338716,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5375548138785065e-07,
+      "loss": 0.8243,
+      "step": 33873
+    },
+    {
+      "epoch": 0.9303232539617149,
+      "grad_norm": 0.0,
+      "learning_rate": 2.535564050544481e-07,
+      "loss": 0.7424,
+      "step": 33874
+    },
+    {
+      "epoch": 0.9303507181895581,
+      "grad_norm": 0.0,
+      "learning_rate": 2.533574058386823e-07,
+      "loss": 0.8078,
+      "step": 33875
+    },
+    {
+      "epoch": 0.9303781824174013,
+      "grad_norm": 0.0,
+      "learning_rate": 2.531584837421297e-07,
+      "loss": 0.8533,
+      "step": 33876
+    },
+    {
+      "epoch": 0.9304056466452446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5295963876636245e-07,
+      "loss": 0.7918,
+      "step": 33877
+    },
+    {
+      "epoch": 0.9304331108730878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5276087091295363e-07,
+      "loss": 0.8579,
+      "step": 33878
+    },
+    {
+      "epoch": 0.930460575100931,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5256218018347655e-07,
+      "loss": 0.7892,
+      "step": 33879
+    },
+    {
+      "epoch": 0.9304880393287742,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5236356657950213e-07,
+      "loss": 0.8601,
+      "step": 33880
+    },
+    {
+      "epoch": 0.9305155035566175,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5216503010260463e-07,
+      "loss": 0.8178,
+      "step": 33881
+    },
+    {
+      "epoch": 0.9305429677844608,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5196657075435393e-07,
+      "loss": 0.8028,
+      "step": 33882
+    },
+    {
+      "epoch": 0.9305704320123039,
+      "grad_norm": 0.0,
+      "learning_rate": 2.517681885363177e-07,
+      "loss": 0.8879,
+      "step": 33883
+    },
+    {
+      "epoch": 0.9305978962401472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5156988345006907e-07,
+      "loss": 0.7638,
+      "step": 33884
+    },
+    {
+      "epoch": 0.9306253604679905,
+      "grad_norm": 0.0,
+      "learning_rate": 2.513716554971757e-07,
+      "loss": 0.8654,
+      "step": 33885
+    },
+    {
+      "epoch": 0.9306528246958337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.511735046792052e-07,
+      "loss": 0.894,
+      "step": 33886
+    },
+    {
+      "epoch": 0.9306802889236769,
+      "grad_norm": 0.0,
+      "learning_rate": 2.509754309977264e-07,
+      "loss": 0.8377,
+      "step": 33887
+    },
+    {
+      "epoch": 0.9307077531515201,
+      "grad_norm": 0.0,
+      "learning_rate": 2.507774344543079e-07,
+      "loss": 0.7884,
+      "step": 33888
+    },
+    {
+      "epoch": 0.9307352173793634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.50579515050513e-07,
+      "loss": 0.7855,
+      "step": 33889
+    },
+    {
+      "epoch": 0.9307626816072067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.503816727879105e-07,
+      "loss": 0.7935,
+      "step": 33890
+    },
+    {
+      "epoch": 0.9307901458350498,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5018390766806343e-07,
+      "loss": 0.8594,
+      "step": 33891
+    },
+    {
+      "epoch": 0.9308176100628931,
+      "grad_norm": 0.0,
+      "learning_rate": 2.499862196925407e-07,
+      "loss": 0.777,
+      "step": 33892
+    },
+    {
+      "epoch": 0.9308450742907363,
+      "grad_norm": 0.0,
+      "learning_rate": 2.497886088629009e-07,
+      "loss": 0.8589,
+      "step": 33893
+    },
+    {
+      "epoch": 0.9308725385185795,
+      "grad_norm": 0.0,
+      "learning_rate": 2.495910751807129e-07,
+      "loss": 0.7446,
+      "step": 33894
+    },
+    {
+      "epoch": 0.9309000027464228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4939361864753655e-07,
+      "loss": 0.7514,
+      "step": 33895
+    },
+    {
+      "epoch": 0.930927466974266,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4919623926493385e-07,
+      "loss": 0.8284,
+      "step": 33896
+    },
+    {
+      "epoch": 0.9309549312021093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.489989370344681e-07,
+      "loss": 0.6541,
+      "step": 33897
+    },
+    {
+      "epoch": 0.9309823954299525,
+      "grad_norm": 0.0,
+      "learning_rate": 2.488017119577002e-07,
+      "loss": 0.8706,
+      "step": 33898
+    },
+    {
+      "epoch": 0.9310098596577957,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4860456403619004e-07,
+      "loss": 0.8625,
+      "step": 33899
+    },
+    {
+      "epoch": 0.931037323885639,
+      "grad_norm": 0.0,
+      "learning_rate": 2.484074932714986e-07,
+      "loss": 0.8165,
+      "step": 33900
+    },
+    {
+      "epoch": 0.9310647881134821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4821049966518464e-07,
+      "loss": 0.8392,
+      "step": 33901
+    },
+    {
+      "epoch": 0.9310922523413254,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4801358321880575e-07,
+      "loss": 0.7396,
+      "step": 33902
+    },
+    {
+      "epoch": 0.9311197165691687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4781674393392074e-07,
+      "loss": 0.6616,
+      "step": 33903
+    },
+    {
+      "epoch": 0.9311471807970119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4761998181208837e-07,
+      "loss": 0.8823,
+      "step": 33904
+    },
+    {
+      "epoch": 0.9311746450248551,
+      "grad_norm": 0.0,
+      "learning_rate": 2.474232968548629e-07,
+      "loss": 0.8476,
+      "step": 33905
+    },
+    {
+      "epoch": 0.9312021092526983,
+      "grad_norm": 0.0,
+      "learning_rate": 2.472266890638042e-07,
+      "loss": 0.7761,
+      "step": 33906
+    },
+    {
+      "epoch": 0.9312295734805416,
+      "grad_norm": 0.0,
+      "learning_rate": 2.470301584404644e-07,
+      "loss": 0.6845,
+      "step": 33907
+    },
+    {
+      "epoch": 0.9312570377083849,
+      "grad_norm": 0.0,
+      "learning_rate": 2.468337049864e-07,
+      "loss": 0.7962,
+      "step": 33908
+    },
+    {
+      "epoch": 0.931284501936228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.466373287031665e-07,
+      "loss": 0.8491,
+      "step": 33909
+    },
+    {
+      "epoch": 0.9313119661640713,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4644102959231474e-07,
+      "loss": 0.8251,
+      "step": 33910
+    },
+    {
+      "epoch": 0.9313394303919146,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4624480765540025e-07,
+      "loss": 0.754,
+      "step": 33911
+    },
+    {
+      "epoch": 0.9313668946197577,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4604866289397623e-07,
+      "loss": 0.7594,
+      "step": 33912
+    },
+    {
+      "epoch": 0.931394358847601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.458525953095925e-07,
+      "loss": 0.722,
+      "step": 33913
+    },
+    {
+      "epoch": 0.9314218230754442,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4565660490380116e-07,
+      "loss": 0.8676,
+      "step": 33914
+    },
+    {
+      "epoch": 0.9314492873032875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454606916781543e-07,
+      "loss": 0.9249,
+      "step": 33915
+    },
+    {
+      "epoch": 0.9314767515311307,
+      "grad_norm": 0.0,
+      "learning_rate": 2.452648556342008e-07,
+      "loss": 0.7756,
+      "step": 33916
+    },
+    {
+      "epoch": 0.9315042157589739,
+      "grad_norm": 0.0,
+      "learning_rate": 2.450690967734892e-07,
+      "loss": 0.7933,
+      "step": 33917
+    },
+    {
+      "epoch": 0.9315316799868172,
+      "grad_norm": 0.0,
+      "learning_rate": 2.448734150975718e-07,
+      "loss": 0.8045,
+      "step": 33918
+    },
+    {
+      "epoch": 0.9315591442146604,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4467781060799277e-07,
+      "loss": 0.798,
+      "step": 33919
+    },
+    {
+      "epoch": 0.9315866084425036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.444822833063021e-07,
+      "loss": 0.8637,
+      "step": 33920
+    },
+    {
+      "epoch": 0.9316140726703469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.442868331940462e-07,
+      "loss": 0.7471,
+      "step": 33921
+    },
+    {
+      "epoch": 0.9316415368981901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4409146027277284e-07,
+      "loss": 0.8,
+      "step": 33922
+    },
+    {
+      "epoch": 0.9316690011260333,
+      "grad_norm": 0.0,
+      "learning_rate": 2.438961645440252e-07,
+      "loss": 0.7465,
+      "step": 33923
+    },
+    {
+      "epoch": 0.9316964653538766,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4370094600935313e-07,
+      "loss": 0.8127,
+      "step": 33924
+    },
+    {
+      "epoch": 0.9317239295817198,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4350580467029763e-07,
+      "loss": 0.7982,
+      "step": 33925
+    },
+    {
+      "epoch": 0.9317513938095631,
+      "grad_norm": 0.0,
+      "learning_rate": 2.433107405284019e-07,
+      "loss": 0.7674,
+      "step": 33926
+    },
+    {
+      "epoch": 0.9317788580374062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.431157535852113e-07,
+      "loss": 0.7838,
+      "step": 33927
+    },
+    {
+      "epoch": 0.9318063222652495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4292084384226923e-07,
+      "loss": 0.8991,
+      "step": 33928
+    },
+    {
+      "epoch": 0.9318337864930928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.427260113011154e-07,
+      "loss": 0.8672,
+      "step": 33929
+    },
+    {
+      "epoch": 0.931861250720936,
+      "grad_norm": 0.0,
+      "learning_rate": 2.425312559632953e-07,
+      "loss": 0.7726,
+      "step": 33930
+    },
+    {
+      "epoch": 0.9318887149487792,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4233657783034657e-07,
+      "loss": 0.8459,
+      "step": 33931
+    },
+    {
+      "epoch": 0.9319161791766224,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4214197690380913e-07,
+      "loss": 0.7913,
+      "step": 33932
+    },
+    {
+      "epoch": 0.9319436434044657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.419474531852262e-07,
+      "loss": 0.7785,
+      "step": 33933
+    },
+    {
+      "epoch": 0.931971107632309,
+      "grad_norm": 0.0,
+      "learning_rate": 2.417530066761342e-07,
+      "loss": 0.8204,
+      "step": 33934
+    },
+    {
+      "epoch": 0.9319985718601521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.415586373780721e-07,
+      "loss": 0.6133,
+      "step": 33935
+    },
+    {
+      "epoch": 0.9320260360879954,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4136434529257955e-07,
+      "loss": 0.8157,
+      "step": 33936
+    },
+    {
+      "epoch": 0.9320535003158387,
+      "grad_norm": 0.0,
+      "learning_rate": 2.411701304211911e-07,
+      "loss": 0.8195,
+      "step": 33937
+    },
+    {
+      "epoch": 0.9320809645436818,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4097599276544424e-07,
+      "loss": 0.8214,
+      "step": 33938
+    },
+    {
+      "epoch": 0.9321084287715251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.407819323268767e-07,
+      "loss": 0.7431,
+      "step": 33939
+    },
+    {
+      "epoch": 0.9321358929993683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4058794910702286e-07,
+      "loss": 0.7059,
+      "step": 33940
+    },
+    {
+      "epoch": 0.9321633572272116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.40394043107417e-07,
+      "loss": 0.8461,
+      "step": 33941
+    },
+    {
+      "epoch": 0.9321908214550548,
+      "grad_norm": 0.0,
+      "learning_rate": 2.402002143295945e-07,
+      "loss": 0.7401,
+      "step": 33942
+    },
+    {
+      "epoch": 0.932218285682898,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4000646277508864e-07,
+      "loss": 0.7913,
+      "step": 33943
+    },
+    {
+      "epoch": 0.9322457499107413,
+      "grad_norm": 0.0,
+      "learning_rate": 2.398127884454315e-07,
+      "loss": 0.7618,
+      "step": 33944
+    },
+    {
+      "epoch": 0.9322732141385844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3961919134215637e-07,
+      "loss": 0.7425,
+      "step": 33945
+    },
+    {
+      "epoch": 0.9323006783664277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.394256714667953e-07,
+      "loss": 0.8194,
+      "step": 33946
+    },
+    {
+      "epoch": 0.932328142594271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3923222882088037e-07,
+      "loss": 0.8832,
+      "step": 33947
+    },
+    {
+      "epoch": 0.9323556068221142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.390388634059393e-07,
+      "loss": 0.8143,
+      "step": 33948
+    },
+    {
+      "epoch": 0.9323830710499574,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3884557522350525e-07,
+      "loss": 0.8894,
+      "step": 33949
+    },
+    {
+      "epoch": 0.9324105352778007,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3865236427510485e-07,
+      "loss": 0.7518,
+      "step": 33950
+    },
+    {
+      "epoch": 0.9324379995056439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3845923056226793e-07,
+      "loss": 0.739,
+      "step": 33951
+    },
+    {
+      "epoch": 0.9324654637334872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.382661740865233e-07,
+      "loss": 0.8387,
+      "step": 33952
+    },
+    {
+      "epoch": 0.9324929279613303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.380731948493986e-07,
+      "loss": 0.7112,
+      "step": 33953
+    },
+    {
+      "epoch": 0.9325203921891736,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3788029285241932e-07,
+      "loss": 0.8209,
+      "step": 33954
+    },
+    {
+      "epoch": 0.9325478564170169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3768746809711307e-07,
+      "loss": 0.7192,
+      "step": 33955
+    },
+    {
+      "epoch": 0.93257532064486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3749472058500533e-07,
+      "loss": 0.8167,
+      "step": 33956
+    },
+    {
+      "epoch": 0.9326027848727033,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3730205031762154e-07,
+      "loss": 0.8288,
+      "step": 33957
+    },
+    {
+      "epoch": 0.9326302491005466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3710945729648493e-07,
+      "loss": 0.8447,
+      "step": 33958
+    },
+    {
+      "epoch": 0.9326577133283898,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3691694152311985e-07,
+      "loss": 0.7901,
+      "step": 33959
+    },
+    {
+      "epoch": 0.932685177556233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3672450299905058e-07,
+      "loss": 0.8317,
+      "step": 33960
+    },
+    {
+      "epoch": 0.9327126417840762,
+      "grad_norm": 0.0,
+      "learning_rate": 2.365321417257971e-07,
+      "loss": 0.8045,
+      "step": 33961
+    },
+    {
+      "epoch": 0.9327401060119195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3633985770488477e-07,
+      "loss": 0.8173,
+      "step": 33962
+    },
+    {
+      "epoch": 0.9327675702397628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3614765093783353e-07,
+      "loss": 0.8556,
+      "step": 33963
+    },
+    {
+      "epoch": 0.9327950344676059,
+      "grad_norm": 0.0,
+      "learning_rate": 2.359555214261644e-07,
+      "loss": 0.7641,
+      "step": 33964
+    },
+    {
+      "epoch": 0.9328224986954492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3576346917139725e-07,
+      "loss": 0.7718,
+      "step": 33965
+    },
+    {
+      "epoch": 0.9328499629232924,
+      "grad_norm": 0.0,
+      "learning_rate": 2.35571494175052e-07,
+      "loss": 0.8123,
+      "step": 33966
+    },
+    {
+      "epoch": 0.9328774271511356,
+      "grad_norm": 0.0,
+      "learning_rate": 2.353795964386474e-07,
+      "loss": 0.7926,
+      "step": 33967
+    },
+    {
+      "epoch": 0.9329048913789789,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3518777596370113e-07,
+      "loss": 0.8303,
+      "step": 33968
+    },
+    {
+      "epoch": 0.9329323556068221,
+      "grad_norm": 0.0,
+      "learning_rate": 2.34996032751732e-07,
+      "loss": 0.7672,
+      "step": 33969
+    },
+    {
+      "epoch": 0.9329598198346654,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3480436680425656e-07,
+      "loss": 0.7849,
+      "step": 33970
+    },
+    {
+      "epoch": 0.9329872840625086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.346127781227925e-07,
+      "loss": 0.8134,
+      "step": 33971
+    },
+    {
+      "epoch": 0.9330147482903518,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3442126670885524e-07,
+      "loss": 0.753,
+      "step": 33972
+    },
+    {
+      "epoch": 0.9330422125181951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3422983256395914e-07,
+      "loss": 0.6001,
+      "step": 33973
+    },
+    {
+      "epoch": 0.9330696767460382,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3403847568961968e-07,
+      "loss": 0.8023,
+      "step": 33974
+    },
+    {
+      "epoch": 0.9330971409738815,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3384719608735007e-07,
+      "loss": 0.8381,
+      "step": 33975
+    },
+    {
+      "epoch": 0.9331246052017248,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3365599375866466e-07,
+      "loss": 0.8269,
+      "step": 33976
+    },
+    {
+      "epoch": 0.933152069429568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.334648687050767e-07,
+      "loss": 0.6993,
+      "step": 33977
+    },
+    {
+      "epoch": 0.9331795336574112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3327382092809713e-07,
+      "loss": 0.7915,
+      "step": 33978
+    },
+    {
+      "epoch": 0.9332069978852544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3308285042923927e-07,
+      "loss": 0.7189,
+      "step": 33979
+    },
+    {
+      "epoch": 0.9332344621130977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.328919572100119e-07,
+      "loss": 0.8867,
+      "step": 33980
+    },
+    {
+      "epoch": 0.933261926340941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3270114127192822e-07,
+      "loss": 0.7198,
+      "step": 33981
+    },
+    {
+      "epoch": 0.9332893905687841,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3251040261649592e-07,
+      "loss": 0.7003,
+      "step": 33982
+    },
+    {
+      "epoch": 0.9333168547966274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3231974124522606e-07,
+      "loss": 0.8745,
+      "step": 33983
+    },
+    {
+      "epoch": 0.9333443190244707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3212915715962515e-07,
+      "loss": 0.8014,
+      "step": 33984
+    },
+    {
+      "epoch": 0.9333717832523138,
+      "grad_norm": 0.0,
+      "learning_rate": 2.319386503612009e-07,
+      "loss": 0.7604,
+      "step": 33985
+    },
+    {
+      "epoch": 0.9333992474801571,
+      "grad_norm": 0.0,
+      "learning_rate": 2.317482208514632e-07,
+      "loss": 0.7572,
+      "step": 33986
+    },
+    {
+      "epoch": 0.9334267117080003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3155786863191642e-07,
+      "loss": 0.724,
+      "step": 33987
+    },
+    {
+      "epoch": 0.9334541759358436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3136759370406937e-07,
+      "loss": 0.8318,
+      "step": 33988
+    },
+    {
+      "epoch": 0.9334816401636868,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3117739606942413e-07,
+      "loss": 0.8524,
+      "step": 33989
+    },
+    {
+      "epoch": 0.93350910439153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3098727572948954e-07,
+      "loss": 0.8505,
+      "step": 33990
+    },
+    {
+      "epoch": 0.9335365686193733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.307972326857666e-07,
+      "loss": 0.8551,
+      "step": 33991
+    },
+    {
+      "epoch": 0.9335640328472165,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3060726693975966e-07,
+      "loss": 0.7876,
+      "step": 33992
+    },
+    {
+      "epoch": 0.9335914970750597,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3041737849297197e-07,
+      "loss": 0.9113,
+      "step": 33993
+    },
+    {
+      "epoch": 0.933618961302903,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3022756734690677e-07,
+      "loss": 0.8784,
+      "step": 33994
+    },
+    {
+      "epoch": 0.9336464255307462,
+      "grad_norm": 0.0,
+      "learning_rate": 2.300378335030662e-07,
+      "loss": 0.7713,
+      "step": 33995
+    },
+    {
+      "epoch": 0.9336738897585894,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2984817696295013e-07,
+      "loss": 0.9405,
+      "step": 33996
+    },
+    {
+      "epoch": 0.9337013539864327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2965859772806188e-07,
+      "loss": 0.802,
+      "step": 33997
+    },
+    {
+      "epoch": 0.9337288182142759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2946909579989684e-07,
+      "loss": 0.8007,
+      "step": 33998
+    },
+    {
+      "epoch": 0.9337562824421192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2927967117995831e-07,
+      "loss": 0.7992,
+      "step": 33999
+    },
+    {
+      "epoch": 0.9337837466699623,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2909032386974283e-07,
+      "loss": 0.8167,
+      "step": 34000
+    },
+    {
+      "epoch": 0.9338112108978056,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2890105387075146e-07,
+      "loss": 0.8115,
+      "step": 34001
+    },
+    {
+      "epoch": 0.9338386751256489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2871186118447852e-07,
+      "loss": 0.8688,
+      "step": 34002
+    },
+    {
+      "epoch": 0.933866139353492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2852274581242174e-07,
+      "loss": 0.6717,
+      "step": 34003
+    },
+    {
+      "epoch": 0.9338936035813353,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2833370775607877e-07,
+      "loss": 0.7625,
+      "step": 34004
+    },
+    {
+      "epoch": 0.9339210678091785,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2814474701694512e-07,
+      "loss": 0.9221,
+      "step": 34005
+    },
+    {
+      "epoch": 0.9339485320370218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2795586359651622e-07,
+      "loss": 0.7867,
+      "step": 34006
+    },
+    {
+      "epoch": 0.933975996264865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2776705749628536e-07,
+      "loss": 0.85,
+      "step": 34007
+    },
+    {
+      "epoch": 0.9340034604927082,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2757832871774576e-07,
+      "loss": 0.7863,
+      "step": 34008
+    },
+    {
+      "epoch": 0.9340309247205515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.273896772623918e-07,
+      "loss": 0.8844,
+      "step": 34009
+    },
+    {
+      "epoch": 0.9340583889483948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2720110313171672e-07,
+      "loss": 0.7453,
+      "step": 34010
+    },
+    {
+      "epoch": 0.9340858531762379,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2701260632721266e-07,
+      "loss": 0.7538,
+      "step": 34011
+    },
+    {
+      "epoch": 0.9341133174040812,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2682418685037066e-07,
+      "loss": 0.7094,
+      "step": 34012
+    },
+    {
+      "epoch": 0.9341407816319244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2663584470268284e-07,
+      "loss": 0.8787,
+      "step": 34013
+    },
+    {
+      "epoch": 0.9341682458597677,
+      "grad_norm": 0.0,
+      "learning_rate": 2.264475798856358e-07,
+      "loss": 0.77,
+      "step": 34014
+    },
+    {
+      "epoch": 0.9341957100876109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.262593924007217e-07,
+      "loss": 0.8106,
+      "step": 34015
+    },
+    {
+      "epoch": 0.9342231743154541,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2607128224942932e-07,
+      "loss": 0.8192,
+      "step": 34016
+    },
+    {
+      "epoch": 0.9342506385432974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2588324943324636e-07,
+      "loss": 0.809,
+      "step": 34017
+    },
+    {
+      "epoch": 0.9342781027711405,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2569529395366385e-07,
+      "loss": 0.7646,
+      "step": 34018
+    },
+    {
+      "epoch": 0.9343055669989838,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2550741581216394e-07,
+      "loss": 0.7934,
+      "step": 34019
+    },
+    {
+      "epoch": 0.9343330312268271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.253196150102366e-07,
+      "loss": 0.8439,
+      "step": 34020
+    },
+    {
+      "epoch": 0.9343604954546703,
+      "grad_norm": 0.0,
+      "learning_rate": 2.251318915493672e-07,
+      "loss": 0.8306,
+      "step": 34021
+    },
+    {
+      "epoch": 0.9343879596825135,
+      "grad_norm": 0.0,
+      "learning_rate": 2.249442454310391e-07,
+      "loss": 0.8832,
+      "step": 34022
+    },
+    {
+      "epoch": 0.9344154239103568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2475667665673995e-07,
+      "loss": 0.729,
+      "step": 34023
+    },
+    {
+      "epoch": 0.9344428881382,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2456918522795302e-07,
+      "loss": 0.839,
+      "step": 34024
+    },
+    {
+      "epoch": 0.9344703523660433,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2438177114615934e-07,
+      "loss": 0.8177,
+      "step": 34025
+    },
+    {
+      "epoch": 0.9344978165938864,
+      "grad_norm": 0.0,
+      "learning_rate": 2.241944344128455e-07,
+      "loss": 0.799,
+      "step": 34026
+    },
+    {
+      "epoch": 0.9345252808217297,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2400717502949144e-07,
+      "loss": 0.7452,
+      "step": 34027
+    },
+    {
+      "epoch": 0.934552745049573,
+      "grad_norm": 0.0,
+      "learning_rate": 2.238199929975793e-07,
+      "loss": 0.8273,
+      "step": 34028
+    },
+    {
+      "epoch": 0.9345802092774161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.236328883185912e-07,
+      "loss": 0.725,
+      "step": 34029
+    },
+    {
+      "epoch": 0.9346076735052594,
+      "grad_norm": 0.0,
+      "learning_rate": 2.23445860994006e-07,
+      "loss": 0.804,
+      "step": 34030
+    },
+    {
+      "epoch": 0.9346351377331026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2325891102530473e-07,
+      "loss": 0.7772,
+      "step": 34031
+    },
+    {
+      "epoch": 0.9346626019609459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2307203841396507e-07,
+      "loss": 0.7729,
+      "step": 34032
+    },
+    {
+      "epoch": 0.9346900661887891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.22885243161467e-07,
+      "loss": 0.8162,
+      "step": 34033
+    },
+    {
+      "epoch": 0.9347175304166323,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2269852526928816e-07,
+      "loss": 0.7195,
+      "step": 34034
+    },
+    {
+      "epoch": 0.9347449946444756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2251188473890628e-07,
+      "loss": 0.804,
+      "step": 34035
+    },
+    {
+      "epoch": 0.9347724588723189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2232532157179798e-07,
+      "loss": 0.8258,
+      "step": 34036
+    },
+    {
+      "epoch": 0.934799923100162,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2213883576943984e-07,
+      "loss": 0.7753,
+      "step": 34037
+    },
+    {
+      "epoch": 0.9348273873280053,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2195242733330624e-07,
+      "loss": 0.8278,
+      "step": 34038
+    },
+    {
+      "epoch": 0.9348548515558485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2176609626487155e-07,
+      "loss": 0.7098,
+      "step": 34039
+    },
+    {
+      "epoch": 0.9348823157836917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2157984256561238e-07,
+      "loss": 0.7603,
+      "step": 34040
+    },
+    {
+      "epoch": 0.934909780011535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2139366623700086e-07,
+      "loss": 0.7922,
+      "step": 34041
+    },
+    {
+      "epoch": 0.9349372442393782,
+      "grad_norm": 0.0,
+      "learning_rate": 2.212075672805114e-07,
+      "loss": 0.8022,
+      "step": 34042
+    },
+    {
+      "epoch": 0.9349647084672215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2102154569761503e-07,
+      "loss": 0.8246,
+      "step": 34043
+    },
+    {
+      "epoch": 0.9349921726950646,
+      "grad_norm": 0.0,
+      "learning_rate": 2.208356014897839e-07,
+      "loss": 0.7913,
+      "step": 34044
+    },
+    {
+      "epoch": 0.9350196369229079,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2064973465849016e-07,
+      "loss": 0.7634,
+      "step": 34045
+    },
+    {
+      "epoch": 0.9350471011507512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2046394520520265e-07,
+      "loss": 0.7646,
+      "step": 34046
+    },
+    {
+      "epoch": 0.9350745653785943,
+      "grad_norm": 0.0,
+      "learning_rate": 2.202782331313935e-07,
+      "loss": 0.7473,
+      "step": 34047
+    },
+    {
+      "epoch": 0.9351020296064376,
+      "grad_norm": 0.0,
+      "learning_rate": 2.200925984385316e-07,
+      "loss": 0.8245,
+      "step": 34048
+    },
+    {
+      "epoch": 0.9351294938342809,
+      "grad_norm": 0.0,
+      "learning_rate": 2.199070411280857e-07,
+      "loss": 0.8661,
+      "step": 34049
+    },
+    {
+      "epoch": 0.9351569580621241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1972156120152244e-07,
+      "loss": 0.8284,
+      "step": 34050
+    },
+    {
+      "epoch": 0.9351844222899673,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1953615866031063e-07,
+      "loss": 0.8302,
+      "step": 34051
+    },
+    {
+      "epoch": 0.9352118865178105,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1935083350591911e-07,
+      "loss": 0.8877,
+      "step": 34052
+    },
+    {
+      "epoch": 0.9352393507456538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1916558573981117e-07,
+      "loss": 0.725,
+      "step": 34053
+    },
+    {
+      "epoch": 0.9352668149734971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.189804153634545e-07,
+      "loss": 0.8011,
+      "step": 34054
+    },
+    {
+      "epoch": 0.9352942792013402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1879532237831347e-07,
+      "loss": 0.7902,
+      "step": 34055
+    },
+    {
+      "epoch": 0.9353217434291835,
+      "grad_norm": 0.0,
+      "learning_rate": 2.186103067858525e-07,
+      "loss": 0.7335,
+      "step": 34056
+    },
+    {
+      "epoch": 0.9353492076570267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1842536858753483e-07,
+      "loss": 0.7084,
+      "step": 34057
+    },
+    {
+      "epoch": 0.93537667188487,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1824050778482596e-07,
+      "loss": 0.8239,
+      "step": 34058
+    },
+    {
+      "epoch": 0.9354041361127132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1805572437918587e-07,
+      "loss": 0.7749,
+      "step": 34059
+    },
+    {
+      "epoch": 0.9354316003405564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1787101837208002e-07,
+      "loss": 0.8219,
+      "step": 34060
+    },
+    {
+      "epoch": 0.9354590645683997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1768638976496725e-07,
+      "loss": 0.8434,
+      "step": 34061
+    },
+    {
+      "epoch": 0.9354865287962429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1750183855930862e-07,
+      "loss": 0.868,
+      "step": 34062
+    },
+    {
+      "epoch": 0.9355139930240861,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1731736475656518e-07,
+      "loss": 0.8403,
+      "step": 34063
+    },
+    {
+      "epoch": 0.9355414572519294,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1713296835819576e-07,
+      "loss": 0.7959,
+      "step": 34064
+    },
+    {
+      "epoch": 0.9355689214797726,
+      "grad_norm": 0.0,
+      "learning_rate": 2.169486493656603e-07,
+      "loss": 0.7562,
+      "step": 34065
+    },
+    {
+      "epoch": 0.9355963857076158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1676440778041652e-07,
+      "loss": 0.8162,
+      "step": 34066
+    },
+    {
+      "epoch": 0.9356238499354591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1658024360392105e-07,
+      "loss": 0.6579,
+      "step": 34067
+    },
+    {
+      "epoch": 0.9356513141633023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1639615683763383e-07,
+      "loss": 0.7455,
+      "step": 34068
+    },
+    {
+      "epoch": 0.9356787783911455,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1621214748301033e-07,
+      "loss": 0.8274,
+      "step": 34069
+    },
+    {
+      "epoch": 0.9357062426189887,
+      "grad_norm": 0.0,
+      "learning_rate": 2.160282155415061e-07,
+      "loss": 0.8324,
+      "step": 34070
+    },
+    {
+      "epoch": 0.935733706846832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.158443610145766e-07,
+      "loss": 0.7821,
+      "step": 34071
+    },
+    {
+      "epoch": 0.9357611710746753,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1566058390367627e-07,
+      "loss": 0.839,
+      "step": 34072
+    },
+    {
+      "epoch": 0.9357886353025184,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1547688421025837e-07,
+      "loss": 0.8637,
+      "step": 34073
+    },
+    {
+      "epoch": 0.9358160995303617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1529326193577838e-07,
+      "loss": 0.7946,
+      "step": 34074
+    },
+    {
+      "epoch": 0.935843563758205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.151097170816885e-07,
+      "loss": 0.7838,
+      "step": 34075
+    },
+    {
+      "epoch": 0.9358710279860482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1492624964944198e-07,
+      "loss": 0.7967,
+      "step": 34076
+    },
+    {
+      "epoch": 0.9358984922138914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.147428596404877e-07,
+      "loss": 0.8541,
+      "step": 34077
+    },
+    {
+      "epoch": 0.9359259564417346,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1455954705627892e-07,
+      "loss": 0.771,
+      "step": 34078
+    },
+    {
+      "epoch": 0.9359534206695779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1437631189826558e-07,
+      "loss": 0.848,
+      "step": 34079
+    },
+    {
+      "epoch": 0.9359808848974212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1419315416789765e-07,
+      "loss": 0.8578,
+      "step": 34080
+    },
+    {
+      "epoch": 0.9360083491252643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1401007386662288e-07,
+      "loss": 0.8458,
+      "step": 34081
+    },
+    {
+      "epoch": 0.9360358133531076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1382707099589228e-07,
+      "loss": 0.8297,
+      "step": 34082
+    },
+    {
+      "epoch": 0.9360632775809508,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1364414555715252e-07,
+      "loss": 0.7354,
+      "step": 34083
+    },
+    {
+      "epoch": 0.936090741808794,
+      "grad_norm": 0.0,
+      "learning_rate": 2.134612975518513e-07,
+      "loss": 0.8984,
+      "step": 34084
+    },
+    {
+      "epoch": 0.9361182060366373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1327852698143525e-07,
+      "loss": 0.8945,
+      "step": 34085
+    },
+    {
+      "epoch": 0.9361456702644805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.130958338473499e-07,
+      "loss": 0.735,
+      "step": 34086
+    },
+    {
+      "epoch": 0.9361731344923238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1291321815104182e-07,
+      "loss": 0.8321,
+      "step": 34087
+    },
+    {
+      "epoch": 0.936200598720167,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1273067989395545e-07,
+      "loss": 0.7357,
+      "step": 34088
+    },
+    {
+      "epoch": 0.9362280629480102,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1254821907753632e-07,
+      "loss": 0.7912,
+      "step": 34089
+    },
+    {
+      "epoch": 0.9362555271758535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1236583570322543e-07,
+      "loss": 0.7594,
+      "step": 34090
+    },
+    {
+      "epoch": 0.9362829914036966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1218352977246726e-07,
+      "loss": 0.8082,
+      "step": 34091
+    },
+    {
+      "epoch": 0.9363104556315399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1200130128670504e-07,
+      "loss": 0.7444,
+      "step": 34092
+    },
+    {
+      "epoch": 0.9363379198593832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.11819150247381e-07,
+      "loss": 0.8502,
+      "step": 34093
+    },
+    {
+      "epoch": 0.9363653840872264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.116370766559339e-07,
+      "loss": 0.7785,
+      "step": 34094
+    },
+    {
+      "epoch": 0.9363928483150696,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1145508051380715e-07,
+      "loss": 0.8556,
+      "step": 34095
+    },
+    {
+      "epoch": 0.9364203125429128,
+      "grad_norm": 0.0,
+      "learning_rate": 2.112731618224384e-07,
+      "loss": 0.7985,
+      "step": 34096
+    },
+    {
+      "epoch": 0.9364477767707561,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1109132058326876e-07,
+      "loss": 0.7769,
+      "step": 34097
+    },
+    {
+      "epoch": 0.9364752409985994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1090955679773595e-07,
+      "loss": 0.7305,
+      "step": 34098
+    },
+    {
+      "epoch": 0.9365027052264425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1072787046727884e-07,
+      "loss": 0.8269,
+      "step": 34099
+    },
+    {
+      "epoch": 0.9365301694542858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1054626159333512e-07,
+      "loss": 0.8356,
+      "step": 34100
+    },
+    {
+      "epoch": 0.9365576336821291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1036473017734148e-07,
+      "loss": 0.8616,
+      "step": 34101
+    },
+    {
+      "epoch": 0.9365850979099722,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1018327622073453e-07,
+      "loss": 0.8661,
+      "step": 34102
+    },
+    {
+      "epoch": 0.9366125621378155,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1000189972494867e-07,
+      "loss": 0.8148,
+      "step": 34103
+    },
+    {
+      "epoch": 0.9366400263656587,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0982060069142052e-07,
+      "loss": 0.8259,
+      "step": 34104
+    },
+    {
+      "epoch": 0.936667490593502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.096393791215845e-07,
+      "loss": 0.7109,
+      "step": 34105
+    },
+    {
+      "epoch": 0.9366949548213452,
+      "grad_norm": 0.0,
+      "learning_rate": 2.094582350168739e-07,
+      "loss": 0.8418,
+      "step": 34106
+    },
+    {
+      "epoch": 0.9367224190491884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0927716837872204e-07,
+      "loss": 0.7831,
+      "step": 34107
+    },
+    {
+      "epoch": 0.9367498832770317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0909617920856217e-07,
+      "loss": 0.793,
+      "step": 34108
+    },
+    {
+      "epoch": 0.9367773475048748,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0891526750782654e-07,
+      "loss": 0.851,
+      "step": 34109
+    },
+    {
+      "epoch": 0.9368048117327181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0873443327794508e-07,
+      "loss": 0.8161,
+      "step": 34110
+    },
+    {
+      "epoch": 0.9368322759605614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0855367652034885e-07,
+      "loss": 0.7888,
+      "step": 34111
+    },
+    {
+      "epoch": 0.9368597401884046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0837299723647008e-07,
+      "loss": 0.766,
+      "step": 34112
+    },
+    {
+      "epoch": 0.9368872044162478,
+      "grad_norm": 0.0,
+      "learning_rate": 2.081923954277376e-07,
+      "loss": 0.7685,
+      "step": 34113
+    },
+    {
+      "epoch": 0.9369146686440911,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0801187109557806e-07,
+      "loss": 0.7504,
+      "step": 34114
+    },
+    {
+      "epoch": 0.9369421328719343,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0783142424142255e-07,
+      "loss": 0.8777,
+      "step": 34115
+    },
+    {
+      "epoch": 0.9369695970997776,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0765105486669769e-07,
+      "loss": 0.8272,
+      "step": 34116
+    },
+    {
+      "epoch": 0.9369970613276207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0747076297283119e-07,
+      "loss": 0.7501,
+      "step": 34117
+    },
+    {
+      "epoch": 0.937024525555464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0729054856124975e-07,
+      "loss": 0.8356,
+      "step": 34118
+    },
+    {
+      "epoch": 0.9370519897833073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0711041163337885e-07,
+      "loss": 0.9189,
+      "step": 34119
+    },
+    {
+      "epoch": 0.9370794540111504,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0693035219064294e-07,
+      "loss": 0.7672,
+      "step": 34120
+    },
+    {
+      "epoch": 0.9371069182389937,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0675037023446752e-07,
+      "loss": 0.7485,
+      "step": 34121
+    },
+    {
+      "epoch": 0.9371343824668369,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0657046576627592e-07,
+      "loss": 0.7281,
+      "step": 34122
+    },
+    {
+      "epoch": 0.9371618466946802,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0639063878749365e-07,
+      "loss": 0.855,
+      "step": 34123
+    },
+    {
+      "epoch": 0.9371893109225234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0621088929954292e-07,
+      "loss": 0.796,
+      "step": 34124
+    },
+    {
+      "epoch": 0.9372167751503666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0603121730384368e-07,
+      "loss": 0.8253,
+      "step": 34125
+    },
+    {
+      "epoch": 0.9372442393782099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0585162280182147e-07,
+      "loss": 0.783,
+      "step": 34126
+    },
+    {
+      "epoch": 0.9372717036060532,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0567210579489293e-07,
+      "loss": 0.7507,
+      "step": 34127
+    },
+    {
+      "epoch": 0.9372991678338963,
+      "grad_norm": 0.0,
+      "learning_rate": 2.054926662844814e-07,
+      "loss": 0.7664,
+      "step": 34128
+    },
+    {
+      "epoch": 0.9373266320617396,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0531330427200568e-07,
+      "loss": 0.7841,
+      "step": 34129
+    },
+    {
+      "epoch": 0.9373540962895828,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0513401975888692e-07,
+      "loss": 0.747,
+      "step": 34130
+    },
+    {
+      "epoch": 0.937381560517426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0495481274654062e-07,
+      "loss": 0.7654,
+      "step": 34131
+    },
+    {
+      "epoch": 0.9374090247452693,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0477568323638564e-07,
+      "loss": 0.8487,
+      "step": 34132
+    },
+    {
+      "epoch": 0.9374364889731125,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0459663122984086e-07,
+      "loss": 0.8612,
+      "step": 34133
+    },
+    {
+      "epoch": 0.9374639532009558,
+      "grad_norm": 0.0,
+      "learning_rate": 2.044176567283207e-07,
+      "loss": 0.7374,
+      "step": 34134
+    },
+    {
+      "epoch": 0.9374914174287989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0423875973324292e-07,
+      "loss": 0.8537,
+      "step": 34135
+    },
+    {
+      "epoch": 0.9375188816566422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0405994024602306e-07,
+      "loss": 0.6819,
+      "step": 34136
+    },
+    {
+      "epoch": 0.9375463458844855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.038811982680755e-07,
+      "loss": 0.8443,
+      "step": 34137
+    },
+    {
+      "epoch": 0.9375738101123287,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0370253380081362e-07,
+      "loss": 0.7596,
+      "step": 34138
+    },
+    {
+      "epoch": 0.9376012743401719,
+      "grad_norm": 0.0,
+      "learning_rate": 2.035239468456529e-07,
+      "loss": 0.6868,
+      "step": 34139
+    },
+    {
+      "epoch": 0.9376287385680152,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0334543740400448e-07,
+      "loss": 0.7641,
+      "step": 34140
+    },
+    {
+      "epoch": 0.9376562027958584,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0316700547728385e-07,
+      "loss": 0.8868,
+      "step": 34141
+    },
+    {
+      "epoch": 0.9376836670237017,
+      "grad_norm": 0.0,
+      "learning_rate": 2.029886510668999e-07,
+      "loss": 0.7886,
+      "step": 34142
+    },
+    {
+      "epoch": 0.9377111312515448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0281037417426375e-07,
+      "loss": 0.8221,
+      "step": 34143
+    },
+    {
+      "epoch": 0.9377385954793881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0263217480078755e-07,
+      "loss": 0.832,
+      "step": 34144
+    },
+    {
+      "epoch": 0.9377660597072314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0245405294788024e-07,
+      "loss": 0.8455,
+      "step": 34145
+    },
+    {
+      "epoch": 0.9377935239350745,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0227600861695284e-07,
+      "loss": 0.7734,
+      "step": 34146
+    },
+    {
+      "epoch": 0.9378209881629178,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0209804180941206e-07,
+      "loss": 0.7667,
+      "step": 34147
+    },
+    {
+      "epoch": 0.9378484523907611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0192015252666785e-07,
+      "loss": 0.8378,
+      "step": 34148
+    },
+    {
+      "epoch": 0.9378759166186043,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0174234077012577e-07,
+      "loss": 0.7873,
+      "step": 34149
+    },
+    {
+      "epoch": 0.9379033808464475,
+      "grad_norm": 0.0,
+      "learning_rate": 2.015646065411947e-07,
+      "loss": 0.8238,
+      "step": 34150
+    },
+    {
+      "epoch": 0.9379308450742907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0138694984127905e-07,
+      "loss": 0.8184,
+      "step": 34151
+    },
+    {
+      "epoch": 0.937958309302134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.012093706717855e-07,
+      "loss": 0.8136,
+      "step": 34152
+    },
+    {
+      "epoch": 0.9379857735299773,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0103186903411954e-07,
+      "loss": 0.7892,
+      "step": 34153
+    },
+    {
+      "epoch": 0.9380132377578204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0085444492968563e-07,
+      "loss": 0.7983,
+      "step": 34154
+    },
+    {
+      "epoch": 0.9380407019856637,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0067709835988604e-07,
+      "loss": 0.7475,
+      "step": 34155
+    },
+    {
+      "epoch": 0.9380681662135069,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0049982932612512e-07,
+      "loss": 0.8099,
+      "step": 34156
+    },
+    {
+      "epoch": 0.9380956304413501,
+      "grad_norm": 0.0,
+      "learning_rate": 2.003226378298073e-07,
+      "loss": 0.7768,
+      "step": 34157
+    },
+    {
+      "epoch": 0.9381230946691934,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0014552387233156e-07,
+      "loss": 0.7382,
+      "step": 34158
+    },
+    {
+      "epoch": 0.9381505588970366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999684874551e-07,
+      "loss": 0.8203,
+      "step": 34159
+    },
+    {
+      "epoch": 0.9381780231248799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979152857951602e-07,
+      "loss": 0.6477,
+      "step": 34160
+    },
+    {
+      "epoch": 0.9382054873527231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996146472469751e-07,
+      "loss": 0.8326,
+      "step": 34161
+    },
+    {
+      "epoch": 0.9382329515805663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943784345888063e-07,
+      "loss": 0.8617,
+      "step": 34162
+    },
+    {
+      "epoch": 0.9382604158084096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926111721663032e-07,
+      "loss": 0.8449,
+      "step": 34163
+    },
+    {
+      "epoch": 0.9382878800362527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99084468521622e-07,
+      "loss": 0.8367,
+      "step": 34164
+    },
+    {
+      "epoch": 0.938315344264096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890789737525452e-07,
+      "loss": 0.6826,
+      "step": 34165
+    },
+    {
+      "epoch": 0.9383428084919393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873140377892564e-07,
+      "loss": 0.8402,
+      "step": 34166
+    },
+    {
+      "epoch": 0.9383702727197825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985549877340287e-07,
+      "loss": 0.711,
+      "step": 34167
+    },
+    {
+      "epoch": 0.9383977369476257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837864924196148e-07,
+      "loss": 0.8553,
+      "step": 34168
+    },
+    {
+      "epoch": 0.9384252011754689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982023883041184e-07,
+      "loss": 0.7526,
+      "step": 34169
+    },
+    {
+      "epoch": 0.9384526654033122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802620492189618e-07,
+      "loss": 0.8207,
+      "step": 34170
+    },
+    {
+      "epoch": 0.9384801296311555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785009909668696e-07,
+      "loss": 0.829,
+      "step": 34171
+    },
+    {
+      "epoch": 0.9385075938589986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976740708298852e-07,
+      "loss": 0.8644,
+      "step": 34172
+    },
+    {
+      "epoch": 0.9385350580868419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749812012288428e-07,
+      "loss": 0.8146,
+      "step": 34173
+    },
+    {
+      "epoch": 0.9385625223146852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732224697707415e-07,
+      "loss": 0.743,
+      "step": 34174
+    },
+    {
+      "epoch": 0.9385899865425283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714645139384703e-07,
+      "loss": 0.7901,
+      "step": 34175
+    },
+    {
+      "epoch": 0.9386174507703716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9697073337459517e-07,
+      "loss": 0.8971,
+      "step": 34176
+    },
+    {
+      "epoch": 0.9386449149982148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679509292070742e-07,
+      "loss": 0.746,
+      "step": 34177
+    },
+    {
+      "epoch": 0.9386723792260581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661953003357603e-07,
+      "loss": 0.7791,
+      "step": 34178
+    },
+    {
+      "epoch": 0.9386998434539013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644404471458654e-07,
+      "loss": 0.8776,
+      "step": 34179
+    },
+    {
+      "epoch": 0.9387273076817445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962686369651301e-07,
+      "loss": 0.7987,
+      "step": 34180
+    },
+    {
+      "epoch": 0.9387547719095878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9609330678659445e-07,
+      "loss": 0.8605,
+      "step": 34181
+    },
+    {
+      "epoch": 0.938782236137431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9591805418036514e-07,
+      "loss": 0.7265,
+      "step": 34182
+    },
+    {
+      "epoch": 0.9388097003652742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574287914783106e-07,
+      "loss": 0.7142,
+      "step": 34183
+    },
+    {
+      "epoch": 0.9388371645931175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955677816903756e-07,
+      "loss": 0.9262,
+      "step": 34184
+    },
+    {
+      "epoch": 0.9388646288209607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9539276180938648e-07,
+      "loss": 0.8409,
+      "step": 34185
+    },
+    {
+      "epoch": 0.9388920930488039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952178195062482e-07,
+      "loss": 0.7641,
+      "step": 34186
+    },
+    {
+      "epoch": 0.9389195572766472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504295478234292e-07,
+      "loss": 0.8206,
+      "step": 34187
+    },
+    {
+      "epoch": 0.9389470215044904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9486816763905737e-07,
+      "loss": 0.8489,
+      "step": 34188
+    },
+    {
+      "epoch": 0.9389744857323337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9469345807777375e-07,
+      "loss": 0.8487,
+      "step": 34189
+    },
+    {
+      "epoch": 0.9390019499601768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451882609987315e-07,
+      "loss": 0.7058,
+      "step": 34190
+    },
+    {
+      "epoch": 0.9390294141880201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434427170673675e-07,
+      "loss": 0.7992,
+      "step": 34191
+    },
+    {
+      "epoch": 0.9390568784158634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9416979489974785e-07,
+      "loss": 0.7758,
+      "step": 34192
+    },
+    {
+      "epoch": 0.9390843426437065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399539568028537e-07,
+      "loss": 0.6928,
+      "step": 34193
+    },
+    {
+      "epoch": 0.9391118068715498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9382107404972928e-07,
+      "loss": 0.7567,
+      "step": 34194
+    },
+    {
+      "epoch": 0.939139271099393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9364683000946073e-07,
+      "loss": 0.8837,
+      "step": 34195
+    },
+    {
+      "epoch": 0.9391667353272363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347266356085638e-07,
+      "loss": 0.747,
+      "step": 34196
+    },
+    {
+      "epoch": 0.9391941995550795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9329857470529623e-07,
+      "loss": 0.7944,
+      "step": 34197
+    },
+    {
+      "epoch": 0.9392216637829227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9312456344415476e-07,
+      "loss": 0.7595,
+      "step": 34198
+    },
+    {
+      "epoch": 0.939249128010766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295062977881085e-07,
+      "loss": 0.8834,
+      "step": 34199
+    },
+    {
+      "epoch": 0.9392765922386093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9277677371064008e-07,
+      "loss": 0.7996,
+      "step": 34200
+    },
+    {
+      "epoch": 0.9393040564664524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926029952410202e-07,
+      "loss": 0.7962,
+      "step": 34201
+    },
+    {
+      "epoch": 0.9393315206942957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924292943713235e-07,
+      "loss": 0.7483,
+      "step": 34202
+    },
+    {
+      "epoch": 0.9393589849221389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9225567110292442e-07,
+      "loss": 0.8772,
+      "step": 34203
+    },
+    {
+      "epoch": 0.9393864491499822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920821254371985e-07,
+      "loss": 0.7765,
+      "step": 34204
+    },
+    {
+      "epoch": 0.9394139133778254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91908657375518e-07,
+      "loss": 0.8667,
+      "step": 34205
+    },
+    {
+      "epoch": 0.9394413776056686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9173526691925516e-07,
+      "loss": 0.864,
+      "step": 34206
+    },
+    {
+      "epoch": 0.9394688418335119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9156195406978328e-07,
+      "loss": 0.7575,
+      "step": 34207
+    },
+    {
+      "epoch": 0.939496306061355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9138871882847242e-07,
+      "loss": 0.8701,
+      "step": 34208
+    },
+    {
+      "epoch": 0.9395237702891983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912155611966926e-07,
+      "loss": 0.84,
+      "step": 34209
+    },
+    {
+      "epoch": 0.9395512345170416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9104248117581492e-07,
+      "loss": 0.748,
+      "step": 34210
+    },
+    {
+      "epoch": 0.9395786987448848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908694787672083e-07,
+      "loss": 0.8586,
+      "step": 34211
+    },
+    {
+      "epoch": 0.939606162972728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9069655397224384e-07,
+      "loss": 0.8137,
+      "step": 34212
+    },
+    {
+      "epoch": 0.9396336272005713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9052370679228604e-07,
+      "loss": 0.8083,
+      "step": 34213
+    },
+    {
+      "epoch": 0.9396610914284145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90350937228706e-07,
+      "loss": 0.8124,
+      "step": 34214
+    },
+    {
+      "epoch": 0.9396885556562578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901782452828693e-07,
+      "loss": 0.7503,
+      "step": 34215
+    },
+    {
+      "epoch": 0.9397160198841009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000563095614156e-07,
+      "loss": 0.8201,
+      "step": 34216
+    },
+    {
+      "epoch": 0.9397434841119442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8983309424988827e-07,
+      "loss": 0.8712,
+      "step": 34217
+    },
+    {
+      "epoch": 0.9397709483397875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8966063516547728e-07,
+      "loss": 0.7834,
+      "step": 34218
+    },
+    {
+      "epoch": 0.9397984125676306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894882537042708e-07,
+      "loss": 0.7884,
+      "step": 34219
+    },
+    {
+      "epoch": 0.9398258767954739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8931594986763334e-07,
+      "loss": 0.8257,
+      "step": 34220
+    },
+    {
+      "epoch": 0.9398533410233171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8914372365692934e-07,
+      "loss": 0.7893,
+      "step": 34221
+    },
+    {
+      "epoch": 0.9398808052511604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8897157507351992e-07,
+      "loss": 0.8485,
+      "step": 34222
+    },
+    {
+      "epoch": 0.9399082694790036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8879950411876736e-07,
+      "loss": 0.7998,
+      "step": 34223
+    },
+    {
+      "epoch": 0.9399357337068468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8862751079403384e-07,
+      "loss": 0.82,
+      "step": 34224
+    },
+    {
+      "epoch": 0.9399631979346901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884555951006817e-07,
+      "loss": 0.9661,
+      "step": 34225
+    },
+    {
+      "epoch": 0.9399906621625334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8828375704006753e-07,
+      "loss": 0.8528,
+      "step": 34226
+    },
+    {
+      "epoch": 0.9400181263903765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881119966135536e-07,
+      "loss": 0.7984,
+      "step": 34227
+    },
+    {
+      "epoch": 0.9400455906182198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879403138224978e-07,
+      "loss": 0.8273,
+      "step": 34228
+    },
+    {
+      "epoch": 0.940073054846063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8776870866826002e-07,
+      "loss": 0.8031,
+      "step": 34229
+    },
+    {
+      "epoch": 0.9401005190739062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8759718115219594e-07,
+      "loss": 0.7273,
+      "step": 34230
+    },
+    {
+      "epoch": 0.9401279833017495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8742573127566555e-07,
+      "loss": 0.8234,
+      "step": 34231
+    },
+    {
+      "epoch": 0.9401554475295927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872543590400222e-07,
+      "loss": 0.8022,
+      "step": 34232
+    },
+    {
+      "epoch": 0.940182911757436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708306444662372e-07,
+      "loss": 0.8205,
+      "step": 34233
+    },
+    {
+      "epoch": 0.9402103759852791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8691184749682567e-07,
+      "loss": 0.7904,
+      "step": 34234
+    },
+    {
+      "epoch": 0.9402378402131224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867407081919814e-07,
+      "loss": 0.7927,
+      "step": 34235
+    },
+    {
+      "epoch": 0.9402653044409657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8656964653344766e-07,
+      "loss": 0.9548,
+      "step": 34236
+    },
+    {
+      "epoch": 0.9402927686688088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8639866252257554e-07,
+      "loss": 0.7832,
+      "step": 34237
+    },
+    {
+      "epoch": 0.9403202328966521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8622775616071842e-07,
+      "loss": 0.8325,
+      "step": 34238
+    },
+    {
+      "epoch": 0.9403476971244954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8605692744922854e-07,
+      "loss": 0.8265,
+      "step": 34239
+    },
+    {
+      "epoch": 0.9403751613523386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8588617638945816e-07,
+      "loss": 0.7516,
+      "step": 34240
+    },
+    {
+      "epoch": 0.9404026255801818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857155029827573e-07,
+      "loss": 0.7123,
+      "step": 34241
+    },
+    {
+      "epoch": 0.940430089808025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554490723047824e-07,
+      "loss": 0.881,
+      "step": 34242
+    },
+    {
+      "epoch": 0.9404575540358683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8537438913396878e-07,
+      "loss": 0.8163,
+      "step": 34243
+    },
+    {
+      "epoch": 0.9404850182637116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8520394869457892e-07,
+      "loss": 0.9012,
+      "step": 34244
+    },
+    {
+      "epoch": 0.9405124824915547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8503358591365872e-07,
+      "loss": 0.7839,
+      "step": 34245
+    },
+    {
+      "epoch": 0.940539946719398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8486330079255267e-07,
+      "loss": 0.8302,
+      "step": 34246
+    },
+    {
+      "epoch": 0.9405674109472412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8469309333261188e-07,
+      "loss": 0.8067,
+      "step": 34247
+    },
+    {
+      "epoch": 0.9405948751750844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8452296353518084e-07,
+      "loss": 0.7693,
+      "step": 34248
+    },
+    {
+      "epoch": 0.9406223394029277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8435291140160628e-07,
+      "loss": 0.8503,
+      "step": 34249
+    },
+    {
+      "epoch": 0.9406498036307709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8418293693323264e-07,
+      "loss": 0.812,
+      "step": 34250
+    },
+    {
+      "epoch": 0.9406772678586142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8401304013140663e-07,
+      "loss": 0.7817,
+      "step": 34251
+    },
+    {
+      "epoch": 0.9407047320864574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8384322099747165e-07,
+      "loss": 0.7709,
+      "step": 34252
+    },
+    {
+      "epoch": 0.9407321963143006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8367347953277215e-07,
+      "loss": 0.8149,
+      "step": 34253
+    },
+    {
+      "epoch": 0.9407596605421439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835038157386515e-07,
+      "loss": 0.8191,
+      "step": 34254
+    },
+    {
+      "epoch": 0.940787124769987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8333422961644975e-07,
+      "loss": 0.7205,
+      "step": 34255
+    },
+    {
+      "epoch": 0.9408145889978303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8316472116751027e-07,
+      "loss": 0.8509,
+      "step": 34256
+    },
+    {
+      "epoch": 0.9408420532256736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8299529039317422e-07,
+      "loss": 0.7501,
+      "step": 34257
+    },
+    {
+      "epoch": 0.9408695174535168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828259372947816e-07,
+      "loss": 0.7479,
+      "step": 34258
+    },
+    {
+      "epoch": 0.94089698168136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826566618736736e-07,
+      "loss": 0.8268,
+      "step": 34259
+    },
+    {
+      "epoch": 0.9409244459092032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8248746413118913e-07,
+      "loss": 0.8708,
+      "step": 34260
+    },
+    {
+      "epoch": 0.9409519101370465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8231834406866712e-07,
+      "loss": 0.7932,
+      "step": 34261
+    },
+    {
+      "epoch": 0.9409793743648898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8214930168744538e-07,
+      "loss": 0.8241,
+      "step": 34262
+    },
+    {
+      "epoch": 0.9410068385927329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8198033698886064e-07,
+      "loss": 0.8159,
+      "step": 34263
+    },
+    {
+      "epoch": 0.9410343028205762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818114499742507e-07,
+      "loss": 0.843,
+      "step": 34264
+    },
+    {
+      "epoch": 0.9410617670484195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816426406449512e-07,
+      "loss": 0.8092,
+      "step": 34265
+    },
+    {
+      "epoch": 0.9410892312762626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814739090022999e-07,
+      "loss": 0.8008,
+      "step": 34266
+    },
+    {
+      "epoch": 0.9411166955041059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813052550476302e-07,
+      "loss": 0.8159,
+      "step": 34267
+    },
+    {
+      "epoch": 0.9411441597319491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8113667878227546e-07,
+      "loss": 0.826,
+      "step": 34268
+    },
+    {
+      "epoch": 0.9411716239597924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8096818020757247e-07,
+      "loss": 0.8197,
+      "step": 34269
+    },
+    {
+      "epoch": 0.9411990881876356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8079975932485117e-07,
+      "loss": 0.75,
+      "step": 34270
+    },
+    {
+      "epoch": 0.9412265524154788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806314161354461e-07,
+      "loss": 0.8469,
+      "step": 34271
+    },
+    {
+      "epoch": 0.9412540166433221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8046315064068953e-07,
+      "loss": 0.8135,
+      "step": 34272
+    },
+    {
+      "epoch": 0.9412814808711653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8029496284191262e-07,
+      "loss": 0.8033,
+      "step": 34273
+    },
+    {
+      "epoch": 0.9413089450990085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8012685274044428e-07,
+      "loss": 0.8075,
+      "step": 34274
+    },
+    {
+      "epoch": 0.9413364093268518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799588203376168e-07,
+      "loss": 0.827,
+      "step": 34275
+    },
+    {
+      "epoch": 0.941363873554695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7979086563475912e-07,
+      "loss": 0.7681,
+      "step": 34276
+    },
+    {
+      "epoch": 0.9413913377825383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7962298863320014e-07,
+      "loss": 0.8449,
+      "step": 34277
+    },
+    {
+      "epoch": 0.9414188020103815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7945518933426886e-07,
+      "loss": 0.8666,
+      "step": 34278
+    },
+    {
+      "epoch": 0.9414462662382247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7928746773929195e-07,
+      "loss": 0.8182,
+      "step": 34279
+    },
+    {
+      "epoch": 0.941473730466068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7911982384959615e-07,
+      "loss": 0.9294,
+      "step": 34280
+    },
+    {
+      "epoch": 0.9415011946939111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7895225766650816e-07,
+      "loss": 0.8042,
+      "step": 34281
+    },
+    {
+      "epoch": 0.9415286589217544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7878476919135467e-07,
+      "loss": 0.7813,
+      "step": 34282
+    },
+    {
+      "epoch": 0.9415561231495977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7861735842546025e-07,
+      "loss": 0.8643,
+      "step": 34283
+    },
+    {
+      "epoch": 0.9415835873774409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7845002537015043e-07,
+      "loss": 0.8065,
+      "step": 34284
+    },
+    {
+      "epoch": 0.9416110516052841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7828277002674755e-07,
+      "loss": 0.7917,
+      "step": 34285
+    },
+    {
+      "epoch": 0.9416385158331273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7811559239657715e-07,
+      "loss": 0.7388,
+      "step": 34286
+    },
+    {
+      "epoch": 0.9416659800609706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7794849248096046e-07,
+      "loss": 0.8468,
+      "step": 34287
+    },
+    {
+      "epoch": 0.9416934442888139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7778147028121974e-07,
+      "loss": 0.741,
+      "step": 34288
+    },
+    {
+      "epoch": 0.941720908516657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7761452579867612e-07,
+      "loss": 0.7743,
+      "step": 34289
+    },
+    {
+      "epoch": 0.9417483727445003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77447659034653e-07,
+      "loss": 0.8192,
+      "step": 34290
+    },
+    {
+      "epoch": 0.9417758369723436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772808699904671e-07,
+      "loss": 0.7932,
+      "step": 34291
+    },
+    {
+      "epoch": 0.9418033012001867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7711415866744076e-07,
+      "loss": 0.8097,
+      "step": 34292
+    },
+    {
+      "epoch": 0.94183076542803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7694752506689284e-07,
+      "loss": 0.8119,
+      "step": 34293
+    },
+    {
+      "epoch": 0.9418582296558732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767809691901401e-07,
+      "loss": 0.7809,
+      "step": 34294
+    },
+    {
+      "epoch": 0.9418856938837165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766144910385015e-07,
+      "loss": 0.7325,
+      "step": 34295
+    },
+    {
+      "epoch": 0.9419131581115597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7644809061329482e-07,
+      "loss": 0.7675,
+      "step": 34296
+    },
+    {
+      "epoch": 0.9419406223394029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7628176791583573e-07,
+      "loss": 0.837,
+      "step": 34297
+    },
+    {
+      "epoch": 0.9419680865672462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7611552294744095e-07,
+      "loss": 0.8367,
+      "step": 34298
+    },
+    {
+      "epoch": 0.9419955507950893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7594935570942495e-07,
+      "loss": 0.7939,
+      "step": 34299
+    },
+    {
+      "epoch": 0.9420230150229326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7578326620310338e-07,
+      "loss": 0.8624,
+      "step": 34300
+    },
+    {
+      "epoch": 0.9420504792507759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756172544297907e-07,
+      "loss": 0.8792,
+      "step": 34301
+    },
+    {
+      "epoch": 0.9420779434786191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7545132039080038e-07,
+      "loss": 0.8253,
+      "step": 34302
+    },
+    {
+      "epoch": 0.9421054077064623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752854640874435e-07,
+      "loss": 0.7197,
+      "step": 34303
+    },
+    {
+      "epoch": 0.9421328719343056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7511968552103465e-07,
+      "loss": 0.6834,
+      "step": 34304
+    },
+    {
+      "epoch": 0.9421603361621488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7495398469288382e-07,
+      "loss": 0.8034,
+      "step": 34305
+    },
+    {
+      "epoch": 0.9421878003899921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7478836160430335e-07,
+      "loss": 0.672,
+      "step": 34306
+    },
+    {
+      "epoch": 0.9422152646178352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7462281625660327e-07,
+      "loss": 0.724,
+      "step": 34307
+    },
+    {
+      "epoch": 0.9422427288456785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445734865109366e-07,
+      "loss": 0.8035,
+      "step": 34308
+    },
+    {
+      "epoch": 0.9422701930735218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7429195878908344e-07,
+      "loss": 0.7809,
+      "step": 34309
+    },
+    {
+      "epoch": 0.9422976573013649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7412664667188162e-07,
+      "loss": 0.8975,
+      "step": 34310
+    },
+    {
+      "epoch": 0.9423251215292082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739614123007949e-07,
+      "loss": 0.8248,
+      "step": 34311
+    },
+    {
+      "epoch": 0.9423525857570514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737962556771322e-07,
+      "loss": 0.7595,
+      "step": 34312
+    },
+    {
+      "epoch": 0.9423800499848947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7363117680220143e-07,
+      "loss": 0.7589,
+      "step": 34313
+    },
+    {
+      "epoch": 0.9424075142127379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7346617567730594e-07,
+      "loss": 0.8761,
+      "step": 34314
+    },
+    {
+      "epoch": 0.9424349784405811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733012523037514e-07,
+      "loss": 0.8608,
+      "step": 34315
+    },
+    {
+      "epoch": 0.9424624426684244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731364066828445e-07,
+      "loss": 0.8304,
+      "step": 34316
+    },
+    {
+      "epoch": 0.9424899068962677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7297163881588975e-07,
+      "loss": 0.8315,
+      "step": 34317
+    },
+    {
+      "epoch": 0.9425173711241108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7280694870418946e-07,
+      "loss": 0.7745,
+      "step": 34318
+    },
+    {
+      "epoch": 0.9425448353519541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264233634904814e-07,
+      "loss": 0.7056,
+      "step": 34319
+    },
+    {
+      "epoch": 0.9425722995797973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7247780175176698e-07,
+      "loss": 0.7739,
+      "step": 34320
+    },
+    {
+      "epoch": 0.9425997638076405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7231334491364716e-07,
+      "loss": 0.8531,
+      "step": 34321
+    },
+    {
+      "epoch": 0.9426272280354838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7214896583599205e-07,
+      "loss": 0.8202,
+      "step": 34322
+    },
+    {
+      "epoch": 0.942654692263327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7198466452010066e-07,
+      "loss": 0.7438,
+      "step": 34323
+    },
+    {
+      "epoch": 0.9426821564911703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7182044096727303e-07,
+      "loss": 0.8029,
+      "step": 34324
+    },
+    {
+      "epoch": 0.9427096207190134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7165629517881143e-07,
+      "loss": 0.8464,
+      "step": 34325
+    },
+    {
+      "epoch": 0.9427370849468567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7149222715601155e-07,
+      "loss": 0.8421,
+      "step": 34326
+    },
+    {
+      "epoch": 0.9427645491747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713282369001712e-07,
+      "loss": 0.7676,
+      "step": 34327
+    },
+    {
+      "epoch": 0.9427920134025431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7116432441258935e-07,
+      "loss": 0.8586,
+      "step": 34328
+    },
+    {
+      "epoch": 0.9428194776303864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7100048969456274e-07,
+      "loss": 0.8535,
+      "step": 34329
+    },
+    {
+      "epoch": 0.9428469418582297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7083673274738811e-07,
+      "loss": 0.7371,
+      "step": 34330
+    },
+    {
+      "epoch": 0.9428744060860729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706730535723611e-07,
+      "loss": 0.8086,
+      "step": 34331
+    },
+    {
+      "epoch": 0.9429018703139161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705094521707762e-07,
+      "loss": 0.8173,
+      "step": 34332
+    },
+    {
+      "epoch": 0.9429293345417593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7034592854392796e-07,
+      "loss": 0.7688,
+      "step": 34333
+    },
+    {
+      "epoch": 0.9429567987696026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7018248269310979e-07,
+      "loss": 0.9036,
+      "step": 34334
+    },
+    {
+      "epoch": 0.9429842629974459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7001911461961617e-07,
+      "loss": 0.8177,
+      "step": 34335
+    },
+    {
+      "epoch": 0.943011727225289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6985582432473836e-07,
+      "loss": 0.8834,
+      "step": 34336
+    },
+    {
+      "epoch": 0.9430391914531323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696926118097697e-07,
+      "loss": 0.8156,
+      "step": 34337
+    },
+    {
+      "epoch": 0.9430666556809755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6952947707600143e-07,
+      "loss": 0.8261,
+      "step": 34338
+    },
+    {
+      "epoch": 0.9430941199088188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6936642012472248e-07,
+      "loss": 0.7592,
+      "step": 34339
+    },
+    {
+      "epoch": 0.943121584136662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6920344095722518e-07,
+      "loss": 0.8402,
+      "step": 34340
+    },
+    {
+      "epoch": 0.9431490483645052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690405395747996e-07,
+      "loss": 0.7993,
+      "step": 34341
+    },
+    {
+      "epoch": 0.9431765125923485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6887771597873136e-07,
+      "loss": 0.8185,
+      "step": 34342
+    },
+    {
+      "epoch": 0.9432039768201917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687149701703128e-07,
+      "loss": 0.8553,
+      "step": 34343
+    },
+    {
+      "epoch": 0.9432314410480349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685523021508284e-07,
+      "loss": 0.7194,
+      "step": 34344
+    },
+    {
+      "epoch": 0.9432589052758782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6838971192156605e-07,
+      "loss": 0.817,
+      "step": 34345
+    },
+    {
+      "epoch": 0.9432863695037214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6822719948381362e-07,
+      "loss": 0.8292,
+      "step": 34346
+    },
+    {
+      "epoch": 0.9433138337315646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680647648388556e-07,
+      "loss": 0.7339,
+      "step": 34347
+    },
+    {
+      "epoch": 0.9433412979594079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6790240798797874e-07,
+      "loss": 0.694,
+      "step": 34348
+    },
+    {
+      "epoch": 0.9433687621872511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6774012893246538e-07,
+      "loss": 0.7436,
+      "step": 34349
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6757792767360226e-07,
+      "loss": 0.7248,
+      "step": 34350
+    },
+    {
+      "epoch": 0.9434236906429376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6741580421266946e-07,
+      "loss": 0.7585,
+      "step": 34351
+    },
+    {
+      "epoch": 0.9434511548707808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672537585509526e-07,
+      "loss": 0.8528,
+      "step": 34352
+    },
+    {
+      "epoch": 0.9434786190986241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670917906897318e-07,
+      "loss": 0.7675,
+      "step": 34353
+    },
+    {
+      "epoch": 0.9435060833264672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6692990063029046e-07,
+      "loss": 0.8141,
+      "step": 34354
+    },
+    {
+      "epoch": 0.9435335475543105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6676808837390978e-07,
+      "loss": 0.8458,
+      "step": 34355
+    },
+    {
+      "epoch": 0.9435610117821538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6660635392186763e-07,
+      "loss": 0.8298,
+      "step": 34356
+    },
+    {
+      "epoch": 0.943588476009997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6644469727544632e-07,
+      "loss": 0.8439,
+      "step": 34357
+    },
+    {
+      "epoch": 0.9436159402378402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662831184359226e-07,
+      "loss": 0.7528,
+      "step": 34358
+    },
+    {
+      "epoch": 0.9436434044656834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6612161740457544e-07,
+      "loss": 0.7974,
+      "step": 34359
+    },
+    {
+      "epoch": 0.9436708686935267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6596019418268383e-07,
+      "loss": 0.7912,
+      "step": 34360
+    },
+    {
+      "epoch": 0.94369833292137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6579884877152564e-07,
+      "loss": 0.7566,
+      "step": 34361
+    },
+    {
+      "epoch": 0.9437257971492131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656375811723754e-07,
+      "loss": 0.8022,
+      "step": 34362
+    },
+    {
+      "epoch": 0.9437532613770564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6547639138650985e-07,
+      "loss": 0.7998,
+      "step": 34363
+    },
+    {
+      "epoch": 0.9437807256048997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6531527941520465e-07,
+      "loss": 0.7773,
+      "step": 34364
+    },
+    {
+      "epoch": 0.9438081898327428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6515424525973544e-07,
+      "loss": 0.7567,
+      "step": 34365
+    },
+    {
+      "epoch": 0.9438356540605861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6499328892137567e-07,
+      "loss": 0.8474,
+      "step": 34366
+    },
+    {
+      "epoch": 0.9438631182884293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6483241040139876e-07,
+      "loss": 0.7561,
+      "step": 34367
+    },
+    {
+      "epoch": 0.9438905825162726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6467160970107587e-07,
+      "loss": 0.7519,
+      "step": 34368
+    },
+    {
+      "epoch": 0.9439180467441158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645108868216816e-07,
+      "loss": 0.8674,
+      "step": 34369
+    },
+    {
+      "epoch": 0.943945510971959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6435024176448822e-07,
+      "loss": 0.8898,
+      "step": 34370
+    },
+    {
+      "epoch": 0.9439729751998023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6418967453076474e-07,
+      "loss": 0.8117,
+      "step": 34371
+    },
+    {
+      "epoch": 0.9440004394276454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6402918512178346e-07,
+      "loss": 0.7417,
+      "step": 34372
+    },
+    {
+      "epoch": 0.9440279036554887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6386877353881226e-07,
+      "loss": 0.7734,
+      "step": 34373
+    },
+    {
+      "epoch": 0.944055367883332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6370843978312233e-07,
+      "loss": 0.8689,
+      "step": 34374
+    },
+    {
+      "epoch": 0.9440828321111752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635481838559816e-07,
+      "loss": 0.7486,
+      "step": 34375
+    },
+    {
+      "epoch": 0.9441102963390184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6338800575865676e-07,
+      "loss": 0.7835,
+      "step": 34376
+    },
+    {
+      "epoch": 0.9441377605668617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322790549241684e-07,
+      "loss": 0.7919,
+      "step": 34377
+    },
+    {
+      "epoch": 0.9441652247947049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6306788305852861e-07,
+      "loss": 0.833,
+      "step": 34378
+    },
+    {
+      "epoch": 0.9441926890225482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629079384582577e-07,
+      "loss": 0.914,
+      "step": 34379
+    },
+    {
+      "epoch": 0.9442201532503913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6274807169286865e-07,
+      "loss": 0.7588,
+      "step": 34380
+    },
+    {
+      "epoch": 0.9442476174782346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6258828276362936e-07,
+      "loss": 0.7596,
+      "step": 34381
+    },
+    {
+      "epoch": 0.9442750817060779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6242857167180103e-07,
+      "loss": 0.7839,
+      "step": 34382
+    },
+    {
+      "epoch": 0.944302545933921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6226893841864823e-07,
+      "loss": 0.7356,
+      "step": 34383
+    },
+    {
+      "epoch": 0.9443300101617643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6210938300543544e-07,
+      "loss": 0.8171,
+      "step": 34384
+    },
+    {
+      "epoch": 0.9443574743896075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619499054334228e-07,
+      "loss": 0.7172,
+      "step": 34385
+    },
+    {
+      "epoch": 0.9443849386174508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617905057038738e-07,
+      "loss": 0.8047,
+      "step": 34386
+    },
+    {
+      "epoch": 0.944412402845294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6163118381804955e-07,
+      "loss": 0.7494,
+      "step": 34387
+    },
+    {
+      "epoch": 0.9444398670731372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6147193977721022e-07,
+      "loss": 0.7327,
+      "step": 34388
+    },
+    {
+      "epoch": 0.9444673313009805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6131277358261587e-07,
+      "loss": 0.8354,
+      "step": 34389
+    },
+    {
+      "epoch": 0.9444947955288238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6115368523552665e-07,
+      "loss": 0.7806,
+      "step": 34390
+    },
+    {
+      "epoch": 0.9445222597566669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6099467473720044e-07,
+      "loss": 0.7663,
+      "step": 34391
+    },
+    {
+      "epoch": 0.9445497239845102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6083574208889506e-07,
+      "loss": 0.7899,
+      "step": 34392
+    },
+    {
+      "epoch": 0.9445771882123534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6067688729186849e-07,
+      "loss": 0.7066,
+      "step": 34393
+    },
+    {
+      "epoch": 0.9446046524401966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6051811034737852e-07,
+      "loss": 0.7892,
+      "step": 34394
+    },
+    {
+      "epoch": 0.9446321166680399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6035941125667975e-07,
+      "loss": 0.7888,
+      "step": 34395
+    },
+    {
+      "epoch": 0.9446595808958831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6020079002103005e-07,
+      "loss": 0.7915,
+      "step": 34396
+    },
+    {
+      "epoch": 0.9446870451237264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6004224664168288e-07,
+      "loss": 0.8197,
+      "step": 34397
+    },
+    {
+      "epoch": 0.9447145093515695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598837811198939e-07,
+      "loss": 0.7565,
+      "step": 34398
+    },
+    {
+      "epoch": 0.9447419735794128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597253934569143e-07,
+      "loss": 0.8337,
+      "step": 34399
+    },
+    {
+      "epoch": 0.9447694378072561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5956708365399975e-07,
+      "loss": 0.7236,
+      "step": 34400
+    },
+    {
+      "epoch": 0.9447969020350993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594088517124026e-07,
+      "loss": 0.7424,
+      "step": 34401
+    },
+    {
+      "epoch": 0.9448243662629425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592506976333752e-07,
+      "loss": 0.892,
+      "step": 34402
+    },
+    {
+      "epoch": 0.9448518304907858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590926214181676e-07,
+      "loss": 0.8044,
+      "step": 34403
+    },
+    {
+      "epoch": 0.944879294718629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5893462306802998e-07,
+      "loss": 0.8841,
+      "step": 34404
+    },
+    {
+      "epoch": 0.9449067589464722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5877670258421575e-07,
+      "loss": 0.725,
+      "step": 34405
+    },
+    {
+      "epoch": 0.9449342231743154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5861885996797167e-07,
+      "loss": 0.7982,
+      "step": 34406
+    },
+    {
+      "epoch": 0.9449616874021587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5846109522054677e-07,
+      "loss": 0.7064,
+      "step": 34407
+    },
+    {
+      "epoch": 0.944989151630002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5830340834319002e-07,
+      "loss": 0.8338,
+      "step": 34408
+    },
+    {
+      "epoch": 0.9450166158578451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5814579933714935e-07,
+      "loss": 0.7366,
+      "step": 34409
+    },
+    {
+      "epoch": 0.9450440800856884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5798826820367152e-07,
+      "loss": 0.7559,
+      "step": 34410
+    },
+    {
+      "epoch": 0.9450715443135316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578308149440022e-07,
+      "loss": 0.7886,
+      "step": 34411
+    },
+    {
+      "epoch": 0.9450990085413749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576734395593882e-07,
+      "loss": 0.9033,
+      "step": 34412
+    },
+    {
+      "epoch": 0.9451264727692181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5751614205107514e-07,
+      "loss": 0.8016,
+      "step": 34413
+    },
+    {
+      "epoch": 0.9451539369970613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573589224203076e-07,
+      "loss": 0.7937,
+      "step": 34414
+    },
+    {
+      "epoch": 0.9451814012249046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5720178066832792e-07,
+      "loss": 0.8437,
+      "step": 34415
+    },
+    {
+      "epoch": 0.9452088654527478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5704471679638066e-07,
+      "loss": 0.8076,
+      "step": 34416
+    },
+    {
+      "epoch": 0.945236329680591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568877308057082e-07,
+      "loss": 0.8116,
+      "step": 34417
+    },
+    {
+      "epoch": 0.9452637939084343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567308226975517e-07,
+      "loss": 0.7386,
+      "step": 34418
+    },
+    {
+      "epoch": 0.9452912581362775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5657399247315573e-07,
+      "loss": 0.8443,
+      "step": 34419
+    },
+    {
+      "epoch": 0.9453187223641207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5641724013375935e-07,
+      "loss": 0.7789,
+      "step": 34420
+    },
+    {
+      "epoch": 0.945346186591964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5626056568060154e-07,
+      "loss": 0.8493,
+      "step": 34421
+    },
+    {
+      "epoch": 0.9453736508198072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5610396911492464e-07,
+      "loss": 0.7511,
+      "step": 34422
+    },
+    {
+      "epoch": 0.9454011150476505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5594745043796543e-07,
+      "loss": 0.8638,
+      "step": 34423
+    },
+    {
+      "epoch": 0.9454285792754936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5579100965096295e-07,
+      "loss": 0.8345,
+      "step": 34424
+    },
+    {
+      "epoch": 0.9454560435033369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5563464675515615e-07,
+      "loss": 0.8762,
+      "step": 34425
+    },
+    {
+      "epoch": 0.9454835077311802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5547836175178078e-07,
+      "loss": 0.6844,
+      "step": 34426
+    },
+    {
+      "epoch": 0.9455109719590233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5532215464207357e-07,
+      "loss": 0.918,
+      "step": 34427
+    },
+    {
+      "epoch": 0.9455384361868666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5516602542727133e-07,
+      "loss": 0.7089,
+      "step": 34428
+    },
+    {
+      "epoch": 0.9455659004147099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5500997410860974e-07,
+      "loss": 0.7219,
+      "step": 34429
+    },
+    {
+      "epoch": 0.9455933646425531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5485400068732226e-07,
+      "loss": 0.8204,
+      "step": 34430
+    },
+    {
+      "epoch": 0.9456208288703963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5469810516464344e-07,
+      "loss": 0.8799,
+      "step": 34431
+    },
+    {
+      "epoch": 0.9456482930982395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5454228754180679e-07,
+      "loss": 0.7421,
+      "step": 34432
+    },
+    {
+      "epoch": 0.9456757573260828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.543865478200457e-07,
+      "loss": 0.783,
+      "step": 34433
+    },
+    {
+      "epoch": 0.945703221553926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5423088600059145e-07,
+      "loss": 0.7373,
+      "step": 34434
+    },
+    {
+      "epoch": 0.9457306857817692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5407530208467748e-07,
+      "loss": 0.8443,
+      "step": 34435
+    },
+    {
+      "epoch": 0.9457581500096125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539197960735317e-07,
+      "loss": 0.7445,
+      "step": 34436
+    },
+    {
+      "epoch": 0.9457856142374557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5376436796838866e-07,
+      "loss": 0.755,
+      "step": 34437
+    },
+    {
+      "epoch": 0.9458130784652989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5360901777047633e-07,
+      "loss": 0.8039,
+      "step": 34438
+    },
+    {
+      "epoch": 0.9458405426931422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5345374548102143e-07,
+      "loss": 0.7664,
+      "step": 34439
+    },
+    {
+      "epoch": 0.9458680069209854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5329855110125523e-07,
+      "loss": 0.7086,
+      "step": 34440
+    },
+    {
+      "epoch": 0.9458954711488287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5314343463240567e-07,
+      "loss": 0.7562,
+      "step": 34441
+    },
+    {
+      "epoch": 0.9459229353766719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529883960756995e-07,
+      "loss": 0.7823,
+      "step": 34442
+    },
+    {
+      "epoch": 0.9459503996045151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5283343543236351e-07,
+      "loss": 0.7792,
+      "step": 34443
+    },
+    {
+      "epoch": 0.9459778638323584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5267855270362342e-07,
+      "loss": 0.6725,
+      "step": 34444
+    },
+    {
+      "epoch": 0.9460053280602015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52523747890706e-07,
+      "loss": 0.8649,
+      "step": 34445
+    },
+    {
+      "epoch": 0.9460327922880448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5236902099483474e-07,
+      "loss": 0.8663,
+      "step": 34446
+    },
+    {
+      "epoch": 0.9460602565158881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5221437201723422e-07,
+      "loss": 0.6689,
+      "step": 34447
+    },
+    {
+      "epoch": 0.9460877207437313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5205980095912787e-07,
+      "loss": 0.7789,
+      "step": 34448
+    },
+    {
+      "epoch": 0.9461151849715745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519053078217403e-07,
+      "loss": 0.8507,
+      "step": 34449
+    },
+    {
+      "epoch": 0.9461426491994177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5175089260629162e-07,
+      "loss": 0.7757,
+      "step": 34450
+    },
+    {
+      "epoch": 0.946170113427261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515965553140053e-07,
+      "loss": 0.8475,
+      "step": 34451
+    },
+    {
+      "epoch": 0.9461975776551043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514422959461015e-07,
+      "loss": 0.8433,
+      "step": 34452
+    },
+    {
+      "epoch": 0.9462250418829474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5128811450380142e-07,
+      "loss": 0.785,
+      "step": 34453
+    },
+    {
+      "epoch": 0.9462525061107907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5113401098832526e-07,
+      "loss": 0.7671,
+      "step": 34454
+    },
+    {
+      "epoch": 0.946279970338634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5097998540089198e-07,
+      "loss": 0.8948,
+      "step": 34455
+    },
+    {
+      "epoch": 0.9463074345664771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5082603774271952e-07,
+      "loss": 0.7545,
+      "step": 34456
+    },
+    {
+      "epoch": 0.9463348987943204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5067216801502694e-07,
+      "loss": 0.8713,
+      "step": 34457
+    },
+    {
+      "epoch": 0.9463623630221636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5051837621903098e-07,
+      "loss": 0.7785,
+      "step": 34458
+    },
+    {
+      "epoch": 0.9463898272500069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503646623559485e-07,
+      "loss": 0.7871,
+      "step": 34459
+    },
+    {
+      "epoch": 0.9464172914778501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5021102642699736e-07,
+      "loss": 0.9256,
+      "step": 34460
+    },
+    {
+      "epoch": 0.9464447557056933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500574684333922e-07,
+      "loss": 0.7922,
+      "step": 34461
+    },
+    {
+      "epoch": 0.9464722199335366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4990398837634756e-07,
+      "loss": 0.7757,
+      "step": 34462
+    },
+    {
+      "epoch": 0.9464996841613798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4975058625707806e-07,
+      "loss": 0.7876,
+      "step": 34463
+    },
+    {
+      "epoch": 0.946527148389223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4959726207679715e-07,
+      "loss": 0.6998,
+      "step": 34464
+    },
+    {
+      "epoch": 0.9465546126170663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494440158367183e-07,
+      "loss": 0.8572,
+      "step": 34465
+    },
+    {
+      "epoch": 0.9465820768449095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492908475380539e-07,
+      "loss": 0.8721,
+      "step": 34466
+    },
+    {
+      "epoch": 0.9466095410727527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491377571820174e-07,
+      "loss": 0.7665,
+      "step": 34467
+    },
+    {
+      "epoch": 0.946637005300596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4898474476981783e-07,
+      "loss": 0.8604,
+      "step": 34468
+    },
+    {
+      "epoch": 0.9466644695284392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4883181030266757e-07,
+      "loss": 0.7683,
+      "step": 34469
+    },
+    {
+      "epoch": 0.9466919337562825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4867895378177567e-07,
+      "loss": 0.825,
+      "step": 34470
+    },
+    {
+      "epoch": 0.9467193979841256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4852617520835112e-07,
+      "loss": 0.8623,
+      "step": 34471
+    },
+    {
+      "epoch": 0.9467468622119689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4837347458360406e-07,
+      "loss": 0.7949,
+      "step": 34472
+    },
+    {
+      "epoch": 0.9467743264398122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.482208519087436e-07,
+      "loss": 0.8097,
+      "step": 34473
+    },
+    {
+      "epoch": 0.9468017906676554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4806830718497424e-07,
+      "loss": 0.7498,
+      "step": 34474
+    },
+    {
+      "epoch": 0.9468292548954986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4791584041350503e-07,
+      "loss": 0.752,
+      "step": 34475
+    },
+    {
+      "epoch": 0.9468567191233418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4776345159554174e-07,
+      "loss": 0.7449,
+      "step": 34476
+    },
+    {
+      "epoch": 0.9468841833511851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476111407322911e-07,
+      "loss": 0.7327,
+      "step": 34477
+    },
+    {
+      "epoch": 0.9469116475790283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4745890782495776e-07,
+      "loss": 0.7722,
+      "step": 34478
+    },
+    {
+      "epoch": 0.9469391118068715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4730675287474626e-07,
+      "loss": 0.8444,
+      "step": 34479
+    },
+    {
+      "epoch": 0.9469665760347148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4715467588285904e-07,
+      "loss": 0.7967,
+      "step": 34480
+    },
+    {
+      "epoch": 0.9469940402625581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4700267685050173e-07,
+      "loss": 0.7757,
+      "step": 34481
+    },
+    {
+      "epoch": 0.9470215044904012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4685075577887454e-07,
+      "loss": 0.8787,
+      "step": 34482
+    },
+    {
+      "epoch": 0.9470489687182445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46698912669182e-07,
+      "loss": 0.8389,
+      "step": 34483
+    },
+    {
+      "epoch": 0.9470764329460877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4654714752262544e-07,
+      "loss": 0.8245,
+      "step": 34484
+    },
+    {
+      "epoch": 0.947103897173931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4639546034040275e-07,
+      "loss": 0.6478,
+      "step": 34485
+    },
+    {
+      "epoch": 0.9471313614017742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.462438511237174e-07,
+      "loss": 0.8018,
+      "step": 34486
+    },
+    {
+      "epoch": 0.9471588256296174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4609231987376737e-07,
+      "loss": 0.811,
+      "step": 34487
+    },
+    {
+      "epoch": 0.9471862898574607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4594086659175166e-07,
+      "loss": 0.7311,
+      "step": 34488
+    },
+    {
+      "epoch": 0.9472137540853038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457894912788693e-07,
+      "loss": 0.8001,
+      "step": 34489
+    },
+    {
+      "epoch": 0.9472412183131471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4563819393631828e-07,
+      "loss": 0.7164,
+      "step": 34490
+    },
+    {
+      "epoch": 0.9472686825409904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4548697456529314e-07,
+      "loss": 0.789,
+      "step": 34491
+    },
+    {
+      "epoch": 0.9472961467688336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4533583316699406e-07,
+      "loss": 0.8801,
+      "step": 34492
+    },
+    {
+      "epoch": 0.9473236109966768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4518476974261453e-07,
+      "loss": 0.8604,
+      "step": 34493
+    },
+    {
+      "epoch": 0.9473510752245201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4503378429335023e-07,
+      "loss": 0.917,
+      "step": 34494
+    },
+    {
+      "epoch": 0.9473785394523633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4488287682039693e-07,
+      "loss": 0.8615,
+      "step": 34495
+    },
+    {
+      "epoch": 0.9474060036802066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4473204732494695e-07,
+      "loss": 0.8521,
+      "step": 34496
+    },
+    {
+      "epoch": 0.9474334679080497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445812958081949e-07,
+      "loss": 0.7831,
+      "step": 34497
+    },
+    {
+      "epoch": 0.947460932135893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444306222713332e-07,
+      "loss": 0.7557,
+      "step": 34498
+    },
+    {
+      "epoch": 0.9474883963637363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442800267155542e-07,
+      "loss": 0.828,
+      "step": 34499
+    },
+    {
+      "epoch": 0.9475158605915794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4412950914204915e-07,
+      "loss": 0.8003,
+      "step": 34500
+    },
+    {
+      "epoch": 0.9475433248194227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4397906955200937e-07,
+      "loss": 0.7791,
+      "step": 34501
+    },
+    {
+      "epoch": 0.9475707890472659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4382870794662606e-07,
+      "loss": 0.815,
+      "step": 34502
+    },
+    {
+      "epoch": 0.9475982532751092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4367842432708611e-07,
+      "loss": 0.826,
+      "step": 34503
+    },
+    {
+      "epoch": 0.9476257175029524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4352821869458188e-07,
+      "loss": 0.7273,
+      "step": 34504
+    },
+    {
+      "epoch": 0.9476531817307956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433780910502991e-07,
+      "loss": 0.8875,
+      "step": 34505
+    },
+    {
+      "epoch": 0.9476806459586389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432280413954279e-07,
+      "loss": 0.7979,
+      "step": 34506
+    },
+    {
+      "epoch": 0.9477081101864822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4307806973115401e-07,
+      "loss": 0.6744,
+      "step": 34507
+    },
+    {
+      "epoch": 0.9477355744143253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4292817605866537e-07,
+      "loss": 0.788,
+      "step": 34508
+    },
+    {
+      "epoch": 0.9477630386421686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427783603791466e-07,
+      "loss": 0.7916,
+      "step": 34509
+    },
+    {
+      "epoch": 0.9477905028700118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4262862269378452e-07,
+      "loss": 0.8941,
+      "step": 34510
+    },
+    {
+      "epoch": 0.947817967097855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4247896300376264e-07,
+      "loss": 0.8389,
+      "step": 34511
+    },
+    {
+      "epoch": 0.9478454313256983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4232938131026553e-07,
+      "loss": 0.7434,
+      "step": 34512
+    },
+    {
+      "epoch": 0.9478728955535415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4217987761447671e-07,
+      "loss": 0.7995,
+      "step": 34513
+    },
+    {
+      "epoch": 0.9479003597813848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420304519175797e-07,
+      "loss": 0.8288,
+      "step": 34514
+    },
+    {
+      "epoch": 0.9479278240092279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4188110422075686e-07,
+      "loss": 0.7771,
+      "step": 34515
+    },
+    {
+      "epoch": 0.9479552882370712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4173183452518835e-07,
+      "loss": 0.8533,
+      "step": 34516
+    },
+    {
+      "epoch": 0.9479827524649145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415826428320566e-07,
+      "loss": 0.8068,
+      "step": 34517
+    },
+    {
+      "epoch": 0.9480102166927576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414335291425417e-07,
+      "loss": 0.781,
+      "step": 34518
+    },
+    {
+      "epoch": 0.9480376809206009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412844934578239e-07,
+      "loss": 0.7616,
+      "step": 34519
+    },
+    {
+      "epoch": 0.9480651451484442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4113553577908223e-07,
+      "loss": 0.8531,
+      "step": 34520
+    },
+    {
+      "epoch": 0.9480926093762874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4098665610749464e-07,
+      "loss": 0.7593,
+      "step": 34521
+    },
+    {
+      "epoch": 0.9481200736041306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4083785444424015e-07,
+      "loss": 0.7433,
+      "step": 34522
+    },
+    {
+      "epoch": 0.9481475378319738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4068913079049452e-07,
+      "loss": 0.7447,
+      "step": 34523
+    },
+    {
+      "epoch": 0.9481750020598171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4054048514743678e-07,
+      "loss": 0.8469,
+      "step": 34524
+    },
+    {
+      "epoch": 0.9482024662876604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.403919175162416e-07,
+      "loss": 0.7749,
+      "step": 34525
+    },
+    {
+      "epoch": 0.9482299305155035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4024342789808575e-07,
+      "loss": 0.7327,
+      "step": 34526
+    },
+    {
+      "epoch": 0.9482573947433468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4009501629414168e-07,
+      "loss": 0.8164,
+      "step": 34527
+    },
+    {
+      "epoch": 0.94828485897119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.399466827055851e-07,
+      "loss": 0.8287,
+      "step": 34528
+    },
+    {
+      "epoch": 0.9483123231990332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.397984271335906e-07,
+      "loss": 0.73,
+      "step": 34529
+    },
+    {
+      "epoch": 0.9483397874268765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.396502495793306e-07,
+      "loss": 0.7198,
+      "step": 34530
+    },
+    {
+      "epoch": 0.9483672516547197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3950215004397638e-07,
+      "loss": 0.7996,
+      "step": 34531
+    },
+    {
+      "epoch": 0.948394715882563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3935412852870145e-07,
+      "loss": 0.8522,
+      "step": 34532
+    },
+    {
+      "epoch": 0.9484221801104062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3920618503467599e-07,
+      "loss": 0.7381,
+      "step": 34533
+    },
+    {
+      "epoch": 0.9484496443382494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3905831956307237e-07,
+      "loss": 0.8233,
+      "step": 34534
+    },
+    {
+      "epoch": 0.9484771085660927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3891053211505745e-07,
+      "loss": 0.754,
+      "step": 34535
+    },
+    {
+      "epoch": 0.9485045727939359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387628226918025e-07,
+      "loss": 0.8434,
+      "step": 34536
+    },
+    {
+      "epoch": 0.9485320370217791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3861519129447554e-07,
+      "loss": 0.6985,
+      "step": 34537
+    },
+    {
+      "epoch": 0.9485595012496224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3846763792424667e-07,
+      "loss": 0.7504,
+      "step": 34538
+    },
+    {
+      "epoch": 0.9485869654774656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3832016258228053e-07,
+      "loss": 0.7061,
+      "step": 34539
+    },
+    {
+      "epoch": 0.9486144297053088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381727652697451e-07,
+      "loss": 0.735,
+      "step": 34540
+    },
+    {
+      "epoch": 0.948641893933152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.380254459878083e-07,
+      "loss": 0.8799,
+      "step": 34541
+    },
+    {
+      "epoch": 0.9486693581609953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.378782047376326e-07,
+      "loss": 0.7722,
+      "step": 34542
+    },
+    {
+      "epoch": 0.9486968223888386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3773104152038697e-07,
+      "loss": 0.8235,
+      "step": 34543
+    },
+    {
+      "epoch": 0.9487242866166817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3758395633723166e-07,
+      "loss": 0.8075,
+      "step": 34544
+    },
+    {
+      "epoch": 0.948751750844525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3743694918933237e-07,
+      "loss": 0.8163,
+      "step": 34545
+    },
+    {
+      "epoch": 0.9487792150723683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3729002007785262e-07,
+      "loss": 0.7808,
+      "step": 34546
+    },
+    {
+      "epoch": 0.9488066793002115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3714316900395485e-07,
+      "loss": 0.7212,
+      "step": 34547
+    },
+    {
+      "epoch": 0.9488341435280547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3699639596880142e-07,
+      "loss": 0.8403,
+      "step": 34548
+    },
+    {
+      "epoch": 0.9488616077558979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3684970097355255e-07,
+      "loss": 0.8049,
+      "step": 34549
+    },
+    {
+      "epoch": 0.9488890719837412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.367030840193706e-07,
+      "loss": 0.8124,
+      "step": 34550
+    },
+    {
+      "epoch": 0.9489165362115844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3655654510741356e-07,
+      "loss": 0.7755,
+      "step": 34551
+    },
+    {
+      "epoch": 0.9489440004394276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3641008423884161e-07,
+      "loss": 0.7778,
+      "step": 34552
+    },
+    {
+      "epoch": 0.9489714646672709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3626370141481383e-07,
+      "loss": 0.892,
+      "step": 34553
+    },
+    {
+      "epoch": 0.9489989288951142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3611739663648815e-07,
+      "loss": 0.8808,
+      "step": 34554
+    },
+    {
+      "epoch": 0.9490263931229573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3597116990502368e-07,
+      "loss": 0.8007,
+      "step": 34555
+    },
+    {
+      "epoch": 0.9490538573508006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3582502122157504e-07,
+      "loss": 0.7417,
+      "step": 34556
+    },
+    {
+      "epoch": 0.9490813215786438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3567895058730018e-07,
+      "loss": 0.7565,
+      "step": 34557
+    },
+    {
+      "epoch": 0.949108785806487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3553295800335598e-07,
+      "loss": 0.7978,
+      "step": 34558
+    },
+    {
+      "epoch": 0.9491362500343303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3538704347089483e-07,
+      "loss": 0.7777,
+      "step": 34559
+    },
+    {
+      "epoch": 0.9491637142621735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3524120699107245e-07,
+      "loss": 0.7671,
+      "step": 34560
+    },
+    {
+      "epoch": 0.9491911784900168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3509544856504352e-07,
+      "loss": 0.8332,
+      "step": 34561
+    },
+    {
+      "epoch": 0.9492186427178599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3494976819396044e-07,
+      "loss": 0.7191,
+      "step": 34562
+    },
+    {
+      "epoch": 0.9492461069457032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.348041658789756e-07,
+      "loss": 0.7889,
+      "step": 34563
+    },
+    {
+      "epoch": 0.9492735711735465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3465864162124253e-07,
+      "loss": 0.6688,
+      "step": 34564
+    },
+    {
+      "epoch": 0.9493010354013897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3451319542191143e-07,
+      "loss": 0.8353,
+      "step": 34565
+    },
+    {
+      "epoch": 0.9493284996292329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343678272821336e-07,
+      "loss": 0.8962,
+      "step": 34566
+    },
+    {
+      "epoch": 0.9493559638570762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3422253720305922e-07,
+      "loss": 0.8347,
+      "step": 34567
+    },
+    {
+      "epoch": 0.9493834280849194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340773251858374e-07,
+      "loss": 0.771,
+      "step": 34568
+    },
+    {
+      "epoch": 0.9494108923127627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3393219123161715e-07,
+      "loss": 0.8981,
+      "step": 34569
+    },
+    {
+      "epoch": 0.9494383565406058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3378713534154652e-07,
+      "loss": 0.7554,
+      "step": 34570
+    },
+    {
+      "epoch": 0.9494658207684491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3364215751677455e-07,
+      "loss": 0.784,
+      "step": 34571
+    },
+    {
+      "epoch": 0.9494932849962924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3349725775844812e-07,
+      "loss": 0.8918,
+      "step": 34572
+    },
+    {
+      "epoch": 0.9495207492241355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.333524360677141e-07,
+      "loss": 0.8827,
+      "step": 34573
+    },
+    {
+      "epoch": 0.9495482134519788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3320769244571708e-07,
+      "loss": 0.8161,
+      "step": 34574
+    },
+    {
+      "epoch": 0.949575677679822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3306302689360173e-07,
+      "loss": 0.733,
+      "step": 34575
+    },
+    {
+      "epoch": 0.9496031419076653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.329184394125138e-07,
+      "loss": 0.7976,
+      "step": 34576
+    },
+    {
+      "epoch": 0.9496306061355085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3277393000359796e-07,
+      "loss": 0.8183,
+      "step": 34577
+    },
+    {
+      "epoch": 0.9496580703633517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3262949866799657e-07,
+      "loss": 0.8424,
+      "step": 34578
+    },
+    {
+      "epoch": 0.949685534591195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324851454068543e-07,
+      "loss": 0.8146,
+      "step": 34579
+    },
+    {
+      "epoch": 0.9497129988190383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3234087022131027e-07,
+      "loss": 0.822,
+      "step": 34580
+    },
+    {
+      "epoch": 0.9497404630468814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3219667311250795e-07,
+      "loss": 0.8063,
+      "step": 34581
+    },
+    {
+      "epoch": 0.9497679272747247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3205255408158867e-07,
+      "loss": 0.7439,
+      "step": 34582
+    },
+    {
+      "epoch": 0.9497953915025679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3190851312969154e-07,
+      "loss": 0.818,
+      "step": 34583
+    },
+    {
+      "epoch": 0.9498228557304111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3176455025795676e-07,
+      "loss": 0.7894,
+      "step": 34584
+    },
+    {
+      "epoch": 0.9498503199582544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.316206654675245e-07,
+      "loss": 0.8391,
+      "step": 34585
+    },
+    {
+      "epoch": 0.9498777841860976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3147685875953055e-07,
+      "loss": 0.7638,
+      "step": 34586
+    },
+    {
+      "epoch": 0.9499052484139409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3133313013511507e-07,
+      "loss": 0.8508,
+      "step": 34587
+    },
+    {
+      "epoch": 0.949932712641784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3118947959541494e-07,
+      "loss": 0.8191,
+      "step": 34588
+    },
+    {
+      "epoch": 0.9499601768696273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3104590714156594e-07,
+      "loss": 0.82,
+      "step": 34589
+    },
+    {
+      "epoch": 0.9499876410974706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090241277470494e-07,
+      "loss": 0.7744,
+      "step": 34590
+    },
+    {
+      "epoch": 0.9500151053253137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3075899649596767e-07,
+      "loss": 0.7411,
+      "step": 34591
+    },
+    {
+      "epoch": 0.950042569553157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.306156583064866e-07,
+      "loss": 0.8326,
+      "step": 34592
+    },
+    {
+      "epoch": 0.9500700337810003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3047239820739855e-07,
+      "loss": 0.8,
+      "step": 34593
+    },
+    {
+      "epoch": 0.9500974980088435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3032921619983486e-07,
+      "loss": 0.6774,
+      "step": 34594
+    },
+    {
+      "epoch": 0.9501249622366867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3018611228493017e-07,
+      "loss": 0.9249,
+      "step": 34595
+    },
+    {
+      "epoch": 0.9501524264645299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.300430864638158e-07,
+      "loss": 0.8076,
+      "step": 34596
+    },
+    {
+      "epoch": 0.9501798906923732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.299001387376253e-07,
+      "loss": 0.7392,
+      "step": 34597
+    },
+    {
+      "epoch": 0.9502073549202165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2975726910748775e-07,
+      "loss": 0.7778,
+      "step": 34598
+    },
+    {
+      "epoch": 0.9502348191480596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296144775745334e-07,
+      "loss": 0.7615,
+      "step": 34599
+    },
+    {
+      "epoch": 0.9502622833759029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2947176413989236e-07,
+      "loss": 0.7902,
+      "step": 34600
+    },
+    {
+      "epoch": 0.9502897476037461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2932912880469495e-07,
+      "loss": 0.7624,
+      "step": 34601
+    },
+    {
+      "epoch": 0.9503172118315893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2918657157006798e-07,
+      "loss": 0.8102,
+      "step": 34602
+    },
+    {
+      "epoch": 0.9503446760594326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2904409243714168e-07,
+      "loss": 0.8727,
+      "step": 34603
+    },
+    {
+      "epoch": 0.9503721402872758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2890169140704178e-07,
+      "loss": 0.8132,
+      "step": 34604
+    },
+    {
+      "epoch": 0.9503996045151191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.287593684808952e-07,
+      "loss": 0.8782,
+      "step": 34605
+    },
+    {
+      "epoch": 0.9504270687429623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.286171236598288e-07,
+      "loss": 0.7685,
+      "step": 34606
+    },
+    {
+      "epoch": 0.9504545329708055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2847495694496726e-07,
+      "loss": 0.8328,
+      "step": 34607
+    },
+    {
+      "epoch": 0.9504819971986488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283328683374363e-07,
+      "loss": 0.8322,
+      "step": 34608
+    },
+    {
+      "epoch": 0.950509461426492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2819085783835837e-07,
+      "loss": 0.8945,
+      "step": 34609
+    },
+    {
+      "epoch": 0.9505369256543352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2804892544885817e-07,
+      "loss": 0.7453,
+      "step": 34610
+    },
+    {
+      "epoch": 0.9505643898821785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2790707117005918e-07,
+      "loss": 0.8013,
+      "step": 34611
+    },
+    {
+      "epoch": 0.9505918541100217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2776529500308388e-07,
+      "loss": 0.808,
+      "step": 34612
+    },
+    {
+      "epoch": 0.950619318337865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2762359694905467e-07,
+      "loss": 0.7317,
+      "step": 34613
+    },
+    {
+      "epoch": 0.9506467825657081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.274819770090907e-07,
+      "loss": 0.8884,
+      "step": 34614
+    },
+    {
+      "epoch": 0.9506742467935514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2734043518431439e-07,
+      "loss": 0.8307,
+      "step": 34615
+    },
+    {
+      "epoch": 0.9507017110213947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271989714758437e-07,
+      "loss": 0.8876,
+      "step": 34616
+    },
+    {
+      "epoch": 0.9507291752492378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.270575858847989e-07,
+      "loss": 0.7316,
+      "step": 34617
+    },
+    {
+      "epoch": 0.9507566394770811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2691627841230014e-07,
+      "loss": 0.81,
+      "step": 34618
+    },
+    {
+      "epoch": 0.9507841037049244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2677504905946326e-07,
+      "loss": 0.7511,
+      "step": 34619
+    },
+    {
+      "epoch": 0.9508115679327676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2663389782740732e-07,
+      "loss": 0.9026,
+      "step": 34620
+    },
+    {
+      "epoch": 0.9508390321606108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2649282471724811e-07,
+      "loss": 0.7376,
+      "step": 34621
+    },
+    {
+      "epoch": 0.950866496388454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2635182973010363e-07,
+      "loss": 0.7775,
+      "step": 34622
+    },
+    {
+      "epoch": 0.9508939606162973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262109128670863e-07,
+      "loss": 0.8733,
+      "step": 34623
+    },
+    {
+      "epoch": 0.9509214248441406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2607007412931416e-07,
+      "loss": 0.8075,
+      "step": 34624
+    },
+    {
+      "epoch": 0.9509488890719837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2592931351789962e-07,
+      "loss": 0.7658,
+      "step": 34625
+    },
+    {
+      "epoch": 0.950976353299827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2578863103395734e-07,
+      "loss": 0.7876,
+      "step": 34626
+    },
+    {
+      "epoch": 0.9510038175276702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2564802667860087e-07,
+      "loss": 0.8096,
+      "step": 34627
+    },
+    {
+      "epoch": 0.9510312817555134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2550750045294158e-07,
+      "loss": 0.7013,
+      "step": 34628
+    },
+    {
+      "epoch": 0.9510587459833567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25367052358093e-07,
+      "loss": 0.7963,
+      "step": 34629
+    },
+    {
+      "epoch": 0.9510862102111999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2522668239516421e-07,
+      "loss": 0.8563,
+      "step": 34630
+    },
+    {
+      "epoch": 0.9511136744390432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2508639056526661e-07,
+      "loss": 0.7231,
+      "step": 34631
+    },
+    {
+      "epoch": 0.9511411386668864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.249461768695126e-07,
+      "loss": 0.759,
+      "step": 34632
+    },
+    {
+      "epoch": 0.9511686028947296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2480604130900797e-07,
+      "loss": 0.7805,
+      "step": 34633
+    },
+    {
+      "epoch": 0.9511960671225729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2466598388486406e-07,
+      "loss": 0.8528,
+      "step": 34634
+    },
+    {
+      "epoch": 0.951223531350416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2452600459818774e-07,
+      "loss": 0.8251,
+      "step": 34635
+    },
+    {
+      "epoch": 0.9512509955782593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2438610345008707e-07,
+      "loss": 0.8853,
+      "step": 34636
+    },
+    {
+      "epoch": 0.9512784598061026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2424628044166997e-07,
+      "loss": 0.8305,
+      "step": 34637
+    },
+    {
+      "epoch": 0.9513059240339458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2410653557404118e-07,
+      "loss": 0.8578,
+      "step": 34638
+    },
+    {
+      "epoch": 0.951333388261789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2396686884830645e-07,
+      "loss": 0.7336,
+      "step": 34639
+    },
+    {
+      "epoch": 0.9513608524896322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2382728026557156e-07,
+      "loss": 0.7288,
+      "step": 34640
+    },
+    {
+      "epoch": 0.9513883167174755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236877698269412e-07,
+      "loss": 0.8046,
+      "step": 34641
+    },
+    {
+      "epoch": 0.9514157809453188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2354833753351892e-07,
+      "loss": 0.7492,
+      "step": 34642
+    },
+    {
+      "epoch": 0.9514432451731619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2340898338640829e-07,
+      "loss": 0.7969,
+      "step": 34643
+    },
+    {
+      "epoch": 0.9514707094010052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2326970738671174e-07,
+      "loss": 0.7749,
+      "step": 34644
+    },
+    {
+      "epoch": 0.9514981736288485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2313050953552952e-07,
+      "loss": 0.7336,
+      "step": 34645
+    },
+    {
+      "epoch": 0.9515256378566916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.229913898339674e-07,
+      "loss": 0.7519,
+      "step": 34646
+    },
+    {
+      "epoch": 0.9515531020845349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.228523482831212e-07,
+      "loss": 0.744,
+      "step": 34647
+    },
+    {
+      "epoch": 0.9515805663123781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2271338488409335e-07,
+      "loss": 0.737,
+      "step": 34648
+    },
+    {
+      "epoch": 0.9516080305402214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2257449963798406e-07,
+      "loss": 0.7931,
+      "step": 34649
+    },
+    {
+      "epoch": 0.9516354947680646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.224356925458914e-07,
+      "loss": 0.8167,
+      "step": 34650
+    },
+    {
+      "epoch": 0.9516629589959078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.222969636089133e-07,
+      "loss": 0.8343,
+      "step": 34651
+    },
+    {
+      "epoch": 0.9516904232237511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2215831282814784e-07,
+      "loss": 0.7711,
+      "step": 34652
+    },
+    {
+      "epoch": 0.9517178874515942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22019740204693e-07,
+      "loss": 0.8306,
+      "step": 34653
+    },
+    {
+      "epoch": 0.9517453516794375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2188124573964454e-07,
+      "loss": 0.8064,
+      "step": 34654
+    },
+    {
+      "epoch": 0.9517728159072808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217428294340961e-07,
+      "loss": 0.7818,
+      "step": 34655
+    },
+    {
+      "epoch": 0.951800280135124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2160449128914676e-07,
+      "loss": 0.8034,
+      "step": 34656
+    },
+    {
+      "epoch": 0.9518277443629672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2146623130588898e-07,
+      "loss": 0.7243,
+      "step": 34657
+    },
+    {
+      "epoch": 0.9518552085908105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2132804948541632e-07,
+      "loss": 0.7753,
+      "step": 34658
+    },
+    {
+      "epoch": 0.9518826728186537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2118994582882238e-07,
+      "loss": 0.7713,
+      "step": 34659
+    },
+    {
+      "epoch": 0.951910137046497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210519203372007e-07,
+      "loss": 0.7648,
+      "step": 34660
+    },
+    {
+      "epoch": 0.9519376012743401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2091397301164377e-07,
+      "loss": 0.855,
+      "step": 34661
+    },
+    {
+      "epoch": 0.9519650655021834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2077610385324178e-07,
+      "loss": 0.863,
+      "step": 34662
+    },
+    {
+      "epoch": 0.9519925297300267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2063831286308615e-07,
+      "loss": 0.8,
+      "step": 34663
+    },
+    {
+      "epoch": 0.9520199939578698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2050060004226706e-07,
+      "loss": 0.841,
+      "step": 34664
+    },
+    {
+      "epoch": 0.9520474581857131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2036296539187363e-07,
+      "loss": 0.8334,
+      "step": 34665
+    },
+    {
+      "epoch": 0.9520749224135563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2022540891299506e-07,
+      "loss": 0.7387,
+      "step": 34666
+    },
+    {
+      "epoch": 0.9521023866413996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2008793060672151e-07,
+      "loss": 0.7552,
+      "step": 34667
+    },
+    {
+      "epoch": 0.9521298508692428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1995053047413884e-07,
+      "loss": 0.7931,
+      "step": 34668
+    },
+    {
+      "epoch": 0.952157315097086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1981320851633392e-07,
+      "loss": 0.7848,
+      "step": 34669
+    },
+    {
+      "epoch": 0.9521847793249293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196759647343959e-07,
+      "loss": 0.832,
+      "step": 34670
+    },
+    {
+      "epoch": 0.9522122435527726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1953879912940724e-07,
+      "loss": 0.8369,
+      "step": 34671
+    },
+    {
+      "epoch": 0.9522397077806157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1940171170245595e-07,
+      "loss": 0.8841,
+      "step": 34672
+    },
+    {
+      "epoch": 0.952267172008459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1926470245462562e-07,
+      "loss": 0.7806,
+      "step": 34673
+    },
+    {
+      "epoch": 0.9522946362363022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1912777138699982e-07,
+      "loss": 0.7802,
+      "step": 34674
+    },
+    {
+      "epoch": 0.9523221004641454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1899091850066213e-07,
+      "loss": 0.8222,
+      "step": 34675
+    },
+    {
+      "epoch": 0.9523495646919887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1885414379669612e-07,
+      "loss": 0.8312,
+      "step": 34676
+    },
+    {
+      "epoch": 0.9523770289198319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.187174472761854e-07,
+      "loss": 0.8188,
+      "step": 34677
+    },
+    {
+      "epoch": 0.9524044931476752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1858082894020795e-07,
+      "loss": 0.8527,
+      "step": 34678
+    },
+    {
+      "epoch": 0.9524319573755183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1844428878984737e-07,
+      "loss": 0.7644,
+      "step": 34679
+    },
+    {
+      "epoch": 0.9524594216033616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1830782682618391e-07,
+      "loss": 0.7975,
+      "step": 34680
+    },
+    {
+      "epoch": 0.9524868858312049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.181714430502956e-07,
+      "loss": 0.7386,
+      "step": 34681
+    },
+    {
+      "epoch": 0.952514350059048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1803513746326267e-07,
+      "loss": 0.7992,
+      "step": 34682
+    },
+    {
+      "epoch": 0.9525418142868913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1789891006616428e-07,
+      "loss": 0.7856,
+      "step": 34683
+    },
+    {
+      "epoch": 0.9525692785147346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1776276086007732e-07,
+      "loss": 0.7706,
+      "step": 34684
+    },
+    {
+      "epoch": 0.9525967427425778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1762668984607983e-07,
+      "loss": 0.8152,
+      "step": 34685
+    },
+    {
+      "epoch": 0.952624206970421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1749069702524873e-07,
+      "loss": 0.7257,
+      "step": 34686
+    },
+    {
+      "epoch": 0.9526516711982642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173547823986576e-07,
+      "loss": 0.7221,
+      "step": 34687
+    },
+    {
+      "epoch": 0.9526791354261075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1721894596738448e-07,
+      "loss": 0.8204,
+      "step": 34688
+    },
+    {
+      "epoch": 0.9527065996539508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1708318773250293e-07,
+      "loss": 0.7921,
+      "step": 34689
+    },
+    {
+      "epoch": 0.9527340638817939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1694750769508767e-07,
+      "loss": 0.873,
+      "step": 34690
+    },
+    {
+      "epoch": 0.9527615281096372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1681190585621227e-07,
+      "loss": 0.7416,
+      "step": 34691
+    },
+    {
+      "epoch": 0.9527889923374804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1667638221694921e-07,
+      "loss": 0.8039,
+      "step": 34692
+    },
+    {
+      "epoch": 0.9528164565653237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1654093677837098e-07,
+      "loss": 0.767,
+      "step": 34693
+    },
+    {
+      "epoch": 0.9528439207931669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1640556954155003e-07,
+      "loss": 0.8013,
+      "step": 34694
+    },
+    {
+      "epoch": 0.9528713850210101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1627028050755663e-07,
+      "loss": 0.8351,
+      "step": 34695
+    },
+    {
+      "epoch": 0.9528988492488534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1613506967746103e-07,
+      "loss": 0.8909,
+      "step": 34696
+    },
+    {
+      "epoch": 0.9529263134766967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1599993705233348e-07,
+      "loss": 0.6931,
+      "step": 34697
+    },
+    {
+      "epoch": 0.9529537777045398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1586488263324314e-07,
+      "loss": 0.7849,
+      "step": 34698
+    },
+    {
+      "epoch": 0.9529812419323831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1572990642125803e-07,
+      "loss": 0.8857,
+      "step": 34699
+    },
+    {
+      "epoch": 0.9530087061602263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1559500841744731e-07,
+      "loss": 0.7809,
+      "step": 34700
+    },
+    {
+      "epoch": 0.9530361703880695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1546018862287789e-07,
+      "loss": 0.7664,
+      "step": 34701
+    },
+    {
+      "epoch": 0.9530636346159128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1532544703861559e-07,
+      "loss": 0.751,
+      "step": 34702
+    },
+    {
+      "epoch": 0.953091098843756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1519078366572844e-07,
+      "loss": 0.6834,
+      "step": 34703
+    },
+    {
+      "epoch": 0.9531185630715993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1505619850528005e-07,
+      "loss": 0.7182,
+      "step": 34704
+    },
+    {
+      "epoch": 0.9531460272994424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.149216915583351e-07,
+      "loss": 0.8526,
+      "step": 34705
+    },
+    {
+      "epoch": 0.9531734915272857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1478726282596053e-07,
+      "loss": 0.7222,
+      "step": 34706
+    },
+    {
+      "epoch": 0.953200955755129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.146529123092166e-07,
+      "loss": 0.8146,
+      "step": 34707
+    },
+    {
+      "epoch": 0.9532284199829721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1451864000916913e-07,
+      "loss": 0.8739,
+      "step": 34708
+    },
+    {
+      "epoch": 0.9532558842108154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1438444592688059e-07,
+      "loss": 0.8305,
+      "step": 34709
+    },
+    {
+      "epoch": 0.9532833484386587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1425033006341013e-07,
+      "loss": 0.8289,
+      "step": 34710
+    },
+    {
+      "epoch": 0.9533108126665019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1411629241982136e-07,
+      "loss": 0.809,
+      "step": 34711
+    },
+    {
+      "epoch": 0.9533382768943451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1398233299717342e-07,
+      "loss": 0.7151,
+      "step": 34712
+    },
+    {
+      "epoch": 0.9533657411221883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1384845179652548e-07,
+      "loss": 0.8228,
+      "step": 34713
+    },
+    {
+      "epoch": 0.9533932053500316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1371464881894e-07,
+      "loss": 0.8729,
+      "step": 34714
+    },
+    {
+      "epoch": 0.9534206695778749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1358092406547283e-07,
+      "loss": 0.7809,
+      "step": 34715
+    },
+    {
+      "epoch": 0.953448133805718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.134472775371831e-07,
+      "loss": 0.8304,
+      "step": 34716
+    },
+    {
+      "epoch": 0.9534755980335613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1331370923512885e-07,
+      "loss": 0.7511,
+      "step": 34717
+    },
+    {
+      "epoch": 0.9535030622614045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1318021916036593e-07,
+      "loss": 0.8317,
+      "step": 34718
+    },
+    {
+      "epoch": 0.9535305264892477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1304680731395012e-07,
+      "loss": 0.7707,
+      "step": 34719
+    },
+    {
+      "epoch": 0.953557990717091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1291347369693839e-07,
+      "loss": 0.8061,
+      "step": 34720
+    },
+    {
+      "epoch": 0.9535854549449342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1278021831038655e-07,
+      "loss": 0.8125,
+      "step": 34721
+    },
+    {
+      "epoch": 0.9536129191727775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1264704115534597e-07,
+      "loss": 0.7803,
+      "step": 34722
+    },
+    {
+      "epoch": 0.9536403834006207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1251394223287138e-07,
+      "loss": 0.8331,
+      "step": 34723
+    },
+    {
+      "epoch": 0.9536678476284639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1238092154401747e-07,
+      "loss": 0.8616,
+      "step": 34724
+    },
+    {
+      "epoch": 0.9536953118563072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1224797908983675e-07,
+      "loss": 0.76,
+      "step": 34725
+    },
+    {
+      "epoch": 0.9537227760841503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1211511487137838e-07,
+      "loss": 0.8455,
+      "step": 34726
+    },
+    {
+      "epoch": 0.9537502403119936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1198232888969707e-07,
+      "loss": 0.84,
+      "step": 34727
+    },
+    {
+      "epoch": 0.9537777045398369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1184962114584086e-07,
+      "loss": 0.8707,
+      "step": 34728
+    },
+    {
+      "epoch": 0.9538051687676801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1171699164086003e-07,
+      "loss": 0.8671,
+      "step": 34729
+    },
+    {
+      "epoch": 0.9538326329955233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1158444037580596e-07,
+      "loss": 0.7797,
+      "step": 34730
+    },
+    {
+      "epoch": 0.9538600972233665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1145196735172558e-07,
+      "loss": 0.8134,
+      "step": 34731
+    },
+    {
+      "epoch": 0.9538875614512098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1131957256966808e-07,
+      "loss": 0.7513,
+      "step": 34732
+    },
+    {
+      "epoch": 0.9539150256790531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1118725603068037e-07,
+      "loss": 0.7754,
+      "step": 34733
+    },
+    {
+      "epoch": 0.9539424899068962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1105501773581051e-07,
+      "loss": 0.7847,
+      "step": 34734
+    },
+    {
+      "epoch": 0.9539699541347395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1092285768610212e-07,
+      "loss": 0.735,
+      "step": 34735
+    },
+    {
+      "epoch": 0.9539974183625828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1079077588260434e-07,
+      "loss": 0.818,
+      "step": 34736
+    },
+    {
+      "epoch": 0.954024882590426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1065877232635968e-07,
+      "loss": 0.7919,
+      "step": 34737
+    },
+    {
+      "epoch": 0.9540523468182692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1052684701841399e-07,
+      "loss": 0.7678,
+      "step": 34738
+    },
+    {
+      "epoch": 0.9540798110461124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1039499995981084e-07,
+      "loss": 0.7585,
+      "step": 34739
+    },
+    {
+      "epoch": 0.9541072752739557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1026323115159276e-07,
+      "loss": 0.8259,
+      "step": 34740
+    },
+    {
+      "epoch": 0.954134739501799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1013154059480335e-07,
+      "loss": 0.8713,
+      "step": 34741
+    },
+    {
+      "epoch": 0.9541622037296421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0999992829048511e-07,
+      "loss": 0.8289,
+      "step": 34742
+    },
+    {
+      "epoch": 0.9541896679574854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0986839423967721e-07,
+      "loss": 0.8542,
+      "step": 34743
+    },
+    {
+      "epoch": 0.9542171321853287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0973693844342214e-07,
+      "loss": 0.7898,
+      "step": 34744
+    },
+    {
+      "epoch": 0.9542445964131718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0960556090275909e-07,
+      "loss": 0.7512,
+      "step": 34745
+    },
+    {
+      "epoch": 0.9542720606410151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0947426161872832e-07,
+      "loss": 0.7332,
+      "step": 34746
+    },
+    {
+      "epoch": 0.9542995248688583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0934304059236789e-07,
+      "loss": 0.7781,
+      "step": 34747
+    },
+    {
+      "epoch": 0.9543269890967016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0921189782471697e-07,
+      "loss": 0.8296,
+      "step": 34748
+    },
+    {
+      "epoch": 0.9543544533245448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0908083331681363e-07,
+      "loss": 0.8221,
+      "step": 34749
+    },
+    {
+      "epoch": 0.954381917552388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0894984706969369e-07,
+      "loss": 0.7504,
+      "step": 34750
+    },
+    {
+      "epoch": 0.9544093817802313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0881893908439301e-07,
+      "loss": 0.7742,
+      "step": 34751
+    },
+    {
+      "epoch": 0.9544368460080744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0868810936194962e-07,
+      "loss": 0.7951,
+      "step": 34752
+    },
+    {
+      "epoch": 0.9544643102359177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0855735790339605e-07,
+      "loss": 0.6956,
+      "step": 34753
+    },
+    {
+      "epoch": 0.954491774463761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0842668470976925e-07,
+      "loss": 0.8996,
+      "step": 34754
+    },
+    {
+      "epoch": 0.9545192386916042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0829608978210172e-07,
+      "loss": 0.8232,
+      "step": 34755
+    },
+    {
+      "epoch": 0.9545467029194474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0816557312142817e-07,
+      "loss": 0.7595,
+      "step": 34756
+    },
+    {
+      "epoch": 0.9545741671472907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0803513472877891e-07,
+      "loss": 0.7955,
+      "step": 34757
+    },
+    {
+      "epoch": 0.9546016313751339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0790477460518867e-07,
+      "loss": 0.7557,
+      "step": 34758
+    },
+    {
+      "epoch": 0.9546290956029772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0777449275168772e-07,
+      "loss": 0.7929,
+      "step": 34759
+    },
+    {
+      "epoch": 0.9546565598308203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0764428916930525e-07,
+      "loss": 0.8319,
+      "step": 34760
+    },
+    {
+      "epoch": 0.9546840240586636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0751416385907487e-07,
+      "loss": 0.7976,
+      "step": 34761
+    },
+    {
+      "epoch": 0.9547114882865069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0738411682202355e-07,
+      "loss": 0.7872,
+      "step": 34762
+    },
+    {
+      "epoch": 0.95473895251435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0725414805918155e-07,
+      "loss": 0.7575,
+      "step": 34763
+    },
+    {
+      "epoch": 0.9547664167421933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0712425757157585e-07,
+      "loss": 0.8113,
+      "step": 34764
+    },
+    {
+      "epoch": 0.9547938809700365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0699444536023562e-07,
+      "loss": 0.8268,
+      "step": 34765
+    },
+    {
+      "epoch": 0.9548213451978798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0686471142618892e-07,
+      "loss": 0.7392,
+      "step": 34766
+    },
+    {
+      "epoch": 0.954848809425723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0673505577045939e-07,
+      "loss": 0.889,
+      "step": 34767
+    },
+    {
+      "epoch": 0.9548762736535662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0660547839407508e-07,
+      "loss": 0.798,
+      "step": 34768
+    },
+    {
+      "epoch": 0.9549037378814095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0647597929805964e-07,
+      "loss": 0.767,
+      "step": 34769
+    },
+    {
+      "epoch": 0.9549312021092528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0634655848343888e-07,
+      "loss": 0.7767,
+      "step": 34770
+    },
+    {
+      "epoch": 0.9549586663370959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0621721595123647e-07,
+      "loss": 0.8277,
+      "step": 34771
+    },
+    {
+      "epoch": 0.9549861305649392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.060879517024771e-07,
+      "loss": 0.8592,
+      "step": 34772
+    },
+    {
+      "epoch": 0.9550135947927824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059587657381822e-07,
+      "loss": 0.8291,
+      "step": 34773
+    },
+    {
+      "epoch": 0.9550410590206256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0582965805937318e-07,
+      "loss": 0.8274,
+      "step": 34774
+    },
+    {
+      "epoch": 0.9550685232484689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0570062866707254e-07,
+      "loss": 0.7582,
+      "step": 34775
+    },
+    {
+      "epoch": 0.9550959874763121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0557167756230169e-07,
+      "loss": 0.7424,
+      "step": 34776
+    },
+    {
+      "epoch": 0.9551234517041554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0544280474608093e-07,
+      "loss": 0.8456,
+      "step": 34777
+    },
+    {
+      "epoch": 0.9551509159319985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0531401021942833e-07,
+      "loss": 0.7888,
+      "step": 34778
+    },
+    {
+      "epoch": 0.9551783801598418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0518529398336419e-07,
+      "loss": 0.8002,
+      "step": 34779
+    },
+    {
+      "epoch": 0.9552058443876851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0505665603890769e-07,
+      "loss": 0.8522,
+      "step": 34780
+    },
+    {
+      "epoch": 0.9552333086155282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049280963870758e-07,
+      "loss": 0.8,
+      "step": 34781
+    },
+    {
+      "epoch": 0.9552607728433715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0479961502888547e-07,
+      "loss": 0.7174,
+      "step": 34782
+    },
+    {
+      "epoch": 0.9552882370712148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0467121196535368e-07,
+      "loss": 0.8467,
+      "step": 34783
+    },
+    {
+      "epoch": 0.955315701299058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0454288719749628e-07,
+      "loss": 0.8741,
+      "step": 34784
+    },
+    {
+      "epoch": 0.9553431655269012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0441464072632911e-07,
+      "loss": 0.8215,
+      "step": 34785
+    },
+    {
+      "epoch": 0.9553706297547444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0428647255286583e-07,
+      "loss": 0.7651,
+      "step": 34786
+    },
+    {
+      "epoch": 0.9553980939825877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0415838267812228e-07,
+      "loss": 0.7564,
+      "step": 34787
+    },
+    {
+      "epoch": 0.955425558210431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0403037110310987e-07,
+      "loss": 0.7419,
+      "step": 34788
+    },
+    {
+      "epoch": 0.9554530224382741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0390243782884335e-07,
+      "loss": 0.7565,
+      "step": 34789
+    },
+    {
+      "epoch": 0.9554804866661174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0377458285633302e-07,
+      "loss": 0.7948,
+      "step": 34790
+    },
+    {
+      "epoch": 0.9555079508939606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.036468061865925e-07,
+      "loss": 0.7482,
+      "step": 34791
+    },
+    {
+      "epoch": 0.9555354151218038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0351910782063102e-07,
+      "loss": 0.7033,
+      "step": 34792
+    },
+    {
+      "epoch": 0.9555628793496471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0339148775946107e-07,
+      "loss": 0.7428,
+      "step": 34793
+    },
+    {
+      "epoch": 0.9555903435774903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0326394600408963e-07,
+      "loss": 0.7468,
+      "step": 34794
+    },
+    {
+      "epoch": 0.9556178078053336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0313648255552921e-07,
+      "loss": 0.7341,
+      "step": 34795
+    },
+    {
+      "epoch": 0.9556452720331768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0300909741478571e-07,
+      "loss": 0.8565,
+      "step": 34796
+    },
+    {
+      "epoch": 0.95567273626102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0288179058286829e-07,
+      "loss": 0.8236,
+      "step": 34797
+    },
+    {
+      "epoch": 0.9557002004888633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0275456206078394e-07,
+      "loss": 0.7473,
+      "step": 34798
+    },
+    {
+      "epoch": 0.9557276647167064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0262741184953961e-07,
+      "loss": 0.7416,
+      "step": 34799
+    },
+    {
+      "epoch": 0.9557551289445497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0250033995014008e-07,
+      "loss": 0.8529,
+      "step": 34800
+    },
+    {
+      "epoch": 0.955782593172393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.023733463635923e-07,
+      "loss": 0.8704,
+      "step": 34801
+    },
+    {
+      "epoch": 0.9558100574002362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0224643109089994e-07,
+      "loss": 0.8314,
+      "step": 34802
+    },
+    {
+      "epoch": 0.9558375216280794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0211959413306882e-07,
+      "loss": 0.7939,
+      "step": 34803
+    },
+    {
+      "epoch": 0.9558649858559226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019928354911015e-07,
+      "loss": 0.864,
+      "step": 34804
+    },
+    {
+      "epoch": 0.9558924500837659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0186615516600051e-07,
+      "loss": 0.8196,
+      "step": 34805
+    },
+    {
+      "epoch": 0.9559199143116092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0173955315876949e-07,
+      "loss": 0.715,
+      "step": 34806
+    },
+    {
+      "epoch": 0.9559473785394523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0161302947040874e-07,
+      "loss": 0.8608,
+      "step": 34807
+    },
+    {
+      "epoch": 0.9559748427672956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0148658410191969e-07,
+      "loss": 0.7935,
+      "step": 34808
+    },
+    {
+      "epoch": 0.9560023069951389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0136021705430377e-07,
+      "loss": 0.7863,
+      "step": 34809
+    },
+    {
+      "epoch": 0.956029771222982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0123392832855905e-07,
+      "loss": 0.8016,
+      "step": 34810
+    },
+    {
+      "epoch": 0.9560572354508253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0110771792568697e-07,
+      "loss": 0.8118,
+      "step": 34811
+    },
+    {
+      "epoch": 0.9560846996786685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.009815858466845e-07,
+      "loss": 0.8269,
+      "step": 34812
+    },
+    {
+      "epoch": 0.9561121639065118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0085553209255084e-07,
+      "loss": 0.7897,
+      "step": 34813
+    },
+    {
+      "epoch": 0.956139628134355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0072955666428186e-07,
+      "loss": 0.833,
+      "step": 34814
+    },
+    {
+      "epoch": 0.9561670923621982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0060365956287677e-07,
+      "loss": 0.8071,
+      "step": 34815
+    },
+    {
+      "epoch": 0.9561945565900415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0047784078932921e-07,
+      "loss": 0.7711,
+      "step": 34816
+    },
+    {
+      "epoch": 0.9562220208178847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0035210034463505e-07,
+      "loss": 0.7951,
+      "step": 34817
+    },
+    {
+      "epoch": 0.9562494850457279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0022643822979017e-07,
+      "loss": 0.8686,
+      "step": 34818
+    },
+    {
+      "epoch": 0.9562769492735712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.001008544457882e-07,
+      "loss": 0.7355,
+      "step": 34819
+    },
+    {
+      "epoch": 0.9563044135014144,
+      "grad_norm": 0.0,
+      "learning_rate": 9.997534899362394e-08,
+      "loss": 0.8599,
+      "step": 34820
+    },
+    {
+      "epoch": 0.9563318777292577,
+      "grad_norm": 0.0,
+      "learning_rate": 9.984992187428987e-08,
+      "loss": 0.8496,
+      "step": 34821
+    },
+    {
+      "epoch": 0.9563593419571009,
+      "grad_norm": 0.0,
+      "learning_rate": 9.972457308877858e-08,
+      "loss": 0.83,
+      "step": 34822
+    },
+    {
+      "epoch": 0.9563868061849441,
+      "grad_norm": 0.0,
+      "learning_rate": 9.959930263808038e-08,
+      "loss": 0.8179,
+      "step": 34823
+    },
+    {
+      "epoch": 0.9564142704127874,
+      "grad_norm": 0.0,
+      "learning_rate": 9.947411052318779e-08,
+      "loss": 0.7941,
+      "step": 34824
+    },
+    {
+      "epoch": 0.9564417346406305,
+      "grad_norm": 0.0,
+      "learning_rate": 9.934899674509001e-08,
+      "loss": 0.8093,
+      "step": 34825
+    },
+    {
+      "epoch": 0.9564691988684738,
+      "grad_norm": 0.0,
+      "learning_rate": 9.922396130477963e-08,
+      "loss": 0.8215,
+      "step": 34826
+    },
+    {
+      "epoch": 0.9564966630963171,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90990042032447e-08,
+      "loss": 0.8347,
+      "step": 34827
+    },
+    {
+      "epoch": 0.9565241273241603,
+      "grad_norm": 0.0,
+      "learning_rate": 9.897412544147222e-08,
+      "loss": 0.816,
+      "step": 34828
+    },
+    {
+      "epoch": 0.9565515915520035,
+      "grad_norm": 0.0,
+      "learning_rate": 9.884932502045253e-08,
+      "loss": 0.8198,
+      "step": 34829
+    },
+    {
+      "epoch": 0.9565790557798467,
+      "grad_norm": 0.0,
+      "learning_rate": 9.872460294117259e-08,
+      "loss": 0.8128,
+      "step": 34830
+    },
+    {
+      "epoch": 0.95660652000769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.85999592046194e-08,
+      "loss": 0.7463,
+      "step": 34831
+    },
+    {
+      "epoch": 0.9566339842355333,
+      "grad_norm": 0.0,
+      "learning_rate": 9.847539381177772e-08,
+      "loss": 0.7216,
+      "step": 34832
+    },
+    {
+      "epoch": 0.9566614484633764,
+      "grad_norm": 0.0,
+      "learning_rate": 9.835090676363457e-08,
+      "loss": 0.8434,
+      "step": 34833
+    },
+    {
+      "epoch": 0.9566889126912197,
+      "grad_norm": 0.0,
+      "learning_rate": 9.822649806117468e-08,
+      "loss": 0.7774,
+      "step": 34834
+    },
+    {
+      "epoch": 0.956716376919063,
+      "grad_norm": 0.0,
+      "learning_rate": 9.810216770538172e-08,
+      "loss": 0.7598,
+      "step": 34835
+    },
+    {
+      "epoch": 0.9567438411469061,
+      "grad_norm": 0.0,
+      "learning_rate": 9.797791569724047e-08,
+      "loss": 0.7609,
+      "step": 34836
+    },
+    {
+      "epoch": 0.9567713053747494,
+      "grad_norm": 0.0,
+      "learning_rate": 9.785374203773456e-08,
+      "loss": 0.7099,
+      "step": 34837
+    },
+    {
+      "epoch": 0.9567987696025926,
+      "grad_norm": 0.0,
+      "learning_rate": 9.772964672784435e-08,
+      "loss": 0.7249,
+      "step": 34838
+    },
+    {
+      "epoch": 0.9568262338304359,
+      "grad_norm": 0.0,
+      "learning_rate": 9.760562976855348e-08,
+      "loss": 0.8265,
+      "step": 34839
+    },
+    {
+      "epoch": 0.9568536980582791,
+      "grad_norm": 0.0,
+      "learning_rate": 9.748169116084227e-08,
+      "loss": 0.7895,
+      "step": 34840
+    },
+    {
+      "epoch": 0.9568811622861223,
+      "grad_norm": 0.0,
+      "learning_rate": 9.735783090569217e-08,
+      "loss": 0.7078,
+      "step": 34841
+    },
+    {
+      "epoch": 0.9569086265139656,
+      "grad_norm": 0.0,
+      "learning_rate": 9.72340490040824e-08,
+      "loss": 0.8491,
+      "step": 34842
+    },
+    {
+      "epoch": 0.9569360907418087,
+      "grad_norm": 0.0,
+      "learning_rate": 9.711034545699328e-08,
+      "loss": 0.7454,
+      "step": 34843
+    },
+    {
+      "epoch": 0.956963554969652,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69867202654029e-08,
+      "loss": 0.7359,
+      "step": 34844
+    },
+    {
+      "epoch": 0.9569910191974953,
+      "grad_norm": 0.0,
+      "learning_rate": 9.686317343028939e-08,
+      "loss": 0.7254,
+      "step": 34845
+    },
+    {
+      "epoch": 0.9570184834253385,
+      "grad_norm": 0.0,
+      "learning_rate": 9.673970495263197e-08,
+      "loss": 0.7748,
+      "step": 34846
+    },
+    {
+      "epoch": 0.9570459476531817,
+      "grad_norm": 0.0,
+      "learning_rate": 9.661631483340428e-08,
+      "loss": 0.8432,
+      "step": 34847
+    },
+    {
+      "epoch": 0.957073411881025,
+      "grad_norm": 0.0,
+      "learning_rate": 9.649300307358555e-08,
+      "loss": 0.8356,
+      "step": 34848
+    },
+    {
+      "epoch": 0.9571008761088682,
+      "grad_norm": 0.0,
+      "learning_rate": 9.636976967414946e-08,
+      "loss": 0.799,
+      "step": 34849
+    },
+    {
+      "epoch": 0.9571283403367115,
+      "grad_norm": 0.0,
+      "learning_rate": 9.62466146360741e-08,
+      "loss": 0.7789,
+      "step": 34850
+    },
+    {
+      "epoch": 0.9571558045645546,
+      "grad_norm": 0.0,
+      "learning_rate": 9.612353796032981e-08,
+      "loss": 0.7626,
+      "step": 34851
+    },
+    {
+      "epoch": 0.9571832687923979,
+      "grad_norm": 0.0,
+      "learning_rate": 9.600053964789358e-08,
+      "loss": 0.7128,
+      "step": 34852
+    },
+    {
+      "epoch": 0.9572107330202412,
+      "grad_norm": 0.0,
+      "learning_rate": 9.587761969973686e-08,
+      "loss": 0.657,
+      "step": 34853
+    },
+    {
+      "epoch": 0.9572381972480843,
+      "grad_norm": 0.0,
+      "learning_rate": 9.575477811683331e-08,
+      "loss": 0.8437,
+      "step": 34854
+    },
+    {
+      "epoch": 0.9572656614759276,
+      "grad_norm": 0.0,
+      "learning_rate": 9.56320149001544e-08,
+      "loss": 0.7061,
+      "step": 34855
+    },
+    {
+      "epoch": 0.9572931257037708,
+      "grad_norm": 0.0,
+      "learning_rate": 9.550933005067154e-08,
+      "loss": 0.7802,
+      "step": 34856
+    },
+    {
+      "epoch": 0.9573205899316141,
+      "grad_norm": 0.0,
+      "learning_rate": 9.538672356935619e-08,
+      "loss": 0.8885,
+      "step": 34857
+    },
+    {
+      "epoch": 0.9573480541594573,
+      "grad_norm": 0.0,
+      "learning_rate": 9.526419545717647e-08,
+      "loss": 0.8824,
+      "step": 34858
+    },
+    {
+      "epoch": 0.9573755183873005,
+      "grad_norm": 0.0,
+      "learning_rate": 9.514174571510381e-08,
+      "loss": 0.764,
+      "step": 34859
+    },
+    {
+      "epoch": 0.9574029826151438,
+      "grad_norm": 0.0,
+      "learning_rate": 9.501937434410635e-08,
+      "loss": 0.9089,
+      "step": 34860
+    },
+    {
+      "epoch": 0.9574304468429871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.489708134515218e-08,
+      "loss": 0.874,
+      "step": 34861
+    },
+    {
+      "epoch": 0.9574579110708302,
+      "grad_norm": 0.0,
+      "learning_rate": 9.477486671920943e-08,
+      "loss": 0.7696,
+      "step": 34862
+    },
+    {
+      "epoch": 0.9574853752986735,
+      "grad_norm": 0.0,
+      "learning_rate": 9.46527304672451e-08,
+      "loss": 0.8312,
+      "step": 34863
+    },
+    {
+      "epoch": 0.9575128395265167,
+      "grad_norm": 0.0,
+      "learning_rate": 9.453067259022508e-08,
+      "loss": 0.8762,
+      "step": 34864
+    },
+    {
+      "epoch": 0.95754030375436,
+      "grad_norm": 0.0,
+      "learning_rate": 9.440869308911527e-08,
+      "loss": 0.8747,
+      "step": 34865
+    },
+    {
+      "epoch": 0.9575677679822032,
+      "grad_norm": 0.0,
+      "learning_rate": 9.428679196488155e-08,
+      "loss": 0.7778,
+      "step": 34866
+    },
+    {
+      "epoch": 0.9575952322100464,
+      "grad_norm": 0.0,
+      "learning_rate": 9.41649692184865e-08,
+      "loss": 0.8001,
+      "step": 34867
+    },
+    {
+      "epoch": 0.9576226964378897,
+      "grad_norm": 0.0,
+      "learning_rate": 9.404322485089712e-08,
+      "loss": 0.7552,
+      "step": 34868
+    },
+    {
+      "epoch": 0.9576501606657328,
+      "grad_norm": 0.0,
+      "learning_rate": 9.392155886307375e-08,
+      "loss": 0.9247,
+      "step": 34869
+    },
+    {
+      "epoch": 0.9576776248935761,
+      "grad_norm": 0.0,
+      "learning_rate": 9.379997125598117e-08,
+      "loss": 0.7542,
+      "step": 34870
+    },
+    {
+      "epoch": 0.9577050891214194,
+      "grad_norm": 0.0,
+      "learning_rate": 9.367846203057974e-08,
+      "loss": 0.7891,
+      "step": 34871
+    },
+    {
+      "epoch": 0.9577325533492625,
+      "grad_norm": 0.0,
+      "learning_rate": 9.355703118783199e-08,
+      "loss": 0.7465,
+      "step": 34872
+    },
+    {
+      "epoch": 0.9577600175771058,
+      "grad_norm": 0.0,
+      "learning_rate": 9.343567872869719e-08,
+      "loss": 0.7811,
+      "step": 34873
+    },
+    {
+      "epoch": 0.9577874818049491,
+      "grad_norm": 0.0,
+      "learning_rate": 9.331440465413899e-08,
+      "loss": 0.8748,
+      "step": 34874
+    },
+    {
+      "epoch": 0.9578149460327923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.31932089651133e-08,
+      "loss": 0.7011,
+      "step": 34875
+    },
+    {
+      "epoch": 0.9578424102606355,
+      "grad_norm": 0.0,
+      "learning_rate": 9.307209166258047e-08,
+      "loss": 0.7798,
+      "step": 34876
+    },
+    {
+      "epoch": 0.9578698744884787,
+      "grad_norm": 0.0,
+      "learning_rate": 9.295105274749861e-08,
+      "loss": 0.8223,
+      "step": 34877
+    },
+    {
+      "epoch": 0.957897338716322,
+      "grad_norm": 0.0,
+      "learning_rate": 9.283009222082695e-08,
+      "loss": 0.6539,
+      "step": 34878
+    },
+    {
+      "epoch": 0.9579248029441653,
+      "grad_norm": 0.0,
+      "learning_rate": 9.270921008351919e-08,
+      "loss": 0.8328,
+      "step": 34879
+    },
+    {
+      "epoch": 0.9579522671720084,
+      "grad_norm": 0.0,
+      "learning_rate": 9.258840633653565e-08,
+      "loss": 0.7256,
+      "step": 34880
+    },
+    {
+      "epoch": 0.9579797313998517,
+      "grad_norm": 0.0,
+      "learning_rate": 9.246768098083003e-08,
+      "loss": 0.7984,
+      "step": 34881
+    },
+    {
+      "epoch": 0.9580071956276949,
+      "grad_norm": 0.0,
+      "learning_rate": 9.234703401735711e-08,
+      "loss": 0.7296,
+      "step": 34882
+    },
+    {
+      "epoch": 0.9580346598555382,
+      "grad_norm": 0.0,
+      "learning_rate": 9.22264654470728e-08,
+      "loss": 0.7221,
+      "step": 34883
+    },
+    {
+      "epoch": 0.9580621240833814,
+      "grad_norm": 0.0,
+      "learning_rate": 9.210597527093079e-08,
+      "loss": 0.8318,
+      "step": 34884
+    },
+    {
+      "epoch": 0.9580895883112246,
+      "grad_norm": 0.0,
+      "learning_rate": 9.198556348988474e-08,
+      "loss": 0.8185,
+      "step": 34885
+    },
+    {
+      "epoch": 0.9581170525390679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.186523010488613e-08,
+      "loss": 0.7116,
+      "step": 34886
+    },
+    {
+      "epoch": 0.9581445167669111,
+      "grad_norm": 0.0,
+      "learning_rate": 9.174497511688862e-08,
+      "loss": 0.807,
+      "step": 34887
+    },
+    {
+      "epoch": 0.9581719809947543,
+      "grad_norm": 0.0,
+      "learning_rate": 9.162479852684258e-08,
+      "loss": 0.8508,
+      "step": 34888
+    },
+    {
+      "epoch": 0.9581994452225976,
+      "grad_norm": 0.0,
+      "learning_rate": 9.150470033569836e-08,
+      "loss": 0.838,
+      "step": 34889
+    },
+    {
+      "epoch": 0.9582269094504408,
+      "grad_norm": 0.0,
+      "learning_rate": 9.138468054440852e-08,
+      "loss": 0.8949,
+      "step": 34890
+    },
+    {
+      "epoch": 0.958254373678284,
+      "grad_norm": 0.0,
+      "learning_rate": 9.126473915392008e-08,
+      "loss": 0.8526,
+      "step": 34891
+    },
+    {
+      "epoch": 0.9582818379061273,
+      "grad_norm": 0.0,
+      "learning_rate": 9.114487616518453e-08,
+      "loss": 0.8382,
+      "step": 34892
+    },
+    {
+      "epoch": 0.9583093021339705,
+      "grad_norm": 0.0,
+      "learning_rate": 9.102509157914774e-08,
+      "loss": 0.8234,
+      "step": 34893
+    },
+    {
+      "epoch": 0.9583367663618138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.090538539676008e-08,
+      "loss": 0.859,
+      "step": 34894
+    },
+    {
+      "epoch": 0.9583642305896569,
+      "grad_norm": 0.0,
+      "learning_rate": 9.078575761896747e-08,
+      "loss": 0.7827,
+      "step": 34895
+    },
+    {
+      "epoch": 0.9583916948175002,
+      "grad_norm": 0.0,
+      "learning_rate": 9.066620824671468e-08,
+      "loss": 0.777,
+      "step": 34896
+    },
+    {
+      "epoch": 0.9584191590453435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.054673728095098e-08,
+      "loss": 0.6899,
+      "step": 34897
+    },
+    {
+      "epoch": 0.9584466232731866,
+      "grad_norm": 0.0,
+      "learning_rate": 9.042734472262004e-08,
+      "loss": 0.8084,
+      "step": 34898
+    },
+    {
+      "epoch": 0.9584740875010299,
+      "grad_norm": 0.0,
+      "learning_rate": 9.030803057266558e-08,
+      "loss": 0.8196,
+      "step": 34899
+    },
+    {
+      "epoch": 0.9585015517288732,
+      "grad_norm": 0.0,
+      "learning_rate": 9.018879483203236e-08,
+      "loss": 0.7146,
+      "step": 34900
+    },
+    {
+      "epoch": 0.9585290159567164,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00696375016652e-08,
+      "loss": 0.8489,
+      "step": 34901
+    },
+    {
+      "epoch": 0.9585564801845596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.995055858250557e-08,
+      "loss": 0.8698,
+      "step": 34902
+    },
+    {
+      "epoch": 0.9585839444124028,
+      "grad_norm": 0.0,
+      "learning_rate": 8.983155807549604e-08,
+      "loss": 0.8025,
+      "step": 34903
+    },
+    {
+      "epoch": 0.9586114086402461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.971263598157809e-08,
+      "loss": 0.7842,
+      "step": 34904
+    },
+    {
+      "epoch": 0.9586388728680894,
+      "grad_norm": 0.0,
+      "learning_rate": 8.959379230169207e-08,
+      "loss": 0.8203,
+      "step": 34905
+    },
+    {
+      "epoch": 0.9586663370959325,
+      "grad_norm": 0.0,
+      "learning_rate": 8.947502703677946e-08,
+      "loss": 0.7676,
+      "step": 34906
+    },
+    {
+      "epoch": 0.9586938013237758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.935634018777839e-08,
+      "loss": 0.8555,
+      "step": 34907
+    },
+    {
+      "epoch": 0.958721265551619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.923773175562922e-08,
+      "loss": 0.8256,
+      "step": 34908
+    },
+    {
+      "epoch": 0.9587487297794622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.91192017412712e-08,
+      "loss": 0.8329,
+      "step": 34909
+    },
+    {
+      "epoch": 0.9587761940073055,
+      "grad_norm": 0.0,
+      "learning_rate": 8.900075014564136e-08,
+      "loss": 0.8528,
+      "step": 34910
+    },
+    {
+      "epoch": 0.9588036582351487,
+      "grad_norm": 0.0,
+      "learning_rate": 8.88823769696756e-08,
+      "loss": 0.8051,
+      "step": 34911
+    },
+    {
+      "epoch": 0.958831122462992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.87640822143121e-08,
+      "loss": 0.8662,
+      "step": 34912
+    },
+    {
+      "epoch": 0.9588585866908352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.864586588048563e-08,
+      "loss": 0.8492,
+      "step": 34913
+    },
+    {
+      "epoch": 0.9588860509186784,
+      "grad_norm": 0.0,
+      "learning_rate": 8.852772796913322e-08,
+      "loss": 0.7954,
+      "step": 34914
+    },
+    {
+      "epoch": 0.9589135151465217,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84096684811897e-08,
+      "loss": 0.8508,
+      "step": 34915
+    },
+    {
+      "epoch": 0.9589409793743648,
+      "grad_norm": 0.0,
+      "learning_rate": 8.829168741758654e-08,
+      "loss": 0.8076,
+      "step": 34916
+    },
+    {
+      "epoch": 0.9589684436022081,
+      "grad_norm": 0.0,
+      "learning_rate": 8.817378477925965e-08,
+      "loss": 0.8242,
+      "step": 34917
+    },
+    {
+      "epoch": 0.9589959078300514,
+      "grad_norm": 0.0,
+      "learning_rate": 8.805596056714161e-08,
+      "loss": 0.8291,
+      "step": 34918
+    },
+    {
+      "epoch": 0.9590233720578946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.793821478216391e-08,
+      "loss": 0.8099,
+      "step": 34919
+    },
+    {
+      "epoch": 0.9590508362857378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.782054742525803e-08,
+      "loss": 0.7204,
+      "step": 34920
+    },
+    {
+      "epoch": 0.959078300513581,
+      "grad_norm": 0.0,
+      "learning_rate": 8.770295849735655e-08,
+      "loss": 0.8079,
+      "step": 34921
+    },
+    {
+      "epoch": 0.9591057647414243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.758544799938873e-08,
+      "loss": 0.8534,
+      "step": 34922
+    },
+    {
+      "epoch": 0.9591332289692676,
+      "grad_norm": 0.0,
+      "learning_rate": 8.746801593228493e-08,
+      "loss": 0.8397,
+      "step": 34923
+    },
+    {
+      "epoch": 0.9591606931971107,
+      "grad_norm": 0.0,
+      "learning_rate": 8.735066229697332e-08,
+      "loss": 0.7248,
+      "step": 34924
+    },
+    {
+      "epoch": 0.959188157424954,
+      "grad_norm": 0.0,
+      "learning_rate": 8.723338709438201e-08,
+      "loss": 0.7168,
+      "step": 34925
+    },
+    {
+      "epoch": 0.9592156216527973,
+      "grad_norm": 0.0,
+      "learning_rate": 8.71161903254425e-08,
+      "loss": 0.7042,
+      "step": 34926
+    },
+    {
+      "epoch": 0.9592430858806404,
+      "grad_norm": 0.0,
+      "learning_rate": 8.699907199107849e-08,
+      "loss": 0.7295,
+      "step": 34927
+    },
+    {
+      "epoch": 0.9592705501084837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.688203209221701e-08,
+      "loss": 0.7798,
+      "step": 34928
+    },
+    {
+      "epoch": 0.9592980143363269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.676507062978512e-08,
+      "loss": 0.7298,
+      "step": 34929
+    },
+    {
+      "epoch": 0.9593254785641702,
+      "grad_norm": 0.0,
+      "learning_rate": 8.664818760470761e-08,
+      "loss": 0.7641,
+      "step": 34930
+    },
+    {
+      "epoch": 0.9593529427920134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.653138301791041e-08,
+      "loss": 0.8372,
+      "step": 34931
+    },
+    {
+      "epoch": 0.9593804070198566,
+      "grad_norm": 0.0,
+      "learning_rate": 8.641465687031725e-08,
+      "loss": 0.884,
+      "step": 34932
+    },
+    {
+      "epoch": 0.9594078712476999,
+      "grad_norm": 0.0,
+      "learning_rate": 8.62980091628518e-08,
+      "loss": 0.783,
+      "step": 34933
+    },
+    {
+      "epoch": 0.959435335475543,
+      "grad_norm": 0.0,
+      "learning_rate": 8.618143989643558e-08,
+      "loss": 0.8467,
+      "step": 34934
+    },
+    {
+      "epoch": 0.9594627997033863,
+      "grad_norm": 0.0,
+      "learning_rate": 8.606494907199336e-08,
+      "loss": 0.7789,
+      "step": 34935
+    },
+    {
+      "epoch": 0.9594902639312296,
+      "grad_norm": 0.0,
+      "learning_rate": 8.594853669044445e-08,
+      "loss": 0.8503,
+      "step": 34936
+    },
+    {
+      "epoch": 0.9595177281590728,
+      "grad_norm": 0.0,
+      "learning_rate": 8.583220275271031e-08,
+      "loss": 0.7035,
+      "step": 34937
+    },
+    {
+      "epoch": 0.959545192386916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.571594725971355e-08,
+      "loss": 0.819,
+      "step": 34938
+    },
+    {
+      "epoch": 0.9595726566147593,
+      "grad_norm": 0.0,
+      "learning_rate": 8.559977021237231e-08,
+      "loss": 0.7812,
+      "step": 34939
+    },
+    {
+      "epoch": 0.9596001208426025,
+      "grad_norm": 0.0,
+      "learning_rate": 8.548367161160587e-08,
+      "loss": 0.7643,
+      "step": 34940
+    },
+    {
+      "epoch": 0.9596275850704458,
+      "grad_norm": 0.0,
+      "learning_rate": 8.536765145833348e-08,
+      "loss": 0.8195,
+      "step": 34941
+    },
+    {
+      "epoch": 0.9596550492982889,
+      "grad_norm": 0.0,
+      "learning_rate": 8.52517097534733e-08,
+      "loss": 0.7939,
+      "step": 34942
+    },
+    {
+      "epoch": 0.9596825135261322,
+      "grad_norm": 0.0,
+      "learning_rate": 8.513584649794126e-08,
+      "loss": 0.8763,
+      "step": 34943
+    },
+    {
+      "epoch": 0.9597099777539755,
+      "grad_norm": 0.0,
+      "learning_rate": 8.502006169265553e-08,
+      "loss": 0.85,
+      "step": 34944
+    },
+    {
+      "epoch": 0.9597374419818187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.490435533853202e-08,
+      "loss": 0.7258,
+      "step": 34945
+    },
+    {
+      "epoch": 0.9597649062096619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.478872743648559e-08,
+      "loss": 0.8471,
+      "step": 34946
+    },
+    {
+      "epoch": 0.9597923704375052,
+      "grad_norm": 0.0,
+      "learning_rate": 8.467317798743103e-08,
+      "loss": 0.7834,
+      "step": 34947
+    },
+    {
+      "epoch": 0.9598198346653484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.455770699228428e-08,
+      "loss": 0.7925,
+      "step": 34948
+    },
+    {
+      "epoch": 0.9598472988931916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.444231445195683e-08,
+      "loss": 0.7607,
+      "step": 34949
+    },
+    {
+      "epoch": 0.9598747631210348,
+      "grad_norm": 0.0,
+      "learning_rate": 8.432700036736352e-08,
+      "loss": 0.7466,
+      "step": 34950
+    },
+    {
+      "epoch": 0.9599022273488781,
+      "grad_norm": 0.0,
+      "learning_rate": 8.421176473941584e-08,
+      "loss": 0.8497,
+      "step": 34951
+    },
+    {
+      "epoch": 0.9599296915767214,
+      "grad_norm": 0.0,
+      "learning_rate": 8.409660756902526e-08,
+      "loss": 0.6757,
+      "step": 34952
+    },
+    {
+      "epoch": 0.9599571558045645,
+      "grad_norm": 0.0,
+      "learning_rate": 8.398152885710331e-08,
+      "loss": 0.6585,
+      "step": 34953
+    },
+    {
+      "epoch": 0.9599846200324078,
+      "grad_norm": 0.0,
+      "learning_rate": 8.386652860456034e-08,
+      "loss": 0.7455,
+      "step": 34954
+    },
+    {
+      "epoch": 0.960012084260251,
+      "grad_norm": 0.0,
+      "learning_rate": 8.375160681230676e-08,
+      "loss": 0.8041,
+      "step": 34955
+    },
+    {
+      "epoch": 0.9600395484880943,
+      "grad_norm": 0.0,
+      "learning_rate": 8.363676348125182e-08,
+      "loss": 0.7779,
+      "step": 34956
+    },
+    {
+      "epoch": 0.9600670127159375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.352199861230481e-08,
+      "loss": 0.8811,
+      "step": 34957
+    },
+    {
+      "epoch": 0.9600944769437807,
+      "grad_norm": 0.0,
+      "learning_rate": 8.340731220637277e-08,
+      "loss": 0.7333,
+      "step": 34958
+    },
+    {
+      "epoch": 0.960121941171624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.329270426436276e-08,
+      "loss": 0.7926,
+      "step": 34959
+    },
+    {
+      "epoch": 0.9601494053994672,
+      "grad_norm": 0.0,
+      "learning_rate": 8.317817478718183e-08,
+      "loss": 0.8527,
+      "step": 34960
+    },
+    {
+      "epoch": 0.9601768696273104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.306372377573702e-08,
+      "loss": 0.8443,
+      "step": 34961
+    },
+    {
+      "epoch": 0.9602043338551537,
+      "grad_norm": 0.0,
+      "learning_rate": 8.294935123093317e-08,
+      "loss": 0.8491,
+      "step": 34962
+    },
+    {
+      "epoch": 0.9602317980829969,
+      "grad_norm": 0.0,
+      "learning_rate": 8.283505715367623e-08,
+      "loss": 0.8717,
+      "step": 34963
+    },
+    {
+      "epoch": 0.9602592623108401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.272084154486993e-08,
+      "loss": 0.7724,
+      "step": 34964
+    },
+    {
+      "epoch": 0.9602867265386834,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260670440541685e-08,
+      "loss": 0.8235,
+      "step": 34965
+    },
+    {
+      "epoch": 0.9603141907665266,
+      "grad_norm": 0.0,
+      "learning_rate": 8.249264573622295e-08,
+      "loss": 0.8571,
+      "step": 34966
+    },
+    {
+      "epoch": 0.9603416549943699,
+      "grad_norm": 0.0,
+      "learning_rate": 8.237866553818752e-08,
+      "loss": 0.8586,
+      "step": 34967
+    },
+    {
+      "epoch": 0.960369119222213,
+      "grad_norm": 0.0,
+      "learning_rate": 8.226476381221427e-08,
+      "loss": 0.8369,
+      "step": 34968
+    },
+    {
+      "epoch": 0.9603965834500563,
+      "grad_norm": 0.0,
+      "learning_rate": 8.215094055920358e-08,
+      "loss": 0.8241,
+      "step": 34969
+    },
+    {
+      "epoch": 0.9604240476778996,
+      "grad_norm": 0.0,
+      "learning_rate": 8.203719578005698e-08,
+      "loss": 0.784,
+      "step": 34970
+    },
+    {
+      "epoch": 0.9604515119057427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.192352947567373e-08,
+      "loss": 0.8501,
+      "step": 34971
+    },
+    {
+      "epoch": 0.960478976133586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.180994164695311e-08,
+      "loss": 0.7681,
+      "step": 34972
+    },
+    {
+      "epoch": 0.9605064403614293,
+      "grad_norm": 0.0,
+      "learning_rate": 8.169643229479441e-08,
+      "loss": 0.7928,
+      "step": 34973
+    },
+    {
+      "epoch": 0.9605339045892725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.158300142009578e-08,
+      "loss": 0.7652,
+      "step": 34974
+    },
+    {
+      "epoch": 0.9605613688171157,
+      "grad_norm": 0.0,
+      "learning_rate": 8.146964902375431e-08,
+      "loss": 0.8077,
+      "step": 34975
+    },
+    {
+      "epoch": 0.9605888330449589,
+      "grad_norm": 0.0,
+      "learning_rate": 8.135637510666705e-08,
+      "loss": 0.7942,
+      "step": 34976
+    },
+    {
+      "epoch": 0.9606162972728022,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124317966972994e-08,
+      "loss": 0.8051,
+      "step": 34977
+    },
+    {
+      "epoch": 0.9606437615006455,
+      "grad_norm": 0.0,
+      "learning_rate": 8.113006271383895e-08,
+      "loss": 0.7549,
+      "step": 34978
+    },
+    {
+      "epoch": 0.9606712257284886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.101702423989e-08,
+      "loss": 0.8143,
+      "step": 34979
+    },
+    {
+      "epoch": 0.9606986899563319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.090406424877572e-08,
+      "loss": 0.8253,
+      "step": 34980
+    },
+    {
+      "epoch": 0.9607261541841751,
+      "grad_norm": 0.0,
+      "learning_rate": 8.079118274139097e-08,
+      "loss": 0.7256,
+      "step": 34981
+    },
+    {
+      "epoch": 0.9607536184120183,
+      "grad_norm": 0.0,
+      "learning_rate": 8.067837971862946e-08,
+      "loss": 0.787,
+      "step": 34982
+    },
+    {
+      "epoch": 0.9607810826398616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05656551813816e-08,
+      "loss": 0.8037,
+      "step": 34983
+    },
+    {
+      "epoch": 0.9608085468677048,
+      "grad_norm": 0.0,
+      "learning_rate": 8.045300913054111e-08,
+      "loss": 0.6967,
+      "step": 34984
+    },
+    {
+      "epoch": 0.9608360110955481,
+      "grad_norm": 0.0,
+      "learning_rate": 8.034044156699949e-08,
+      "loss": 0.7737,
+      "step": 34985
+    },
+    {
+      "epoch": 0.9608634753233913,
+      "grad_norm": 0.0,
+      "learning_rate": 8.022795249164717e-08,
+      "loss": 0.7497,
+      "step": 34986
+    },
+    {
+      "epoch": 0.9608909395512345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.011554190537229e-08,
+      "loss": 0.8617,
+      "step": 34987
+    },
+    {
+      "epoch": 0.9609184037790778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.00032098090664e-08,
+      "loss": 0.7799,
+      "step": 34988
+    },
+    {
+      "epoch": 0.960945868006921,
+      "grad_norm": 0.0,
+      "learning_rate": 7.989095620361875e-08,
+      "loss": 0.7842,
+      "step": 34989
+    },
+    {
+      "epoch": 0.9609733322347642,
+      "grad_norm": 0.0,
+      "learning_rate": 7.977878108991533e-08,
+      "loss": 0.7684,
+      "step": 34990
+    },
+    {
+      "epoch": 0.9610007964626075,
+      "grad_norm": 0.0,
+      "learning_rate": 7.96666844688454e-08,
+      "loss": 0.854,
+      "step": 34991
+    },
+    {
+      "epoch": 0.9610282606904507,
+      "grad_norm": 0.0,
+      "learning_rate": 7.955466634129605e-08,
+      "loss": 0.767,
+      "step": 34992
+    },
+    {
+      "epoch": 0.9610557249182939,
+      "grad_norm": 0.0,
+      "learning_rate": 7.944272670815212e-08,
+      "loss": 0.7932,
+      "step": 34993
+    },
+    {
+      "epoch": 0.9610831891461371,
+      "grad_norm": 0.0,
+      "learning_rate": 7.933086557030068e-08,
+      "loss": 0.7765,
+      "step": 34994
+    },
+    {
+      "epoch": 0.9611106533739804,
+      "grad_norm": 0.0,
+      "learning_rate": 7.921908292862546e-08,
+      "loss": 0.9037,
+      "step": 34995
+    },
+    {
+      "epoch": 0.9611381176018237,
+      "grad_norm": 0.0,
+      "learning_rate": 7.910737878401243e-08,
+      "loss": 0.8398,
+      "step": 34996
+    },
+    {
+      "epoch": 0.9611655818296668,
+      "grad_norm": 0.0,
+      "learning_rate": 7.899575313734531e-08,
+      "loss": 0.8015,
+      "step": 34997
+    },
+    {
+      "epoch": 0.9611930460575101,
+      "grad_norm": 0.0,
+      "learning_rate": 7.888420598950674e-08,
+      "loss": 0.8336,
+      "step": 34998
+    },
+    {
+      "epoch": 0.9612205102853534,
+      "grad_norm": 0.0,
+      "learning_rate": 7.877273734137936e-08,
+      "loss": 0.7807,
+      "step": 34999
+    },
+    {
+      "epoch": 0.9612479745131965,
+      "grad_norm": 0.0,
+      "learning_rate": 7.866134719384466e-08,
+      "loss": 0.8098,
+      "step": 35000
+    },
+    {
+      "epoch": 0.9612754387410398,
+      "grad_norm": 0.0,
+      "learning_rate": 7.855003554778418e-08,
+      "loss": 0.7143,
+      "step": 35001
+    },
+    {
+      "epoch": 0.961302902968883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.843880240407941e-08,
+      "loss": 0.7647,
+      "step": 35002
+    },
+    {
+      "epoch": 0.9613303671967263,
+      "grad_norm": 0.0,
+      "learning_rate": 7.832764776360968e-08,
+      "loss": 0.7675,
+      "step": 35003
+    },
+    {
+      "epoch": 0.9613578314245695,
+      "grad_norm": 0.0,
+      "learning_rate": 7.821657162725648e-08,
+      "loss": 0.7665,
+      "step": 35004
+    },
+    {
+      "epoch": 0.9613852956524127,
+      "grad_norm": 0.0,
+      "learning_rate": 7.81055739958958e-08,
+      "loss": 0.7803,
+      "step": 35005
+    },
+    {
+      "epoch": 0.961412759880256,
+      "grad_norm": 0.0,
+      "learning_rate": 7.799465487040803e-08,
+      "loss": 0.8014,
+      "step": 35006
+    },
+    {
+      "epoch": 0.9614402241080992,
+      "grad_norm": 0.0,
+      "learning_rate": 7.788381425166913e-08,
+      "loss": 0.8756,
+      "step": 35007
+    },
+    {
+      "epoch": 0.9614676883359424,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77730521405573e-08,
+      "loss": 0.8581,
+      "step": 35008
+    },
+    {
+      "epoch": 0.9614951525637857,
+      "grad_norm": 0.0,
+      "learning_rate": 7.76623685379485e-08,
+      "loss": 0.8067,
+      "step": 35009
+    },
+    {
+      "epoch": 0.9615226167916289,
+      "grad_norm": 0.0,
+      "learning_rate": 7.755176344471871e-08,
+      "loss": 0.8948,
+      "step": 35010
+    },
+    {
+      "epoch": 0.9615500810194721,
+      "grad_norm": 0.0,
+      "learning_rate": 7.744123686174277e-08,
+      "loss": 0.8754,
+      "step": 35011
+    },
+    {
+      "epoch": 0.9615775452473154,
+      "grad_norm": 0.0,
+      "learning_rate": 7.733078878989552e-08,
+      "loss": 0.7152,
+      "step": 35012
+    },
+    {
+      "epoch": 0.9616050094751586,
+      "grad_norm": 0.0,
+      "learning_rate": 7.722041923004963e-08,
+      "loss": 0.7338,
+      "step": 35013
+    },
+    {
+      "epoch": 0.9616324737030019,
+      "grad_norm": 0.0,
+      "learning_rate": 7.711012818308106e-08,
+      "loss": 0.7512,
+      "step": 35014
+    },
+    {
+      "epoch": 0.961659937930845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.69999156498602e-08,
+      "loss": 0.7419,
+      "step": 35015
+    },
+    {
+      "epoch": 0.9616874021586883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.688978163126082e-08,
+      "loss": 0.8199,
+      "step": 35016
+    },
+    {
+      "epoch": 0.9617148663865316,
+      "grad_norm": 0.0,
+      "learning_rate": 7.677972612815221e-08,
+      "loss": 0.8262,
+      "step": 35017
+    },
+    {
+      "epoch": 0.9617423306143748,
+      "grad_norm": 0.0,
+      "learning_rate": 7.666974914140702e-08,
+      "loss": 0.8329,
+      "step": 35018
+    },
+    {
+      "epoch": 0.961769794842218,
+      "grad_norm": 0.0,
+      "learning_rate": 7.655985067189342e-08,
+      "loss": 0.8264,
+      "step": 35019
+    },
+    {
+      "epoch": 0.9617972590700612,
+      "grad_norm": 0.0,
+      "learning_rate": 7.645003072048296e-08,
+      "loss": 0.8512,
+      "step": 35020
+    },
+    {
+      "epoch": 0.9618247232979045,
+      "grad_norm": 0.0,
+      "learning_rate": 7.634028928804493e-08,
+      "loss": 0.7719,
+      "step": 35021
+    },
+    {
+      "epoch": 0.9618521875257477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.623062637544531e-08,
+      "loss": 0.7608,
+      "step": 35022
+    },
+    {
+      "epoch": 0.9618796517535909,
+      "grad_norm": 0.0,
+      "learning_rate": 7.612104198355342e-08,
+      "loss": 0.7349,
+      "step": 35023
+    },
+    {
+      "epoch": 0.9619071159814342,
+      "grad_norm": 0.0,
+      "learning_rate": 7.60115361132363e-08,
+      "loss": 0.7854,
+      "step": 35024
+    },
+    {
+      "epoch": 0.9619345802092775,
+      "grad_norm": 0.0,
+      "learning_rate": 7.590210876535997e-08,
+      "loss": 0.7986,
+      "step": 35025
+    },
+    {
+      "epoch": 0.9619620444371206,
+      "grad_norm": 0.0,
+      "learning_rate": 7.579275994079038e-08,
+      "loss": 0.7929,
+      "step": 35026
+    },
+    {
+      "epoch": 0.9619895086649639,
+      "grad_norm": 0.0,
+      "learning_rate": 7.568348964039241e-08,
+      "loss": 0.9593,
+      "step": 35027
+    },
+    {
+      "epoch": 0.9620169728928071,
+      "grad_norm": 0.0,
+      "learning_rate": 7.557429786503089e-08,
+      "loss": 0.8296,
+      "step": 35028
+    },
+    {
+      "epoch": 0.9620444371206504,
+      "grad_norm": 0.0,
+      "learning_rate": 7.546518461557073e-08,
+      "loss": 0.7296,
+      "step": 35029
+    },
+    {
+      "epoch": 0.9620719013484936,
+      "grad_norm": 0.0,
+      "learning_rate": 7.535614989287343e-08,
+      "loss": 0.8182,
+      "step": 35030
+    },
+    {
+      "epoch": 0.9620993655763368,
+      "grad_norm": 0.0,
+      "learning_rate": 7.524719369780387e-08,
+      "loss": 0.8087,
+      "step": 35031
+    },
+    {
+      "epoch": 0.9621268298041801,
+      "grad_norm": 0.0,
+      "learning_rate": 7.513831603122135e-08,
+      "loss": 0.7592,
+      "step": 35032
+    },
+    {
+      "epoch": 0.9621542940320232,
+      "grad_norm": 0.0,
+      "learning_rate": 7.502951689399074e-08,
+      "loss": 0.7963,
+      "step": 35033
+    },
+    {
+      "epoch": 0.9621817582598665,
+      "grad_norm": 0.0,
+      "learning_rate": 7.492079628697024e-08,
+      "loss": 0.9236,
+      "step": 35034
+    },
+    {
+      "epoch": 0.9622092224877098,
+      "grad_norm": 0.0,
+      "learning_rate": 7.481215421102028e-08,
+      "loss": 0.7605,
+      "step": 35035
+    },
+    {
+      "epoch": 0.962236686715553,
+      "grad_norm": 0.0,
+      "learning_rate": 7.47035906670024e-08,
+      "loss": 0.7994,
+      "step": 35036
+    },
+    {
+      "epoch": 0.9622641509433962,
+      "grad_norm": 0.0,
+      "learning_rate": 7.459510565577366e-08,
+      "loss": 0.7494,
+      "step": 35037
+    },
+    {
+      "epoch": 0.9622916151712395,
+      "grad_norm": 0.0,
+      "learning_rate": 7.448669917819451e-08,
+      "loss": 0.8297,
+      "step": 35038
+    },
+    {
+      "epoch": 0.9623190793990827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.437837123512093e-08,
+      "loss": 0.8718,
+      "step": 35039
+    },
+    {
+      "epoch": 0.962346543626926,
+      "grad_norm": 0.0,
+      "learning_rate": 7.427012182740889e-08,
+      "loss": 0.8701,
+      "step": 35040
+    },
+    {
+      "epoch": 0.9623740078547691,
+      "grad_norm": 0.0,
+      "learning_rate": 7.416195095591771e-08,
+      "loss": 0.792,
+      "step": 35041
+    },
+    {
+      "epoch": 0.9624014720826124,
+      "grad_norm": 0.0,
+      "learning_rate": 7.405385862150227e-08,
+      "loss": 0.7387,
+      "step": 35042
+    },
+    {
+      "epoch": 0.9624289363104557,
+      "grad_norm": 0.0,
+      "learning_rate": 7.394584482501743e-08,
+      "loss": 0.8618,
+      "step": 35043
+    },
+    {
+      "epoch": 0.9624564005382988,
+      "grad_norm": 0.0,
+      "learning_rate": 7.383790956731696e-08,
+      "loss": 0.8127,
+      "step": 35044
+    },
+    {
+      "epoch": 0.9624838647661421,
+      "grad_norm": 0.0,
+      "learning_rate": 7.373005284925793e-08,
+      "loss": 0.7627,
+      "step": 35045
+    },
+    {
+      "epoch": 0.9625113289939853,
+      "grad_norm": 0.0,
+      "learning_rate": 7.362227467169081e-08,
+      "loss": 0.7097,
+      "step": 35046
+    },
+    {
+      "epoch": 0.9625387932218286,
+      "grad_norm": 0.0,
+      "learning_rate": 7.351457503546933e-08,
+      "loss": 0.8747,
+      "step": 35047
+    },
+    {
+      "epoch": 0.9625662574496718,
+      "grad_norm": 0.0,
+      "learning_rate": 7.340695394144614e-08,
+      "loss": 0.8469,
+      "step": 35048
+    },
+    {
+      "epoch": 0.962593721677515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.32994113904717e-08,
+      "loss": 0.8628,
+      "step": 35049
+    },
+    {
+      "epoch": 0.9626211859053583,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319194738339752e-08,
+      "loss": 0.7971,
+      "step": 35050
+    },
+    {
+      "epoch": 0.9626486501332016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.308456192107405e-08,
+      "loss": 0.7145,
+      "step": 35051
+    },
+    {
+      "epoch": 0.9626761143610447,
+      "grad_norm": 0.0,
+      "learning_rate": 7.297725500435172e-08,
+      "loss": 0.7779,
+      "step": 35052
+    },
+    {
+      "epoch": 0.962703578588888,
+      "grad_norm": 0.0,
+      "learning_rate": 7.28700266340776e-08,
+      "loss": 0.8493,
+      "step": 35053
+    },
+    {
+      "epoch": 0.9627310428167312,
+      "grad_norm": 0.0,
+      "learning_rate": 7.276287681110217e-08,
+      "loss": 0.7387,
+      "step": 35054
+    },
+    {
+      "epoch": 0.9627585070445744,
+      "grad_norm": 0.0,
+      "learning_rate": 7.265580553627139e-08,
+      "loss": 0.7805,
+      "step": 35055
+    },
+    {
+      "epoch": 0.9627859712724177,
+      "grad_norm": 0.0,
+      "learning_rate": 7.254881281043346e-08,
+      "loss": 0.7943,
+      "step": 35056
+    },
+    {
+      "epoch": 0.9628134355002609,
+      "grad_norm": 0.0,
+      "learning_rate": 7.244189863443552e-08,
+      "loss": 0.797,
+      "step": 35057
+    },
+    {
+      "epoch": 0.9628408997281042,
+      "grad_norm": 0.0,
+      "learning_rate": 7.233506300912241e-08,
+      "loss": 0.6609,
+      "step": 35058
+    },
+    {
+      "epoch": 0.9628683639559473,
+      "grad_norm": 0.0,
+      "learning_rate": 7.222830593533903e-08,
+      "loss": 0.8327,
+      "step": 35059
+    },
+    {
+      "epoch": 0.9628958281837906,
+      "grad_norm": 0.0,
+      "learning_rate": 7.212162741393247e-08,
+      "loss": 0.7502,
+      "step": 35060
+    },
+    {
+      "epoch": 0.9629232924116339,
+      "grad_norm": 0.0,
+      "learning_rate": 7.201502744574429e-08,
+      "loss": 0.943,
+      "step": 35061
+    },
+    {
+      "epoch": 0.962950756639477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.190850603161936e-08,
+      "loss": 0.7697,
+      "step": 35062
+    },
+    {
+      "epoch": 0.9629782208673203,
+      "grad_norm": 0.0,
+      "learning_rate": 7.180206317240035e-08,
+      "loss": 0.8939,
+      "step": 35063
+    },
+    {
+      "epoch": 0.9630056850951636,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16956988689288e-08,
+      "loss": 0.7814,
+      "step": 35064
+    },
+    {
+      "epoch": 0.9630331493230068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.158941312204737e-08,
+      "loss": 0.7922,
+      "step": 35065
+    },
+    {
+      "epoch": 0.96306061355085,
+      "grad_norm": 0.0,
+      "learning_rate": 7.148320593259539e-08,
+      "loss": 0.8651,
+      "step": 35066
+    },
+    {
+      "epoch": 0.9630880777786932,
+      "grad_norm": 0.0,
+      "learning_rate": 7.13770773014144e-08,
+      "loss": 0.8431,
+      "step": 35067
+    },
+    {
+      "epoch": 0.9631155420065365,
+      "grad_norm": 0.0,
+      "learning_rate": 7.127102722934487e-08,
+      "loss": 0.8598,
+      "step": 35068
+    },
+    {
+      "epoch": 0.9631430062343798,
+      "grad_norm": 0.0,
+      "learning_rate": 7.116505571722388e-08,
+      "loss": 0.8496,
+      "step": 35069
+    },
+    {
+      "epoch": 0.9631704704622229,
+      "grad_norm": 0.0,
+      "learning_rate": 7.105916276589187e-08,
+      "loss": 0.9353,
+      "step": 35070
+    },
+    {
+      "epoch": 0.9631979346900662,
+      "grad_norm": 0.0,
+      "learning_rate": 7.095334837618595e-08,
+      "loss": 0.8098,
+      "step": 35071
+    },
+    {
+      "epoch": 0.9632253989179094,
+      "grad_norm": 0.0,
+      "learning_rate": 7.084761254894323e-08,
+      "loss": 0.8038,
+      "step": 35072
+    },
+    {
+      "epoch": 0.9632528631457526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.074195528499971e-08,
+      "loss": 0.8148,
+      "step": 35073
+    },
+    {
+      "epoch": 0.9632803273735959,
+      "grad_norm": 0.0,
+      "learning_rate": 7.063637658519251e-08,
+      "loss": 0.6954,
+      "step": 35074
+    },
+    {
+      "epoch": 0.9633077916014391,
+      "grad_norm": 0.0,
+      "learning_rate": 7.05308764503565e-08,
+      "loss": 0.786,
+      "step": 35075
+    },
+    {
+      "epoch": 0.9633352558292824,
+      "grad_norm": 0.0,
+      "learning_rate": 7.042545488132657e-08,
+      "loss": 0.7458,
+      "step": 35076
+    },
+    {
+      "epoch": 0.9633627200571256,
+      "grad_norm": 0.0,
+      "learning_rate": 7.03201118789365e-08,
+      "loss": 0.8822,
+      "step": 35077
+    },
+    {
+      "epoch": 0.9633901842849688,
+      "grad_norm": 0.0,
+      "learning_rate": 7.021484744402007e-08,
+      "loss": 0.8343,
+      "step": 35078
+    },
+    {
+      "epoch": 0.9634176485128121,
+      "grad_norm": 0.0,
+      "learning_rate": 7.010966157740994e-08,
+      "loss": 0.8233,
+      "step": 35079
+    },
+    {
+      "epoch": 0.9634451127406553,
+      "grad_norm": 0.0,
+      "learning_rate": 7.000455427993991e-08,
+      "loss": 0.8412,
+      "step": 35080
+    },
+    {
+      "epoch": 0.9634725769684985,
+      "grad_norm": 0.0,
+      "learning_rate": 6.989952555243929e-08,
+      "loss": 0.8352,
+      "step": 35081
+    },
+    {
+      "epoch": 0.9635000411963418,
+      "grad_norm": 0.0,
+      "learning_rate": 6.979457539574075e-08,
+      "loss": 0.7666,
+      "step": 35082
+    },
+    {
+      "epoch": 0.963527505424185,
+      "grad_norm": 0.0,
+      "learning_rate": 6.968970381067364e-08,
+      "loss": 0.9061,
+      "step": 35083
+    },
+    {
+      "epoch": 0.9635549696520282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95849107980684e-08,
+      "loss": 0.8702,
+      "step": 35084
+    },
+    {
+      "epoch": 0.9635824338798714,
+      "grad_norm": 0.0,
+      "learning_rate": 6.948019635875325e-08,
+      "loss": 0.743,
+      "step": 35085
+    },
+    {
+      "epoch": 0.9636098981077147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.937556049355865e-08,
+      "loss": 0.7789,
+      "step": 35086
+    },
+    {
+      "epoch": 0.963637362335558,
+      "grad_norm": 0.0,
+      "learning_rate": 6.92710032033117e-08,
+      "loss": 0.8358,
+      "step": 35087
+    },
+    {
+      "epoch": 0.9636648265634011,
+      "grad_norm": 0.0,
+      "learning_rate": 6.916652448883843e-08,
+      "loss": 0.7981,
+      "step": 35088
+    },
+    {
+      "epoch": 0.9636922907912444,
+      "grad_norm": 0.0,
+      "learning_rate": 6.906212435096704e-08,
+      "loss": 0.754,
+      "step": 35089
+    },
+    {
+      "epoch": 0.9637197550190877,
+      "grad_norm": 0.0,
+      "learning_rate": 6.895780279052244e-08,
+      "loss": 0.7667,
+      "step": 35090
+    },
+    {
+      "epoch": 0.9637472192469309,
+      "grad_norm": 0.0,
+      "learning_rate": 6.885355980833064e-08,
+      "loss": 0.7487,
+      "step": 35091
+    },
+    {
+      "epoch": 0.9637746834747741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.874939540521652e-08,
+      "loss": 0.8984,
+      "step": 35092
+    },
+    {
+      "epoch": 0.9638021477026173,
+      "grad_norm": 0.0,
+      "learning_rate": 6.864530958200388e-08,
+      "loss": 0.7764,
+      "step": 35093
+    },
+    {
+      "epoch": 0.9638296119304606,
+      "grad_norm": 0.0,
+      "learning_rate": 6.854130233951761e-08,
+      "loss": 0.8399,
+      "step": 35094
+    },
+    {
+      "epoch": 0.9638570761583038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.843737367857928e-08,
+      "loss": 0.8425,
+      "step": 35095
+    },
+    {
+      "epoch": 0.963884540386147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.833352360001155e-08,
+      "loss": 0.8763,
+      "step": 35096
+    },
+    {
+      "epoch": 0.9639120046139903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.822975210463489e-08,
+      "loss": 0.7822,
+      "step": 35097
+    },
+    {
+      "epoch": 0.9639394688418335,
+      "grad_norm": 0.0,
+      "learning_rate": 6.812605919327309e-08,
+      "loss": 0.7558,
+      "step": 35098
+    },
+    {
+      "epoch": 0.9639669330696767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.802244486674548e-08,
+      "loss": 0.8266,
+      "step": 35099
+    },
+    {
+      "epoch": 0.96399439729752,
+      "grad_norm": 0.0,
+      "learning_rate": 6.791890912587029e-08,
+      "loss": 0.7932,
+      "step": 35100
+    },
+    {
+      "epoch": 0.9640218615253632,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7815451971468e-08,
+      "loss": 0.9131,
+      "step": 35101
+    },
+    {
+      "epoch": 0.9640493257532065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.771207340435793e-08,
+      "loss": 0.8853,
+      "step": 35102
+    },
+    {
+      "epoch": 0.9640767899810497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.760877342535721e-08,
+      "loss": 0.9329,
+      "step": 35103
+    },
+    {
+      "epoch": 0.9641042542088929,
+      "grad_norm": 0.0,
+      "learning_rate": 6.750555203528408e-08,
+      "loss": 0.8409,
+      "step": 35104
+    },
+    {
+      "epoch": 0.9641317184367362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.740240923495345e-08,
+      "loss": 0.7758,
+      "step": 35105
+    },
+    {
+      "epoch": 0.9641591826645793,
+      "grad_norm": 0.0,
+      "learning_rate": 6.729934502518243e-08,
+      "loss": 0.8157,
+      "step": 35106
+    },
+    {
+      "epoch": 0.9641866468924226,
+      "grad_norm": 0.0,
+      "learning_rate": 6.719635940678704e-08,
+      "loss": 0.7337,
+      "step": 35107
+    },
+    {
+      "epoch": 0.9642141111202659,
+      "grad_norm": 0.0,
+      "learning_rate": 6.709345238058106e-08,
+      "loss": 0.7866,
+      "step": 35108
+    },
+    {
+      "epoch": 0.9642415753481091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.699062394737943e-08,
+      "loss": 0.853,
+      "step": 35109
+    },
+    {
+      "epoch": 0.9642690395759523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.688787410799591e-08,
+      "loss": 0.7463,
+      "step": 35110
+    },
+    {
+      "epoch": 0.9642965038037955,
+      "grad_norm": 0.0,
+      "learning_rate": 6.67852028632443e-08,
+      "loss": 0.7981,
+      "step": 35111
+    },
+    {
+      "epoch": 0.9643239680316388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.668261021393396e-08,
+      "loss": 0.7897,
+      "step": 35112
+    },
+    {
+      "epoch": 0.964351432259482,
+      "grad_norm": 0.0,
+      "learning_rate": 6.65800961608798e-08,
+      "loss": 0.7509,
+      "step": 35113
+    },
+    {
+      "epoch": 0.9643788964873252,
+      "grad_norm": 0.0,
+      "learning_rate": 6.647766070489115e-08,
+      "loss": 0.7557,
+      "step": 35114
+    },
+    {
+      "epoch": 0.9644063607151685,
+      "grad_norm": 0.0,
+      "learning_rate": 6.637530384677959e-08,
+      "loss": 0.7701,
+      "step": 35115
+    },
+    {
+      "epoch": 0.9644338249430118,
+      "grad_norm": 0.0,
+      "learning_rate": 6.62730255873545e-08,
+      "loss": 0.7729,
+      "step": 35116
+    },
+    {
+      "epoch": 0.9644612891708549,
+      "grad_norm": 0.0,
+      "learning_rate": 6.617082592742519e-08,
+      "loss": 0.8338,
+      "step": 35117
+    },
+    {
+      "epoch": 0.9644887533986982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.606870486780104e-08,
+      "loss": 0.7152,
+      "step": 35118
+    },
+    {
+      "epoch": 0.9645162176265414,
+      "grad_norm": 0.0,
+      "learning_rate": 6.596666240928806e-08,
+      "loss": 0.8757,
+      "step": 35119
+    },
+    {
+      "epoch": 0.9645436818543847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.586469855269562e-08,
+      "loss": 0.8304,
+      "step": 35120
+    },
+    {
+      "epoch": 0.9645711460822279,
+      "grad_norm": 0.0,
+      "learning_rate": 6.57628132988286e-08,
+      "loss": 0.8766,
+      "step": 35121
+    },
+    {
+      "epoch": 0.9645986103100711,
+      "grad_norm": 0.0,
+      "learning_rate": 6.56610066484964e-08,
+      "loss": 0.735,
+      "step": 35122
+    },
+    {
+      "epoch": 0.9646260745379144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.555927860250056e-08,
+      "loss": 0.8502,
+      "step": 35123
+    },
+    {
+      "epoch": 0.9646535387657575,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545762916164932e-08,
+      "loss": 0.7506,
+      "step": 35124
+    },
+    {
+      "epoch": 0.9646810029936008,
+      "grad_norm": 0.0,
+      "learning_rate": 6.53560583267443e-08,
+      "loss": 0.7707,
+      "step": 35125
+    },
+    {
+      "epoch": 0.9647084672214441,
+      "grad_norm": 0.0,
+      "learning_rate": 6.525456609859148e-08,
+      "loss": 0.8239,
+      "step": 35126
+    },
+    {
+      "epoch": 0.9647359314492873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.515315247799359e-08,
+      "loss": 0.8409,
+      "step": 35127
+    },
+    {
+      "epoch": 0.9647633956771305,
+      "grad_norm": 0.0,
+      "learning_rate": 6.505181746575106e-08,
+      "loss": 0.914,
+      "step": 35128
+    },
+    {
+      "epoch": 0.9647908599049738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.495056106266773e-08,
+      "loss": 0.8283,
+      "step": 35129
+    },
+    {
+      "epoch": 0.964818324132817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.484938326954293e-08,
+      "loss": 0.7654,
+      "step": 35130
+    },
+    {
+      "epoch": 0.9648457883606603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.474828408717937e-08,
+      "loss": 0.7739,
+      "step": 35131
+    },
+    {
+      "epoch": 0.9648732525885034,
+      "grad_norm": 0.0,
+      "learning_rate": 6.464726351637529e-08,
+      "loss": 0.8487,
+      "step": 35132
+    },
+    {
+      "epoch": 0.9649007168163467,
+      "grad_norm": 0.0,
+      "learning_rate": 6.454632155793117e-08,
+      "loss": 0.7485,
+      "step": 35133
+    },
+    {
+      "epoch": 0.96492818104419,
+      "grad_norm": 0.0,
+      "learning_rate": 6.444545821264525e-08,
+      "loss": 0.9106,
+      "step": 35134
+    },
+    {
+      "epoch": 0.9649556452720331,
+      "grad_norm": 0.0,
+      "learning_rate": 6.43446734813158e-08,
+      "loss": 0.7581,
+      "step": 35135
+    },
+    {
+      "epoch": 0.9649831094998764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.424396736473881e-08,
+      "loss": 0.9134,
+      "step": 35136
+    },
+    {
+      "epoch": 0.9650105737277197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.414333986371257e-08,
+      "loss": 0.7278,
+      "step": 35137
+    },
+    {
+      "epoch": 0.9650380379555629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40427909790331e-08,
+      "loss": 0.7689,
+      "step": 35138
+    },
+    {
+      "epoch": 0.9650655021834061,
+      "grad_norm": 0.0,
+      "learning_rate": 6.394232071149641e-08,
+      "loss": 0.8067,
+      "step": 35139
+    },
+    {
+      "epoch": 0.9650929664112493,
+      "grad_norm": 0.0,
+      "learning_rate": 6.384192906189745e-08,
+      "loss": 0.7838,
+      "step": 35140
+    },
+    {
+      "epoch": 0.9651204306390926,
+      "grad_norm": 0.0,
+      "learning_rate": 6.374161603102891e-08,
+      "loss": 0.8363,
+      "step": 35141
+    },
+    {
+      "epoch": 0.9651478948669359,
+      "grad_norm": 0.0,
+      "learning_rate": 6.364138161968569e-08,
+      "loss": 0.8574,
+      "step": 35142
+    },
+    {
+      "epoch": 0.965175359094779,
+      "grad_norm": 0.0,
+      "learning_rate": 6.354122582866162e-08,
+      "loss": 0.7365,
+      "step": 35143
+    },
+    {
+      "epoch": 0.9652028233226223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.344114865874717e-08,
+      "loss": 0.8198,
+      "step": 35144
+    },
+    {
+      "epoch": 0.9652302875504655,
+      "grad_norm": 0.0,
+      "learning_rate": 6.334115011073616e-08,
+      "loss": 0.8862,
+      "step": 35145
+    },
+    {
+      "epoch": 0.9652577517783087,
+      "grad_norm": 0.0,
+      "learning_rate": 6.324123018542016e-08,
+      "loss": 0.8398,
+      "step": 35146
+    },
+    {
+      "epoch": 0.965285216006152,
+      "grad_norm": 0.0,
+      "learning_rate": 6.314138888358745e-08,
+      "loss": 0.9388,
+      "step": 35147
+    },
+    {
+      "epoch": 0.9653126802339952,
+      "grad_norm": 0.0,
+      "learning_rate": 6.304162620602849e-08,
+      "loss": 0.7953,
+      "step": 35148
+    },
+    {
+      "epoch": 0.9653401444618385,
+      "grad_norm": 0.0,
+      "learning_rate": 6.294194215353488e-08,
+      "loss": 0.8275,
+      "step": 35149
+    },
+    {
+      "epoch": 0.9653676086896817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.284233672689266e-08,
+      "loss": 0.7302,
+      "step": 35150
+    },
+    {
+      "epoch": 0.9653950729175249,
+      "grad_norm": 0.0,
+      "learning_rate": 6.27428099268923e-08,
+      "loss": 0.8498,
+      "step": 35151
+    },
+    {
+      "epoch": 0.9654225371453682,
+      "grad_norm": 0.0,
+      "learning_rate": 6.264336175431984e-08,
+      "loss": 0.7177,
+      "step": 35152
+    },
+    {
+      "epoch": 0.9654500013732114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.254399220996244e-08,
+      "loss": 0.8299,
+      "step": 35153
+    },
+    {
+      "epoch": 0.9654774656010546,
+      "grad_norm": 0.0,
+      "learning_rate": 6.244470129460611e-08,
+      "loss": 0.8035,
+      "step": 35154
+    },
+    {
+      "epoch": 0.9655049298288979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.234548900903581e-08,
+      "loss": 0.764,
+      "step": 35155
+    },
+    {
+      "epoch": 0.9655323940567411,
+      "grad_norm": 0.0,
+      "learning_rate": 6.224635535403867e-08,
+      "loss": 0.8599,
+      "step": 35156
+    },
+    {
+      "epoch": 0.9655598582845843,
+      "grad_norm": 0.0,
+      "learning_rate": 6.214730033039851e-08,
+      "loss": 0.6949,
+      "step": 35157
+    },
+    {
+      "epoch": 0.9655873225124275,
+      "grad_norm": 0.0,
+      "learning_rate": 6.204832393889692e-08,
+      "loss": 0.8272,
+      "step": 35158
+    },
+    {
+      "epoch": 0.9656147867402708,
+      "grad_norm": 0.0,
+      "learning_rate": 6.194942618031996e-08,
+      "loss": 0.778,
+      "step": 35159
+    },
+    {
+      "epoch": 0.9656422509681141,
+      "grad_norm": 0.0,
+      "learning_rate": 6.185060705544809e-08,
+      "loss": 0.8433,
+      "step": 35160
+    },
+    {
+      "epoch": 0.9656697151959572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.175186656506404e-08,
+      "loss": 0.7861,
+      "step": 35161
+    },
+    {
+      "epoch": 0.9656971794238005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.165320470994829e-08,
+      "loss": 0.658,
+      "step": 35162
+    },
+    {
+      "epoch": 0.9657246436516438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.155462149088353e-08,
+      "loss": 0.7541,
+      "step": 35163
+    },
+    {
+      "epoch": 0.965752107879487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.145611690864695e-08,
+      "loss": 0.7504,
+      "step": 35164
+    },
+    {
+      "epoch": 0.9657795721073302,
+      "grad_norm": 0.0,
+      "learning_rate": 6.135769096402011e-08,
+      "loss": 0.7719,
+      "step": 35165
+    },
+    {
+      "epoch": 0.9658070363351734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.12593436577813e-08,
+      "loss": 0.8673,
+      "step": 35166
+    },
+    {
+      "epoch": 0.9658345005630167,
+      "grad_norm": 0.0,
+      "learning_rate": 6.116107499070767e-08,
+      "loss": 0.8535,
+      "step": 35167
+    },
+    {
+      "epoch": 0.96586196479086,
+      "grad_norm": 0.0,
+      "learning_rate": 6.106288496357748e-08,
+      "loss": 0.8794,
+      "step": 35168
+    },
+    {
+      "epoch": 0.9658894290187031,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0964773577169e-08,
+      "loss": 0.6927,
+      "step": 35169
+    },
+    {
+      "epoch": 0.9659168932465464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.086674083225608e-08,
+      "loss": 0.7855,
+      "step": 35170
+    },
+    {
+      "epoch": 0.9659443574743896,
+      "grad_norm": 0.0,
+      "learning_rate": 6.076878672961583e-08,
+      "loss": 0.7964,
+      "step": 35171
+    },
+    {
+      "epoch": 0.9659718217022328,
+      "grad_norm": 0.0,
+      "learning_rate": 6.067091127002322e-08,
+      "loss": 0.7927,
+      "step": 35172
+    },
+    {
+      "epoch": 0.9659992859300761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.057311445425207e-08,
+      "loss": 0.9317,
+      "step": 35173
+    },
+    {
+      "epoch": 0.9660267501579193,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04753962830773e-08,
+      "loss": 0.7404,
+      "step": 35174
+    },
+    {
+      "epoch": 0.9660542143857626,
+      "grad_norm": 0.0,
+      "learning_rate": 6.037775675727165e-08,
+      "loss": 0.7825,
+      "step": 35175
+    },
+    {
+      "epoch": 0.9660816786136058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.02801958776067e-08,
+      "loss": 0.7506,
+      "step": 35176
+    },
+    {
+      "epoch": 0.966109142841449,
+      "grad_norm": 0.0,
+      "learning_rate": 6.018271364485518e-08,
+      "loss": 0.8295,
+      "step": 35177
+    },
+    {
+      "epoch": 0.9661366070692923,
+      "grad_norm": 0.0,
+      "learning_rate": 6.008531005978868e-08,
+      "loss": 0.6483,
+      "step": 35178
+    },
+    {
+      "epoch": 0.9661640712971354,
+      "grad_norm": 0.0,
+      "learning_rate": 5.998798512317661e-08,
+      "loss": 0.8426,
+      "step": 35179
+    },
+    {
+      "epoch": 0.9661915355249787,
+      "grad_norm": 0.0,
+      "learning_rate": 5.989073883579055e-08,
+      "loss": 0.8196,
+      "step": 35180
+    },
+    {
+      "epoch": 0.966218999752822,
+      "grad_norm": 0.0,
+      "learning_rate": 5.97935711983999e-08,
+      "loss": 0.7188,
+      "step": 35181
+    },
+    {
+      "epoch": 0.9662464639806652,
+      "grad_norm": 0.0,
+      "learning_rate": 5.969648221177294e-08,
+      "loss": 0.7601,
+      "step": 35182
+    },
+    {
+      "epoch": 0.9662739282085084,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9599471876677914e-08,
+      "loss": 0.8103,
+      "step": 35183
+    },
+    {
+      "epoch": 0.9663013924363516,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9502540193882016e-08,
+      "loss": 0.7944,
+      "step": 35184
+    },
+    {
+      "epoch": 0.9663288566641949,
+      "grad_norm": 0.0,
+      "learning_rate": 5.94056871641524e-08,
+      "loss": 0.8378,
+      "step": 35185
+    },
+    {
+      "epoch": 0.9663563208920382,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9308912788255124e-08,
+      "loss": 0.6832,
+      "step": 35186
+    },
+    {
+      "epoch": 0.9663837851198813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9212217066957346e-08,
+      "loss": 0.7918,
+      "step": 35187
+    },
+    {
+      "epoch": 0.9664112493477246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.91156000010229e-08,
+      "loss": 0.7856,
+      "step": 35188
+    },
+    {
+      "epoch": 0.9664387135755679,
+      "grad_norm": 0.0,
+      "learning_rate": 5.901906159121673e-08,
+      "loss": 0.807,
+      "step": 35189
+    },
+    {
+      "epoch": 0.966466177803411,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892260183830267e-08,
+      "loss": 0.7451,
+      "step": 35190
+    },
+    {
+      "epoch": 0.9664936420312543,
+      "grad_norm": 0.0,
+      "learning_rate": 5.882622074304345e-08,
+      "loss": 0.8446,
+      "step": 35191
+    },
+    {
+      "epoch": 0.9665211062590975,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8729918306201786e-08,
+      "loss": 0.8141,
+      "step": 35192
+    },
+    {
+      "epoch": 0.9665485704869408,
+      "grad_norm": 0.0,
+      "learning_rate": 5.863369452854151e-08,
+      "loss": 0.7442,
+      "step": 35193
+    },
+    {
+      "epoch": 0.966576034714784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.85375494108209e-08,
+      "loss": 0.7695,
+      "step": 35194
+    },
+    {
+      "epoch": 0.9666034989426272,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8441482953801584e-08,
+      "loss": 0.8491,
+      "step": 35195
+    },
+    {
+      "epoch": 0.9666309631704705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.834549515824517e-08,
+      "loss": 0.7382,
+      "step": 35196
+    },
+    {
+      "epoch": 0.9666584273983136,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824958602490993e-08,
+      "loss": 0.7699,
+      "step": 35197
+    },
+    {
+      "epoch": 0.9666858916261569,
+      "grad_norm": 0.0,
+      "learning_rate": 5.815375555455527e-08,
+      "loss": 0.6774,
+      "step": 35198
+    },
+    {
+      "epoch": 0.9667133558540002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.805800374793946e-08,
+      "loss": 0.7759,
+      "step": 35199
+    },
+    {
+      "epoch": 0.9667408200818434,
+      "grad_norm": 0.0,
+      "learning_rate": 5.796233060581968e-08,
+      "loss": 0.7317,
+      "step": 35200
+    },
+    {
+      "epoch": 0.9667682843096866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7866736128952e-08,
+      "loss": 0.8359,
+      "step": 35201
+    },
+    {
+      "epoch": 0.9667957485375299,
+      "grad_norm": 0.0,
+      "learning_rate": 5.777122031809468e-08,
+      "loss": 0.8099,
+      "step": 35202
+    },
+    {
+      "epoch": 0.9668232127653731,
+      "grad_norm": 0.0,
+      "learning_rate": 5.767578317400269e-08,
+      "loss": 0.9005,
+      "step": 35203
+    },
+    {
+      "epoch": 0.9668506769932164,
+      "grad_norm": 0.0,
+      "learning_rate": 5.758042469743208e-08,
+      "loss": 0.8401,
+      "step": 35204
+    },
+    {
+      "epoch": 0.9668781412210595,
+      "grad_norm": 0.0,
+      "learning_rate": 5.748514488913559e-08,
+      "loss": 0.8616,
+      "step": 35205
+    },
+    {
+      "epoch": 0.9669056054489028,
+      "grad_norm": 0.0,
+      "learning_rate": 5.738994374986817e-08,
+      "loss": 0.7796,
+      "step": 35206
+    },
+    {
+      "epoch": 0.9669330696767461,
+      "grad_norm": 0.0,
+      "learning_rate": 5.729482128038366e-08,
+      "loss": 0.8617,
+      "step": 35207
+    },
+    {
+      "epoch": 0.9669605339045892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.719977748143257e-08,
+      "loss": 0.8299,
+      "step": 35208
+    },
+    {
+      "epoch": 0.9669879981324325,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7104812353768744e-08,
+      "loss": 0.8014,
+      "step": 35209
+    },
+    {
+      "epoch": 0.9670154623602757,
+      "grad_norm": 0.0,
+      "learning_rate": 5.70099258981438e-08,
+      "loss": 0.8075,
+      "step": 35210
+    },
+    {
+      "epoch": 0.967042926588119,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6915118115307146e-08,
+      "loss": 0.7895,
+      "step": 35211
+    },
+    {
+      "epoch": 0.9670703908159622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.682038900601039e-08,
+      "loss": 0.7339,
+      "step": 35212
+    },
+    {
+      "epoch": 0.9670978550438054,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6725738571001834e-08,
+      "loss": 0.8075,
+      "step": 35213
+    },
+    {
+      "epoch": 0.9671253192716487,
+      "grad_norm": 0.0,
+      "learning_rate": 5.663116681103087e-08,
+      "loss": 0.7883,
+      "step": 35214
+    },
+    {
+      "epoch": 0.967152783499492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.653667372684579e-08,
+      "loss": 0.8117,
+      "step": 35215
+    },
+    {
+      "epoch": 0.9671802477273351,
+      "grad_norm": 0.0,
+      "learning_rate": 5.644225931919489e-08,
+      "loss": 0.7517,
+      "step": 35216
+    },
+    {
+      "epoch": 0.9672077119551784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.634792358882424e-08,
+      "loss": 0.7591,
+      "step": 35217
+    },
+    {
+      "epoch": 0.9672351761830216,
+      "grad_norm": 0.0,
+      "learning_rate": 5.62536665364799e-08,
+      "loss": 0.8098,
+      "step": 35218
+    },
+    {
+      "epoch": 0.9672626404108648,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6159488162909056e-08,
+      "loss": 0.7322,
+      "step": 35219
+    },
+    {
+      "epoch": 0.9672901046387081,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6065388468855564e-08,
+      "loss": 0.8557,
+      "step": 35220
+    },
+    {
+      "epoch": 0.9673175688665513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5971367455065486e-08,
+      "loss": 0.7272,
+      "step": 35221
+    },
+    {
+      "epoch": 0.9673450330943946,
+      "grad_norm": 0.0,
+      "learning_rate": 5.587742512228156e-08,
+      "loss": 0.8775,
+      "step": 35222
+    },
+    {
+      "epoch": 0.9673724973222377,
+      "grad_norm": 0.0,
+      "learning_rate": 5.578356147124764e-08,
+      "loss": 0.715,
+      "step": 35223
+    },
+    {
+      "epoch": 0.967399961550081,
+      "grad_norm": 0.0,
+      "learning_rate": 5.568977650270646e-08,
+      "loss": 0.7787,
+      "step": 35224
+    },
+    {
+      "epoch": 0.9674274257779243,
+      "grad_norm": 0.0,
+      "learning_rate": 5.559607021739854e-08,
+      "loss": 0.8381,
+      "step": 35225
+    },
+    {
+      "epoch": 0.9674548900057675,
+      "grad_norm": 0.0,
+      "learning_rate": 5.550244261606774e-08,
+      "loss": 0.8222,
+      "step": 35226
+    },
+    {
+      "epoch": 0.9674823542336107,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5408893699453456e-08,
+      "loss": 0.7812,
+      "step": 35227
+    },
+    {
+      "epoch": 0.967509818461454,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5315423468296216e-08,
+      "loss": 0.8203,
+      "step": 35228
+    },
+    {
+      "epoch": 0.9675372826892972,
+      "grad_norm": 0.0,
+      "learning_rate": 5.522203192333542e-08,
+      "loss": 0.8399,
+      "step": 35229
+    },
+    {
+      "epoch": 0.9675647469171405,
+      "grad_norm": 0.0,
+      "learning_rate": 5.512871906531048e-08,
+      "loss": 0.7858,
+      "step": 35230
+    },
+    {
+      "epoch": 0.9675922111449836,
+      "grad_norm": 0.0,
+      "learning_rate": 5.503548489495969e-08,
+      "loss": 0.8433,
+      "step": 35231
+    },
+    {
+      "epoch": 0.9676196753728269,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4942329413019134e-08,
+      "loss": 0.8279,
+      "step": 35232
+    },
+    {
+      "epoch": 0.9676471396006702,
+      "grad_norm": 0.0,
+      "learning_rate": 5.484925262022822e-08,
+      "loss": 0.7857,
+      "step": 35233
+    },
+    {
+      "epoch": 0.9676746038285133,
+      "grad_norm": 0.0,
+      "learning_rate": 5.475625451732192e-08,
+      "loss": 0.8845,
+      "step": 35234
+    },
+    {
+      "epoch": 0.9677020680563566,
+      "grad_norm": 0.0,
+      "learning_rate": 5.466333510503741e-08,
+      "loss": 0.7577,
+      "step": 35235
+    },
+    {
+      "epoch": 0.9677295322841998,
+      "grad_norm": 0.0,
+      "learning_rate": 5.457049438410744e-08,
+      "loss": 0.6893,
+      "step": 35236
+    },
+    {
+      "epoch": 0.967756996512043,
+      "grad_norm": 0.0,
+      "learning_rate": 5.447773235526921e-08,
+      "loss": 0.7789,
+      "step": 35237
+    },
+    {
+      "epoch": 0.9677844607398863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4385049019256565e-08,
+      "loss": 0.813,
+      "step": 35238
+    },
+    {
+      "epoch": 0.9678119249677295,
+      "grad_norm": 0.0,
+      "learning_rate": 5.429244437680115e-08,
+      "loss": 0.815,
+      "step": 35239
+    },
+    {
+      "epoch": 0.9678393891955728,
+      "grad_norm": 0.0,
+      "learning_rate": 5.419991842863681e-08,
+      "loss": 0.7946,
+      "step": 35240
+    },
+    {
+      "epoch": 0.967866853423416,
+      "grad_norm": 0.0,
+      "learning_rate": 5.410747117549409e-08,
+      "loss": 0.7245,
+      "step": 35241
+    },
+    {
+      "epoch": 0.9678943176512592,
+      "grad_norm": 0.0,
+      "learning_rate": 5.401510261810683e-08,
+      "loss": 0.7643,
+      "step": 35242
+    },
+    {
+      "epoch": 0.9679217818791025,
+      "grad_norm": 0.0,
+      "learning_rate": 5.392281275720446e-08,
+      "loss": 0.8316,
+      "step": 35243
+    },
+    {
+      "epoch": 0.9679492461069457,
+      "grad_norm": 0.0,
+      "learning_rate": 5.38306015935175e-08,
+      "loss": 0.7797,
+      "step": 35244
+    },
+    {
+      "epoch": 0.9679767103347889,
+      "grad_norm": 0.0,
+      "learning_rate": 5.373846912777647e-08,
+      "loss": 0.7828,
+      "step": 35245
+    },
+    {
+      "epoch": 0.9680041745626322,
+      "grad_norm": 0.0,
+      "learning_rate": 5.364641536070858e-08,
+      "loss": 0.8969,
+      "step": 35246
+    },
+    {
+      "epoch": 0.9680316387904754,
+      "grad_norm": 0.0,
+      "learning_rate": 5.355444029304324e-08,
+      "loss": 0.8062,
+      "step": 35247
+    },
+    {
+      "epoch": 0.9680591030183187,
+      "grad_norm": 0.0,
+      "learning_rate": 5.346254392550876e-08,
+      "loss": 0.7024,
+      "step": 35248
+    },
+    {
+      "epoch": 0.9680865672461618,
+      "grad_norm": 0.0,
+      "learning_rate": 5.337072625883011e-08,
+      "loss": 0.7787,
+      "step": 35249
+    },
+    {
+      "epoch": 0.9681140314740051,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32789872937356e-08,
+      "loss": 0.744,
+      "step": 35250
+    },
+    {
+      "epoch": 0.9681414957018484,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3187327030951305e-08,
+      "loss": 0.8761,
+      "step": 35251
+    },
+    {
+      "epoch": 0.9681689599296915,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309574547120222e-08,
+      "loss": 0.8792,
+      "step": 35252
+    },
+    {
+      "epoch": 0.9681964241575348,
+      "grad_norm": 0.0,
+      "learning_rate": 5.30042426152122e-08,
+      "loss": 0.7849,
+      "step": 35253
+    },
+    {
+      "epoch": 0.9682238883853781,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2912818463706215e-08,
+      "loss": 0.8095,
+      "step": 35254
+    },
+    {
+      "epoch": 0.9682513526132213,
+      "grad_norm": 0.0,
+      "learning_rate": 5.282147301740592e-08,
+      "loss": 0.7854,
+      "step": 35255
+    },
+    {
+      "epoch": 0.9682788168410645,
+      "grad_norm": 0.0,
+      "learning_rate": 5.273020627703629e-08,
+      "loss": 0.7419,
+      "step": 35256
+    },
+    {
+      "epoch": 0.9683062810689077,
+      "grad_norm": 0.0,
+      "learning_rate": 5.263901824331785e-08,
+      "loss": 0.786,
+      "step": 35257
+    },
+    {
+      "epoch": 0.968333745296751,
+      "grad_norm": 0.0,
+      "learning_rate": 5.254790891697337e-08,
+      "loss": 0.7513,
+      "step": 35258
+    },
+    {
+      "epoch": 0.9683612095245943,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2456878298723366e-08,
+      "loss": 0.8492,
+      "step": 35259
+    },
+    {
+      "epoch": 0.9683886737524374,
+      "grad_norm": 0.0,
+      "learning_rate": 5.236592638928728e-08,
+      "loss": 0.7603,
+      "step": 35260
+    },
+    {
+      "epoch": 0.9684161379802807,
+      "grad_norm": 0.0,
+      "learning_rate": 5.227505318938453e-08,
+      "loss": 0.8375,
+      "step": 35261
+    },
+    {
+      "epoch": 0.9684436022081239,
+      "grad_norm": 0.0,
+      "learning_rate": 5.218425869973676e-08,
+      "loss": 0.8316,
+      "step": 35262
+    },
+    {
+      "epoch": 0.9684710664359671,
+      "grad_norm": 0.0,
+      "learning_rate": 5.209354292105895e-08,
+      "loss": 0.8339,
+      "step": 35263
+    },
+    {
+      "epoch": 0.9684985306638104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.200290585407164e-08,
+      "loss": 0.8537,
+      "step": 35264
+    },
+    {
+      "epoch": 0.9685259948916536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1912347499489813e-08,
+      "loss": 0.8396,
+      "step": 35265
+    },
+    {
+      "epoch": 0.9685534591194969,
+      "grad_norm": 0.0,
+      "learning_rate": 5.182186785803178e-08,
+      "loss": 0.8006,
+      "step": 35266
+    },
+    {
+      "epoch": 0.9685809233473401,
+      "grad_norm": 0.0,
+      "learning_rate": 5.173146693041253e-08,
+      "loss": 0.7792,
+      "step": 35267
+    },
+    {
+      "epoch": 0.9686083875751833,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1641144717347047e-08,
+      "loss": 0.7867,
+      "step": 35268
+    },
+    {
+      "epoch": 0.9686358518030266,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1550901219550306e-08,
+      "loss": 0.8835,
+      "step": 35269
+    },
+    {
+      "epoch": 0.9686633160308697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1460736437737305e-08,
+      "loss": 0.8075,
+      "step": 35270
+    },
+    {
+      "epoch": 0.968690780258713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.137065037262079e-08,
+      "loss": 0.799,
+      "step": 35271
+    },
+    {
+      "epoch": 0.9687182444865563,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1280643024912427e-08,
+      "loss": 0.746,
+      "step": 35272
+    },
+    {
+      "epoch": 0.9687457087143995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.119071439532608e-08,
+      "loss": 0.806,
+      "step": 35273
+    },
+    {
+      "epoch": 0.9687731729422427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.11008644845723e-08,
+      "loss": 0.7478,
+      "step": 35274
+    },
+    {
+      "epoch": 0.9688006371700859,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1011093293362733e-08,
+      "loss": 0.8336,
+      "step": 35275
+    },
+    {
+      "epoch": 0.9688281013979292,
+      "grad_norm": 0.0,
+      "learning_rate": 5.092140082240682e-08,
+      "loss": 0.7227,
+      "step": 35276
+    },
+    {
+      "epoch": 0.9688555656257725,
+      "grad_norm": 0.0,
+      "learning_rate": 5.083178707241509e-08,
+      "loss": 0.8128,
+      "step": 35277
+    },
+    {
+      "epoch": 0.9688830298536156,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0742252044095886e-08,
+      "loss": 0.826,
+      "step": 35278
+    },
+    {
+      "epoch": 0.9689104940814589,
+      "grad_norm": 0.0,
+      "learning_rate": 5.065279573815751e-08,
+      "loss": 0.8574,
+      "step": 35279
+    },
+    {
+      "epoch": 0.9689379583093022,
+      "grad_norm": 0.0,
+      "learning_rate": 5.05634181553083e-08,
+      "loss": 0.7634,
+      "step": 35280
+    },
+    {
+      "epoch": 0.9689654225371453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0474119296256564e-08,
+      "loss": 0.7584,
+      "step": 35281
+    },
+    {
+      "epoch": 0.9689928867649886,
+      "grad_norm": 0.0,
+      "learning_rate": 5.03848991617073e-08,
+      "loss": 0.7512,
+      "step": 35282
+    },
+    {
+      "epoch": 0.9690203509928318,
+      "grad_norm": 0.0,
+      "learning_rate": 5.02957577523655e-08,
+      "loss": 0.8278,
+      "step": 35283
+    },
+    {
+      "epoch": 0.9690478152206751,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0206695068939495e-08,
+      "loss": 0.7237,
+      "step": 35284
+    },
+    {
+      "epoch": 0.9690752794485183,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0117711112130926e-08,
+      "loss": 0.8177,
+      "step": 35285
+    },
+    {
+      "epoch": 0.9691027436763615,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0028805882645916e-08,
+      "loss": 0.8227,
+      "step": 35286
+    },
+    {
+      "epoch": 0.9691302079042048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.993997938118722e-08,
+      "loss": 0.8682,
+      "step": 35287
+    },
+    {
+      "epoch": 0.969157672132048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.985123160845762e-08,
+      "loss": 0.8879,
+      "step": 35288
+    },
+    {
+      "epoch": 0.9691851363598912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.976256256515988e-08,
+      "loss": 0.7309,
+      "step": 35289
+    },
+    {
+      "epoch": 0.9692126005877345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.967397225199455e-08,
+      "loss": 0.8356,
+      "step": 35290
+    },
+    {
+      "epoch": 0.9692400648155777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.95854606696633e-08,
+      "loss": 0.7659,
+      "step": 35291
+    },
+    {
+      "epoch": 0.969267529043421,
+      "grad_norm": 0.0,
+      "learning_rate": 4.949702781886556e-08,
+      "loss": 0.781,
+      "step": 35292
+    },
+    {
+      "epoch": 0.9692949932712642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9408673700303e-08,
+      "loss": 0.7936,
+      "step": 35293
+    },
+    {
+      "epoch": 0.9693224574991074,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9320398314672835e-08,
+      "loss": 0.7506,
+      "step": 35294
+    },
+    {
+      "epoch": 0.9693499217269507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.923220166267451e-08,
+      "loss": 0.7937,
+      "step": 35295
+    },
+    {
+      "epoch": 0.9693773859547938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.914408374500523e-08,
+      "loss": 0.7597,
+      "step": 35296
+    },
+    {
+      "epoch": 0.9694048501826371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.905604456236224e-08,
+      "loss": 0.8598,
+      "step": 35297
+    },
+    {
+      "epoch": 0.9694323144104804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8968084115443847e-08,
+      "loss": 0.8023,
+      "step": 35298
+    },
+    {
+      "epoch": 0.9694597786383236,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8880202404943956e-08,
+      "loss": 0.8365,
+      "step": 35299
+    },
+    {
+      "epoch": 0.9694872428661668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.879239943155756e-08,
+      "loss": 0.8131,
+      "step": 35300
+    },
+    {
+      "epoch": 0.96951470709401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8704675195981875e-08,
+      "loss": 0.8742,
+      "step": 35301
+    },
+    {
+      "epoch": 0.9695421713218533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.861702969890858e-08,
+      "loss": 0.8001,
+      "step": 35302
+    },
+    {
+      "epoch": 0.9695696355496966,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8529462941032666e-08,
+      "loss": 0.8026,
+      "step": 35303
+    },
+    {
+      "epoch": 0.9695970997775397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8441974923045806e-08,
+      "loss": 0.7371,
+      "step": 35304
+    },
+    {
+      "epoch": 0.969624564005383,
+      "grad_norm": 0.0,
+      "learning_rate": 4.835456564564189e-08,
+      "loss": 0.7275,
+      "step": 35305
+    },
+    {
+      "epoch": 0.9696520282332263,
+      "grad_norm": 0.0,
+      "learning_rate": 4.826723510951148e-08,
+      "loss": 0.8194,
+      "step": 35306
+    },
+    {
+      "epoch": 0.9696794924610694,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8179983315344017e-08,
+      "loss": 0.7166,
+      "step": 35307
+    },
+    {
+      "epoch": 0.9697069566889127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.809281026383339e-08,
+      "loss": 0.7819,
+      "step": 35308
+    },
+    {
+      "epoch": 0.9697344209167559,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8005715955666834e-08,
+      "loss": 0.832,
+      "step": 35309
+    },
+    {
+      "epoch": 0.9697618851445992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.791870039153379e-08,
+      "loss": 0.7521,
+      "step": 35310
+    },
+    {
+      "epoch": 0.9697893493724424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7831763572123714e-08,
+      "loss": 0.7819,
+      "step": 35311
+    },
+    {
+      "epoch": 0.9698168136002856,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774490549812272e-08,
+      "loss": 0.8669,
+      "step": 35312
+    },
+    {
+      "epoch": 0.9698442778281289,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7658126170220253e-08,
+      "loss": 0.8646,
+      "step": 35313
+    },
+    {
+      "epoch": 0.969871742055972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7571425589101326e-08,
+      "loss": 0.8069,
+      "step": 35314
+    },
+    {
+      "epoch": 0.9698992062838153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.748480375545206e-08,
+      "loss": 0.88,
+      "step": 35315
+    },
+    {
+      "epoch": 0.9699266705116586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.739826066995856e-08,
+      "loss": 0.8306,
+      "step": 35316
+    },
+    {
+      "epoch": 0.9699541347395018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7311796333305845e-08,
+      "loss": 0.8536,
+      "step": 35317
+    },
+    {
+      "epoch": 0.969981598967345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.722541074617781e-08,
+      "loss": 0.8401,
+      "step": 35318
+    },
+    {
+      "epoch": 0.9700090631951883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7139103909257244e-08,
+      "loss": 0.8228,
+      "step": 35319
+    },
+    {
+      "epoch": 0.9700365274230315,
+      "grad_norm": 0.0,
+      "learning_rate": 4.705287582322693e-08,
+      "loss": 0.7306,
+      "step": 35320
+    },
+    {
+      "epoch": 0.9700639916508748,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6966726488769653e-08,
+      "loss": 0.8571,
+      "step": 35321
+    },
+    {
+      "epoch": 0.9700914558787179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6880655906567095e-08,
+      "loss": 0.8767,
+      "step": 35322
+    },
+    {
+      "epoch": 0.9701189201065612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.679466407730093e-08,
+      "loss": 0.8169,
+      "step": 35323
+    },
+    {
+      "epoch": 0.9701463843344045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6708751001650624e-08,
+      "loss": 0.8473,
+      "step": 35324
+    },
+    {
+      "epoch": 0.9701738485622476,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6622916680295616e-08,
+      "loss": 0.8151,
+      "step": 35325
+    },
+    {
+      "epoch": 0.9702013127900909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6537161113916485e-08,
+      "loss": 0.7556,
+      "step": 35326
+    },
+    {
+      "epoch": 0.9702287770179341,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6451484303190466e-08,
+      "loss": 0.8567,
+      "step": 35327
+    },
+    {
+      "epoch": 0.9702562412457774,
+      "grad_norm": 0.0,
+      "learning_rate": 4.636588624879701e-08,
+      "loss": 0.7993,
+      "step": 35328
+    },
+    {
+      "epoch": 0.9702837054736206,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6280366951411136e-08,
+      "loss": 0.9158,
+      "step": 35329
+    },
+    {
+      "epoch": 0.9703111697014638,
+      "grad_norm": 0.0,
+      "learning_rate": 4.619492641171119e-08,
+      "loss": 0.7625,
+      "step": 35330
+    },
+    {
+      "epoch": 0.9703386339293071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.61095646303722e-08,
+      "loss": 0.871,
+      "step": 35331
+    },
+    {
+      "epoch": 0.9703660981571504,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6024281608070265e-08,
+      "loss": 0.7742,
+      "step": 35332
+    },
+    {
+      "epoch": 0.9703935623849935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.593907734547931e-08,
+      "loss": 0.8067,
+      "step": 35333
+    },
+    {
+      "epoch": 0.9704210266128368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.585395184327546e-08,
+      "loss": 0.7968,
+      "step": 35334
+    },
+    {
+      "epoch": 0.97044849084068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.57689051021315e-08,
+      "loss": 0.7407,
+      "step": 35335
+    },
+    {
+      "epoch": 0.9704759550685232,
+      "grad_norm": 0.0,
+      "learning_rate": 4.568393712271801e-08,
+      "loss": 0.713,
+      "step": 35336
+    },
+    {
+      "epoch": 0.9705034192963665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.559904790571001e-08,
+      "loss": 0.7895,
+      "step": 35337
+    },
+    {
+      "epoch": 0.9705308835242097,
+      "grad_norm": 0.0,
+      "learning_rate": 4.551423745177918e-08,
+      "loss": 0.8249,
+      "step": 35338
+    },
+    {
+      "epoch": 0.970558347752053,
+      "grad_norm": 0.0,
+      "learning_rate": 4.542950576159389e-08,
+      "loss": 0.7664,
+      "step": 35339
+    },
+    {
+      "epoch": 0.9705858119798962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.534485283582801e-08,
+      "loss": 0.7435,
+      "step": 35340
+    },
+    {
+      "epoch": 0.9706132762077394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.526027867514882e-08,
+      "loss": 0.8417,
+      "step": 35341
+    },
+    {
+      "epoch": 0.9706407404355827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.517578328022687e-08,
+      "loss": 0.8211,
+      "step": 35342
+    },
+    {
+      "epoch": 0.9706682046634258,
+      "grad_norm": 0.0,
+      "learning_rate": 4.509136665173053e-08,
+      "loss": 0.7939,
+      "step": 35343
+    },
+    {
+      "epoch": 0.9706956688912691,
+      "grad_norm": 0.0,
+      "learning_rate": 4.500702879032703e-08,
+      "loss": 0.7768,
+      "step": 35344
+    },
+    {
+      "epoch": 0.9707231331191124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4922769696683633e-08,
+      "loss": 0.7717,
+      "step": 35345
+    },
+    {
+      "epoch": 0.9707505973469556,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4838589371468675e-08,
+      "loss": 0.8477,
+      "step": 35346
+    },
+    {
+      "epoch": 0.9707780615747988,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4754487815346085e-08,
+      "loss": 0.8811,
+      "step": 35347
+    },
+    {
+      "epoch": 0.970805525802642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.467046502898198e-08,
+      "loss": 0.8861,
+      "step": 35348
+    },
+    {
+      "epoch": 0.9708329900304853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4586521013042504e-08,
+      "loss": 0.7895,
+      "step": 35349
+    },
+    {
+      "epoch": 0.9708604542583286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.450265576819046e-08,
+      "loss": 0.8544,
+      "step": 35350
+    },
+    {
+      "epoch": 0.9708879184861717,
+      "grad_norm": 0.0,
+      "learning_rate": 4.441886929508976e-08,
+      "loss": 0.7352,
+      "step": 35351
+    },
+    {
+      "epoch": 0.970915382714015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.433516159440321e-08,
+      "loss": 0.7091,
+      "step": 35352
+    },
+    {
+      "epoch": 0.9709428469418583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.425153266679361e-08,
+      "loss": 0.807,
+      "step": 35353
+    },
+    {
+      "epoch": 0.9709703111697014,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4167982512922644e-08,
+      "loss": 0.7399,
+      "step": 35354
+    },
+    {
+      "epoch": 0.9709977753975447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.40845111334498e-08,
+      "loss": 0.8445,
+      "step": 35355
+    },
+    {
+      "epoch": 0.9710252396253879,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4001118529037875e-08,
+      "loss": 0.8149,
+      "step": 35356
+    },
+    {
+      "epoch": 0.9710527038532312,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391780470034524e-08,
+      "loss": 0.7588,
+      "step": 35357
+    },
+    {
+      "epoch": 0.9710801680810744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3834569648032456e-08,
+      "loss": 0.7149,
+      "step": 35358
+    },
+    {
+      "epoch": 0.9711076323089176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.375141337275679e-08,
+      "loss": 0.7176,
+      "step": 35359
+    },
+    {
+      "epoch": 0.9711350965367609,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3668335875176605e-08,
+      "loss": 0.897,
+      "step": 35360
+    },
+    {
+      "epoch": 0.971162560764604,
+      "grad_norm": 0.0,
+      "learning_rate": 4.358533715594915e-08,
+      "loss": 0.8686,
+      "step": 35361
+    },
+    {
+      "epoch": 0.9711900249924473,
+      "grad_norm": 0.0,
+      "learning_rate": 4.350241721573056e-08,
+      "loss": 0.8659,
+      "step": 35362
+    },
+    {
+      "epoch": 0.9712174892202906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.34195760551781e-08,
+      "loss": 0.8099,
+      "step": 35363
+    },
+    {
+      "epoch": 0.9712449534481338,
+      "grad_norm": 0.0,
+      "learning_rate": 4.333681367494791e-08,
+      "loss": 0.7986,
+      "step": 35364
+    },
+    {
+      "epoch": 0.971272417675977,
+      "grad_norm": 0.0,
+      "learning_rate": 4.325413007569168e-08,
+      "loss": 0.782,
+      "step": 35365
+    },
+    {
+      "epoch": 0.9712998819038203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.317152525806778e-08,
+      "loss": 0.7893,
+      "step": 35366
+    },
+    {
+      "epoch": 0.9713273461316635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.308899922272569e-08,
+      "loss": 0.922,
+      "step": 35367
+    },
+    {
+      "epoch": 0.9713548103595068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.300655197032155e-08,
+      "loss": 0.9144,
+      "step": 35368
+    },
+    {
+      "epoch": 0.9713822745873499,
+      "grad_norm": 0.0,
+      "learning_rate": 4.292418350150596e-08,
+      "loss": 0.9238,
+      "step": 35369
+    },
+    {
+      "epoch": 0.9714097388151932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2841893816931715e-08,
+      "loss": 0.7648,
+      "step": 35370
+    },
+    {
+      "epoch": 0.9714372030430365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.275968291724941e-08,
+      "loss": 0.7917,
+      "step": 35371
+    },
+    {
+      "epoch": 0.9714646672708797,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2677550803109646e-08,
+      "loss": 0.7941,
+      "step": 35372
+    },
+    {
+      "epoch": 0.9714921314987229,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2595497475161894e-08,
+      "loss": 0.7438,
+      "step": 35373
+    },
+    {
+      "epoch": 0.9715195957265661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.251352293405564e-08,
+      "loss": 0.6684,
+      "step": 35374
+    },
+    {
+      "epoch": 0.9715470599544094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.243162718043925e-08,
+      "loss": 0.8195,
+      "step": 35375
+    },
+    {
+      "epoch": 0.9715745241822527,
+      "grad_norm": 0.0,
+      "learning_rate": 4.23498102149622e-08,
+      "loss": 0.7886,
+      "step": 35376
+    },
+    {
+      "epoch": 0.9716019884100958,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2268072038269546e-08,
+      "loss": 0.7554,
+      "step": 35377
+    },
+    {
+      "epoch": 0.9716294526379391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.218641265100965e-08,
+      "loss": 0.8033,
+      "step": 35378
+    },
+    {
+      "epoch": 0.9716569168657824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.210483205382754e-08,
+      "loss": 0.7401,
+      "step": 35379
+    },
+    {
+      "epoch": 0.9716843810936255,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2023330247370484e-08,
+      "loss": 0.8533,
+      "step": 35380
+    },
+    {
+      "epoch": 0.9717118453214688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.194190723228131e-08,
+      "loss": 0.8629,
+      "step": 35381
+    },
+    {
+      "epoch": 0.971739309549312,
+      "grad_norm": 0.0,
+      "learning_rate": 4.186056300920505e-08,
+      "loss": 0.8491,
+      "step": 35382
+    },
+    {
+      "epoch": 0.9717667737771553,
+      "grad_norm": 0.0,
+      "learning_rate": 4.177929757878674e-08,
+      "loss": 0.783,
+      "step": 35383
+    },
+    {
+      "epoch": 0.9717942380049985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.169811094166698e-08,
+      "loss": 0.8175,
+      "step": 35384
+    },
+    {
+      "epoch": 0.9718217022328417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1617003098488596e-08,
+      "loss": 0.6981,
+      "step": 35385
+    },
+    {
+      "epoch": 0.971849166460685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.153597404989551e-08,
+      "loss": 0.7722,
+      "step": 35386
+    },
+    {
+      "epoch": 0.9718766306885281,
+      "grad_norm": 0.0,
+      "learning_rate": 4.145502379652611e-08,
+      "loss": 0.8887,
+      "step": 35387
+    },
+    {
+      "epoch": 0.9719040949163714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1374152339023196e-08,
+      "loss": 0.743,
+      "step": 35388
+    },
+    {
+      "epoch": 0.9719315591442147,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1293359678024056e-08,
+      "loss": 0.7718,
+      "step": 35389
+    },
+    {
+      "epoch": 0.9719590233720579,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1212645814171506e-08,
+      "loss": 0.8311,
+      "step": 35390
+    },
+    {
+      "epoch": 0.9719864875999011,
+      "grad_norm": 0.0,
+      "learning_rate": 4.113201074810058e-08,
+      "loss": 0.7524,
+      "step": 35391
+    },
+    {
+      "epoch": 0.9720139518277444,
+      "grad_norm": 0.0,
+      "learning_rate": 4.105145448045078e-08,
+      "loss": 0.7141,
+      "step": 35392
+    },
+    {
+      "epoch": 0.9720414160555876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.097097701186048e-08,
+      "loss": 0.8418,
+      "step": 35393
+    },
+    {
+      "epoch": 0.9720688802834309,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0890578342965834e-08,
+      "loss": 0.9109,
+      "step": 35394
+    },
+    {
+      "epoch": 0.972096344511274,
+      "grad_norm": 0.0,
+      "learning_rate": 4.08102584744019e-08,
+      "loss": 0.7539,
+      "step": 35395
+    },
+    {
+      "epoch": 0.9721238087391173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.073001740680482e-08,
+      "loss": 0.7197,
+      "step": 35396
+    },
+    {
+      "epoch": 0.9721512729669606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.064985514081077e-08,
+      "loss": 0.7372,
+      "step": 35397
+    },
+    {
+      "epoch": 0.9721787371948037,
+      "grad_norm": 0.0,
+      "learning_rate": 4.056977167705256e-08,
+      "loss": 0.7889,
+      "step": 35398
+    },
+    {
+      "epoch": 0.972206201422647,
+      "grad_norm": 0.0,
+      "learning_rate": 4.048976701616414e-08,
+      "loss": 0.8141,
+      "step": 35399
+    },
+    {
+      "epoch": 0.9722336656504902,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0409841158778326e-08,
+      "loss": 0.756,
+      "step": 35400
+    },
+    {
+      "epoch": 0.9722611298783335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.032999410552796e-08,
+      "loss": 0.8265,
+      "step": 35401
+    },
+    {
+      "epoch": 0.9722885941061767,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0250225857044746e-08,
+      "loss": 0.7728,
+      "step": 35402
+    },
+    {
+      "epoch": 0.9723160583340199,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0170536413959296e-08,
+      "loss": 0.7555,
+      "step": 35403
+    },
+    {
+      "epoch": 0.9723435225618632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.009092577690332e-08,
+      "loss": 0.8924,
+      "step": 35404
+    },
+    {
+      "epoch": 0.9723709867897065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.001139394650633e-08,
+      "loss": 0.8474,
+      "step": 35405
+    },
+    {
+      "epoch": 0.9723984510175496,
+      "grad_norm": 0.0,
+      "learning_rate": 3.99319409233967e-08,
+      "loss": 0.7828,
+      "step": 35406
+    },
+    {
+      "epoch": 0.9724259152453929,
+      "grad_norm": 0.0,
+      "learning_rate": 3.985256670820392e-08,
+      "loss": 0.8241,
+      "step": 35407
+    },
+    {
+      "epoch": 0.9724533794732361,
+      "grad_norm": 0.0,
+      "learning_rate": 3.977327130155528e-08,
+      "loss": 0.8871,
+      "step": 35408
+    },
+    {
+      "epoch": 0.9724808437010793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9694054704079166e-08,
+      "loss": 0.8145,
+      "step": 35409
+    },
+    {
+      "epoch": 0.9725083079289226,
+      "grad_norm": 0.0,
+      "learning_rate": 3.961491691640174e-08,
+      "loss": 0.7792,
+      "step": 35410
+    },
+    {
+      "epoch": 0.9725357721567658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.953585793915027e-08,
+      "loss": 0.8275,
+      "step": 35411
+    },
+    {
+      "epoch": 0.9725632363846091,
+      "grad_norm": 0.0,
+      "learning_rate": 3.945687777294871e-08,
+      "loss": 0.8191,
+      "step": 35412
+    },
+    {
+      "epoch": 0.9725907006124522,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9377976418422116e-08,
+      "loss": 0.8026,
+      "step": 35413
+    },
+    {
+      "epoch": 0.9726181648402955,
+      "grad_norm": 0.0,
+      "learning_rate": 3.929915387619665e-08,
+      "loss": 0.8332,
+      "step": 35414
+    },
+    {
+      "epoch": 0.9726456290681388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.922041014689404e-08,
+      "loss": 0.7474,
+      "step": 35415
+    },
+    {
+      "epoch": 0.972673093295982,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9141745231137115e-08,
+      "loss": 0.8252,
+      "step": 35416
+    },
+    {
+      "epoch": 0.9727005575238252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.906315912955094e-08,
+      "loss": 0.7993,
+      "step": 35417
+    },
+    {
+      "epoch": 0.9727280217516685,
+      "grad_norm": 0.0,
+      "learning_rate": 3.898465184275391e-08,
+      "loss": 0.7253,
+      "step": 35418
+    },
+    {
+      "epoch": 0.9727554859795117,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8906223371369955e-08,
+      "loss": 0.8481,
+      "step": 35419
+    },
+    {
+      "epoch": 0.972782950207355,
+      "grad_norm": 0.0,
+      "learning_rate": 3.882787371601748e-08,
+      "loss": 0.772,
+      "step": 35420
+    },
+    {
+      "epoch": 0.9728104144351981,
+      "grad_norm": 0.0,
+      "learning_rate": 3.874960287731822e-08,
+      "loss": 0.8795,
+      "step": 35421
+    },
+    {
+      "epoch": 0.9728378786630414,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8671410855890544e-08,
+      "loss": 0.8358,
+      "step": 35422
+    },
+    {
+      "epoch": 0.9728653428908847,
+      "grad_norm": 0.0,
+      "learning_rate": 3.859329765235398e-08,
+      "loss": 0.8382,
+      "step": 35423
+    },
+    {
+      "epoch": 0.9728928071187278,
+      "grad_norm": 0.0,
+      "learning_rate": 3.851526326732469e-08,
+      "loss": 0.8223,
+      "step": 35424
+    },
+    {
+      "epoch": 0.9729202713465711,
+      "grad_norm": 0.0,
+      "learning_rate": 3.843730770142218e-08,
+      "loss": 0.7564,
+      "step": 35425
+    },
+    {
+      "epoch": 0.9729477355744143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.835943095526262e-08,
+      "loss": 0.8241,
+      "step": 35426
+    },
+    {
+      "epoch": 0.9729751998022576,
+      "grad_norm": 0.0,
+      "learning_rate": 3.828163302946109e-08,
+      "loss": 0.7983,
+      "step": 35427
+    },
+    {
+      "epoch": 0.9730026640301008,
+      "grad_norm": 0.0,
+      "learning_rate": 3.820391392463485e-08,
+      "loss": 0.7494,
+      "step": 35428
+    },
+    {
+      "epoch": 0.973030128257944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.812627364139787e-08,
+      "loss": 0.7855,
+      "step": 35429
+    },
+    {
+      "epoch": 0.9730575924857873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80487121803641e-08,
+      "loss": 0.8278,
+      "step": 35430
+    },
+    {
+      "epoch": 0.9730850567136305,
+      "grad_norm": 0.0,
+      "learning_rate": 3.79712295421486e-08,
+      "loss": 0.8268,
+      "step": 35431
+    },
+    {
+      "epoch": 0.9731125209414737,
+      "grad_norm": 0.0,
+      "learning_rate": 3.789382572736311e-08,
+      "loss": 0.8166,
+      "step": 35432
+    },
+    {
+      "epoch": 0.973139985169317,
+      "grad_norm": 0.0,
+      "learning_rate": 3.781650073662046e-08,
+      "loss": 0.7903,
+      "step": 35433
+    },
+    {
+      "epoch": 0.9731674493971602,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7739254570533514e-08,
+      "loss": 0.7774,
+      "step": 35434
+    },
+    {
+      "epoch": 0.9731949136250034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7662087229712875e-08,
+      "loss": 0.8334,
+      "step": 35435
+    },
+    {
+      "epoch": 0.9732223778528467,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7584998714766954e-08,
+      "loss": 0.8261,
+      "step": 35436
+    },
+    {
+      "epoch": 0.9732498420806899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.75079890263097e-08,
+      "loss": 0.7808,
+      "step": 35437
+    },
+    {
+      "epoch": 0.9732773063085332,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7431058164947297e-08,
+      "loss": 0.8725,
+      "step": 35438
+    },
+    {
+      "epoch": 0.9733047705363763,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7354206131289264e-08,
+      "loss": 0.7568,
+      "step": 35439
+    },
+    {
+      "epoch": 0.9733322347642196,
+      "grad_norm": 0.0,
+      "learning_rate": 3.727743292594399e-08,
+      "loss": 0.7423,
+      "step": 35440
+    },
+    {
+      "epoch": 0.9733596989920629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.720073854951989e-08,
+      "loss": 0.8375,
+      "step": 35441
+    },
+    {
+      "epoch": 0.973387163219906,
+      "grad_norm": 0.0,
+      "learning_rate": 3.712412300262203e-08,
+      "loss": 0.8191,
+      "step": 35442
+    },
+    {
+      "epoch": 0.9734146274477493,
+      "grad_norm": 0.0,
+      "learning_rate": 3.704758628585658e-08,
+      "loss": 0.7853,
+      "step": 35443
+    },
+    {
+      "epoch": 0.9734420916755926,
+      "grad_norm": 0.0,
+      "learning_rate": 3.697112839983086e-08,
+      "loss": 0.7458,
+      "step": 35444
+    },
+    {
+      "epoch": 0.9734695559034358,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68947493451488e-08,
+      "loss": 0.7813,
+      "step": 35445
+    },
+    {
+      "epoch": 0.973497020131279,
+      "grad_norm": 0.0,
+      "learning_rate": 3.681844912241439e-08,
+      "loss": 0.8824,
+      "step": 35446
+    },
+    {
+      "epoch": 0.9735244843591222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.674222773223268e-08,
+      "loss": 0.8242,
+      "step": 35447
+    },
+    {
+      "epoch": 0.9735519485869655,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6666085175204314e-08,
+      "loss": 0.7823,
+      "step": 35448
+    },
+    {
+      "epoch": 0.9735794128148088,
+      "grad_norm": 0.0,
+      "learning_rate": 3.659002145193435e-08,
+      "loss": 0.7352,
+      "step": 35449
+    },
+    {
+      "epoch": 0.9736068770426519,
+      "grad_norm": 0.0,
+      "learning_rate": 3.651403656302344e-08,
+      "loss": 0.7852,
+      "step": 35450
+    },
+    {
+      "epoch": 0.9736343412704952,
+      "grad_norm": 0.0,
+      "learning_rate": 3.64381305090733e-08,
+      "loss": 0.756,
+      "step": 35451
+    },
+    {
+      "epoch": 0.9736618054983384,
+      "grad_norm": 0.0,
+      "learning_rate": 3.636230329068346e-08,
+      "loss": 0.8821,
+      "step": 35452
+    },
+    {
+      "epoch": 0.9736892697261816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6286554908455676e-08,
+      "loss": 0.8329,
+      "step": 35453
+    },
+    {
+      "epoch": 0.9737167339540249,
+      "grad_norm": 0.0,
+      "learning_rate": 3.621088536298723e-08,
+      "loss": 0.8311,
+      "step": 35454
+    },
+    {
+      "epoch": 0.9737441981818681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.613529465487875e-08,
+      "loss": 0.8981,
+      "step": 35455
+    },
+    {
+      "epoch": 0.9737716624097114,
+      "grad_norm": 0.0,
+      "learning_rate": 3.605978278472644e-08,
+      "loss": 0.8152,
+      "step": 35456
+    },
+    {
+      "epoch": 0.9737991266375546,
+      "grad_norm": 0.0,
+      "learning_rate": 3.598434975312981e-08,
+      "loss": 0.7993,
+      "step": 35457
+    },
+    {
+      "epoch": 0.9738265908653978,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5908995560683946e-08,
+      "loss": 0.8065,
+      "step": 35458
+    },
+    {
+      "epoch": 0.9738540550932411,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5833720207986146e-08,
+      "loss": 0.9044,
+      "step": 35459
+    },
+    {
+      "epoch": 0.9738815193210842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5758523695631485e-08,
+      "loss": 0.8269,
+      "step": 35460
+    },
+    {
+      "epoch": 0.9739089835489275,
+      "grad_norm": 0.0,
+      "learning_rate": 3.568340602421505e-08,
+      "loss": 0.8091,
+      "step": 35461
+    },
+    {
+      "epoch": 0.9739364477767708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5608367194331915e-08,
+      "loss": 0.8904,
+      "step": 35462
+    },
+    {
+      "epoch": 0.973963912004614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.553340720657494e-08,
+      "loss": 0.7458,
+      "step": 35463
+    },
+    {
+      "epoch": 0.9739913762324572,
+      "grad_norm": 0.0,
+      "learning_rate": 3.545852606153699e-08,
+      "loss": 0.8691,
+      "step": 35464
+    },
+    {
+      "epoch": 0.9740188404603004,
+      "grad_norm": 0.0,
+      "learning_rate": 3.538372375981203e-08,
+      "loss": 0.768,
+      "step": 35465
+    },
+    {
+      "epoch": 0.9740463046881437,
+      "grad_norm": 0.0,
+      "learning_rate": 3.530900030198958e-08,
+      "loss": 0.7637,
+      "step": 35466
+    },
+    {
+      "epoch": 0.974073768915987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.523435568866362e-08,
+      "loss": 0.81,
+      "step": 35467
+    },
+    {
+      "epoch": 0.9741012331438301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.515978992042257e-08,
+      "loss": 0.8729,
+      "step": 35468
+    },
+    {
+      "epoch": 0.9741286973716734,
+      "grad_norm": 0.0,
+      "learning_rate": 3.508530299785817e-08,
+      "loss": 0.8491,
+      "step": 35469
+    },
+    {
+      "epoch": 0.9741561615995167,
+      "grad_norm": 0.0,
+      "learning_rate": 3.501089492155774e-08,
+      "loss": 0.7987,
+      "step": 35470
+    },
+    {
+      "epoch": 0.9741836258273598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.493656569211301e-08,
+      "loss": 0.8083,
+      "step": 35471
+    },
+    {
+      "epoch": 0.9742110900552031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4862315310107976e-08,
+      "loss": 0.8228,
+      "step": 35472
+    },
+    {
+      "epoch": 0.9742385542830463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.478814377613437e-08,
+      "loss": 0.767,
+      "step": 35473
+    },
+    {
+      "epoch": 0.9742660185108896,
+      "grad_norm": 0.0,
+      "learning_rate": 3.471405109077508e-08,
+      "loss": 0.8026,
+      "step": 35474
+    },
+    {
+      "epoch": 0.9742934827387328,
+      "grad_norm": 0.0,
+      "learning_rate": 3.464003725461962e-08,
+      "loss": 0.8455,
+      "step": 35475
+    },
+    {
+      "epoch": 0.974320946966576,
+      "grad_norm": 0.0,
+      "learning_rate": 3.456610226825308e-08,
+      "loss": 0.8697,
+      "step": 35476
+    },
+    {
+      "epoch": 0.9743484111944193,
+      "grad_norm": 0.0,
+      "learning_rate": 3.449224613225943e-08,
+      "loss": 0.8059,
+      "step": 35477
+    },
+    {
+      "epoch": 0.9743758754222624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4418468847222664e-08,
+      "loss": 0.7421,
+      "step": 35478
+    },
+    {
+      "epoch": 0.9744033396501057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.434477041372786e-08,
+      "loss": 0.7919,
+      "step": 35479
+    },
+    {
+      "epoch": 0.974430803877949,
+      "grad_norm": 0.0,
+      "learning_rate": 3.427115083235677e-08,
+      "loss": 0.7375,
+      "step": 35480
+    },
+    {
+      "epoch": 0.9744582681057922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.419761010369338e-08,
+      "loss": 0.8003,
+      "step": 35481
+    },
+    {
+      "epoch": 0.9744857323336354,
+      "grad_norm": 0.0,
+      "learning_rate": 3.412414822831944e-08,
+      "loss": 0.8722,
+      "step": 35482
+    },
+    {
+      "epoch": 0.9745131965614787,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4050765206814495e-08,
+      "loss": 0.7287,
+      "step": 35483
+    },
+    {
+      "epoch": 0.9745406607893219,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3977461039760294e-08,
+      "loss": 0.8572,
+      "step": 35484
+    },
+    {
+      "epoch": 0.9745681250171652,
+      "grad_norm": 0.0,
+      "learning_rate": 3.390423572773749e-08,
+      "loss": 0.8597,
+      "step": 35485
+    },
+    {
+      "epoch": 0.9745955892450083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.383108927132561e-08,
+      "loss": 0.7886,
+      "step": 35486
+    },
+    {
+      "epoch": 0.9746230534728516,
+      "grad_norm": 0.0,
+      "learning_rate": 3.375802167110087e-08,
+      "loss": 0.877,
+      "step": 35487
+    },
+    {
+      "epoch": 0.9746505177006949,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3685032927645024e-08,
+      "loss": 0.8483,
+      "step": 35488
+    },
+    {
+      "epoch": 0.974677981928538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.361212304153316e-08,
+      "loss": 0.8889,
+      "step": 35489
+    },
+    {
+      "epoch": 0.9747054461563813,
+      "grad_norm": 0.0,
+      "learning_rate": 3.35392920133415e-08,
+      "loss": 0.8387,
+      "step": 35490
+    },
+    {
+      "epoch": 0.9747329103842245,
+      "grad_norm": 0.0,
+      "learning_rate": 3.346653984364956e-08,
+      "loss": 0.8156,
+      "step": 35491
+    },
+    {
+      "epoch": 0.9747603746120678,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3393866533029117e-08,
+      "loss": 0.8397,
+      "step": 35492
+    },
+    {
+      "epoch": 0.974787838839911,
+      "grad_norm": 0.0,
+      "learning_rate": 3.33212720820586e-08,
+      "loss": 0.6633,
+      "step": 35493
+    },
+    {
+      "epoch": 0.9748153030677542,
+      "grad_norm": 0.0,
+      "learning_rate": 3.324875649131088e-08,
+      "loss": 0.8283,
+      "step": 35494
+    },
+    {
+      "epoch": 0.9748427672955975,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3176319761358825e-08,
+      "loss": 0.8056,
+      "step": 35495
+    },
+    {
+      "epoch": 0.9748702315234408,
+      "grad_norm": 0.0,
+      "learning_rate": 3.310396189277754e-08,
+      "loss": 0.6962,
+      "step": 35496
+    },
+    {
+      "epoch": 0.9748976957512839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.303168288613767e-08,
+      "loss": 0.7457,
+      "step": 35497
+    },
+    {
+      "epoch": 0.9749251599791272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2959482742012104e-08,
+      "loss": 0.7615,
+      "step": 35498
+    },
+    {
+      "epoch": 0.9749526242069704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.288736146097149e-08,
+      "loss": 0.8139,
+      "step": 35499
+    },
+    {
+      "epoch": 0.9749800884348137,
+      "grad_norm": 0.0,
+      "learning_rate": 3.28153190435887e-08,
+      "loss": 0.7799,
+      "step": 35500
+    },
+    {
+      "epoch": 0.9750075526626569,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2743355490429954e-08,
+      "loss": 0.8214,
+      "step": 35501
+    },
+    {
+      "epoch": 0.9750350168905001,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2671470802068117e-08,
+      "loss": 0.8663,
+      "step": 35502
+    },
+    {
+      "epoch": 0.9750624811183434,
+      "grad_norm": 0.0,
+      "learning_rate": 3.259966497907052e-08,
+      "loss": 0.8867,
+      "step": 35503
+    },
+    {
+      "epoch": 0.9750899453461865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.252793802200449e-08,
+      "loss": 0.7481,
+      "step": 35504
+    },
+    {
+      "epoch": 0.9751174095740298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2456289931438456e-08,
+      "loss": 0.7806,
+      "step": 35505
+    },
+    {
+      "epoch": 0.9751448738018731,
+      "grad_norm": 0.0,
+      "learning_rate": 3.238472070794085e-08,
+      "loss": 0.7316,
+      "step": 35506
+    },
+    {
+      "epoch": 0.9751723380297163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2313230352074565e-08,
+      "loss": 0.8119,
+      "step": 35507
+    },
+    {
+      "epoch": 0.9751998022575595,
+      "grad_norm": 0.0,
+      "learning_rate": 3.224181886440803e-08,
+      "loss": 0.8258,
+      "step": 35508
+    },
+    {
+      "epoch": 0.9752272664854028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2170486245505226e-08,
+      "loss": 0.7552,
+      "step": 35509
+    },
+    {
+      "epoch": 0.975254730713246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2099232495931275e-08,
+      "loss": 0.8365,
+      "step": 35510
+    },
+    {
+      "epoch": 0.9752821949410893,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2028057616249056e-08,
+      "loss": 0.793,
+      "step": 35511
+    },
+    {
+      "epoch": 0.9753096591689324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.195696160702144e-08,
+      "loss": 0.8018,
+      "step": 35512
+    },
+    {
+      "epoch": 0.9753371233967757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1885944468812434e-08,
+      "loss": 0.8122,
+      "step": 35513
+    },
+    {
+      "epoch": 0.975364587624619,
+      "grad_norm": 0.0,
+      "learning_rate": 3.181500620218159e-08,
+      "loss": 0.8426,
+      "step": 35514
+    },
+    {
+      "epoch": 0.9753920518524621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.174414680769289e-08,
+      "loss": 0.7613,
+      "step": 35515
+    },
+    {
+      "epoch": 0.9754195160803054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.167336628590589e-08,
+      "loss": 0.7754,
+      "step": 35516
+    },
+    {
+      "epoch": 0.9754469803081486,
+      "grad_norm": 0.0,
+      "learning_rate": 3.160266463738126e-08,
+      "loss": 0.8147,
+      "step": 35517
+    },
+    {
+      "epoch": 0.9754744445359919,
+      "grad_norm": 0.0,
+      "learning_rate": 3.153204186267633e-08,
+      "loss": 0.853,
+      "step": 35518
+    },
+    {
+      "epoch": 0.9755019087638351,
+      "grad_norm": 0.0,
+      "learning_rate": 3.146149796235176e-08,
+      "loss": 0.8508,
+      "step": 35519
+    },
+    {
+      "epoch": 0.9755293729916783,
+      "grad_norm": 0.0,
+      "learning_rate": 3.139103293696599e-08,
+      "loss": 0.8326,
+      "step": 35520
+    },
+    {
+      "epoch": 0.9755568372195216,
+      "grad_norm": 0.0,
+      "learning_rate": 3.132064678707525e-08,
+      "loss": 0.7708,
+      "step": 35521
+    },
+    {
+      "epoch": 0.9755843014473649,
+      "grad_norm": 0.0,
+      "learning_rate": 3.125033951323686e-08,
+      "loss": 0.8743,
+      "step": 35522
+    },
+    {
+      "epoch": 0.975611765675208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.118011111600705e-08,
+      "loss": 0.8396,
+      "step": 35523
+    },
+    {
+      "epoch": 0.9756392299030513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.110996159594204e-08,
+      "loss": 0.8562,
+      "step": 35524
+    },
+    {
+      "epoch": 0.9756666941308945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103989095359694e-08,
+      "loss": 0.803,
+      "step": 35525
+    },
+    {
+      "epoch": 0.9756941583587377,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0969899189525756e-08,
+      "loss": 0.8914,
+      "step": 35526
+    },
+    {
+      "epoch": 0.975721622586581,
+      "grad_norm": 0.0,
+      "learning_rate": 3.089998630428248e-08,
+      "loss": 0.8411,
+      "step": 35527
+    },
+    {
+      "epoch": 0.9757490868144242,
+      "grad_norm": 0.0,
+      "learning_rate": 3.083015229842001e-08,
+      "loss": 0.6902,
+      "step": 35528
+    },
+    {
+      "epoch": 0.9757765510422675,
+      "grad_norm": 0.0,
+      "learning_rate": 3.076039717249124e-08,
+      "loss": 0.8381,
+      "step": 35529
+    },
+    {
+      "epoch": 0.9758040152701107,
+      "grad_norm": 0.0,
+      "learning_rate": 3.069072092704683e-08,
+      "loss": 0.7771,
+      "step": 35530
+    },
+    {
+      "epoch": 0.9758314794979539,
+      "grad_norm": 0.0,
+      "learning_rate": 3.06211235626408e-08,
+      "loss": 0.9008,
+      "step": 35531
+    },
+    {
+      "epoch": 0.9758589437257972,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0551605079820465e-08,
+      "loss": 0.8486,
+      "step": 35532
+    },
+    {
+      "epoch": 0.9758864079536403,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0482165479138736e-08,
+      "loss": 0.7807,
+      "step": 35533
+    },
+    {
+      "epoch": 0.9759138721814836,
+      "grad_norm": 0.0,
+      "learning_rate": 3.041280476114406e-08,
+      "loss": 0.7175,
+      "step": 35534
+    },
+    {
+      "epoch": 0.9759413364093269,
+      "grad_norm": 0.0,
+      "learning_rate": 3.034352292638487e-08,
+      "loss": 0.8082,
+      "step": 35535
+    },
+    {
+      "epoch": 0.9759688006371701,
+      "grad_norm": 0.0,
+      "learning_rate": 3.027431997540964e-08,
+      "loss": 0.7113,
+      "step": 35536
+    },
+    {
+      "epoch": 0.9759962648650133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.020519590876569e-08,
+      "loss": 0.78,
+      "step": 35537
+    },
+    {
+      "epoch": 0.9760237290928565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.013615072700038e-08,
+      "loss": 0.7413,
+      "step": 35538
+    },
+    {
+      "epoch": 0.9760511933206998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0067184430659926e-08,
+      "loss": 0.8456,
+      "step": 35539
+    },
+    {
+      "epoch": 0.9760786575485431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9998297020289444e-08,
+      "loss": 0.8268,
+      "step": 35540
+    },
+    {
+      "epoch": 0.9761061217763862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.992948849643518e-08,
+      "loss": 0.843,
+      "step": 35541
+    },
+    {
+      "epoch": 0.9761335860042295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9860758859640014e-08,
+      "loss": 0.9128,
+      "step": 35542
+    },
+    {
+      "epoch": 0.9761610502320728,
+      "grad_norm": 0.0,
+      "learning_rate": 2.979210811044908e-08,
+      "loss": 0.8144,
+      "step": 35543
+    },
+    {
+      "epoch": 0.976188514459916,
+      "grad_norm": 0.0,
+      "learning_rate": 2.972353624940527e-08,
+      "loss": 0.8181,
+      "step": 35544
+    },
+    {
+      "epoch": 0.9762159786877592,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9655043277050376e-08,
+      "loss": 0.8386,
+      "step": 35545
+    },
+    {
+      "epoch": 0.9762434429156024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9586629193926188e-08,
+      "loss": 0.7767,
+      "step": 35546
+    },
+    {
+      "epoch": 0.9762709071434457,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9518294000576708e-08,
+      "loss": 0.7538,
+      "step": 35547
+    },
+    {
+      "epoch": 0.9762983713712889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.945003769753929e-08,
+      "loss": 0.8853,
+      "step": 35548
+    },
+    {
+      "epoch": 0.9763258355991321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9381860285355725e-08,
+      "loss": 0.8884,
+      "step": 35549
+    },
+    {
+      "epoch": 0.9763532998269754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9313761764565575e-08,
+      "loss": 0.7749,
+      "step": 35550
+    },
+    {
+      "epoch": 0.9763807640548186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.924574213570619e-08,
+      "loss": 0.8733,
+      "step": 35551
+    },
+    {
+      "epoch": 0.9764082282826618,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9177801399318252e-08,
+      "loss": 0.801,
+      "step": 35552
+    },
+    {
+      "epoch": 0.9764356925105051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9109939555936884e-08,
+      "loss": 0.7346,
+      "step": 35553
+    },
+    {
+      "epoch": 0.9764631567383483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9042156606100546e-08,
+      "loss": 0.8206,
+      "step": 35554
+    },
+    {
+      "epoch": 0.9764906209661915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8974452550344368e-08,
+      "loss": 0.7545,
+      "step": 35555
+    },
+    {
+      "epoch": 0.9765180851940348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8906827389204585e-08,
+      "loss": 0.7317,
+      "step": 35556
+    },
+    {
+      "epoch": 0.976545549421878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8839281123217434e-08,
+      "loss": 0.7654,
+      "step": 35557
+    },
+    {
+      "epoch": 0.9765730136497213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8771813752915822e-08,
+      "loss": 0.7701,
+      "step": 35558
+    },
+    {
+      "epoch": 0.9766004778775644,
+      "grad_norm": 0.0,
+      "learning_rate": 2.870442527883377e-08,
+      "loss": 0.8008,
+      "step": 35559
+    },
+    {
+      "epoch": 0.9766279421054077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8637115701505292e-08,
+      "loss": 0.8802,
+      "step": 35560
+    },
+    {
+      "epoch": 0.976655406333251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8569885021462184e-08,
+      "loss": 0.8486,
+      "step": 35561
+    },
+    {
+      "epoch": 0.9766828705610942,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8502733239237357e-08,
+      "loss": 0.7626,
+      "step": 35562
+    },
+    {
+      "epoch": 0.9767103347889374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8435660355360383e-08,
+      "loss": 0.7793,
+      "step": 35563
+    },
+    {
+      "epoch": 0.9767377990167806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8368666370364174e-08,
+      "loss": 0.8218,
+      "step": 35564
+    },
+    {
+      "epoch": 0.9767652632446239,
+      "grad_norm": 0.0,
+      "learning_rate": 2.830175128477719e-08,
+      "loss": 0.8254,
+      "step": 35565
+    },
+    {
+      "epoch": 0.9767927274724671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.823491509912901e-08,
+      "loss": 0.7177,
+      "step": 35566
+    },
+    {
+      "epoch": 0.9768201917003103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8168157813949215e-08,
+      "loss": 0.836,
+      "step": 35567
+    },
+    {
+      "epoch": 0.9768476559281536,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8101479429766265e-08,
+      "loss": 0.9419,
+      "step": 35568
+    },
+    {
+      "epoch": 0.9768751201559969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.803487994710641e-08,
+      "loss": 0.7967,
+      "step": 35569
+    },
+    {
+      "epoch": 0.97690258438384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.796835936649811e-08,
+      "loss": 0.8203,
+      "step": 35570
+    },
+    {
+      "epoch": 0.9769300486116833,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7901917688466507e-08,
+      "loss": 0.7487,
+      "step": 35571
+    },
+    {
+      "epoch": 0.9769575128395265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7835554913537844e-08,
+      "loss": 0.8095,
+      "step": 35572
+    },
+    {
+      "epoch": 0.9769849770673698,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7769271042236146e-08,
+      "loss": 0.7942,
+      "step": 35573
+    },
+    {
+      "epoch": 0.977012441295213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.770306607508766e-08,
+      "loss": 0.6541,
+      "step": 35574
+    },
+    {
+      "epoch": 0.9770399055230562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7636940012615297e-08,
+      "loss": 0.681,
+      "step": 35575
+    },
+    {
+      "epoch": 0.9770673697508995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.757089285534309e-08,
+      "loss": 0.7941,
+      "step": 35576
+    },
+    {
+      "epoch": 0.9770948339787426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7504924603792838e-08,
+      "loss": 0.8779,
+      "step": 35577
+    },
+    {
+      "epoch": 0.9771222982065859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7439035258485236e-08,
+      "loss": 0.9411,
+      "step": 35578
+    },
+    {
+      "epoch": 0.9771497624344292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.737322481994431e-08,
+      "loss": 0.7814,
+      "step": 35579
+    },
+    {
+      "epoch": 0.9771772266622724,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7307493288688534e-08,
+      "loss": 0.8644,
+      "step": 35580
+    },
+    {
+      "epoch": 0.9772046908901156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7241840665239717e-08,
+      "loss": 0.7757,
+      "step": 35581
+    },
+    {
+      "epoch": 0.9772321551179589,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7176266950116324e-08,
+      "loss": 0.8287,
+      "step": 35582
+    },
+    {
+      "epoch": 0.9772596193458021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7110772143837948e-08,
+      "loss": 0.7146,
+      "step": 35583
+    },
+    {
+      "epoch": 0.9772870835736454,
+      "grad_norm": 0.0,
+      "learning_rate": 2.704535624692084e-08,
+      "loss": 0.7506,
+      "step": 35584
+    },
+    {
+      "epoch": 0.9773145478014885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.698001925988458e-08,
+      "loss": 0.825,
+      "step": 35585
+    },
+    {
+      "epoch": 0.9773420120293318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6914761183246542e-08,
+      "loss": 0.851,
+      "step": 35586
+    },
+    {
+      "epoch": 0.9773694762571751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.684958201752075e-08,
+      "loss": 0.8285,
+      "step": 35587
+    },
+    {
+      "epoch": 0.9773969404850182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.678448176322568e-08,
+      "loss": 0.8593,
+      "step": 35588
+    },
+    {
+      "epoch": 0.9774244047128615,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6719460420874254e-08,
+      "loss": 0.8436,
+      "step": 35589
+    },
+    {
+      "epoch": 0.9774518689407047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6654517990981622e-08,
+      "loss": 0.9134,
+      "step": 35590
+    },
+    {
+      "epoch": 0.977479333168548,
+      "grad_norm": 0.0,
+      "learning_rate": 2.658965447406181e-08,
+      "loss": 0.8169,
+      "step": 35591
+    },
+    {
+      "epoch": 0.9775067973963912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6524869870627744e-08,
+      "loss": 0.8406,
+      "step": 35592
+    },
+    {
+      "epoch": 0.9775342616242344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.646016418119346e-08,
+      "loss": 0.8529,
+      "step": 35593
+    },
+    {
+      "epoch": 0.9775617258520777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6395537406268545e-08,
+      "loss": 0.6757,
+      "step": 35594
+    },
+    {
+      "epoch": 0.977589190079921,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6330989546365926e-08,
+      "loss": 0.7354,
+      "step": 35595
+    },
+    {
+      "epoch": 0.9776166543077641,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6266520601995195e-08,
+      "loss": 0.843,
+      "step": 35596
+    },
+    {
+      "epoch": 0.9776441185356074,
+      "grad_norm": 0.0,
+      "learning_rate": 2.620213057366816e-08,
+      "loss": 0.7803,
+      "step": 35597
+    },
+    {
+      "epoch": 0.9776715827634506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.613781946189331e-08,
+      "loss": 0.7006,
+      "step": 35598
+    },
+    {
+      "epoch": 0.9776990469912938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6073587267179123e-08,
+      "loss": 0.8266,
+      "step": 35599
+    },
+    {
+      "epoch": 0.9777265112191371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6009433990035194e-08,
+      "loss": 0.8131,
+      "step": 35600
+    },
+    {
+      "epoch": 0.9777539754469803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.594535963096667e-08,
+      "loss": 0.833,
+      "step": 35601
+    },
+    {
+      "epoch": 0.9777814396748236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.588136419048315e-08,
+      "loss": 0.9041,
+      "step": 35602
+    },
+    {
+      "epoch": 0.9778089039026667,
+      "grad_norm": 0.0,
+      "learning_rate": 2.581744766908978e-08,
+      "loss": 0.7754,
+      "step": 35603
+    },
+    {
+      "epoch": 0.97783636813051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5753610067291714e-08,
+      "loss": 0.8749,
+      "step": 35604
+    },
+    {
+      "epoch": 0.9778638323583533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.568985138559521e-08,
+      "loss": 0.7726,
+      "step": 35605
+    },
+    {
+      "epoch": 0.9778912965861964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5626171624504315e-08,
+      "loss": 0.8903,
+      "step": 35606
+    },
+    {
+      "epoch": 0.9779187608140397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5562570784523068e-08,
+      "loss": 0.9065,
+      "step": 35607
+    },
+    {
+      "epoch": 0.977946225041883,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5499048866154396e-08,
+      "loss": 0.8116,
+      "step": 35608
+    },
+    {
+      "epoch": 0.9779736892697262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5435605869900127e-08,
+      "loss": 0.8527,
+      "step": 35609
+    },
+    {
+      "epoch": 0.9780011534975694,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5372241796264295e-08,
+      "loss": 0.8671,
+      "step": 35610
+    },
+    {
+      "epoch": 0.9780286177254126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5308956645746507e-08,
+      "loss": 0.7962,
+      "step": 35611
+    },
+    {
+      "epoch": 0.9780560819532559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.524575041884858e-08,
+      "loss": 0.7833,
+      "step": 35612
+    },
+    {
+      "epoch": 0.9780835461810992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5182623116069004e-08,
+      "loss": 0.7987,
+      "step": 35613
+    },
+    {
+      "epoch": 0.9781110104089423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5119574737909602e-08,
+      "loss": 0.7979,
+      "step": 35614
+    },
+    {
+      "epoch": 0.9781384746367856,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5056605284867752e-08,
+      "loss": 0.7177,
+      "step": 35615
+    },
+    {
+      "epoch": 0.9781659388646288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4993714757440833e-08,
+      "loss": 0.8037,
+      "step": 35616
+    },
+    {
+      "epoch": 0.978193403092472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.493090315612845e-08,
+      "loss": 0.7959,
+      "step": 35617
+    },
+    {
+      "epoch": 0.9782208673203153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4868170481427977e-08,
+      "loss": 0.8347,
+      "step": 35618
+    },
+    {
+      "epoch": 0.9782483315481585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.480551673383236e-08,
+      "loss": 0.8424,
+      "step": 35619
+    },
+    {
+      "epoch": 0.9782757957760018,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4742941913841188e-08,
+      "loss": 0.8694,
+      "step": 35620
+    },
+    {
+      "epoch": 0.978303260003845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4680446021946303e-08,
+      "loss": 0.8531,
+      "step": 35621
+    },
+    {
+      "epoch": 0.9783307242316882,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4618029058645076e-08,
+      "loss": 0.8853,
+      "step": 35622
+    },
+    {
+      "epoch": 0.9783581884595315,
+      "grad_norm": 0.0,
+      "learning_rate": 2.455569102443045e-08,
+      "loss": 0.7879,
+      "step": 35623
+    },
+    {
+      "epoch": 0.9783856526873747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4493431919794253e-08,
+      "loss": 0.8854,
+      "step": 35624
+    },
+    {
+      "epoch": 0.9784131169152179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4431251745230532e-08,
+      "loss": 0.847,
+      "step": 35625
+    },
+    {
+      "epoch": 0.9784405811430612,
+      "grad_norm": 0.0,
+      "learning_rate": 2.436915050123112e-08,
+      "loss": 0.7995,
+      "step": 35626
+    },
+    {
+      "epoch": 0.9784680453709044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4307128188286734e-08,
+      "loss": 0.7821,
+      "step": 35627
+    },
+    {
+      "epoch": 0.9784955095987476,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4245184806888088e-08,
+      "loss": 0.7752,
+      "step": 35628
+    },
+    {
+      "epoch": 0.9785229738265908,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4183320357527017e-08,
+      "loss": 0.8089,
+      "step": 35629
+    },
+    {
+      "epoch": 0.9785504380544341,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4121534840690907e-08,
+      "loss": 0.802,
+      "step": 35630
+    },
+    {
+      "epoch": 0.9785779022822774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4059828256870477e-08,
+      "loss": 0.7946,
+      "step": 35631
+    },
+    {
+      "epoch": 0.9786053665101205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3998200606552004e-08,
+      "loss": 0.8511,
+      "step": 35632
+    },
+    {
+      "epoch": 0.9786328307379638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3936651890223984e-08,
+      "loss": 0.7965,
+      "step": 35633
+    },
+    {
+      "epoch": 0.9786602949658071,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3875182108373808e-08,
+      "loss": 0.8514,
+      "step": 35634
+    },
+    {
+      "epoch": 0.9786877591936503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.381379126148886e-08,
+      "loss": 0.785,
+      "step": 35635
+    },
+    {
+      "epoch": 0.9787152234214935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3752479350052092e-08,
+      "loss": 0.7744,
+      "step": 35636
+    },
+    {
+      "epoch": 0.9787426876493367,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3691246374549783e-08,
+      "loss": 0.8022,
+      "step": 35637
+    },
+    {
+      "epoch": 0.97877015187718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3630092335468203e-08,
+      "loss": 0.8318,
+      "step": 35638
+    },
+    {
+      "epoch": 0.9787976161050232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3569017233289192e-08,
+      "loss": 0.8284,
+      "step": 35639
+    },
+    {
+      "epoch": 0.9788250803328664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3508021068496813e-08,
+      "loss": 0.8179,
+      "step": 35640
+    },
+    {
+      "epoch": 0.9788525445607097,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3447103841572895e-08,
+      "loss": 0.7375,
+      "step": 35641
+    },
+    {
+      "epoch": 0.9788800087885529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3386265553000385e-08,
+      "loss": 0.8422,
+      "step": 35642
+    },
+    {
+      "epoch": 0.9789074730163961,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3325506203260016e-08,
+      "loss": 0.7961,
+      "step": 35643
+    },
+    {
+      "epoch": 0.9789349372442394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3264825792833622e-08,
+      "loss": 0.9047,
+      "step": 35644
+    },
+    {
+      "epoch": 0.9789624014720826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.320422432219971e-08,
+      "loss": 0.7986,
+      "step": 35645
+    },
+    {
+      "epoch": 0.9789898656999259,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3143701791839003e-08,
+      "loss": 0.6917,
+      "step": 35646
+    },
+    {
+      "epoch": 0.9790173299277691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3083258202230008e-08,
+      "loss": 0.795,
+      "step": 35647
+    },
+    {
+      "epoch": 0.9790447941556123,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3022893553851233e-08,
+      "loss": 0.8055,
+      "step": 35648
+    },
+    {
+      "epoch": 0.9790722583834556,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2962607847178964e-08,
+      "loss": 0.8835,
+      "step": 35649
+    },
+    {
+      "epoch": 0.9790997226112987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2902401082692817e-08,
+      "loss": 0.7848,
+      "step": 35650
+    },
+    {
+      "epoch": 0.979127186839142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2842273260865744e-08,
+      "loss": 0.7949,
+      "step": 35651
+    },
+    {
+      "epoch": 0.9791546510669853,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2782224382177365e-08,
+      "loss": 0.7047,
+      "step": 35652
+    },
+    {
+      "epoch": 0.9791821152948285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2722254447099523e-08,
+      "loss": 0.8129,
+      "step": 35653
+    },
+    {
+      "epoch": 0.9792095795226717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2662363456108506e-08,
+      "loss": 0.7963,
+      "step": 35654
+    },
+    {
+      "epoch": 0.9792370437505149,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2602551409678374e-08,
+      "loss": 0.8778,
+      "step": 35655
+    },
+    {
+      "epoch": 0.9792645079783582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2542818308280977e-08,
+      "loss": 0.839,
+      "step": 35656
+    },
+    {
+      "epoch": 0.9792919722062015,
+      "grad_norm": 0.0,
+      "learning_rate": 2.248316415238927e-08,
+      "loss": 0.7242,
+      "step": 35657
+    },
+    {
+      "epoch": 0.9793194364340446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2423588942476203e-08,
+      "loss": 0.7792,
+      "step": 35658
+    },
+    {
+      "epoch": 0.9793469006618879,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2364092679012518e-08,
+      "loss": 0.8188,
+      "step": 35659
+    },
+    {
+      "epoch": 0.9793743648897312,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2304675362468942e-08,
+      "loss": 0.8445,
+      "step": 35660
+    },
+    {
+      "epoch": 0.9794018291175743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2245336993316212e-08,
+      "loss": 0.9426,
+      "step": 35661
+    },
+    {
+      "epoch": 0.9794292933454176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2186077572022845e-08,
+      "loss": 0.8157,
+      "step": 35662
+    },
+    {
+      "epoch": 0.9794567575732608,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2126897099058465e-08,
+      "loss": 0.7697,
+      "step": 35663
+    },
+    {
+      "epoch": 0.9794842218011041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.206779557489158e-08,
+      "loss": 0.7528,
+      "step": 35664
+    },
+    {
+      "epoch": 0.9795116860289473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2008772999989603e-08,
+      "loss": 0.768,
+      "step": 35665
+    },
+    {
+      "epoch": 0.9795391502567905,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1949829374818822e-08,
+      "loss": 0.8858,
+      "step": 35666
+    },
+    {
+      "epoch": 0.9795666144846338,
+      "grad_norm": 0.0,
+      "learning_rate": 2.189096469984553e-08,
+      "loss": 0.8118,
+      "step": 35667
+    },
+    {
+      "epoch": 0.979594078712477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1832178975537133e-08,
+      "loss": 0.761,
+      "step": 35668
+    },
+    {
+      "epoch": 0.9796215429403202,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1773472202356592e-08,
+      "loss": 0.7856,
+      "step": 35669
+    },
+    {
+      "epoch": 0.9796490071681635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1714844380770204e-08,
+      "loss": 0.8194,
+      "step": 35670
+    },
+    {
+      "epoch": 0.9796764713960067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1656295511242042e-08,
+      "loss": 0.7958,
+      "step": 35671
+    },
+    {
+      "epoch": 0.9797039356238499,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1597825594232847e-08,
+      "loss": 0.6917,
+      "step": 35672
+    },
+    {
+      "epoch": 0.9797313998516932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.15394346302078e-08,
+      "loss": 0.7966,
+      "step": 35673
+    },
+    {
+      "epoch": 0.9797588640795364,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1481122619627648e-08,
+      "loss": 0.8497,
+      "step": 35674
+    },
+    {
+      "epoch": 0.9797863283073797,
+      "grad_norm": 0.0,
+      "learning_rate": 2.142288956295424e-08,
+      "loss": 0.8264,
+      "step": 35675
+    },
+    {
+      "epoch": 0.9798137925352228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.136473546064832e-08,
+      "loss": 0.8304,
+      "step": 35676
+    },
+    {
+      "epoch": 0.9798412567630661,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1306660313170634e-08,
+      "loss": 0.7532,
+      "step": 35677
+    },
+    {
+      "epoch": 0.9798687209909094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.12486641209797e-08,
+      "loss": 0.8541,
+      "step": 35678
+    },
+    {
+      "epoch": 0.9798961852187525,
+      "grad_norm": 0.0,
+      "learning_rate": 2.119074688453404e-08,
+      "loss": 0.8148,
+      "step": 35679
+    },
+    {
+      "epoch": 0.9799236494465958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.113290860429329e-08,
+      "loss": 0.7275,
+      "step": 35680
+    },
+    {
+      "epoch": 0.979951113674439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1075149280713746e-08,
+      "loss": 0.7584,
+      "step": 35681
+    },
+    {
+      "epoch": 0.9799785779022823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1017468914253934e-08,
+      "loss": 0.7556,
+      "step": 35682
+    },
+    {
+      "epoch": 0.9800060421301255,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0959867505369047e-08,
+      "loss": 0.7654,
+      "step": 35683
+    },
+    {
+      "epoch": 0.9800335063579687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0902345054515383e-08,
+      "loss": 0.9327,
+      "step": 35684
+    },
+    {
+      "epoch": 0.980060970585812,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0844901562147025e-08,
+      "loss": 0.7786,
+      "step": 35685
+    },
+    {
+      "epoch": 0.9800884348136553,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0787537028720272e-08,
+      "loss": 0.747,
+      "step": 35686
+    },
+    {
+      "epoch": 0.9801158990414984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0730251454688098e-08,
+      "loss": 0.7779,
+      "step": 35687
+    },
+    {
+      "epoch": 0.9801433632693417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0673044840503475e-08,
+      "loss": 0.8002,
+      "step": 35688
+    },
+    {
+      "epoch": 0.9801708274971849,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0615917186619372e-08,
+      "loss": 0.7547,
+      "step": 35689
+    },
+    {
+      "epoch": 0.9801982917250281,
+      "grad_norm": 0.0,
+      "learning_rate": 2.055886849348765e-08,
+      "loss": 0.7668,
+      "step": 35690
+    },
+    {
+      "epoch": 0.9802257559528714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0501898761560167e-08,
+      "loss": 0.7754,
+      "step": 35691
+    },
+    {
+      "epoch": 0.9802532201807146,
+      "grad_norm": 0.0,
+      "learning_rate": 2.044500799128657e-08,
+      "loss": 0.8487,
+      "step": 35692
+    },
+    {
+      "epoch": 0.9802806844085579,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0388196183118713e-08,
+      "loss": 0.8012,
+      "step": 35693
+    },
+    {
+      "epoch": 0.980308148636401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0331463337505132e-08,
+      "loss": 0.762,
+      "step": 35694
+    },
+    {
+      "epoch": 0.9803356128642443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0274809454894352e-08,
+      "loss": 0.8678,
+      "step": 35695
+    },
+    {
+      "epoch": 0.9803630770920876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0218234535734904e-08,
+      "loss": 0.7279,
+      "step": 35696
+    },
+    {
+      "epoch": 0.9803905413199308,
+      "grad_norm": 0.0,
+      "learning_rate": 2.016173858047532e-08,
+      "loss": 0.7866,
+      "step": 35697
+    },
+    {
+      "epoch": 0.980418005547774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.01053215895608e-08,
+      "loss": 0.8802,
+      "step": 35698
+    },
+    {
+      "epoch": 0.9804454697756173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0048983563439872e-08,
+      "loss": 0.8025,
+      "step": 35699
+    },
+    {
+      "epoch": 0.9804729340034605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992724502556626e-08,
+      "loss": 0.7425,
+      "step": 35700
+    },
+    {
+      "epoch": 0.9805003982313037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993654440735626e-08,
+      "loss": 0.7986,
+      "step": 35701
+    },
+    {
+      "epoch": 0.9805278624591469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988044327828509e-08,
+      "loss": 0.7833,
+      "step": 35702
+    },
+    {
+      "epoch": 0.9805553266869902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824421115784975e-08,
+      "loss": 0.844,
+      "step": 35703
+    },
+    {
+      "epoch": 0.9805827909148335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976847792030112e-08,
+      "loss": 0.7009,
+      "step": 35704
+    },
+    {
+      "epoch": 0.9806102551426766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712613692273174e-08,
+      "loss": 0.7981,
+      "step": 35705
+    },
+    {
+      "epoch": 0.9806377193705199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9656828432146336e-08,
+      "loss": 0.8349,
+      "step": 35706
+    },
+    {
+      "epoch": 0.9806651835983631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601122140360252e-08,
+      "loss": 0.7586,
+      "step": 35707
+    },
+    {
+      "epoch": 0.9806926478262064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954549481735679e-08,
+      "loss": 0.7956,
+      "step": 35708
+    },
+    {
+      "epoch": 0.9807201120540496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489946463574495e-08,
+      "loss": 0.7749,
+      "step": 35709
+    },
+    {
+      "epoch": 0.9807475762818928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434477079455228e-08,
+      "loss": 0.7775,
+      "step": 35710
+    },
+    {
+      "epoch": 0.9807750405097361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937908666543531e-08,
+      "loss": 0.8458,
+      "step": 35711
+    },
+    {
+      "epoch": 0.9808025047375794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932377522195439e-08,
+      "loss": 0.8096,
+      "step": 35712
+    },
+    {
+      "epoch": 0.9808299689654225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268542749449893e-08,
+      "loss": 0.7924,
+      "step": 35713
+    },
+    {
+      "epoch": 0.9808574331932658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921338924835814e-08,
+      "loss": 0.8011,
+      "step": 35714
+    },
+    {
+      "epoch": 0.980884897421109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9158314719117664e-08,
+      "loss": 0.9462,
+      "step": 35715
+    },
+    {
+      "epoch": 0.9809123616489522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910331916216146e-08,
+      "loss": 0.8349,
+      "step": 35716
+    },
+    {
+      "epoch": 0.9809398258767955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904840257792695e-08,
+      "loss": 0.7265,
+      "step": 35717
+    },
+    {
+      "epoch": 0.9809672901046387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993564966847122e-08,
+      "loss": 0.7942,
+      "step": 35718
+    },
+    {
+      "epoch": 0.980994754332482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8938806329357184e-08,
+      "loss": 0.8918,
+      "step": 35719
+    },
+    {
+      "epoch": 0.9810222185603251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884126665889012e-08,
+      "loss": 0.7976,
+      "step": 35720
+    },
+    {
+      "epoch": 0.9810496827881684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829525976876707e-08,
+      "loss": 0.8424,
+      "step": 35721
+    },
+    {
+      "epoch": 0.9810771470160117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877500426275103e-08,
+      "loss": 0.7958,
+      "step": 35722
+    },
+    {
+      "epoch": 0.9811046112438548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8720561523943858e-08,
+      "loss": 0.8281,
+      "step": 35723
+    },
+    {
+      "epoch": 0.9811320754716981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866619776088596e-08,
+      "loss": 0.7888,
+      "step": 35724
+    },
+    {
+      "epoch": 0.9811595396995414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8611912974008106e-08,
+      "loss": 0.7941,
+      "step": 35725
+    },
+    {
+      "epoch": 0.9811870039273846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8557707163737725e-08,
+      "loss": 0.8092,
+      "step": 35726
+    },
+    {
+      "epoch": 0.9812144681552278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8503580330506697e-08,
+      "loss": 0.7442,
+      "step": 35727
+    },
+    {
+      "epoch": 0.981241932383071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844953247474135e-08,
+      "loss": 0.8531,
+      "step": 35728
+    },
+    {
+      "epoch": 0.9812693966109143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8395563596869116e-08,
+      "loss": 0.8059,
+      "step": 35729
+    },
+    {
+      "epoch": 0.9812968608387576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8341673697318542e-08,
+      "loss": 0.8316,
+      "step": 35730
+    },
+    {
+      "epoch": 0.9813243250666007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828786277651484e-08,
+      "loss": 0.7745,
+      "step": 35731
+    },
+    {
+      "epoch": 0.981351789294444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8234130834884346e-08,
+      "loss": 0.8323,
+      "step": 35732
+    },
+    {
+      "epoch": 0.9813792535222873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8180477872852263e-08,
+      "loss": 0.7656,
+      "step": 35733
+    },
+    {
+      "epoch": 0.9814067177501304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8126903890842705e-08,
+      "loss": 0.8521,
+      "step": 35734
+    },
+    {
+      "epoch": 0.9814341819779737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8073408889278666e-08,
+      "loss": 0.6946,
+      "step": 35735
+    },
+    {
+      "epoch": 0.9814616462058169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8019992868585356e-08,
+      "loss": 0.7841,
+      "step": 35736
+    },
+    {
+      "epoch": 0.9814891104336602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966655829184663e-08,
+      "loss": 0.8566,
+      "step": 35737
+    },
+    {
+      "epoch": 0.9815165746615034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7913397771497365e-08,
+      "loss": 0.8726,
+      "step": 35738
+    },
+    {
+      "epoch": 0.9815440388893466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786021869594756e-08,
+      "loss": 0.7574,
+      "step": 35739
+    },
+    {
+      "epoch": 0.9815715031171899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78071186029527e-08,
+      "loss": 0.763,
+      "step": 35740
+    },
+    {
+      "epoch": 0.981598967345033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7754097492935773e-08,
+      "loss": 0.7799,
+      "step": 35741
+    },
+    {
+      "epoch": 0.9816264315728763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7701155366315337e-08,
+      "loss": 0.7734,
+      "step": 35742
+    },
+    {
+      "epoch": 0.9816538958007196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7648292223509943e-08,
+      "loss": 0.8454,
+      "step": 35743
+    },
+    {
+      "epoch": 0.9816813600285628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759550806493704e-08,
+      "loss": 0.7194,
+      "step": 35744
+    },
+    {
+      "epoch": 0.981708824256406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7542802891016285e-08,
+      "loss": 0.9198,
+      "step": 35745
+    },
+    {
+      "epoch": 0.9817362884842493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749017670216291e-08,
+      "loss": 0.7845,
+      "step": 35746
+    },
+    {
+      "epoch": 0.9817637527120925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743762949879546e-08,
+      "loss": 0.7979,
+      "step": 35747
+    },
+    {
+      "epoch": 0.9817912169399358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7385161281326947e-08,
+      "loss": 0.7784,
+      "step": 35748
+    },
+    {
+      "epoch": 0.9818186811677789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7332772050174808e-08,
+      "loss": 0.6983,
+      "step": 35749
+    },
+    {
+      "epoch": 0.9818461453956222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7280461805752047e-08,
+      "loss": 0.7957,
+      "step": 35750
+    },
+    {
+      "epoch": 0.9818736096234655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722823054847389e-08,
+      "loss": 0.7586,
+      "step": 35751
+    },
+    {
+      "epoch": 0.9819010738513086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7176078278753338e-08,
+      "loss": 0.7243,
+      "step": 35752
+    },
+    {
+      "epoch": 0.9819285380791519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7124004997002287e-08,
+      "loss": 0.8216,
+      "step": 35753
+    },
+    {
+      "epoch": 0.9819560023069951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7072010703632624e-08,
+      "loss": 0.7686,
+      "step": 35754
+    },
+    {
+      "epoch": 0.9819834665348384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7020095399056247e-08,
+      "loss": 0.7296,
+      "step": 35755
+    },
+    {
+      "epoch": 0.9820109307626816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6968259083683937e-08,
+      "loss": 0.7211,
+      "step": 35756
+    },
+    {
+      "epoch": 0.9820383949905248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6916501757925364e-08,
+      "loss": 0.8228,
+      "step": 35757
+    },
+    {
+      "epoch": 0.9820658592183681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6864823422191313e-08,
+      "loss": 0.7389,
+      "step": 35758
+    },
+    {
+      "epoch": 0.9820933234462114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6813224076889235e-08,
+      "loss": 0.7766,
+      "step": 35759
+    },
+    {
+      "epoch": 0.9821207876740545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6761703722428802e-08,
+      "loss": 0.8132,
+      "step": 35760
+    },
+    {
+      "epoch": 0.9821482519018978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6710262359216356e-08,
+      "loss": 0.8142,
+      "step": 35761
+    },
+    {
+      "epoch": 0.982175716129741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665889998766046e-08,
+      "loss": 0.7915,
+      "step": 35762
+    },
+    {
+      "epoch": 0.9822031803575842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6607616608165235e-08,
+      "loss": 0.7373,
+      "step": 35763
+    },
+    {
+      "epoch": 0.9822306445854275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655641222113813e-08,
+      "loss": 0.8453,
+      "step": 35764
+    },
+    {
+      "epoch": 0.9822581088132707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650528682698438e-08,
+      "loss": 0.8522,
+      "step": 35765
+    },
+    {
+      "epoch": 0.982285573041114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645424042610699e-08,
+      "loss": 0.8192,
+      "step": 35766
+    },
+    {
+      "epoch": 0.9823130372689571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6403273018912312e-08,
+      "loss": 0.8678,
+      "step": 35767
+    },
+    {
+      "epoch": 0.9823405014968004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635238460580113e-08,
+      "loss": 0.9337,
+      "step": 35768
+    },
+    {
+      "epoch": 0.9823679657246437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6301575187177565e-08,
+      "loss": 0.7176,
+      "step": 35769
+    },
+    {
+      "epoch": 0.9823954299524869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625084476344352e-08,
+      "loss": 0.7248,
+      "step": 35770
+    },
+    {
+      "epoch": 0.9824228941803301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6200193334999782e-08,
+      "loss": 0.8064,
+      "step": 35771
+    },
+    {
+      "epoch": 0.9824503584081734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6149620902247144e-08,
+      "loss": 0.7394,
+      "step": 35772
+    },
+    {
+      "epoch": 0.9824778226360166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6099127465586395e-08,
+      "loss": 0.8567,
+      "step": 35773
+    },
+    {
+      "epoch": 0.9825052868638599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6048713025417217e-08,
+      "loss": 0.8486,
+      "step": 35774
+    },
+    {
+      "epoch": 0.982532751091703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599837758213707e-08,
+      "loss": 0.7682,
+      "step": 35775
+    },
+    {
+      "epoch": 0.9825602153195463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5948121136145634e-08,
+      "loss": 0.7414,
+      "step": 35776
+    },
+    {
+      "epoch": 0.9825876795473896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5897943687840367e-08,
+      "loss": 0.9304,
+      "step": 35777
+    },
+    {
+      "epoch": 0.9826151437752327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584784523761762e-08,
+      "loss": 0.7685,
+      "step": 35778
+    },
+    {
+      "epoch": 0.982642608003076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5797825785874853e-08,
+      "loss": 0.8323,
+      "step": 35779
+    },
+    {
+      "epoch": 0.9826700722309192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5747885333006195e-08,
+      "loss": 0.7517,
+      "step": 35780
+    },
+    {
+      "epoch": 0.9826975364587625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5698023879409108e-08,
+      "loss": 0.7612,
+      "step": 35781
+    },
+    {
+      "epoch": 0.9827250006866057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5648241425476607e-08,
+      "loss": 0.787,
+      "step": 35782
+    },
+    {
+      "epoch": 0.9827524649144489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5598537971601713e-08,
+      "loss": 0.7897,
+      "step": 35783
+    },
+    {
+      "epoch": 0.9827799291422922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5548913518179666e-08,
+      "loss": 0.7304,
+      "step": 35784
+    },
+    {
+      "epoch": 0.9828073933701355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5499368065601263e-08,
+      "loss": 0.7513,
+      "step": 35785
+    },
+    {
+      "epoch": 0.9828348575979786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544990161426063e-08,
+      "loss": 0.8659,
+      "step": 35786
+    },
+    {
+      "epoch": 0.9828623218258219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5400514164547463e-08,
+      "loss": 0.8355,
+      "step": 35787
+    },
+    {
+      "epoch": 0.9828897860536651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5351205716852558e-08,
+      "loss": 0.9044,
+      "step": 35788
+    },
+    {
+      "epoch": 0.9829172502815083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530197627156782e-08,
+      "loss": 0.8539,
+      "step": 35789
+    },
+    {
+      "epoch": 0.9829447145093516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.525282582908072e-08,
+      "loss": 0.6931,
+      "step": 35790
+    },
+    {
+      "epoch": 0.9829721787371948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5203754389780944e-08,
+      "loss": 0.835,
+      "step": 35791
+    },
+    {
+      "epoch": 0.9829996429650381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5154761954057075e-08,
+      "loss": 0.8806,
+      "step": 35792
+    },
+    {
+      "epoch": 0.9830271071928812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5105848522296573e-08,
+      "loss": 0.9185,
+      "step": 35793
+    },
+    {
+      "epoch": 0.9830545714207245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5057014094886912e-08,
+      "loss": 0.8701,
+      "step": 35794
+    },
+    {
+      "epoch": 0.9830820356485678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5008258672213338e-08,
+      "loss": 0.8322,
+      "step": 35795
+    },
+    {
+      "epoch": 0.9831094998764109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4959582254662208e-08,
+      "loss": 0.8019,
+      "step": 35796
+    },
+    {
+      "epoch": 0.9831369641042542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491098484261766e-08,
+      "loss": 0.8406,
+      "step": 35797
+    },
+    {
+      "epoch": 0.9831644283320975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486246643646716e-08,
+      "loss": 0.8376,
+      "step": 35798
+    },
+    {
+      "epoch": 0.9831918925599407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4814027036591515e-08,
+      "loss": 0.7834,
+      "step": 35799
+    },
+    {
+      "epoch": 0.9832193567877839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4765666643375976e-08,
+      "loss": 0.8408,
+      "step": 35800
+    },
+    {
+      "epoch": 0.9832468210156271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4717385257201345e-08,
+      "loss": 0.7462,
+      "step": 35801
+    },
+    {
+      "epoch": 0.9832742852434704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4669182878451759e-08,
+      "loss": 0.778,
+      "step": 35802
+    },
+    {
+      "epoch": 0.9833017494713137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4621059507506919e-08,
+      "loss": 0.7657,
+      "step": 35803
+    },
+    {
+      "epoch": 0.9833292136991568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4573015144747626e-08,
+      "loss": 0.8036,
+      "step": 35804
+    },
+    {
+      "epoch": 0.9833566779270001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4525049790555801e-08,
+      "loss": 0.8142,
+      "step": 35805
+    },
+    {
+      "epoch": 0.9833841421548433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4477163445308918e-08,
+      "loss": 0.8157,
+      "step": 35806
+    },
+    {
+      "epoch": 0.9834116063826865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4429356109387783e-08,
+      "loss": 0.8397,
+      "step": 35807
+    },
+    {
+      "epoch": 0.9834390706105298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4381627783168762e-08,
+      "loss": 0.8181,
+      "step": 35808
+    },
+    {
+      "epoch": 0.983466534838373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4333978467030441e-08,
+      "loss": 0.844,
+      "step": 35809
+    },
+    {
+      "epoch": 0.9834939990662163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4286408161350296e-08,
+      "loss": 0.8059,
+      "step": 35810
+    },
+    {
+      "epoch": 0.9835214632940595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4238916866503583e-08,
+      "loss": 0.7346,
+      "step": 35811
+    },
+    {
+      "epoch": 0.9835489275219027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4191504582866666e-08,
+      "loss": 0.8023,
+      "step": 35812
+    },
+    {
+      "epoch": 0.983576391749746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4144171310814803e-08,
+      "loss": 0.8168,
+      "step": 35813
+    },
+    {
+      "epoch": 0.9836038559775891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4096917050722137e-08,
+      "loss": 0.7711,
+      "step": 35814
+    },
+    {
+      "epoch": 0.9836313202054324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4049741802963923e-08,
+      "loss": 0.8547,
+      "step": 35815
+    },
+    {
+      "epoch": 0.9836587844332757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4002645567910977e-08,
+      "loss": 0.8204,
+      "step": 35816
+    },
+    {
+      "epoch": 0.9836862486611189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3955628345938554e-08,
+      "loss": 0.7371,
+      "step": 35817
+    },
+    {
+      "epoch": 0.9837137128889621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3908690137416358e-08,
+      "loss": 0.7497,
+      "step": 35818
+    },
+    {
+      "epoch": 0.9837411771168053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3861830942717424e-08,
+      "loss": 0.8683,
+      "step": 35819
+    },
+    {
+      "epoch": 0.9837686413446486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3815050762212568e-08,
+      "loss": 0.8556,
+      "step": 35820
+    },
+    {
+      "epoch": 0.9837961055724919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.376834959627038e-08,
+      "loss": 0.8203,
+      "step": 35821
+    },
+    {
+      "epoch": 0.983823569800335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3721727445260568e-08,
+      "loss": 0.832,
+      "step": 35822
+    },
+    {
+      "epoch": 0.9838510340281783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3675184309553945e-08,
+      "loss": 0.8015,
+      "step": 35823
+    },
+    {
+      "epoch": 0.9838784982560216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3628720189517996e-08,
+      "loss": 0.7307,
+      "step": 35824
+    },
+    {
+      "epoch": 0.9839059624838647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3582335085519093e-08,
+      "loss": 0.8556,
+      "step": 35825
+    },
+    {
+      "epoch": 0.983933426711708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3536028997925832e-08,
+      "loss": 0.8022,
+      "step": 35826
+    },
+    {
+      "epoch": 0.9839608909395512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3489801927102364e-08,
+      "loss": 0.8939,
+      "step": 35827
+    },
+    {
+      "epoch": 0.9839883551673945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3443653873417284e-08,
+      "loss": 0.9226,
+      "step": 35828
+    },
+    {
+      "epoch": 0.9840158193952377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3397584837234745e-08,
+      "loss": 0.8242,
+      "step": 35829
+    },
+    {
+      "epoch": 0.9840432836230809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335159481891779e-08,
+      "loss": 0.7379,
+      "step": 35830
+    },
+    {
+      "epoch": 0.9840707478509242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3305683818832794e-08,
+      "loss": 0.7937,
+      "step": 35831
+    },
+    {
+      "epoch": 0.9840982120787674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3259851837340576e-08,
+      "loss": 0.7481,
+      "step": 35832
+    },
+    {
+      "epoch": 0.9841256763066106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3214098874805293e-08,
+      "loss": 0.7675,
+      "step": 35833
+    },
+    {
+      "epoch": 0.9841531405344539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3168424931588875e-08,
+      "loss": 0.735,
+      "step": 35834
+    },
+    {
+      "epoch": 0.9841806047622971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3122830008052145e-08,
+      "loss": 0.8407,
+      "step": 35835
+    },
+    {
+      "epoch": 0.9842080689901404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3077314104555927e-08,
+      "loss": 0.7421,
+      "step": 35836
+    },
+    {
+      "epoch": 0.9842355332179836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3031877221461042e-08,
+      "loss": 0.7497,
+      "step": 35837
+    },
+    {
+      "epoch": 0.9842629974458268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2986519359126092e-08,
+      "loss": 0.8003,
+      "step": 35838
+    },
+    {
+      "epoch": 0.9842904616736701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2941240517910792e-08,
+      "loss": 0.6814,
+      "step": 35839
+    },
+    {
+      "epoch": 0.9843179259015132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2896040698172629e-08,
+      "loss": 0.7509,
+      "step": 35840
+    },
+    {
+      "epoch": 0.9843453901293565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2850919900269098e-08,
+      "loss": 0.748,
+      "step": 35841
+    },
+    {
+      "epoch": 0.9843728543571998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2805878124558802e-08,
+      "loss": 0.7625,
+      "step": 35842
+    },
+    {
+      "epoch": 0.984400318585043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276091537139701e-08,
+      "loss": 0.7565,
+      "step": 35843
+    },
+    {
+      "epoch": 0.9844277828128862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2716031641138992e-08,
+      "loss": 0.8551,
+      "step": 35844
+    },
+    {
+      "epoch": 0.9844552470407294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2671226934140025e-08,
+      "loss": 0.7812,
+      "step": 35845
+    },
+    {
+      "epoch": 0.9844827112685727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2626501250756485e-08,
+      "loss": 0.7888,
+      "step": 35846
+    },
+    {
+      "epoch": 0.984510175496416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2581854591339204e-08,
+      "loss": 0.8098,
+      "step": 35847
+    },
+    {
+      "epoch": 0.9845376397242591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2537286956244565e-08,
+      "loss": 0.8192,
+      "step": 35848
+    },
+    {
+      "epoch": 0.9845651039521024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2492798345822287e-08,
+      "loss": 0.8919,
+      "step": 35849
+    },
+    {
+      "epoch": 0.9845925681799457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2448388760426533e-08,
+      "loss": 0.7049,
+      "step": 35850
+    },
+    {
+      "epoch": 0.9846200324077888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2404058200408131e-08,
+      "loss": 0.8655,
+      "step": 35851
+    },
+    {
+      "epoch": 0.9846474966356321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2359806666116802e-08,
+      "loss": 0.856,
+      "step": 35852
+    },
+    {
+      "epoch": 0.9846749608634753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2315634157904488e-08,
+      "loss": 0.7888,
+      "step": 35853
+    },
+    {
+      "epoch": 0.9847024250913186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2271540676118687e-08,
+      "loss": 0.8794,
+      "step": 35854
+    },
+    {
+      "epoch": 0.9847298893191618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2227526221110231e-08,
+      "loss": 0.8535,
+      "step": 35855
+    },
+    {
+      "epoch": 0.984757353547005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2183590793226619e-08,
+      "loss": 0.7964,
+      "step": 35856
+    },
+    {
+      "epoch": 0.9847848177748483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.213973439281424e-08,
+      "loss": 0.7883,
+      "step": 35857
+    },
+    {
+      "epoch": 0.9848122820026914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2095957020221705e-08,
+      "loss": 0.7553,
+      "step": 35858
+    },
+    {
+      "epoch": 0.9848397462305347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2052258675795402e-08,
+      "loss": 0.8383,
+      "step": 35859
+    },
+    {
+      "epoch": 0.984867210458378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2008639359880615e-08,
+      "loss": 0.747,
+      "step": 35860
+    },
+    {
+      "epoch": 0.9848946746862212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1965099072821507e-08,
+      "loss": 0.7506,
+      "step": 35861
+    },
+    {
+      "epoch": 0.9849221389140644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1921637814964471e-08,
+      "loss": 0.7618,
+      "step": 35862
+    },
+    {
+      "epoch": 0.9849496031419077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1878255586652565e-08,
+      "loss": 0.8646,
+      "step": 35863
+    },
+    {
+      "epoch": 0.9849770673697509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1834952388227739e-08,
+      "loss": 0.8453,
+      "step": 35864
+    },
+    {
+      "epoch": 0.9850045315975942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.179172822003527e-08,
+      "loss": 0.8623,
+      "step": 35865
+    },
+    {
+      "epoch": 0.9850319958254373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1748583082414889e-08,
+      "loss": 0.8238,
+      "step": 35866
+    },
+    {
+      "epoch": 0.9850594600532806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1705516975708542e-08,
+      "loss": 0.7803,
+      "step": 35867
+    },
+    {
+      "epoch": 0.9850869242811239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1662529900258179e-08,
+      "loss": 0.74,
+      "step": 35868
+    },
+    {
+      "epoch": 0.985114388508967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.161962185640242e-08,
+      "loss": 0.813,
+      "step": 35869
+    },
+    {
+      "epoch": 0.9851418527368103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1576792844480989e-08,
+      "loss": 0.8084,
+      "step": 35870
+    },
+    {
+      "epoch": 0.9851693169646535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1534042864833617e-08,
+      "loss": 0.8197,
+      "step": 35871
+    },
+    {
+      "epoch": 0.9851967811924968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1491371917797811e-08,
+      "loss": 0.8111,
+      "step": 35872
+    },
+    {
+      "epoch": 0.98522424542034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.144878000371108e-08,
+      "loss": 0.7971,
+      "step": 35873
+    },
+    {
+      "epoch": 0.9852517096481832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.140626712291204e-08,
+      "loss": 0.8184,
+      "step": 35874
+    },
+    {
+      "epoch": 0.9852791738760265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136383327573487e-08,
+      "loss": 0.8219,
+      "step": 35875
+    },
+    {
+      "epoch": 0.9853066381038698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1321478462515966e-08,
+      "loss": 0.7938,
+      "step": 35876
+    },
+    {
+      "epoch": 0.9853341023317129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1279202683591727e-08,
+      "loss": 0.7877,
+      "step": 35877
+    },
+    {
+      "epoch": 0.9853615665595562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123700593929522e-08,
+      "loss": 0.816,
+      "step": 35878
+    },
+    {
+      "epoch": 0.9853890307873994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.119488822996062e-08,
+      "loss": 0.7926,
+      "step": 35879
+    },
+    {
+      "epoch": 0.9854164950152426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1152849555922107e-08,
+      "loss": 0.8499,
+      "step": 35880
+    },
+    {
+      "epoch": 0.9854439592430859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1110889917510526e-08,
+      "loss": 0.7802,
+      "step": 35881
+    },
+    {
+      "epoch": 0.9854714234709291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1069009315058943e-08,
+      "loss": 0.875,
+      "step": 35882
+    },
+    {
+      "epoch": 0.9854988876987724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1027207748899316e-08,
+      "loss": 0.8011,
+      "step": 35883
+    },
+    {
+      "epoch": 0.9855263519266155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0985485219361381e-08,
+      "loss": 0.7933,
+      "step": 35884
+    },
+    {
+      "epoch": 0.9855538161544588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0943841726775983e-08,
+      "loss": 0.7995,
+      "step": 35885
+    },
+    {
+      "epoch": 0.9855812803823021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0902277271472861e-08,
+      "loss": 0.8123,
+      "step": 35886
+    },
+    {
+      "epoch": 0.9856087446101452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0860791853779528e-08,
+      "loss": 0.9043,
+      "step": 35887
+    },
+    {
+      "epoch": 0.9856362088379885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0819385474024613e-08,
+      "loss": 0.724,
+      "step": 35888
+    },
+    {
+      "epoch": 0.9856636730658318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077805813253674e-08,
+      "loss": 0.8139,
+      "step": 35889
+    },
+    {
+      "epoch": 0.985691137293675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0736809829642314e-08,
+      "loss": 0.7796,
+      "step": 35890
+    },
+    {
+      "epoch": 0.9857186015215182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0695640565667743e-08,
+      "loss": 0.7303,
+      "step": 35891
+    },
+    {
+      "epoch": 0.9857460657493614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0654550340938319e-08,
+      "loss": 0.7959,
+      "step": 35892
+    },
+    {
+      "epoch": 0.9857735299772047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061353915578045e-08,
+      "loss": 0.7701,
+      "step": 35893
+    },
+    {
+      "epoch": 0.985800994205048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0572607010517211e-08,
+      "loss": 0.8613,
+      "step": 35894
+    },
+    {
+      "epoch": 0.9858284584328911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0531753905472786e-08,
+      "loss": 0.8788,
+      "step": 35895
+    },
+    {
+      "epoch": 0.9858559226607344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.049097984097136e-08,
+      "loss": 0.7837,
+      "step": 35896
+    },
+    {
+      "epoch": 0.9858833868885776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0450284817334899e-08,
+      "loss": 0.7453,
+      "step": 35897
+    },
+    {
+      "epoch": 0.9859108511164208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0409668834885367e-08,
+      "loss": 0.7313,
+      "step": 35898
+    },
+    {
+      "epoch": 0.9859383153442641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0369131893943618e-08,
+      "loss": 0.8195,
+      "step": 35899
+    },
+    {
+      "epoch": 0.9859657795721073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0328673994830506e-08,
+      "loss": 0.8471,
+      "step": 35900
+    },
+    {
+      "epoch": 0.9859932437999506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0288295137866888e-08,
+      "loss": 0.8285,
+      "step": 35901
+    },
+    {
+      "epoch": 0.9860207080277938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247995323372506e-08,
+      "loss": 0.7755,
+      "step": 35902
+    },
+    {
+      "epoch": 0.986048172255637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0207774551665994e-08,
+      "loss": 0.8377,
+      "step": 35903
+    },
+    {
+      "epoch": 0.9860756364834803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0167632823063767e-08,
+      "loss": 0.7882,
+      "step": 35904
+    },
+    {
+      "epoch": 0.9861031007113235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127570137885567e-08,
+      "loss": 0.895,
+      "step": 35905
+    },
+    {
+      "epoch": 0.9861305649391667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.008758649644781e-08,
+      "loss": 0.7757,
+      "step": 35906
+    },
+    {
+      "epoch": 0.98615802916701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0047681899066908e-08,
+      "loss": 0.8444,
+      "step": 35907
+    },
+    {
+      "epoch": 0.9861854933948532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0007856346058164e-08,
+      "loss": 0.7144,
+      "step": 35908
+    },
+    {
+      "epoch": 0.9862129576226965,
+      "grad_norm": 0.0,
+      "learning_rate": 9.968109837737993e-09,
+      "loss": 0.8326,
+      "step": 35909
+    },
+    {
+      "epoch": 0.9862404218505396,
+      "grad_norm": 0.0,
+      "learning_rate": 9.928442374418368e-09,
+      "loss": 0.7976,
+      "step": 35910
+    },
+    {
+      "epoch": 0.9862678860783829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.888853956415701e-09,
+      "loss": 0.8047,
+      "step": 35911
+    },
+    {
+      "epoch": 0.9862953503062262,
+      "grad_norm": 0.0,
+      "learning_rate": 9.849344584041964e-09,
+      "loss": 0.795,
+      "step": 35912
+    },
+    {
+      "epoch": 0.9863228145340693,
+      "grad_norm": 0.0,
+      "learning_rate": 9.809914257610242e-09,
+      "loss": 0.782,
+      "step": 35913
+    },
+    {
+      "epoch": 0.9863502787619126,
+      "grad_norm": 0.0,
+      "learning_rate": 9.770562977432508e-09,
+      "loss": 0.8436,
+      "step": 35914
+    },
+    {
+      "epoch": 0.9863777429897559,
+      "grad_norm": 0.0,
+      "learning_rate": 9.731290743819621e-09,
+      "loss": 0.7695,
+      "step": 35915
+    },
+    {
+      "epoch": 0.9864052072175991,
+      "grad_norm": 0.0,
+      "learning_rate": 9.692097557082447e-09,
+      "loss": 0.7863,
+      "step": 35916
+    },
+    {
+      "epoch": 0.9864326714454423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.652983417530737e-09,
+      "loss": 0.797,
+      "step": 35917
+    },
+    {
+      "epoch": 0.9864601356732855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.613948325475352e-09,
+      "loss": 0.8627,
+      "step": 35918
+    },
+    {
+      "epoch": 0.9864875999011288,
+      "grad_norm": 0.0,
+      "learning_rate": 9.574992281224938e-09,
+      "loss": 0.7739,
+      "step": 35919
+    },
+    {
+      "epoch": 0.986515064128972,
+      "grad_norm": 0.0,
+      "learning_rate": 9.536115285085913e-09,
+      "loss": 0.8571,
+      "step": 35920
+    },
+    {
+      "epoch": 0.9865425283568152,
+      "grad_norm": 0.0,
+      "learning_rate": 9.497317337368029e-09,
+      "loss": 0.8461,
+      "step": 35921
+    },
+    {
+      "epoch": 0.9865699925846585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.458598438376598e-09,
+      "loss": 0.8046,
+      "step": 35922
+    },
+    {
+      "epoch": 0.9865974568125018,
+      "grad_norm": 0.0,
+      "learning_rate": 9.419958588418043e-09,
+      "loss": 0.8048,
+      "step": 35923
+    },
+    {
+      "epoch": 0.9866249210403449,
+      "grad_norm": 0.0,
+      "learning_rate": 9.381397787799896e-09,
+      "loss": 0.8101,
+      "step": 35924
+    },
+    {
+      "epoch": 0.9866523852681882,
+      "grad_norm": 0.0,
+      "learning_rate": 9.342916036826355e-09,
+      "loss": 0.7677,
+      "step": 35925
+    },
+    {
+      "epoch": 0.9866798494960314,
+      "grad_norm": 0.0,
+      "learning_rate": 9.304513335800514e-09,
+      "loss": 0.8355,
+      "step": 35926
+    },
+    {
+      "epoch": 0.9867073137238747,
+      "grad_norm": 0.0,
+      "learning_rate": 9.266189685027683e-09,
+      "loss": 0.7878,
+      "step": 35927
+    },
+    {
+      "epoch": 0.9867347779517179,
+      "grad_norm": 0.0,
+      "learning_rate": 9.227945084812063e-09,
+      "loss": 0.8591,
+      "step": 35928
+    },
+    {
+      "epoch": 0.9867622421795611,
+      "grad_norm": 0.0,
+      "learning_rate": 9.189779535453413e-09,
+      "loss": 0.8747,
+      "step": 35929
+    },
+    {
+      "epoch": 0.9867897064074044,
+      "grad_norm": 0.0,
+      "learning_rate": 9.151693037255938e-09,
+      "loss": 0.725,
+      "step": 35930
+    },
+    {
+      "epoch": 0.9868171706352475,
+      "grad_norm": 0.0,
+      "learning_rate": 9.113685590520504e-09,
+      "loss": 0.7993,
+      "step": 35931
+    },
+    {
+      "epoch": 0.9868446348630908,
+      "grad_norm": 0.0,
+      "learning_rate": 9.075757195547985e-09,
+      "loss": 0.767,
+      "step": 35932
+    },
+    {
+      "epoch": 0.9868720990909341,
+      "grad_norm": 0.0,
+      "learning_rate": 9.03790785263703e-09,
+      "loss": 0.9043,
+      "step": 35933
+    },
+    {
+      "epoch": 0.9868995633187773,
+      "grad_norm": 0.0,
+      "learning_rate": 9.000137562089616e-09,
+      "loss": 0.8497,
+      "step": 35934
+    },
+    {
+      "epoch": 0.9869270275466205,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96244632420218e-09,
+      "loss": 0.7632,
+      "step": 35935
+    },
+    {
+      "epoch": 0.9869544917744638,
+      "grad_norm": 0.0,
+      "learning_rate": 8.924834139274475e-09,
+      "loss": 0.8416,
+      "step": 35936
+    },
+    {
+      "epoch": 0.986981956002307,
+      "grad_norm": 0.0,
+      "learning_rate": 8.887301007604044e-09,
+      "loss": 0.9154,
+      "step": 35937
+    },
+    {
+      "epoch": 0.9870094202301503,
+      "grad_norm": 0.0,
+      "learning_rate": 8.849846929486205e-09,
+      "loss": 0.8135,
+      "step": 35938
+    },
+    {
+      "epoch": 0.9870368844579934,
+      "grad_norm": 0.0,
+      "learning_rate": 8.81247190521961e-09,
+      "loss": 0.7455,
+      "step": 35939
+    },
+    {
+      "epoch": 0.9870643486858367,
+      "grad_norm": 0.0,
+      "learning_rate": 8.775175935099579e-09,
+      "loss": 0.7986,
+      "step": 35940
+    },
+    {
+      "epoch": 0.98709181291368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.737959019419207e-09,
+      "loss": 0.8133,
+      "step": 35941
+    },
+    {
+      "epoch": 0.9871192771415231,
+      "grad_norm": 0.0,
+      "learning_rate": 8.700821158474927e-09,
+      "loss": 0.8081,
+      "step": 35942
+    },
+    {
+      "epoch": 0.9871467413693664,
+      "grad_norm": 0.0,
+      "learning_rate": 8.663762352559835e-09,
+      "loss": 0.7663,
+      "step": 35943
+    },
+    {
+      "epoch": 0.9871742055972096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.626782601968143e-09,
+      "loss": 0.8438,
+      "step": 35944
+    },
+    {
+      "epoch": 0.9872016698250529,
+      "grad_norm": 0.0,
+      "learning_rate": 8.589881906990727e-09,
+      "loss": 0.7815,
+      "step": 35945
+    },
+    {
+      "epoch": 0.9872291340528961,
+      "grad_norm": 0.0,
+      "learning_rate": 8.553060267920688e-09,
+      "loss": 0.826,
+      "step": 35946
+    },
+    {
+      "epoch": 0.9872565982807393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.516317685048903e-09,
+      "loss": 0.7326,
+      "step": 35947
+    },
+    {
+      "epoch": 0.9872840625085826,
+      "grad_norm": 0.0,
+      "learning_rate": 8.479654158667361e-09,
+      "loss": 0.8674,
+      "step": 35948
+    },
+    {
+      "epoch": 0.9873115267364259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.443069689063609e-09,
+      "loss": 0.7788,
+      "step": 35949
+    },
+    {
+      "epoch": 0.987338990964269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.406564276529639e-09,
+      "loss": 0.8452,
+      "step": 35950
+    },
+    {
+      "epoch": 0.9873664551921123,
+      "grad_norm": 0.0,
+      "learning_rate": 8.370137921352995e-09,
+      "loss": 0.7505,
+      "step": 35951
+    },
+    {
+      "epoch": 0.9873939194199555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.333790623822335e-09,
+      "loss": 0.9017,
+      "step": 35952
+    },
+    {
+      "epoch": 0.9874213836477987,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29752238422521e-09,
+      "loss": 0.7246,
+      "step": 35953
+    },
+    {
+      "epoch": 0.987448847875642,
+      "grad_norm": 0.0,
+      "learning_rate": 8.261333202849165e-09,
+      "loss": 0.7962,
+      "step": 35954
+    },
+    {
+      "epoch": 0.9874763121034852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.225223079978417e-09,
+      "loss": 0.8228,
+      "step": 35955
+    },
+    {
+      "epoch": 0.9875037763313285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.189192015901626e-09,
+      "loss": 0.8159,
+      "step": 35956
+    },
+    {
+      "epoch": 0.9875312405591716,
+      "grad_norm": 0.0,
+      "learning_rate": 8.153240010900787e-09,
+      "loss": 0.8593,
+      "step": 35957
+    },
+    {
+      "epoch": 0.9875587047870149,
+      "grad_norm": 0.0,
+      "learning_rate": 8.117367065263449e-09,
+      "loss": 0.833,
+      "step": 35958
+    },
+    {
+      "epoch": 0.9875861690148582,
+      "grad_norm": 0.0,
+      "learning_rate": 8.081573179271607e-09,
+      "loss": 0.7711,
+      "step": 35959
+    },
+    {
+      "epoch": 0.9876136332427013,
+      "grad_norm": 0.0,
+      "learning_rate": 8.045858353208369e-09,
+      "loss": 0.8198,
+      "step": 35960
+    },
+    {
+      "epoch": 0.9876410974705446,
+      "grad_norm": 0.0,
+      "learning_rate": 8.010222587356842e-09,
+      "loss": 0.7695,
+      "step": 35961
+    },
+    {
+      "epoch": 0.9876685616983879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.974665881999023e-09,
+      "loss": 0.7646,
+      "step": 35962
+    },
+    {
+      "epoch": 0.9876960259262311,
+      "grad_norm": 0.0,
+      "learning_rate": 7.939188237416906e-09,
+      "loss": 0.7978,
+      "step": 35963
+    },
+    {
+      "epoch": 0.9877234901540743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903789653890271e-09,
+      "loss": 0.7711,
+      "step": 35964
+    },
+    {
+      "epoch": 0.9877509543819175,
+      "grad_norm": 0.0,
+      "learning_rate": 7.86847013169889e-09,
+      "loss": 0.8449,
+      "step": 35965
+    },
+    {
+      "epoch": 0.9877784186097608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.833229671122544e-09,
+      "loss": 0.7292,
+      "step": 35966
+    },
+    {
+      "epoch": 0.9878058828376041,
+      "grad_norm": 0.0,
+      "learning_rate": 7.798068272441006e-09,
+      "loss": 0.8271,
+      "step": 35967
+    },
+    {
+      "epoch": 0.9878333470654472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.762985935931833e-09,
+      "loss": 0.8469,
+      "step": 35968
+    },
+    {
+      "epoch": 0.9878608112932905,
+      "grad_norm": 0.0,
+      "learning_rate": 7.72798266187147e-09,
+      "loss": 0.7051,
+      "step": 35969
+    },
+    {
+      "epoch": 0.9878882755211337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.693058450538582e-09,
+      "loss": 0.7086,
+      "step": 35970
+    },
+    {
+      "epoch": 0.987915739748977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.658213302209616e-09,
+      "loss": 0.8314,
+      "step": 35971
+    },
+    {
+      "epoch": 0.9879432039768202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.623447217158797e-09,
+      "loss": 0.9042,
+      "step": 35972
+    },
+    {
+      "epoch": 0.9879706682046634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.58876019566146e-09,
+      "loss": 0.7815,
+      "step": 35973
+    },
+    {
+      "epoch": 0.9879981324325067,
+      "grad_norm": 0.0,
+      "learning_rate": 7.55415223799294e-09,
+      "loss": 0.7694,
+      "step": 35974
+    },
+    {
+      "epoch": 0.98802559666035,
+      "grad_norm": 0.0,
+      "learning_rate": 7.519623344427463e-09,
+      "loss": 0.8451,
+      "step": 35975
+    },
+    {
+      "epoch": 0.9880530608881931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.485173515235922e-09,
+      "loss": 0.8183,
+      "step": 35976
+    },
+    {
+      "epoch": 0.9880805251160364,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450802750693653e-09,
+      "loss": 0.7437,
+      "step": 35977
+    },
+    {
+      "epoch": 0.9881079893438796,
+      "grad_norm": 0.0,
+      "learning_rate": 7.416511051071551e-09,
+      "loss": 0.8013,
+      "step": 35978
+    },
+    {
+      "epoch": 0.9881354535717228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.382298416639399e-09,
+      "loss": 0.8018,
+      "step": 35979
+    },
+    {
+      "epoch": 0.9881629177995661,
+      "grad_norm": 0.0,
+      "learning_rate": 7.348164847670314e-09,
+      "loss": 0.8396,
+      "step": 35980
+    },
+    {
+      "epoch": 0.9881903820274093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3141103444329676e-09,
+      "loss": 0.7922,
+      "step": 35981
+    },
+    {
+      "epoch": 0.9882178462552526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.280134907197145e-09,
+      "loss": 0.8636,
+      "step": 35982
+    },
+    {
+      "epoch": 0.9882453104830957,
+      "grad_norm": 0.0,
+      "learning_rate": 7.24623853623263e-09,
+      "loss": 0.8328,
+      "step": 35983
+    },
+    {
+      "epoch": 0.988272774710939,
+      "grad_norm": 0.0,
+      "learning_rate": 7.212421231805878e-09,
+      "loss": 0.7352,
+      "step": 35984
+    },
+    {
+      "epoch": 0.9883002389387823,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17868299418556e-09,
+      "loss": 0.8316,
+      "step": 35985
+    },
+    {
+      "epoch": 0.9883277031666254,
+      "grad_norm": 0.0,
+      "learning_rate": 7.145023823638131e-09,
+      "loss": 0.826,
+      "step": 35986
+    },
+    {
+      "epoch": 0.9883551673944687,
+      "grad_norm": 0.0,
+      "learning_rate": 7.111443720430044e-09,
+      "loss": 0.8184,
+      "step": 35987
+    },
+    {
+      "epoch": 0.988382631622312,
+      "grad_norm": 0.0,
+      "learning_rate": 7.077942684826644e-09,
+      "loss": 0.8549,
+      "step": 35988
+    },
+    {
+      "epoch": 0.9884100958501552,
+      "grad_norm": 0.0,
+      "learning_rate": 7.044520717094383e-09,
+      "loss": 0.7457,
+      "step": 35989
+    },
+    {
+      "epoch": 0.9884375600779984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.011177817496384e-09,
+      "loss": 0.8264,
+      "step": 35990
+    },
+    {
+      "epoch": 0.9884650243058416,
+      "grad_norm": 0.0,
+      "learning_rate": 6.97791398629688e-09,
+      "loss": 0.8286,
+      "step": 35991
+    },
+    {
+      "epoch": 0.9884924885336849,
+      "grad_norm": 0.0,
+      "learning_rate": 6.944729223758995e-09,
+      "loss": 0.8582,
+      "step": 35992
+    },
+    {
+      "epoch": 0.9885199527615282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.91162353014474e-09,
+      "loss": 0.6945,
+      "step": 35993
+    },
+    {
+      "epoch": 0.9885474169893713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.878596905717239e-09,
+      "loss": 0.8754,
+      "step": 35994
+    },
+    {
+      "epoch": 0.9885748812172146,
+      "grad_norm": 0.0,
+      "learning_rate": 6.845649350737393e-09,
+      "loss": 0.7625,
+      "step": 35995
+    },
+    {
+      "epoch": 0.9886023454450578,
+      "grad_norm": 0.0,
+      "learning_rate": 6.812780865466107e-09,
+      "loss": 0.7202,
+      "step": 35996
+    },
+    {
+      "epoch": 0.988629809672901,
+      "grad_norm": 0.0,
+      "learning_rate": 6.77999145016206e-09,
+      "loss": 0.8059,
+      "step": 35997
+    },
+    {
+      "epoch": 0.9886572739007443,
+      "grad_norm": 0.0,
+      "learning_rate": 6.747281105087267e-09,
+      "loss": 0.7527,
+      "step": 35998
+    },
+    {
+      "epoch": 0.9886847381285875,
+      "grad_norm": 0.0,
+      "learning_rate": 6.714649830498187e-09,
+      "loss": 0.6805,
+      "step": 35999
+    },
+    {
+      "epoch": 0.9887122023564308,
+      "grad_norm": 0.0,
+      "learning_rate": 6.682097626653505e-09,
+      "loss": 0.7436,
+      "step": 36000
+    },
+    {
+      "epoch": 0.988739666584274,
+      "grad_norm": 0.0,
+      "learning_rate": 6.649624493810791e-09,
+      "loss": 0.8126,
+      "step": 36001
+    },
+    {
+      "epoch": 0.9887671308121172,
+      "grad_norm": 0.0,
+      "learning_rate": 6.617230432227617e-09,
+      "loss": 0.7854,
+      "step": 36002
+    },
+    {
+      "epoch": 0.9887945950399605,
+      "grad_norm": 0.0,
+      "learning_rate": 6.584915442160445e-09,
+      "loss": 0.7859,
+      "step": 36003
+    },
+    {
+      "epoch": 0.9888220592678036,
+      "grad_norm": 0.0,
+      "learning_rate": 6.552679523864624e-09,
+      "loss": 0.8313,
+      "step": 36004
+    },
+    {
+      "epoch": 0.9888495234956469,
+      "grad_norm": 0.0,
+      "learning_rate": 6.520522677594399e-09,
+      "loss": 0.8524,
+      "step": 36005
+    },
+    {
+      "epoch": 0.9888769877234902,
+      "grad_norm": 0.0,
+      "learning_rate": 6.488444903605118e-09,
+      "loss": 0.8269,
+      "step": 36006
+    },
+    {
+      "epoch": 0.9889044519513334,
+      "grad_norm": 0.0,
+      "learning_rate": 6.456446202151024e-09,
+      "loss": 0.863,
+      "step": 36007
+    },
+    {
+      "epoch": 0.9889319161791766,
+      "grad_norm": 0.0,
+      "learning_rate": 6.424526573484136e-09,
+      "loss": 0.7714,
+      "step": 36008
+    },
+    {
+      "epoch": 0.9889593804070198,
+      "grad_norm": 0.0,
+      "learning_rate": 6.392686017856475e-09,
+      "loss": 0.7498,
+      "step": 36009
+    },
+    {
+      "epoch": 0.9889868446348631,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3609245355222835e-09,
+      "loss": 0.8346,
+      "step": 36010
+    },
+    {
+      "epoch": 0.9890143088627064,
+      "grad_norm": 0.0,
+      "learning_rate": 6.329242126731361e-09,
+      "loss": 0.7929,
+      "step": 36011
+    },
+    {
+      "epoch": 0.9890417730905495,
+      "grad_norm": 0.0,
+      "learning_rate": 6.297638791734617e-09,
+      "loss": 0.7993,
+      "step": 36012
+    },
+    {
+      "epoch": 0.9890692373183928,
+      "grad_norm": 0.0,
+      "learning_rate": 6.266114530782963e-09,
+      "loss": 0.7015,
+      "step": 36013
+    },
+    {
+      "epoch": 0.9890967015462361,
+      "grad_norm": 0.0,
+      "learning_rate": 6.234669344123978e-09,
+      "loss": 0.8129,
+      "step": 36014
+    },
+    {
+      "epoch": 0.9891241657740792,
+      "grad_norm": 0.0,
+      "learning_rate": 6.203303232007463e-09,
+      "loss": 0.8351,
+      "step": 36015
+    },
+    {
+      "epoch": 0.9891516300019225,
+      "grad_norm": 0.0,
+      "learning_rate": 6.172016194682106e-09,
+      "loss": 0.71,
+      "step": 36016
+    },
+    {
+      "epoch": 0.9891790942297657,
+      "grad_norm": 0.0,
+      "learning_rate": 6.14080823239438e-09,
+      "loss": 0.7923,
+      "step": 36017
+    },
+    {
+      "epoch": 0.989206558457609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.109679345392971e-09,
+      "loss": 0.8713,
+      "step": 36018
+    },
+    {
+      "epoch": 0.9892340226854522,
+      "grad_norm": 0.0,
+      "learning_rate": 6.078629533922131e-09,
+      "loss": 0.7804,
+      "step": 36019
+    },
+    {
+      "epoch": 0.9892614869132954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.047658798228328e-09,
+      "loss": 0.7325,
+      "step": 36020
+    },
+    {
+      "epoch": 0.9892889511411387,
+      "grad_norm": 0.0,
+      "learning_rate": 6.016767138556923e-09,
+      "loss": 0.8123,
+      "step": 36021
+    },
+    {
+      "epoch": 0.9893164153689818,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9859545551532726e-09,
+      "loss": 0.7727,
+      "step": 36022
+    },
+    {
+      "epoch": 0.9893438795968251,
+      "grad_norm": 0.0,
+      "learning_rate": 5.955221048258297e-09,
+      "loss": 0.8028,
+      "step": 36023
+    },
+    {
+      "epoch": 0.9893713438246684,
+      "grad_norm": 0.0,
+      "learning_rate": 5.924566618118466e-09,
+      "loss": 0.8359,
+      "step": 36024
+    },
+    {
+      "epoch": 0.9893988080525116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8939912649735865e-09,
+      "loss": 0.8925,
+      "step": 36025
+    },
+    {
+      "epoch": 0.9894262722803548,
+      "grad_norm": 0.0,
+      "learning_rate": 5.863494989067908e-09,
+      "loss": 0.8123,
+      "step": 36026
+    },
+    {
+      "epoch": 0.9894537365081981,
+      "grad_norm": 0.0,
+      "learning_rate": 5.83307779064013e-09,
+      "loss": 0.8245,
+      "step": 36027
+    },
+    {
+      "epoch": 0.9894812007360413,
+      "grad_norm": 0.0,
+      "learning_rate": 5.802739669933388e-09,
+      "loss": 0.816,
+      "step": 36028
+    },
+    {
+      "epoch": 0.9895086649638846,
+      "grad_norm": 0.0,
+      "learning_rate": 5.772480627187493e-09,
+      "loss": 0.8172,
+      "step": 36029
+    },
+    {
+      "epoch": 0.9895361291917277,
+      "grad_norm": 0.0,
+      "learning_rate": 5.742300662640032e-09,
+      "loss": 0.7831,
+      "step": 36030
+    },
+    {
+      "epoch": 0.989563593419571,
+      "grad_norm": 0.0,
+      "learning_rate": 5.712199776530813e-09,
+      "loss": 0.8854,
+      "step": 36031
+    },
+    {
+      "epoch": 0.9895910576474143,
+      "grad_norm": 0.0,
+      "learning_rate": 5.682177969098535e-09,
+      "loss": 0.8438,
+      "step": 36032
+    },
+    {
+      "epoch": 0.9896185218752575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.652235240580783e-09,
+      "loss": 0.8399,
+      "step": 36033
+    },
+    {
+      "epoch": 0.9896459861031007,
+      "grad_norm": 0.0,
+      "learning_rate": 5.622371591212927e-09,
+      "loss": 0.8961,
+      "step": 36034
+    },
+    {
+      "epoch": 0.9896734503309439,
+      "grad_norm": 0.0,
+      "learning_rate": 5.592587021232554e-09,
+      "loss": 0.8007,
+      "step": 36035
+    },
+    {
+      "epoch": 0.9897009145587872,
+      "grad_norm": 0.0,
+      "learning_rate": 5.56288153087503e-09,
+      "loss": 0.7571,
+      "step": 36036
+    },
+    {
+      "epoch": 0.9897283787866304,
+      "grad_norm": 0.0,
+      "learning_rate": 5.533255120375724e-09,
+      "loss": 0.7302,
+      "step": 36037
+    },
+    {
+      "epoch": 0.9897558430144736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.503707789968893e-09,
+      "loss": 0.8093,
+      "step": 36038
+    },
+    {
+      "epoch": 0.9897833072423169,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4742395398887925e-09,
+      "loss": 0.7644,
+      "step": 36039
+    },
+    {
+      "epoch": 0.9898107714701602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.44485037036746e-09,
+      "loss": 0.7812,
+      "step": 36040
+    },
+    {
+      "epoch": 0.9898382356980033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.415540281636933e-09,
+      "loss": 0.8318,
+      "step": 36041
+    },
+    {
+      "epoch": 0.9898656999258466,
+      "grad_norm": 0.0,
+      "learning_rate": 5.386309273931467e-09,
+      "loss": 0.7904,
+      "step": 36042
+    },
+    {
+      "epoch": 0.9898931641536898,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3571573474808795e-09,
+      "loss": 0.7821,
+      "step": 36043
+    },
+    {
+      "epoch": 0.989920628381533,
+      "grad_norm": 0.0,
+      "learning_rate": 5.328084502516095e-09,
+      "loss": 0.7659,
+      "step": 36044
+    },
+    {
+      "epoch": 0.9899480926093763,
+      "grad_norm": 0.0,
+      "learning_rate": 5.299090739266932e-09,
+      "loss": 0.6693,
+      "step": 36045
+    },
+    {
+      "epoch": 0.9899755568372195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.270176057962095e-09,
+      "loss": 0.8919,
+      "step": 36046
+    },
+    {
+      "epoch": 0.9900030210650628,
+      "grad_norm": 0.0,
+      "learning_rate": 5.241340458832511e-09,
+      "loss": 0.7407,
+      "step": 36047
+    },
+    {
+      "epoch": 0.9900304852929059,
+      "grad_norm": 0.0,
+      "learning_rate": 5.212583942104665e-09,
+      "loss": 0.8168,
+      "step": 36048
+    },
+    {
+      "epoch": 0.9900579495207492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.183906508006153e-09,
+      "loss": 0.7388,
+      "step": 36049
+    },
+    {
+      "epoch": 0.9900854137485925,
+      "grad_norm": 0.0,
+      "learning_rate": 5.155308156764572e-09,
+      "loss": 0.7975,
+      "step": 36050
+    },
+    {
+      "epoch": 0.9901128779764357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.126788888605294e-09,
+      "loss": 0.8581,
+      "step": 36051
+    },
+    {
+      "epoch": 0.9901403422042789,
+      "grad_norm": 0.0,
+      "learning_rate": 5.098348703754808e-09,
+      "loss": 0.8133,
+      "step": 36052
+    },
+    {
+      "epoch": 0.9901678064321222,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0699876024373764e-09,
+      "loss": 0.8178,
+      "step": 36053
+    },
+    {
+      "epoch": 0.9901952706599654,
+      "grad_norm": 0.0,
+      "learning_rate": 5.041705584879486e-09,
+      "loss": 0.7815,
+      "step": 36054
+    },
+    {
+      "epoch": 0.9902227348878087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.013502651302071e-09,
+      "loss": 0.7006,
+      "step": 36055
+    },
+    {
+      "epoch": 0.9902501991156518,
+      "grad_norm": 0.0,
+      "learning_rate": 4.985378801929397e-09,
+      "loss": 0.7692,
+      "step": 36056
+    },
+    {
+      "epoch": 0.9902776633434951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.957334036984618e-09,
+      "loss": 0.8786,
+      "step": 36057
+    },
+    {
+      "epoch": 0.9903051275713384,
+      "grad_norm": 0.0,
+      "learning_rate": 4.929368356688668e-09,
+      "loss": 0.752,
+      "step": 36058
+    },
+    {
+      "epoch": 0.9903325917991815,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9014817612624835e-09,
+      "loss": 0.873,
+      "step": 36059
+    },
+    {
+      "epoch": 0.9903600560270248,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8736742509292165e-09,
+      "loss": 0.9536,
+      "step": 36060
+    },
+    {
+      "epoch": 0.990387520254868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.845945825906473e-09,
+      "loss": 0.849,
+      "step": 36061
+    },
+    {
+      "epoch": 0.9904149844827113,
+      "grad_norm": 0.0,
+      "learning_rate": 4.818296486414076e-09,
+      "loss": 0.8328,
+      "step": 36062
+    },
+    {
+      "epoch": 0.9904424487105545,
+      "grad_norm": 0.0,
+      "learning_rate": 4.790726232670739e-09,
+      "loss": 0.8205,
+      "step": 36063
+    },
+    {
+      "epoch": 0.9904699129383977,
+      "grad_norm": 0.0,
+      "learning_rate": 4.763235064896288e-09,
+      "loss": 0.7881,
+      "step": 36064
+    },
+    {
+      "epoch": 0.990497377166241,
+      "grad_norm": 0.0,
+      "learning_rate": 4.735822983306104e-09,
+      "loss": 0.828,
+      "step": 36065
+    },
+    {
+      "epoch": 0.9905248413940843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.708489988117793e-09,
+      "loss": 0.7641,
+      "step": 36066
+    },
+    {
+      "epoch": 0.9905523056219274,
+      "grad_norm": 0.0,
+      "learning_rate": 4.681236079547846e-09,
+      "loss": 0.7526,
+      "step": 36067
+    },
+    {
+      "epoch": 0.9905797698497707,
+      "grad_norm": 0.0,
+      "learning_rate": 4.654061257811648e-09,
+      "loss": 0.7746,
+      "step": 36068
+    },
+    {
+      "epoch": 0.9906072340776139,
+      "grad_norm": 0.0,
+      "learning_rate": 4.626965523124582e-09,
+      "loss": 0.7814,
+      "step": 36069
+    },
+    {
+      "epoch": 0.9906346983054571,
+      "grad_norm": 0.0,
+      "learning_rate": 4.599948875699811e-09,
+      "loss": 0.8253,
+      "step": 36070
+    },
+    {
+      "epoch": 0.9906621625333004,
+      "grad_norm": 0.0,
+      "learning_rate": 4.573011315753828e-09,
+      "loss": 0.7844,
+      "step": 36071
+    },
+    {
+      "epoch": 0.9906896267611436,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546152843496465e-09,
+      "loss": 0.8575,
+      "step": 36072
+    },
+    {
+      "epoch": 0.9907170909889869,
+      "grad_norm": 0.0,
+      "learning_rate": 4.519373459143106e-09,
+      "loss": 0.7958,
+      "step": 36073
+    },
+    {
+      "epoch": 0.99074455521683,
+      "grad_norm": 0.0,
+      "learning_rate": 4.492673162903583e-09,
+      "loss": 0.7968,
+      "step": 36074
+    },
+    {
+      "epoch": 0.9907720194446733,
+      "grad_norm": 0.0,
+      "learning_rate": 4.466051954989947e-09,
+      "loss": 0.6919,
+      "step": 36075
+    },
+    {
+      "epoch": 0.9907994836725166,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4395098356131426e-09,
+      "loss": 0.7686,
+      "step": 36076
+    },
+    {
+      "epoch": 0.9908269479003597,
+      "grad_norm": 0.0,
+      "learning_rate": 4.413046804981891e-09,
+      "loss": 0.7401,
+      "step": 36077
+    },
+    {
+      "epoch": 0.990854412128203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.386662863308244e-09,
+      "loss": 0.7784,
+      "step": 36078
+    },
+    {
+      "epoch": 0.9908818763560463,
+      "grad_norm": 0.0,
+      "learning_rate": 4.360358010797594e-09,
+      "loss": 0.7412,
+      "step": 36079
+    },
+    {
+      "epoch": 0.9909093405838895,
+      "grad_norm": 0.0,
+      "learning_rate": 4.334132247660883e-09,
+      "loss": 0.879,
+      "step": 36080
+    },
+    {
+      "epoch": 0.9909368048117327,
+      "grad_norm": 0.0,
+      "learning_rate": 4.307985574103501e-09,
+      "loss": 0.7076,
+      "step": 36081
+    },
+    {
+      "epoch": 0.9909642690395759,
+      "grad_norm": 0.0,
+      "learning_rate": 4.281917990334172e-09,
+      "loss": 0.7006,
+      "step": 36082
+    },
+    {
+      "epoch": 0.9909917332674192,
+      "grad_norm": 0.0,
+      "learning_rate": 4.255929496558287e-09,
+      "loss": 0.6937,
+      "step": 36083
+    },
+    {
+      "epoch": 0.9910191974952625,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2300200929812355e-09,
+      "loss": 0.8052,
+      "step": 36084
+    },
+    {
+      "epoch": 0.9910466617231056,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2041897798072996e-09,
+      "loss": 0.8,
+      "step": 36085
+    },
+    {
+      "epoch": 0.9910741259509489,
+      "grad_norm": 0.0,
+      "learning_rate": 4.178438557242981e-09,
+      "loss": 0.8303,
+      "step": 36086
+    },
+    {
+      "epoch": 0.9911015901787921,
+      "grad_norm": 0.0,
+      "learning_rate": 4.15276642549034e-09,
+      "loss": 0.8118,
+      "step": 36087
+    },
+    {
+      "epoch": 0.9911290544066353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.127173384753658e-09,
+      "loss": 0.7666,
+      "step": 36088
+    },
+    {
+      "epoch": 0.9911565186344786,
+      "grad_norm": 0.0,
+      "learning_rate": 4.101659435233885e-09,
+      "loss": 0.7072,
+      "step": 36089
+    },
+    {
+      "epoch": 0.9911839828623218,
+      "grad_norm": 0.0,
+      "learning_rate": 4.076224577134191e-09,
+      "loss": 0.7928,
+      "step": 36090
+    },
+    {
+      "epoch": 0.9912114470901651,
+      "grad_norm": 0.0,
+      "learning_rate": 4.050868810655529e-09,
+      "loss": 0.7352,
+      "step": 36091
+    },
+    {
+      "epoch": 0.9912389113180083,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025592135998846e-09,
+      "loss": 0.7947,
+      "step": 36092
+    },
+    {
+      "epoch": 0.9912663755458515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.000394553362874e-09,
+      "loss": 0.8775,
+      "step": 36093
+    },
+    {
+      "epoch": 0.9912938397736948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.975276062948563e-09,
+      "loss": 0.7996,
+      "step": 36094
+    },
+    {
+      "epoch": 0.991321304001538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.950236664954643e-09,
+      "loss": 0.7938,
+      "step": 36095
+    },
+    {
+      "epoch": 0.9913487682293812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.925276359577623e-09,
+      "loss": 0.7804,
+      "step": 36096
+    },
+    {
+      "epoch": 0.9913762324572245,
+      "grad_norm": 0.0,
+      "learning_rate": 3.900395147015123e-09,
+      "loss": 0.8795,
+      "step": 36097
+    },
+    {
+      "epoch": 0.9914036966850677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.875593027465873e-09,
+      "loss": 0.7882,
+      "step": 36098
+    },
+    {
+      "epoch": 0.991431160912911,
+      "grad_norm": 0.0,
+      "learning_rate": 3.850870001125273e-09,
+      "loss": 0.7835,
+      "step": 36099
+    },
+    {
+      "epoch": 0.9914586251407541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8262260681876105e-09,
+      "loss": 0.8199,
+      "step": 36100
+    },
+    {
+      "epoch": 0.9914860893685974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.801661228850506e-09,
+      "loss": 0.7908,
+      "step": 36101
+    },
+    {
+      "epoch": 0.9915135535964407,
+      "grad_norm": 0.0,
+      "learning_rate": 3.777175483306028e-09,
+      "loss": 0.7991,
+      "step": 36102
+    },
+    {
+      "epoch": 0.9915410178242838,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7527688317484654e-09,
+      "loss": 0.8667,
+      "step": 36103
+    },
+    {
+      "epoch": 0.9915684820521271,
+      "grad_norm": 0.0,
+      "learning_rate": 3.728441274370998e-09,
+      "loss": 0.7971,
+      "step": 36104
+    },
+    {
+      "epoch": 0.9915959462799704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7041928113668023e-09,
+      "loss": 0.8141,
+      "step": 36105
+    },
+    {
+      "epoch": 0.9916234105078136,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6800234429268387e-09,
+      "loss": 0.7727,
+      "step": 36106
+    },
+    {
+      "epoch": 0.9916508747356568,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6559331692431753e-09,
+      "loss": 0.8009,
+      "step": 36107
+    },
+    {
+      "epoch": 0.9916783389635,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6319219905067705e-09,
+      "loss": 0.7074,
+      "step": 36108
+    },
+    {
+      "epoch": 0.9917058031913433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6079899069052517e-09,
+      "loss": 0.7706,
+      "step": 36109
+    },
+    {
+      "epoch": 0.9917332674191865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.584136918630687e-09,
+      "loss": 0.7529,
+      "step": 36110
+    },
+    {
+      "epoch": 0.9917607316470297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5603630258695956e-09,
+      "loss": 0.7724,
+      "step": 36111
+    },
+    {
+      "epoch": 0.991788195874873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.536668228811824e-09,
+      "loss": 0.8035,
+      "step": 36112
+    },
+    {
+      "epoch": 0.9918156601027162,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5130525276438896e-09,
+      "loss": 0.7643,
+      "step": 36113
+    },
+    {
+      "epoch": 0.9918431243305594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4895159225534215e-09,
+      "loss": 0.7874,
+      "step": 36114
+    },
+    {
+      "epoch": 0.9918705885584027,
+      "grad_norm": 0.0,
+      "learning_rate": 3.466058413725826e-09,
+      "loss": 0.7324,
+      "step": 36115
+    },
+    {
+      "epoch": 0.9918980527862459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4426800013476203e-09,
+      "loss": 0.7493,
+      "step": 36116
+    },
+    {
+      "epoch": 0.9919255170140892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4193806856019917e-09,
+      "loss": 0.7789,
+      "step": 36117
+    },
+    {
+      "epoch": 0.9919529812419324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.396160466675458e-09,
+      "loss": 0.7861,
+      "step": 36118
+    },
+    {
+      "epoch": 0.9919804454697756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.373019344751205e-09,
+      "loss": 0.8781,
+      "step": 36119
+    },
+    {
+      "epoch": 0.9920079096976189,
+      "grad_norm": 0.0,
+      "learning_rate": 3.34995732001131e-09,
+      "loss": 0.8355,
+      "step": 36120
+    },
+    {
+      "epoch": 0.992035373925462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3269743926389597e-09,
+      "loss": 0.8405,
+      "step": 36121
+    },
+    {
+      "epoch": 0.9920628381533053,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3040705628151204e-09,
+      "loss": 0.7488,
+      "step": 36122
+    },
+    {
+      "epoch": 0.9920903023811486,
+      "grad_norm": 0.0,
+      "learning_rate": 3.281245830722979e-09,
+      "loss": 0.8109,
+      "step": 36123
+    },
+    {
+      "epoch": 0.9921177666089918,
+      "grad_norm": 0.0,
+      "learning_rate": 3.258500196541281e-09,
+      "loss": 0.7133,
+      "step": 36124
+    },
+    {
+      "epoch": 0.992145230836835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.235833660450993e-09,
+      "loss": 0.783,
+      "step": 36125
+    },
+    {
+      "epoch": 0.9921726950646783,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2132462226319716e-09,
+      "loss": 0.9088,
+      "step": 36126
+    },
+    {
+      "epoch": 0.9922001592925215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1907378832607416e-09,
+      "loss": 0.8183,
+      "step": 36127
+    },
+    {
+      "epoch": 0.9922276235203648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.16830864251827e-09,
+      "loss": 0.7487,
+      "step": 36128
+    },
+    {
+      "epoch": 0.9922550877482079,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1459585005788607e-09,
+      "loss": 0.8885,
+      "step": 36129
+    },
+    {
+      "epoch": 0.9922825519760512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.123687457622371e-09,
+      "loss": 0.8834,
+      "step": 36130
+    },
+    {
+      "epoch": 0.9923100162038945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1014955138231053e-09,
+      "loss": 0.7673,
+      "step": 36131
+    },
+    {
+      "epoch": 0.9923374804317376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.079382669357589e-09,
+      "loss": 0.7687,
+      "step": 36132
+    },
+    {
+      "epoch": 0.9923649446595809,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0573489244001274e-09,
+      "loss": 0.8347,
+      "step": 36133
+    },
+    {
+      "epoch": 0.9923924088874241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0353942791250256e-09,
+      "loss": 0.9409,
+      "step": 36134
+    },
+    {
+      "epoch": 0.9924198731152674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0135187337065885e-09,
+      "loss": 0.8865,
+      "step": 36135
+    },
+    {
+      "epoch": 0.9924473373431106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9917222883180107e-09,
+      "loss": 0.8657,
+      "step": 36136
+    },
+    {
+      "epoch": 0.9924748015709538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.970004943131377e-09,
+      "loss": 0.8369,
+      "step": 36137
+    },
+    {
+      "epoch": 0.9925022657987971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.948366698317662e-09,
+      "loss": 0.7555,
+      "step": 36138
+    },
+    {
+      "epoch": 0.9925297300266404,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9268075540500597e-09,
+      "loss": 0.8001,
+      "step": 36139
+    },
+    {
+      "epoch": 0.9925571942544835,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9053275104973244e-09,
+      "loss": 0.7344,
+      "step": 36140
+    },
+    {
+      "epoch": 0.9925846584823268,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8839265678304308e-09,
+      "loss": 0.8479,
+      "step": 36141
+    },
+    {
+      "epoch": 0.99261212271017,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8626047262192426e-09,
+      "loss": 0.8201,
+      "step": 36142
+    },
+    {
+      "epoch": 0.9926395869380132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.841361985830293e-09,
+      "loss": 0.8353,
+      "step": 36143
+    },
+    {
+      "epoch": 0.9926670511658565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.820198346833447e-09,
+      "loss": 0.7523,
+      "step": 36144
+    },
+    {
+      "epoch": 0.9926945153936997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7991138093963477e-09,
+      "loss": 0.8193,
+      "step": 36145
+    },
+    {
+      "epoch": 0.992721979621543,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7781083736844184e-09,
+      "loss": 0.8211,
+      "step": 36146
+    },
+    {
+      "epoch": 0.9927494438493861,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7571820398653025e-09,
+      "loss": 0.6685,
+      "step": 36147
+    },
+    {
+      "epoch": 0.9927769080772294,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7363348081033137e-09,
+      "loss": 0.8214,
+      "step": 36148
+    },
+    {
+      "epoch": 0.9928043723050727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7155666785638747e-09,
+      "loss": 0.7857,
+      "step": 36149
+    },
+    {
+      "epoch": 0.9928318365329158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6948776514124086e-09,
+      "loss": 0.7722,
+      "step": 36150
+    },
+    {
+      "epoch": 0.9928593007607591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6742677268110084e-09,
+      "loss": 0.765,
+      "step": 36151
+    },
+    {
+      "epoch": 0.9928867649886024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6537369049228767e-09,
+      "loss": 0.7655,
+      "step": 36152
+    },
+    {
+      "epoch": 0.9929142292164456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6332851859123265e-09,
+      "loss": 0.8275,
+      "step": 36153
+    },
+    {
+      "epoch": 0.9929416934442888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6129125699381198e-09,
+      "loss": 0.8306,
+      "step": 36154
+    },
+    {
+      "epoch": 0.992969157672132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5926190571645692e-09,
+      "loss": 0.7962,
+      "step": 36155
+    },
+    {
+      "epoch": 0.9929966218999753,
+      "grad_norm": 0.0,
+      "learning_rate": 2.572404647750437e-09,
+      "loss": 0.7887,
+      "step": 36156
+    },
+    {
+      "epoch": 0.9930240861278186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5522693418555955e-09,
+      "loss": 0.8182,
+      "step": 36157
+    },
+    {
+      "epoch": 0.9930515503556617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5322131396410266e-09,
+      "loss": 0.8245,
+      "step": 36158
+    },
+    {
+      "epoch": 0.993079014583505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5122360412632716e-09,
+      "loss": 0.7982,
+      "step": 36159
+    },
+    {
+      "epoch": 0.9931064788113482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.492338046881093e-09,
+      "loss": 0.7728,
+      "step": 36160
+    },
+    {
+      "epoch": 0.9931339430391914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4725191566532525e-09,
+      "loss": 0.8999,
+      "step": 36161
+    },
+    {
+      "epoch": 0.9931614072670347,
+      "grad_norm": 0.0,
+      "learning_rate": 2.452779370735181e-09,
+      "loss": 0.7978,
+      "step": 36162
+    },
+    {
+      "epoch": 0.9931888714948779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.43311868928231e-09,
+      "loss": 0.8448,
+      "step": 36163
+    },
+    {
+      "epoch": 0.9932163357227212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.413537112452291e-09,
+      "loss": 0.7896,
+      "step": 36164
+    },
+    {
+      "epoch": 0.9932437999505644,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3940346403994454e-09,
+      "loss": 0.8417,
+      "step": 36165
+    },
+    {
+      "epoch": 0.9932712641784076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.374611273276983e-09,
+      "loss": 0.7786,
+      "step": 36166
+    },
+    {
+      "epoch": 0.9932987284062509,
+      "grad_norm": 0.0,
+      "learning_rate": 2.355267011239226e-09,
+      "loss": 0.7546,
+      "step": 36167
+    },
+    {
+      "epoch": 0.993326192634094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.336001854439385e-09,
+      "loss": 0.8121,
+      "step": 36168
+    },
+    {
+      "epoch": 0.9933536568619373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3168158030306696e-09,
+      "loss": 0.8245,
+      "step": 36169
+    },
+    {
+      "epoch": 0.9933811210897806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2977088571629614e-09,
+      "loss": 0.8572,
+      "step": 36170
+    },
+    {
+      "epoch": 0.9934085853176238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.27868101698947e-09,
+      "loss": 0.8435,
+      "step": 36171
+    },
+    {
+      "epoch": 0.993436049545467,
+      "grad_norm": 0.0,
+      "learning_rate": 2.259732282658966e-09,
+      "loss": 0.7413,
+      "step": 36172
+    },
+    {
+      "epoch": 0.9934635137733102,
+      "grad_norm": 0.0,
+      "learning_rate": 2.24086265432244e-09,
+      "loss": 0.707,
+      "step": 36173
+    },
+    {
+      "epoch": 0.9934909780011535,
+      "grad_norm": 0.0,
+      "learning_rate": 2.222072132128661e-09,
+      "loss": 0.8551,
+      "step": 36174
+    },
+    {
+      "epoch": 0.9935184422289968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2033607162275097e-09,
+      "loss": 0.7876,
+      "step": 36175
+    },
+    {
+      "epoch": 0.9935459064568399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.184728406765535e-09,
+      "loss": 0.7665,
+      "step": 36176
+    },
+    {
+      "epoch": 0.9935733706846832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1661752038903973e-09,
+      "loss": 0.8835,
+      "step": 36177
+    },
+    {
+      "epoch": 0.9936008349125265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1477011077497558e-09,
+      "loss": 0.8495,
+      "step": 36178
+    },
+    {
+      "epoch": 0.9936282991403697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1293061184901597e-09,
+      "loss": 0.8472,
+      "step": 36179
+    },
+    {
+      "epoch": 0.9936557633682129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1109902362548283e-09,
+      "loss": 0.8381,
+      "step": 36180
+    },
+    {
+      "epoch": 0.9936832275960561,
+      "grad_norm": 0.0,
+      "learning_rate": 2.092753461190311e-09,
+      "loss": 0.8915,
+      "step": 36181
+    },
+    {
+      "epoch": 0.9937106918238994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.074595793442047e-09,
+      "loss": 0.8148,
+      "step": 36182
+    },
+    {
+      "epoch": 0.9937381560517426,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0565172331510342e-09,
+      "loss": 0.8502,
+      "step": 36183
+    },
+    {
+      "epoch": 0.9937656202795858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.038517780462712e-09,
+      "loss": 0.8734,
+      "step": 36184
+    },
+    {
+      "epoch": 0.9937930845074291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.020597435518079e-09,
+      "loss": 0.8058,
+      "step": 36185
+    },
+    {
+      "epoch": 0.9938205487352723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0027561984592436e-09,
+      "loss": 0.7409,
+      "step": 36186
+    },
+    {
+      "epoch": 0.9938480129631155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849940694272042e-09,
+      "loss": 0.831,
+      "step": 36187
+    },
+    {
+      "epoch": 0.9938754771909588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967311048562959e-09,
+      "loss": 0.8722,
+      "step": 36188
+    },
+    {
+      "epoch": 0.993902941418802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9497071360063955e-09,
+      "loss": 0.8275,
+      "step": 36189
+    },
+    {
+      "epoch": 0.9939304056466453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932182331896293e-09,
+      "loss": 0.8341,
+      "step": 36190
+    },
+    {
+      "epoch": 0.9939578698744885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914736636371428e-09,
+      "loss": 0.8506,
+      "step": 36191
+    },
+    {
+      "epoch": 0.9939853341023317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897370049570579e-09,
+      "loss": 0.7758,
+      "step": 36192
+    },
+    {
+      "epoch": 0.994012798330175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880082571630304e-09,
+      "loss": 0.8017,
+      "step": 36193
+    },
+    {
+      "epoch": 0.9940402625580181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8628742026882695e-09,
+      "loss": 0.8699,
+      "step": 36194
+    },
+    {
+      "epoch": 0.9940677267858614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8457449428799234e-09,
+      "loss": 0.8544,
+      "step": 36195
+    },
+    {
+      "epoch": 0.9940951910137047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8286947923407127e-09,
+      "loss": 0.8381,
+      "step": 36196
+    },
+    {
+      "epoch": 0.9941226552415479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8117237512060847e-09,
+      "loss": 0.776,
+      "step": 36197
+    },
+    {
+      "epoch": 0.9941501194693911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7948318196092663e-09,
+      "loss": 0.9012,
+      "step": 36198
+    },
+    {
+      "epoch": 0.9941775836972343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7780189976857043e-09,
+      "loss": 0.8446,
+      "step": 36199
+    },
+    {
+      "epoch": 0.9942050479250776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7612852855664053e-09,
+      "loss": 0.7955,
+      "step": 36200
+    },
+    {
+      "epoch": 0.9942325121529209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7446306833857062e-09,
+      "loss": 0.8369,
+      "step": 36201
+    },
+    {
+      "epoch": 0.994259976380764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7280551912735033e-09,
+      "loss": 0.755,
+      "step": 36202
+    },
+    {
+      "epoch": 0.9942874406086073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7115588093619128e-09,
+      "loss": 0.7419,
+      "step": 36203
+    },
+    {
+      "epoch": 0.9943149048364506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6951415377819413e-09,
+      "loss": 0.8181,
+      "step": 36204
+    },
+    {
+      "epoch": 0.9943423690642937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678803376663485e-09,
+      "loss": 0.8315,
+      "step": 36205
+    },
+    {
+      "epoch": 0.994369833292137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662544326134219e-09,
+      "loss": 0.8492,
+      "step": 36206
+    },
+    {
+      "epoch": 0.9943972975199802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64636438632515e-09,
+      "loss": 0.8274,
+      "step": 36207
+    },
+    {
+      "epoch": 0.9944247617478235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6302635573617331e-09,
+      "loss": 0.8257,
+      "step": 36208
+    },
+    {
+      "epoch": 0.9944522259756667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6142418393738645e-09,
+      "loss": 0.7391,
+      "step": 36209
+    },
+    {
+      "epoch": 0.9944796902035099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5982992324858893e-09,
+      "loss": 0.7344,
+      "step": 36210
+    },
+    {
+      "epoch": 0.9945071544313532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824357368254829e-09,
+      "loss": 0.7893,
+      "step": 36211
+    },
+    {
+      "epoch": 0.9945346186591963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5666513525181005e-09,
+      "loss": 0.7648,
+      "step": 36212
+    },
+    {
+      "epoch": 0.9945620828870396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5509460796891972e-09,
+      "loss": 0.8279,
+      "step": 36213
+    },
+    {
+      "epoch": 0.9945895471148829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535319918460898e-09,
+      "loss": 0.7888,
+      "step": 36214
+    },
+    {
+      "epoch": 0.9946170113427261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5197728689586578e-09,
+      "loss": 0.8563,
+      "step": 36215
+    },
+    {
+      "epoch": 0.9946444755705693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5043049313057111e-09,
+      "loss": 0.8543,
+      "step": 36216
+    },
+    {
+      "epoch": 0.9946719397984126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4889161056230727e-09,
+      "loss": 0.7855,
+      "step": 36217
+    },
+    {
+      "epoch": 0.9946994040262558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473606392032867e-09,
+      "loss": 0.7988,
+      "step": 36218
+    },
+    {
+      "epoch": 0.9947268682540991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4583757906572183e-09,
+      "loss": 0.7809,
+      "step": 36219
+    },
+    {
+      "epoch": 0.9947543324819422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4432243016160307e-09,
+      "loss": 0.8181,
+      "step": 36220
+    },
+    {
+      "epoch": 0.9947817967097855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4281519250292086e-09,
+      "loss": 0.8279,
+      "step": 36221
+    },
+    {
+      "epoch": 0.9948092609376288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4131586610155457e-09,
+      "loss": 0.7656,
+      "step": 36222
+    },
+    {
+      "epoch": 0.994836725165472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3982445096938358e-09,
+      "loss": 0.8431,
+      "step": 36223
+    },
+    {
+      "epoch": 0.9948641893933152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3834094711817625e-09,
+      "loss": 0.8773,
+      "step": 36224
+    },
+    {
+      "epoch": 0.9948916536211584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3686535455981198e-09,
+      "loss": 0.8718,
+      "step": 36225
+    },
+    {
+      "epoch": 0.9949191178490017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3539767330594812e-09,
+      "loss": 0.8008,
+      "step": 36226
+    },
+    {
+      "epoch": 0.9949465820768449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3393790336801993e-09,
+      "loss": 0.8058,
+      "step": 36227
+    },
+    {
+      "epoch": 0.9949740463046881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3248604475768479e-09,
+      "loss": 0.9029,
+      "step": 36228
+    },
+    {
+      "epoch": 0.9950015105325314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.31042097486489e-09,
+      "loss": 0.8411,
+      "step": 36229
+    },
+    {
+      "epoch": 0.9950289747603747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2960606156575684e-09,
+      "loss": 0.7128,
+      "step": 36230
+    },
+    {
+      "epoch": 0.9950564389882178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.281779370069236e-09,
+      "loss": 0.8692,
+      "step": 36231
+    },
+    {
+      "epoch": 0.9950839032160611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2675772382120255e-09,
+      "loss": 0.8486,
+      "step": 36232
+    },
+    {
+      "epoch": 0.9951113674439043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2534542201991795e-09,
+      "loss": 0.7262,
+      "step": 36233
+    },
+    {
+      "epoch": 0.9951388316717475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2394103161428305e-09,
+      "loss": 0.7293,
+      "step": 36234
+    },
+    {
+      "epoch": 0.9951662958995908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2254455261528908e-09,
+      "loss": 0.7917,
+      "step": 36235
+    },
+    {
+      "epoch": 0.995193760127434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2115598503403824e-09,
+      "loss": 0.7443,
+      "step": 36236
+    },
+    {
+      "epoch": 0.9952212243552773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1977532888152177e-09,
+      "loss": 0.7821,
+      "step": 36237
+    },
+    {
+      "epoch": 0.9952486885831204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1840258416861983e-09,
+      "loss": 0.8777,
+      "step": 36238
+    },
+    {
+      "epoch": 0.9952761528109637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1703775090621262e-09,
+      "loss": 0.8845,
+      "step": 36239
+    },
+    {
+      "epoch": 0.995303617038807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.156808291051803e-09,
+      "loss": 0.8084,
+      "step": 36240
+    },
+    {
+      "epoch": 0.9953310812666502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1433181877618105e-09,
+      "loss": 0.9406,
+      "step": 36241
+    },
+    {
+      "epoch": 0.9953585454944934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.12990719929984e-09,
+      "loss": 0.7818,
+      "step": 36242
+    },
+    {
+      "epoch": 0.9953860097223367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1165753257702526e-09,
+      "loss": 0.774,
+      "step": 36243
+    },
+    {
+      "epoch": 0.9954134739501799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10332256727963e-09,
+      "loss": 0.7624,
+      "step": 36244
+    },
+    {
+      "epoch": 0.9954409381780231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.090148923933443e-09,
+      "loss": 0.7307,
+      "step": 36245
+    },
+    {
+      "epoch": 0.9954684024058663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0770543958349422e-09,
+      "loss": 0.7287,
+      "step": 36246
+    },
+    {
+      "epoch": 0.9954958666337096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0640389830873788e-09,
+      "loss": 0.7681,
+      "step": 36247
+    },
+    {
+      "epoch": 0.9955233308615529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0511026857951134e-09,
+      "loss": 0.8395,
+      "step": 36248
+    },
+    {
+      "epoch": 0.995550795089396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0382455040591765e-09,
+      "loss": 0.7336,
+      "step": 36249
+    },
+    {
+      "epoch": 0.9955782593172393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0254674379817086e-09,
+      "loss": 0.7933,
+      "step": 36250
+    },
+    {
+      "epoch": 0.9956057235450825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127684876648502e-09,
+      "loss": 0.7505,
+      "step": 36251
+    },
+    {
+      "epoch": 0.9956331877729258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0001486532074111e-09,
+      "loss": 0.7556,
+      "step": 36252
+    },
+    {
+      "epoch": 0.995660652000769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.876079347104217e-10,
+      "loss": 0.7421,
+      "step": 36253
+    },
+    {
+      "epoch": 0.9956881162286122,
+      "grad_norm": 0.0,
+      "learning_rate": 9.751463322726916e-10,
+      "loss": 0.8154,
+      "step": 36254
+    },
+    {
+      "epoch": 0.9957155804564555,
+      "grad_norm": 0.0,
+      "learning_rate": 9.627638459919208e-10,
+      "loss": 0.7755,
+      "step": 36255
+    },
+    {
+      "epoch": 0.9957430446842988,
+      "grad_norm": 0.0,
+      "learning_rate": 9.504604759680292e-10,
+      "loss": 0.8125,
+      "step": 36256
+    },
+    {
+      "epoch": 0.9957705089121419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.38236222297606e-10,
+      "loss": 0.7873,
+      "step": 36257
+    },
+    {
+      "epoch": 0.9957979731399852,
+      "grad_norm": 0.0,
+      "learning_rate": 9.260910850761306e-10,
+      "loss": 0.7567,
+      "step": 36258
+    },
+    {
+      "epoch": 0.9958254373678284,
+      "grad_norm": 0.0,
+      "learning_rate": 9.140250644013026e-10,
+      "loss": 0.7363,
+      "step": 36259
+    },
+    {
+      "epoch": 0.9958529015956716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.020381603686013e-10,
+      "loss": 0.7158,
+      "step": 36260
+    },
+    {
+      "epoch": 0.9958803658235149,
+      "grad_norm": 0.0,
+      "learning_rate": 8.901303730712851e-10,
+      "loss": 0.7638,
+      "step": 36261
+    },
+    {
+      "epoch": 0.9959078300513581,
+      "grad_norm": 0.0,
+      "learning_rate": 8.783017026048335e-10,
+      "loss": 0.79,
+      "step": 36262
+    },
+    {
+      "epoch": 0.9959352942792014,
+      "grad_norm": 0.0,
+      "learning_rate": 8.665521490625051e-10,
+      "loss": 0.8991,
+      "step": 36263
+    },
+    {
+      "epoch": 0.9959627585070445,
+      "grad_norm": 0.0,
+      "learning_rate": 8.548817125375586e-10,
+      "loss": 0.8328,
+      "step": 36264
+    },
+    {
+      "epoch": 0.9959902227348878,
+      "grad_norm": 0.0,
+      "learning_rate": 8.432903931221425e-10,
+      "loss": 0.7893,
+      "step": 36265
+    },
+    {
+      "epoch": 0.9960176869627311,
+      "grad_norm": 0.0,
+      "learning_rate": 8.317781909072953e-10,
+      "loss": 0.7126,
+      "step": 36266
+    },
+    {
+      "epoch": 0.9960451511905742,
+      "grad_norm": 0.0,
+      "learning_rate": 8.203451059851653e-10,
+      "loss": 0.7526,
+      "step": 36267
+    },
+    {
+      "epoch": 0.9960726154184175,
+      "grad_norm": 0.0,
+      "learning_rate": 8.089911384456805e-10,
+      "loss": 0.8585,
+      "step": 36268
+    },
+    {
+      "epoch": 0.9961000796462608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.977162883787692e-10,
+      "loss": 0.7933,
+      "step": 36269
+    },
+    {
+      "epoch": 0.996127543874104,
+      "grad_norm": 0.0,
+      "learning_rate": 7.865205558743594e-10,
+      "loss": 0.8658,
+      "step": 36270
+    },
+    {
+      "epoch": 0.9961550081019472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.754039410190484e-10,
+      "loss": 0.7916,
+      "step": 36271
+    },
+    {
+      "epoch": 0.9961824723297904,
+      "grad_norm": 0.0,
+      "learning_rate": 7.643664439027643e-10,
+      "loss": 0.8515,
+      "step": 36272
+    },
+    {
+      "epoch": 0.9962099365576337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.534080646121045e-10,
+      "loss": 0.8984,
+      "step": 36273
+    },
+    {
+      "epoch": 0.996237400785477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.425288032336664e-10,
+      "loss": 0.8235,
+      "step": 36274
+    },
+    {
+      "epoch": 0.9962648650133201,
+      "grad_norm": 0.0,
+      "learning_rate": 7.317286598540474e-10,
+      "loss": 0.6645,
+      "step": 36275
+    },
+    {
+      "epoch": 0.9962923292411634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.210076345576244e-10,
+      "loss": 0.7347,
+      "step": 36276
+    },
+    {
+      "epoch": 0.9963197934690066,
+      "grad_norm": 0.0,
+      "learning_rate": 7.10365727430995e-10,
+      "loss": 0.8281,
+      "step": 36277
+    },
+    {
+      "epoch": 0.9963472576968498,
+      "grad_norm": 0.0,
+      "learning_rate": 6.998029385563154e-10,
+      "loss": 0.7151,
+      "step": 36278
+    },
+    {
+      "epoch": 0.9963747219246931,
+      "grad_norm": 0.0,
+      "learning_rate": 6.89319268019073e-10,
+      "loss": 0.8701,
+      "step": 36279
+    },
+    {
+      "epoch": 0.9964021861525363,
+      "grad_norm": 0.0,
+      "learning_rate": 6.78914715900314e-10,
+      "loss": 0.8806,
+      "step": 36280
+    },
+    {
+      "epoch": 0.9964296503803796,
+      "grad_norm": 0.0,
+      "learning_rate": 6.685892822844154e-10,
+      "loss": 0.8314,
+      "step": 36281
+    },
+    {
+      "epoch": 0.9964571146082228,
+      "grad_norm": 0.0,
+      "learning_rate": 6.583429672513131e-10,
+      "loss": 0.7626,
+      "step": 36282
+    },
+    {
+      "epoch": 0.996484578836066,
+      "grad_norm": 0.0,
+      "learning_rate": 6.481757708831638e-10,
+      "loss": 0.7072,
+      "step": 36283
+    },
+    {
+      "epoch": 0.9965120430639093,
+      "grad_norm": 0.0,
+      "learning_rate": 6.380876932587932e-10,
+      "loss": 0.8435,
+      "step": 36284
+    },
+    {
+      "epoch": 0.9965395072917524,
+      "grad_norm": 0.0,
+      "learning_rate": 6.280787344603578e-10,
+      "loss": 0.8959,
+      "step": 36285
+    },
+    {
+      "epoch": 0.9965669715195957,
+      "grad_norm": 0.0,
+      "learning_rate": 6.181488945655733e-10,
+      "loss": 0.7374,
+      "step": 36286
+    },
+    {
+      "epoch": 0.996594435747439,
+      "grad_norm": 0.0,
+      "learning_rate": 6.082981736543759e-10,
+      "loss": 0.7964,
+      "step": 36287
+    },
+    {
+      "epoch": 0.9966218999752822,
+      "grad_norm": 0.0,
+      "learning_rate": 5.985265718022604e-10,
+      "loss": 0.8454,
+      "step": 36288
+    },
+    {
+      "epoch": 0.9966493642031254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.88834089089163e-10,
+      "loss": 0.811,
+      "step": 36289
+    },
+    {
+      "epoch": 0.9966768284309686,
+      "grad_norm": 0.0,
+      "learning_rate": 5.792207255894689e-10,
+      "loss": 0.8432,
+      "step": 36290
+    },
+    {
+      "epoch": 0.9967042926588119,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696864813820036e-10,
+      "loss": 0.7428,
+      "step": 36291
+    },
+    {
+      "epoch": 0.9967317568866552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.602313565389317e-10,
+      "loss": 0.8523,
+      "step": 36292
+    },
+    {
+      "epoch": 0.9967592211144983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.508553511379689e-10,
+      "loss": 0.8323,
+      "step": 36293
+    },
+    {
+      "epoch": 0.9967866853423416,
+      "grad_norm": 0.0,
+      "learning_rate": 5.415584652512795e-10,
+      "loss": 0.8275,
+      "step": 36294
+    },
+    {
+      "epoch": 0.9968141495701849,
+      "grad_norm": 0.0,
+      "learning_rate": 5.323406989532487e-10,
+      "loss": 0.7876,
+      "step": 36295
+    },
+    {
+      "epoch": 0.996841613798028,
+      "grad_norm": 0.0,
+      "learning_rate": 5.23202052317151e-10,
+      "loss": 0.9454,
+      "step": 36296
+    },
+    {
+      "epoch": 0.9968690780258713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.141425254151511e-10,
+      "loss": 0.7899,
+      "step": 36297
+    },
+    {
+      "epoch": 0.9968965422537145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.051621183171928e-10,
+      "loss": 0.8076,
+      "step": 36298
+    },
+    {
+      "epoch": 0.9969240064815578,
+      "grad_norm": 0.0,
+      "learning_rate": 4.962608310976613e-10,
+      "loss": 0.7384,
+      "step": 36299
+    },
+    {
+      "epoch": 0.996951470709401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.874386638242801e-10,
+      "loss": 0.7378,
+      "step": 36300
+    },
+    {
+      "epoch": 0.9969789349372442,
+      "grad_norm": 0.0,
+      "learning_rate": 4.786956165681034e-10,
+      "loss": 0.7375,
+      "step": 36301
+    },
+    {
+      "epoch": 0.9970063991650875,
+      "grad_norm": 0.0,
+      "learning_rate": 4.700316893968548e-10,
+      "loss": 0.8204,
+      "step": 36302
+    },
+    {
+      "epoch": 0.9970338633929307,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6144688238158875e-10,
+      "loss": 0.7303,
+      "step": 36303
+    },
+    {
+      "epoch": 0.9970613276207739,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5294119558780823e-10,
+      "loss": 0.8976,
+      "step": 36304
+    },
+    {
+      "epoch": 0.9970887918486172,
+      "grad_norm": 0.0,
+      "learning_rate": 4.445146290843472e-10,
+      "loss": 0.8163,
+      "step": 36305
+    },
+    {
+      "epoch": 0.9971162560764604,
+      "grad_norm": 0.0,
+      "learning_rate": 4.361671829367087e-10,
+      "loss": 0.7586,
+      "step": 36306
+    },
+    {
+      "epoch": 0.9971437203043036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.278988572126164e-10,
+      "loss": 0.9345,
+      "step": 36307
+    },
+    {
+      "epoch": 0.9971711845321469,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1970965197535294e-10,
+      "loss": 0.8015,
+      "step": 36308
+    },
+    {
+      "epoch": 0.9971986487599901,
+      "grad_norm": 0.0,
+      "learning_rate": 4.115995672915318e-10,
+      "loss": 0.7573,
+      "step": 36309
+    },
+    {
+      "epoch": 0.9972261129878334,
+      "grad_norm": 0.0,
+      "learning_rate": 4.035686032244357e-10,
+      "loss": 0.8385,
+      "step": 36310
+    },
+    {
+      "epoch": 0.9972535772156765,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9561675983734725e-10,
+      "loss": 0.8554,
+      "step": 36311
+    },
+    {
+      "epoch": 0.9972810414435198,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8774403719465945e-10,
+      "loss": 0.8037,
+      "step": 36312
+    },
+    {
+      "epoch": 0.9973085056713631,
+      "grad_norm": 0.0,
+      "learning_rate": 3.799504353563244e-10,
+      "loss": 0.7664,
+      "step": 36313
+    },
+    {
+      "epoch": 0.9973359698992063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7223595438562465e-10,
+      "loss": 0.7123,
+      "step": 36314
+    },
+    {
+      "epoch": 0.9973634341270495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.646005943436226e-10,
+      "loss": 0.8741,
+      "step": 36315
+    },
+    {
+      "epoch": 0.9973908983548928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5704435529027024e-10,
+      "loss": 0.8045,
+      "step": 36316
+    },
+    {
+      "epoch": 0.997418362582736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.495672372855197e-10,
+      "loss": 0.812,
+      "step": 36317
+    },
+    {
+      "epoch": 0.9974458268105793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.421692403893229e-10,
+      "loss": 0.8049,
+      "step": 36318
+    },
+    {
+      "epoch": 0.9974732910384224,
+      "grad_norm": 0.0,
+      "learning_rate": 3.348503646583012e-10,
+      "loss": 0.772,
+      "step": 36319
+    },
+    {
+      "epoch": 0.9975007552662657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.276106101512966e-10,
+      "loss": 0.7632,
+      "step": 36320
+    },
+    {
+      "epoch": 0.997528219494109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.204499769271507e-10,
+      "loss": 0.8229,
+      "step": 36321
+    },
+    {
+      "epoch": 0.9975556837219521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.133684650391544e-10,
+      "loss": 0.8282,
+      "step": 36322
+    },
+    {
+      "epoch": 0.9975831479497954,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0636607454725964e-10,
+      "loss": 0.8546,
+      "step": 36323
+    },
+    {
+      "epoch": 0.9976106121776386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.994428055036469e-10,
+      "loss": 0.6901,
+      "step": 36324
+    },
+    {
+      "epoch": 0.9976380764054819,
+      "grad_norm": 0.0,
+      "learning_rate": 2.925986579638274e-10,
+      "loss": 0.7951,
+      "step": 36325
+    },
+    {
+      "epoch": 0.9976655406333251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8583363198331216e-10,
+      "loss": 0.8388,
+      "step": 36326
+    },
+    {
+      "epoch": 0.9976930048611683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.79147727615392e-10,
+      "loss": 0.8451,
+      "step": 36327
+    },
+    {
+      "epoch": 0.9977204690890116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.725409449111371e-10,
+      "loss": 0.7328,
+      "step": 36328
+    },
+    {
+      "epoch": 0.9977479333168549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.660132839249485e-10,
+      "loss": 0.8582,
+      "step": 36329
+    },
+    {
+      "epoch": 0.997775397544698,
+      "grad_norm": 0.0,
+      "learning_rate": 2.595647447078964e-10,
+      "loss": 0.7224,
+      "step": 36330
+    },
+    {
+      "epoch": 0.9978028617725413,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5319532730994076e-10,
+      "loss": 0.7423,
+      "step": 36331
+    },
+    {
+      "epoch": 0.9978303260003845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4690503178215196e-10,
+      "loss": 0.8176,
+      "step": 36332
+    },
+    {
+      "epoch": 0.9978577902282277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4069385817449e-10,
+      "loss": 0.8416,
+      "step": 36333
+    },
+    {
+      "epoch": 0.997885254456071,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3456180653691486e-10,
+      "loss": 0.8342,
+      "step": 36334
+    },
+    {
+      "epoch": 0.9979127186839142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2850887691605594e-10,
+      "loss": 0.7946,
+      "step": 36335
+    },
+    {
+      "epoch": 0.9979401829117575,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2253506936076307e-10,
+      "loss": 0.7816,
+      "step": 36336
+    },
+    {
+      "epoch": 0.9979676471396006,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1664038391877586e-10,
+      "loss": 0.7644,
+      "step": 36337
+    },
+    {
+      "epoch": 0.9979951113674439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1082482063561338e-10,
+      "loss": 0.7966,
+      "step": 36338
+    },
+    {
+      "epoch": 0.9980225755952872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.050883795590153e-10,
+      "loss": 0.8562,
+      "step": 36339
+    },
+    {
+      "epoch": 0.9980500398231303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943106073228024e-10,
+      "loss": 0.7379,
+      "step": 36340
+    },
+    {
+      "epoch": 0.9980775040509736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385286420092742e-10,
+      "loss": 0.8296,
+      "step": 36341
+    },
+    {
+      "epoch": 0.9981049682788169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8835379001047594e-10,
+      "loss": 0.8068,
+      "step": 36342
+    },
+    {
+      "epoch": 0.9981324325066601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829338382031143e-10,
+      "loss": 0.772,
+      "step": 36343
+    },
+    {
+      "epoch": 0.9981598967345033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7759300882214115e-10,
+      "loss": 0.7682,
+      "step": 36344
+    },
+    {
+      "epoch": 0.9981873609623465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7233130190863478e-10,
+      "loss": 0.7969,
+      "step": 36345
+    },
+    {
+      "epoch": 0.9982148251901898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671487175058939e-10,
+      "loss": 0.746,
+      "step": 36346
+    },
+    {
+      "epoch": 0.9982422894180331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6204525565388651e-10,
+      "loss": 0.8542,
+      "step": 36347
+    },
+    {
+      "epoch": 0.9982697536458762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5702091639369088e-10,
+      "loss": 0.8643,
+      "step": 36348
+    },
+    {
+      "epoch": 0.9982972178737195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5207569976416482e-10,
+      "loss": 0.8436,
+      "step": 36349
+    },
+    {
+      "epoch": 0.9983246821015627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4720960580527633e-10,
+      "loss": 0.6491,
+      "step": 36350
+    },
+    {
+      "epoch": 0.9983521463294059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4242263455588323e-10,
+      "loss": 0.7774,
+      "step": 36351
+    },
+    {
+      "epoch": 0.9983796105572492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.377147860526229e-10,
+      "loss": 0.8933,
+      "step": 36352
+    },
+    {
+      "epoch": 0.9984070747850924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330860603332429e-10,
+      "loss": 0.8359,
+      "step": 36353
+    },
+    {
+      "epoch": 0.9984345390129357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2853645743438058e-10,
+      "loss": 0.7842,
+      "step": 36354
+    },
+    {
+      "epoch": 0.9984620032407789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2406597739267335e-10,
+      "loss": 0.7859,
+      "step": 36355
+    },
+    {
+      "epoch": 0.9984894674686221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1967462024253806e-10,
+      "loss": 0.8103,
+      "step": 36356
+    },
+    {
+      "epoch": 0.9985169316964654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1536238601950189e-10,
+      "loss": 0.7644,
+      "step": 36357
+    },
+    {
+      "epoch": 0.9985443959243085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1112927475687152e-10,
+      "loss": 0.8383,
+      "step": 36358
+    },
+    {
+      "epoch": 0.9985718601521518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0697528648906386e-10,
+      "loss": 0.7932,
+      "step": 36359
+    },
+    {
+      "epoch": 0.9985993243799951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0290042124716515e-10,
+      "loss": 0.8282,
+      "step": 36360
+    },
+    {
+      "epoch": 0.9986267886078383,
+      "grad_norm": 0.0,
+      "learning_rate": 9.890467906670253e-11,
+      "loss": 0.8526,
+      "step": 36361
+    },
+    {
+      "epoch": 0.9986542528356815,
+      "grad_norm": 0.0,
+      "learning_rate": 9.498805997543159e-11,
+      "loss": 0.8639,
+      "step": 36362
+    },
+    {
+      "epoch": 0.9986817170635247,
+      "grad_norm": 0.0,
+      "learning_rate": 9.115056400776922e-11,
+      "loss": 0.6883,
+      "step": 36363
+    },
+    {
+      "epoch": 0.998709181291368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.739219119147102e-11,
+      "loss": 0.9062,
+      "step": 36364
+    },
+    {
+      "epoch": 0.9987366455192113,
+      "grad_norm": 0.0,
+      "learning_rate": 8.371294155762322e-11,
+      "loss": 0.794,
+      "step": 36365
+    },
+    {
+      "epoch": 0.9987641097470544,
+      "grad_norm": 0.0,
+      "learning_rate": 8.011281513509162e-11,
+      "loss": 0.8009,
+      "step": 36366
+    },
+    {
+      "epoch": 0.9987915739748977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.659181195163179e-11,
+      "loss": 0.7701,
+      "step": 36367
+    },
+    {
+      "epoch": 0.998819038202741,
+      "grad_norm": 0.0,
+      "learning_rate": 7.314993203721977e-11,
+      "loss": 0.806,
+      "step": 36368
+    },
+    {
+      "epoch": 0.9988465024305841,
+      "grad_norm": 0.0,
+      "learning_rate": 6.978717541628044e-11,
+      "loss": 0.8115,
+      "step": 36369
+    },
+    {
+      "epoch": 0.9988739666584274,
+      "grad_norm": 0.0,
+      "learning_rate": 6.650354211767962e-11,
+      "loss": 0.8138,
+      "step": 36370
+    },
+    {
+      "epoch": 0.9989014308862706,
+      "grad_norm": 0.0,
+      "learning_rate": 6.329903216584222e-11,
+      "loss": 0.7833,
+      "step": 36371
+    },
+    {
+      "epoch": 0.9989288951141139,
+      "grad_norm": 0.0,
+      "learning_rate": 6.017364558741356e-11,
+      "loss": 0.9071,
+      "step": 36372
+    },
+    {
+      "epoch": 0.9989563593419571,
+      "grad_norm": 0.0,
+      "learning_rate": 5.712738240570836e-11,
+      "loss": 0.8181,
+      "step": 36373
+    },
+    {
+      "epoch": 0.9989838235698003,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416024264626174e-11,
+      "loss": 0.8276,
+      "step": 36374
+    },
+    {
+      "epoch": 0.9990112877976436,
+      "grad_norm": 0.0,
+      "learning_rate": 5.127222633127815e-11,
+      "loss": 0.806,
+      "step": 36375
+    },
+    {
+      "epoch": 0.9990387520254868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.846333348518251e-11,
+      "loss": 0.8936,
+      "step": 36376
+    },
+    {
+      "epoch": 0.99906621625333,
+      "grad_norm": 0.0,
+      "learning_rate": 4.573356412906904e-11,
+      "loss": 0.7706,
+      "step": 36377
+    },
+    {
+      "epoch": 0.9990936804811733,
+      "grad_norm": 0.0,
+      "learning_rate": 4.308291828403199e-11,
+      "loss": 0.8715,
+      "step": 36378
+    },
+    {
+      "epoch": 0.9991211447090165,
+      "grad_norm": 0.0,
+      "learning_rate": 4.051139597227582e-11,
+      "loss": 0.878,
+      "step": 36379
+    },
+    {
+      "epoch": 0.9991486089368598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.801899721378455e-11,
+      "loss": 0.7783,
+      "step": 36380
+    },
+    {
+      "epoch": 0.999176073164703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.560572202743195e-11,
+      "loss": 0.8105,
+      "step": 36381
+    },
+    {
+      "epoch": 0.9992035373925462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.327157043320206e-11,
+      "loss": 0.7741,
+      "step": 36382
+    },
+    {
+      "epoch": 0.9992310016203895,
+      "grad_norm": 0.0,
+      "learning_rate": 3.101654244996866e-11,
+      "loss": 0.7017,
+      "step": 36383
+    },
+    {
+      "epoch": 0.9992584658482326,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8840638094385086e-11,
+      "loss": 0.8004,
+      "step": 36384
+    },
+    {
+      "epoch": 0.9992859300760759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6743857384214922e-11,
+      "loss": 0.7519,
+      "step": 36385
+    },
+    {
+      "epoch": 0.9993133943039192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4726200336111505e-11,
+      "loss": 0.7473,
+      "step": 36386
+    },
+    {
+      "epoch": 0.9993408585317624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.278766696561796e-11,
+      "loss": 0.7292,
+      "step": 36387
+    },
+    {
+      "epoch": 0.9993683227596056,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0928257289387633e-11,
+      "loss": 0.7773,
+      "step": 36388
+    },
+    {
+      "epoch": 0.9993957869874488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147971320743197e-11,
+      "loss": 0.8154,
+      "step": 36389
+    },
+    {
+      "epoch": 0.9994232512152921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7446809074117556e-11,
+      "loss": 0.7156,
+      "step": 36390
+    },
+    {
+      "epoch": 0.9994507154431354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582477056283338e-11,
+      "loss": 0.7972,
+      "step": 36391
+    },
+    {
+      "epoch": 0.9994781796709785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4281855800213351e-11,
+      "loss": 0.8254,
+      "step": 36392
+    },
+    {
+      "epoch": 0.9995056438988218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2818064798469921e-11,
+      "loss": 0.858,
+      "step": 36393
+    },
+    {
+      "epoch": 0.9995331081266651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1433397568705317e-11,
+      "loss": 0.8011,
+      "step": 36394
+    },
+    {
+      "epoch": 0.9995605723545082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127854122021774e-11,
+      "loss": 0.7189,
+      "step": 36395
+    },
+    {
+      "epoch": 0.9995880365823515,
+      "grad_norm": 0.0,
+      "learning_rate": 8.901434468411296e-12,
+      "loss": 0.7022,
+      "step": 36396
+    },
+    {
+      "epoch": 0.9996155008101947,
+      "grad_norm": 0.0,
+      "learning_rate": 7.754138618976115e-12,
+      "loss": 0.7851,
+      "step": 36397
+    },
+    {
+      "epoch": 0.999642965038038,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6859665814877905e-12,
+      "loss": 0.7788,
+      "step": 36398
+    },
+    {
+      "epoch": 0.9996704292658812,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696918363717885e-12,
+      "loss": 0.8723,
+      "step": 36399
+    },
+    {
+      "epoch": 0.9996978934937244,
+      "grad_norm": 0.0,
+      "learning_rate": 4.786993975658405e-12,
+      "loss": 0.9216,
+      "step": 36400
+    },
+    {
+      "epoch": 0.9997253577215677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.956193422860466e-12,
+      "loss": 0.8919,
+      "step": 36401
+    },
+    {
+      "epoch": 0.9997528219494108,
+      "grad_norm": 0.0,
+      "learning_rate": 3.204516713095629e-12,
+      "loss": 0.8252,
+      "step": 36402
+    },
+    {
+      "epoch": 0.9997802861772541,
+      "grad_norm": 0.0,
+      "learning_rate": 2.531963850804786e-12,
+      "loss": 0.7879,
+      "step": 36403
+    },
+    {
+      "epoch": 0.9998077504050974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385348426492754e-12,
+      "loss": 0.8693,
+      "step": 36404
+    },
+    {
+      "epoch": 0.9998352146329406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4242296919597664e-12,
+      "loss": 0.7652,
+      "step": 36405
+    },
+    {
+      "epoch": 0.9998626788607838,
+      "grad_norm": 0.0,
+      "learning_rate": 9.890484042873738e-13,
+      "loss": 0.8045,
+      "step": 36406
+    },
+    {
+      "epoch": 0.9998901430886271,
+      "grad_norm": 0.0,
+      "learning_rate": 6.329909829627667e-13,
+      "loss": 0.7816,
+      "step": 36407
+    },
+    {
+      "epoch": 0.9999176073164703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5605742909616823e-13,
+      "loss": 0.7745,
+      "step": 36408
+    },
+    {
+      "epoch": 0.9999450715443136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824774712847047e-13,
+      "loss": 0.8871,
+      "step": 36409
+    },
+    {
+      "epoch": 0.9999725357721567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.956193705967337e-14,
+      "loss": 0.7702,
+      "step": 36410
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.7414,
+      "step": 36411
+    },
     {
       "epoch": 1.0,
-      "step": 3060,
-      "total_flos": 1.5807535160498323e+19,
-      "train_loss": 1.0534319706602033,
-      "train_runtime": 57512.2695,
-      "train_samples_per_second": 34.052,
-      "train_steps_per_second": 0.053
+      "step": 36411,
+      "total_flos": 9.93833455314503e+19,
+      "train_loss": 0.9174524423369619,
+      "train_runtime": 276687.3272,
+      "train_samples_per_second": 16.844,
+      "train_steps_per_second": 0.132
     }
   ],
   "logging_steps": 1.0,
-  "max_steps": 3060,
+  "max_steps": 36411,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 3000,
@@ -21455,8 +254912,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.5807535160498323e+19,
-  "train_batch_size": 8,
+  "total_flos": 9.93833455314503e+19,
+  "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null
 }